Fault diagnosis system of pipe network
Through the multi-channel data acquisition, preprocessing and multi-model analysis pipeline fault diagnosis system, the traditional diagnosis system is easily vulnerable to noise interference and insufficient real-time performance, and the accurate monitoring and timely warning of pipeline faults is realized, ensuring the safe and stable operation of the pipeline network.
Patent Information
- Application Number
- CN202510576853.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-05-06
- Publication Date
- 2025-08-22
- Estimated Expiration
- 2045-05-06
AI Technical Summary
The existing pipeline fault diagnosis technology is susceptible to noise interference and causes frequent false alarms, lacks real-time monitoring and precise positioning capabilities, resulting in lagging maintenance response and posing safety hazards.
Multi-channel data collection, data preprocessing, multi-model analysis and real-time early warning mechanisms are adopted, combined with dynamic threshold determination and fault early warning analysis, and a scoring model is built for real-time monitoring and early warning.
It improves the accuracy and real-time nature of fault diagnosis, reduces false alarms, and realizes timely warning of pipeline faults, avoids leakage accidents and ecological disasters.
Smart Images

Figure CN120524347A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of fault detection, and in particular to a fault diagnosis system for a pipe network. Background Art
[0002] Oil pipelines are the invisible arteries that sustain a nation's energy lifeline. Their safe and stable operation is directly related to the continuity of energy supply and national security strategy. However, current pipeline fault diagnosis technology faces many challenges, becoming a bottleneck restricting the efficient transportation of energy:
[0003] On the one hand, traditional diagnostic systems often rely on single sensors or simple threshold alarm mechanisms. In complex environments, they are easily affected by noise factors such as electromagnetic interference and temperature fluctuations, resulting in frequent false alarms. This not only consumes a large amount of operation and maintenance resources, but also may weaken staff's trust in the early warning system.
[0004] On the other hand, due to the lack of efficient data collection and analysis methods, there is a significant time delay in fault detection, making it difficult to achieve real-time monitoring and precise positioning of faults, resulting in delayed maintenance responses, posing a safety hazard to pipeline operations, and even potentially causing serious leakage accidents and ecological disasters. Summary of the Invention
[0005] The purpose of the present invention is to provide a fault diagnosis system for a pipe network to solve the above technical problems.
[0006] The purpose of the present invention can be achieved through the following technical solutions:
[0007] A pipe network fault diagnosis system, comprising:
[0008] Data acquisition module, used to obtain pipeline operation data from SCADA system, simulation environment, intermediate database, API transmission and manual inspection data entry;
[0009] The data preprocessing module is used to clean, normalize and structure the collected data, including data availability testing, basic statistical calculation and box plot drawing; the cleaning process uses methods based on statistical analysis and machine learning to identify and process outliers and missing values; normalization uses standardization or normalization functions to unify the data into the same scale range.
[0010] The data analysis module builds a scoring model based on pipeline operation data, and then determines the data distribution type based on the scoring model and the dynamic threshold judgment mechanism;
[0011] The model development module selects the optimal model based on data distribution characteristics, supports the integration of multiple models, generates and evaluates multiple model results, and uses a hybrid modeling strategy to output visual model results;
[0012] The real-time early warning module is used to monitor the operation status of the pipeline network in real time and issue early warning information based on the diagnosis results for visual display.
[0013] In a further solution, the model development module uses convolutional neural networks, recurrent neural networks and their variants, and gradient boosting tree models for model selection, and uses stacking integration and hybrid models for model integration.
[0014] In a further solution, the pipeline operation data includes pressure, flow, temperature, vibration frequency, fluid density, corrosion rate, valve opening, pump speed, acoustic wave signal, and gas concentration parameters;
[0015] The process of building a scoring model is:
[0016] Calculate the mean of various parameters based on historical pipeline operation data and standard deviation σ i , thus obtaining the normal range of each parameter i represents the i-th parameter;
[0017] Calculate the parameters x i The degree of deviation d i ,when When d i =0, when hour, when hour, And according to the weight distribution, the comprehensive score S is obtained, which is expressed as: θ i Indicates the weight coefficient corresponding to each parameter, and n indicates the total number of parameters;
[0018] Combined with the dynamic threshold judgment mechanism, the data distribution type and anomaly level are determined, including concentrated distribution, positive deviation concentration, positive long tail, negative deviation concentration, and negative long tail;
[0019] According to the comprehensive score and fault prediction value, the fault probability is output.
[0020] In a further solution, the dynamic threshold determination mechanism includes:
[0021] Obtain standard deviation thresholds, interquartile range thresholds, and data concentration thresholds using historical pipeline operation data;
[0022] According to the real-time range of data R i =max(x i )-min(x i ) and the average range of historical data Dynamically adjust the threshold ratio. The adjusted standard deviation threshold, interquartile range threshold, and data concentration threshold are calculated using the formula: adjusted threshold = initial threshold × [1 + k × (real-time data range / historical data average range)], where k is the adjustment coefficient, determined through experiments or experience.
[0023] The final data distribution type is determined based on the adjusted standard deviation threshold, interquartile range threshold, and data concentration threshold. The comprehensive score output by the scoring model is then compared with the preset scoring range.
[0024] When the comprehensive score is lower than the scoring interval and the data distribution type is concentrated distribution, the anomaly level is low; when the comprehensive score is higher than the scoring interval and the data distribution type is positive deviation concentrated or negative deviation concentrated, the anomaly level is high; otherwise, the anomaly level is medium.
[0025] In a further solution, the decision logic of the data distribution type is as follows:
[0026] When the ratio of the absolute value of the difference between the mean and the median to the median is less than or equal to the set ratio, and the standard deviation and interquartile range are both lower than the corresponding standard deviation threshold and interquartile range threshold, the data type is determined to be concentrated distribution;
[0027] When the data is positively skewed and the proportion of data exceeding the upper bound is higher than the concentration threshold, the data type is determined to be positively deviating from concentration;
[0028] When the data is positively skewed, but the proportion of data exceeding the upper bound is lower than the concentration threshold, the data type is determined to be positive long tail;
[0029] When the data is negatively skewed and the proportion of data below the lower bound is higher than the concentration threshold, the data type is determined to be negatively skewed;
[0030] When the data is negatively skewed, but the proportion of data below the lower bound is lower than the concentration threshold, the data type is determined to be negative long tail.
[0031] In a further embodiment, the working process of the real-time warning module includes:
[0032] Based on the cumulative change in historical pipeline operation data, the comprehensive score output by the scoring model, and the abnormality level, the fault prediction value is obtained through the fault warning analysis model. Then, the fault warning level is divided according to the comparison result between the fault prediction value and the fault prediction interval. The fault warning levels are blue, yellow, and red from low to high.
[0033] Differentiated early warning measures are taken for different fault levels, including sound and light alarms, SMS notifications, and automatic shutdown.
[0034] In a further solution, the expression of the fault warning analysis model is:
[0035]
[0036] In the formula, σ(S) represents the transformation result of the Sigmoid function on the comprehensive score S, ρ1, ρ2, and ρ3 represent exponential coefficients, C represents the cumulative change in historical pipeline operation data, ta and tb represent the start and end times, respectively, and x i (t) represents the value of the i-th pipeline operating parameter at time t, ψ i represents the weight factor corresponding to the change in the i-th pipeline operating parameter, L represents the abnormality level, with low level being 1, medium level being 2, and high level being 3, Γ represents the intermediate impact index, and ξ j represents the comprehensive score characteristic value obtained based on historical data statistics, ω j represents the weight factor of the jth Sigmoid function variant, m represents the number of Sigmoid function variants, and P represents the fault prediction value;
[0037] When P exceeds the fault prediction interval, the fault warning level is red; when P falls within the fault prediction interval, the fault warning level is yellow; when P is lower than the fault prediction interval, the fault warning level is blue.
[0038] In a further embodiment, the model of the model development module includes:
[0039] Residual models are suitable for data with a central distribution or positive deviation from central distribution;
[0040] Isolation forest, suitable for data with positive long tail distribution;
[0041] Support vector machines are suitable for data with negative deviation concentration or negative long tail distribution.
[0042] Beneficial effects of the present invention:
[0043] (1) By collecting data through multiple channels, including SCADA systems, simulation environments, intermediate databases, API transmission, and manual inspection data entry, more comprehensive pipeline operation information can be obtained; in the data preprocessing stage, statistical analysis and machine learning-based methods are used to identify and process outliers and missing values, and standardization or normalization functions are used to unify data scales and improve data quality; when constructing the scoring model, the degree of deviation of multiple parameters such as pressure and flow from the normal range is comprehensively considered, and the data distribution type and abnormality level are determined in combination with the dynamic threshold judgment mechanism, which can more accurately assess the probability of failure, effectively reduce false alarms, and greatly improve the accuracy of fault diagnosis;
[0044] (2) The data acquisition module acquires pipeline operation data in real time, and the data analysis module quickly builds a scoring model and determines the data distribution type and abnormality level; the real-time warning module obtains the fault prediction value through the fault warning analysis model based on the cumulative change of historical pipeline operation data, the comprehensive score output by the scoring model and the abnormality level, and divides the fault warning level according to the comparison result between the prediction value and the fault prediction interval. It can promptly issue early warning information such as sound and light alarms and SMS notifications in the early stage of the fault, and can automatically shut down for serious faults, thus realizing real-time monitoring and rapid warning of pipeline network faults, effectively avoiding serious leakage accidents and ecological disasters caused by untimely fault detection;
[0045] (3) Select the optimal model based on the data distribution characteristics. For example, the residual model is suitable for data with concentrated distribution or positive deviation concentration, the isolation forest is suitable for data with positive long-tail distribution, and the support vector machine is suitable for data with negative deviation concentration or negative long-tail distribution. This can give full play to the advantages of different models, better adapt to complex pipeline network operation data, improve the diagnosis ability of various fault types, and provide more reliable technical support for the safe and stable operation of the pipeline network. BRIEF DESCRIPTION OF THE DRAWINGS
[0046] The present invention will be further described below with reference to the accompanying drawings.
[0047] Figure 1 This is a system structure diagram of the present invention. DETAILED DESCRIPTION
[0048] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative efforts shall fall within the scope of protection of the present invention.
[0049] See also Figure 1 As shown, the present invention is a fault diagnosis system for a pipe network, comprising:
[0050] Data acquisition module, used to obtain pipeline operation data from SCADA system, simulation environment, intermediate database, API transmission and manual inspection data entry;
[0051] The data preprocessing module is used to clean, normalize, and structure the collected data for storage, including data availability testing, basic statistical calculations, and boxplot drawing. For example, the cleaning process uses statistical analysis and machine learning methods such as the isolation forest algorithm to identify and handle outliers and missing values. Normalization uses standardization or normalization functions to unify the data to the same scale range. Data availability testing: Checks data integrity to ensure data usability; tests are performed by setting data validity rules such as numerical ranges and data types. Basic statistical calculations: Calculates statistics such as mean, median, and standard deviation to quantify data characteristics. The mean reflects the center position of the data, the median is insensitive to outliers, and the standard deviation measures the degree of dispersion of the data. Mean: Calculates the average value of the data, reflecting the center position of the data. Median: The value in the middle after the data is arranged in order of size and is insensitive to outliers. Standard deviation: Measures the degree of dispersion of the data. The larger the standard deviation, the more dispersed the data. Range: The difference between the maximum and minimum values, reflecting the degree of dispersion of the data. Box plot drawing: intuitively display data distribution and identify outliers; box plots show the quartiles and outliers of the data, helping to quickly determine whether there are outliers that deviate in the positive or negative direction; quartiles: arrange the data in order of size and divide them into four equal parts, Q1 (25% quantile), Q2 (50% quantile, i.e. median), Q3 (75% quantile); IQR: interquartile range, Q3-Q1, reflects the degree of dispersion of the data; outliers: data points beyond the upper edge (Q3+1.5IQR) or lower edge (Q1-1.5IQR) may be outliers.
[0052] The data analysis module builds a scoring model based on pipeline operation data, and then determines the data distribution type based on the scoring model and the dynamic threshold judgment mechanism;
[0053] The model development module selects the optimal model based on data distribution characteristics, supports the integration of multiple models, generates and evaluates multiple model results, and uses a hybrid modeling strategy to output visual model results. When selecting a model, if the data has a concentrated distribution or positive deviation from the concentrated distribution, a residual model can be selected; if the distribution has a positive long tail, an isolation forest can be selected; if the distribution has a negative deviation from the concentrated distribution or a negative long tail, a support vector machine can be selected. Regarding the integration of multiple models, when using stacking integration, the outputs of multiple base models are input as new features into another model for training; when using hybrid models, the advantages of different models are combined, such as combining the learning ability of neural networks with the interpretability of decision trees. The selected model is trained using training data, and the model's accuracy, recall, and other indicators are evaluated using test data. The model is then optimized based on the evaluation results.
[0054] The real-time warning module monitors the network's operating status in real time and issues warnings based on diagnostic results, providing a visual display. When the predicted fault value exceeds the predicted range, the warning level changes to red, triggering audible and visual alarms, SMS notifications, and, for severe faults, automatic shutdown. When the predicted fault value falls within the predicted range, the warning level changes to yellow, with corresponding prompts issued. When the predicted fault value falls below the predicted range, the warning level changes to blue, providing a general prompt. The network's operating status and warning information are displayed in a visual interface. The interface design can take the form of dashboards, map annotations, and other tools to intuitively present the network's real-time status.
[0055] The present invention collects data through multiple channels to ensure the comprehensiveness and diversity of data, providing rich information for accurate diagnosis; data preprocessing improves data quality and makes subsequent analysis more reliable; constructing a scoring model and determining the data distribution type help to accurately evaluate the operating status of the pipeline network; the model development module selects the optimal model and integrates it to improve the accuracy and adaptability of fault diagnosis; the real-time warning module realizes real-time monitoring and visual warning, promptly detects faults and notifies relevant personnel, reducing the risk of accidents.
[0056] The model development module uses convolutional neural networks, recurrent neural networks and their variants, and gradient boosting tree models for model selection, and stacking integration and hybrid models for model integration. Among them, stacking integration: first, divide the historical pipeline operation data into training sets and test sets; use different basic models such as CNN, RNN, and gradient boosting tree models to train on the training set to obtain their respective prediction results; then, use the prediction results of these basic models as new features to build a meta-model such as a logistic regression model or a multi-layer perceptron; on the test set, first use the basic model to make predictions, and then input the prediction results into the meta-model for secondary predictions, combining the advantages of multiple basic models to improve the overall fault diagnosis accuracy;
[0057] Hybrid model: Multiple non-overlapping subsets are divided from historical data, and different models such as CNN, RNN, and gradient boosting tree models are used for training on these subsets. Based on the characteristics of the data or the current operating environment, the appropriate model or model combination is dynamically selected for fault diagnosis. For example, when the data fluctuates greatly, models that are sensitive to changes are used first; when the data is relatively stable, models with better stability are selected, so that better diagnostic results can be obtained under different working conditions.
[0058] In the present invention, convolutional neural networks are good at processing the spatial characteristics of data, recurrent neural networks and their variants have good processing capabilities for time series data, and gradient boosting tree models perform well in processing complex nonlinear relationships; multiple model selection and integration methods can give full play to the advantages of different models, better adapt to the complexity of pipeline operation data, and improve the accuracy and reliability of fault diagnosis.
[0059] The pipeline operation data includes pressure, flow, temperature, vibration frequency, fluid density, corrosion rate, valve opening, pump speed, acoustic wave signal, and gas concentration parameters;
[0060] The process of building a scoring model is:
[0061] Calculate the mean of various parameters based on historical pipeline operation data and standard deviation σ i , thus obtaining the normal range of each parameter i represents the i-th parameter;
[0062] Calculate the parameters x i The degree of deviation d i ,when When d i =0, when hour, when hour, And according to the weight distribution, the comprehensive score S is obtained, which is expressed as: θ i The weight coefficients for each parameter are determined using the analytic hierarchy process (AHP), and n represents the total number of parameters. After constructing the scoring model, it is verified using cross-validation. The specific process is as follows: historical pipeline operation data is divided into multiple subsets, one of which is selected as the test set each time, and the remaining subsets are used as the training set. The scoring model is trained and evaluated on the test set. The model's performance is evaluated by calculating the accuracy and recall rate of the model on multiple test sets. If the model performance does not meet expectations, the reasons are analyzed and optimized. Optimization measures include adjusting parameter weights and adding or removing some parameters. After multiple verifications and optimizations, the scoring model can stably and accurately reflect the degree of abnormality in the pipeline network's operating status.
[0063] Combined with the dynamic threshold judgment mechanism, the data distribution type and anomaly level are determined, including concentrated distribution, positive deviation concentration, positive long tail, negative deviation concentration, and negative long tail;
[0064] Based on the comprehensive score and fault prediction value, the fault probability F is output; F = a × P + (1-a) × S. a is the weight coefficient, which is obtained based on historical data analysis;
[0065] In the present invention, multiple pipeline operating parameters are taken into consideration to comprehensively reflect the operating status of the pipeline network; the normal range is determined by calculating the mean and standard deviation, the degree of parameter deviation is accurately measured, and a comprehensive score is obtained by combining the weights to make the score more scientific; the data distribution type and abnormality level are determined to more carefully evaluate the operating status of the pipeline network, and the output failure probability provides a quantitative basis for maintenance decisions.
[0066] The dynamic threshold determination mechanism includes:
[0067] Obtain standard deviation thresholds, interquartile range thresholds, and data concentration thresholds using historical pipeline operation data;
[0068] According to the real-time range of data R i =max(x i )-min(x i ) and the average range of historical data Dynamically adjust the threshold ratio. The adjusted standard deviation threshold, interquartile range threshold, and data concentration threshold are calculated using the formula: adjusted threshold = initial threshold × [1 + k × (real-time data range / historical data average range)], where k is the adjustment coefficient, determined through experiments or experience.
[0069] The final data distribution type is determined based on the adjusted standard deviation threshold, interquartile range threshold, and data concentration threshold. The comprehensive score output by the scoring model is then compared with the preset scoring range.
[0070] When the comprehensive score is lower than the scoring interval and the data distribution type is concentrated distribution, the anomaly level is low; when the comprehensive score is higher than the scoring interval and the data distribution type is positive deviation concentrated or negative deviation concentrated, the anomaly level is high; otherwise, the anomaly level is medium.
[0071] In this invention, the dynamic threshold determination mechanism can dynamically adjust the threshold according to the changes in real-time data and historical data, and more flexibly adapt to the dynamic changes in pipe network operation. Accurately determine the data distribution type and abnormal level, improve the accuracy and timeliness of fault diagnosis, and provide an accurate basis for taking appropriate maintenance measures.
[0072] The decision logic of the data distribution type is as follows:
[0073] When the ratio of the absolute value of the difference between the mean and the median to the median is less than or equal to the set ratio, and the standard deviation and interquartile range are both lower than the corresponding standard deviation threshold and interquartile range threshold, the data type is determined to be concentrated distribution;
[0074] When the data is positively skewed and the proportion of data exceeding the upper bound is higher than the concentration threshold, the data type is determined to be positively deviating from concentration;
[0075] When the data is positively skewed, but the proportion of data exceeding the upper bound is lower than the concentration threshold, the data type is determined to be positive long tail;
[0076] When the data is negatively skewed and the proportion of data below the lower bound is higher than the concentration threshold, the data type is determined to be negatively skewed;
[0077] When the data is negatively skewed, but the proportion of data below the lower bound is lower than the concentration threshold, the data type is determined to be a negative long tail. In this invention, the clear data distribution type determination logic provides a unified standard for data analysis, making the process of determining the data distribution type more scientific and accurate, and helping to further improve the reliability of fault diagnosis.
[0078] In a further embodiment, the working process of the real-time warning module includes:
[0079] Based on the cumulative change in historical pipeline operation data, the comprehensive score output by the scoring model, and the abnormality level, the fault prediction value is obtained through the fault warning analysis model. Then, the fault warning level is divided according to the comparison result between the fault prediction value and the fault prediction interval. The fault warning levels are blue, yellow, and red from low to high.
[0080] Differentiated early warning measures are taken for different fault levels, including sound and light alarms, SMS notifications, and automatic shutdown.
[0081] In the present invention, a variety of information is combined to obtain a fault prediction value, and fault warning levels are divided and differentiated warning measures are taken. This can promptly and accurately convey fault information to operation and maintenance personnel, allowing them to take corresponding measures according to different fault levels, effectively reducing accident losses.
[0082] In a further solution, the expression of the fault warning analysis model is:
[0083]
[0084] In the formula, σ(S) represents the transformation result of the Sigmoid function on the comprehensive score S, The comprehensive score S comprehensively considers the degree of deviation of multiple pipeline operating parameters such as pressure and flow from the normal range, and comprehensively reflects the degree of abnormality of the pipeline network operation status; when multiple parameters deviate from the normal range, the comprehensive score will increase accordingly, indicating that the abnormal operation of the pipeline network is more serious; σ(S) maps the comprehensive score S to the (0,1) interval, performs a nonlinear transformation on the comprehensive score, and highlights the different effects of different score intervals on fault prediction values; for example, when the comprehensive score is low, after the Sigmoid function transformation, its value is close to 0; when the comprehensive score is high, the transformed value is close to 1, making the model more sensitive to comprehensive scores of different abnormality levels. ρ1, ρ2, and ρ3 represent exponential coefficients, and C represents the cumulative change of historical pipeline operation data, reflecting the trend of changes in pipeline operation parameters over time. For example, for pressure parameters, if the pressure continues to rise over a period of time, the C value will increase accordingly, indicating that the pressure change trend has an important impact on fault prediction. ta and tb represent the starting time and the ending time, respectively, and x i (t) represents the value of the i-th pipeline operating parameter at time t, ψ i It represents the weight factor corresponding to the change in the i-th pipeline operating parameter. L represents the abnormality level, with low level being 1, medium level being 2, and high level being 3. It directly reflects the severity of the abnormal operation of the pipeline network and provides an important reference for fault prediction. A low abnormality level indicates that the pipeline network operation status is slightly abnormal, but may not affect normal operation. A high abnormality level indicates that there is a serious abnormality in the pipeline network and the possibility of a fault is high. Γ represents the intermediate impact index, ξ j represents the comprehensive score characteristic value obtained based on historical data statistics, ω j represents the weight factor of the jth Sigmoid function variant, m represents the number of Sigmoid function variants, and P represents the fault prediction value;
[0085] When P exceeds the fault prediction interval, the fault warning level is red; when P falls within the fault prediction interval, the fault warning level is yellow; when P is lower than the fault prediction interval, the fault warning level is blue. By multiplying multiple Sigmoid function variants with comprehensive scores as variables, the complex impact of comprehensive scores on fault prediction values under different value ranges and changing trends is captured; After processing the abnormality level, an exponential operation is performed to highlight the importance of the abnormality level in fault prediction; The nonlinear transformation result of the comprehensive score is multiplied by the logarithmic transformation result of the cumulative change, and an exponential operation is performed to reflect the joint effect of the comprehensive score and the cumulative change.
[0086] In the present invention, the fault warning analysis model comprehensively considers multiple factors such as comprehensive score, cumulative change of historical data, abnormal level, etc., and more accurately calculates the fault prediction value through the above expression. The fault warning level is divided according to the fault prediction value, providing a more scientific basis for real-time warning, further improving the accuracy and reliability of fault warning.
[0087] In a further embodiment, the model of the model development module includes:
[0088] Residual models are suitable for data with a central distribution or positive deviation from central distribution;
[0089] Isolation forest, suitable for data with positive long tail distribution;
[0090] Support vector machines are suitable for data with negative deviation concentration or negative long tail distribution.
[0091] In the present invention, by selecting appropriate models for different data distribution types, the advantages of each model can be fully utilized, the model's ability to identify different fault modes can be improved, and thus the diagnostic effect of the entire fault diagnosis system can be improved.
[0092] It should be noted that the calculation formulas and various parameters involved in the calculations in the present invention have been dimensionally processed in advance, and the process of dimensionless processing is well known in the industry and will not be described here.
[0093] The above is a detailed description of an embodiment of the present invention. However, the content described is only a preferred embodiment of the present invention and should not be considered to limit the scope of the present invention. All equivalent changes and improvements made within the scope of the present invention should still fall within the scope of the patent coverage of the present invention.
Claims
1. A pipe network fault diagnosis system, characterized in that: include: Data acquisition module, used to obtain pipeline operation data from SCADA system, simulation environment, intermediate database, API transmission and manual inspection data entry; The data preprocessing module is used to clean, normalize and structure the collected data, including data availability testing, basic statistics calculation and box plot drawing; The data analysis module builds a scoring model based on pipeline operation data, and then determines the data distribution type based on the scoring model and the dynamic threshold judgment mechanism; The model development module selects the optimal model based on data distribution characteristics, supports the integration of multiple models, generates and evaluates multiple model results, and uses a hybrid modeling strategy to output visual model results; The real-time early warning module is used to monitor the operation status of the pipeline network in real time and issue early warning information based on the diagnosis results for visual display.
2. The pipe network fault diagnosis system according to claim 1, characterized in that: The model development module uses convolutional neural networks, recurrent neural networks and their variants, and gradient boosting tree models for model selection, and uses stacking integration and hybrid models for model integration.
3. The pipe network fault diagnosis system according to claim 2, characterized in that: The pipeline operation data includes pressure, flow, temperature, vibration frequency, fluid density, corrosion rate, valve opening, pump speed, acoustic wave signal, and gas concentration parameters; The process of building a scoring model is: Calculate the mean of various parameters based on historical pipeline operation data and standard deviation σ i , thus obtaining the normal range of each parameter i represents the i-th parameter; Calculate the parameters x i The degree of deviation d i ,when When d i =0, when hour, when hour, And according to the weight distribution, the comprehensive score S is obtained, which is expressed as: θ i Indicates the weight coefficient corresponding to each parameter, and n indicates the total number of parameters; Combined with the dynamic threshold judgment mechanism, the data distribution type and anomaly level are determined, including concentrated distribution, positive deviation concentration, positive long tail, negative deviation concentration, and negative long tail; According to the comprehensive score and fault prediction value, the fault probability is output.
4. The pipe network fault diagnosis system according to claim 3, characterized in that: The dynamic threshold determination mechanism includes: Obtain standard deviation thresholds, interquartile range thresholds, and data concentration thresholds using historical pipeline operation data; According to the real-time range of data R i =max(x i )-min(x i ) and the average range of historical data Dynamically adjust the threshold ratio. The adjusted standard deviation threshold, interquartile range threshold, and data concentration threshold are calculated using the formula: adjusted threshold = initial threshold × [1 + k × (real-time data range / historical data average range)], where k is the adjustment coefficient, determined through experiments or experience. The final data distribution type is determined based on the adjusted standard deviation threshold, interquartile range threshold, and data concentration threshold. The comprehensive score output by the scoring model is then compared with the preset scoring range. When the comprehensive score is lower than the scoring interval and the data distribution type is concentrated distribution, the anomaly level is low; when the comprehensive score is higher than the scoring interval and the data distribution type is positive deviation concentrated or negative deviation concentrated, the anomaly level is high; otherwise, the anomaly level is medium.
5. The pipe network fault diagnosis system according to claim 4, characterized in that: The decision logic of the data distribution type is as follows: When the ratio of the absolute value of the difference between the mean and the median to the median is less than or equal to the set ratio, and the standard deviation and interquartile range are both lower than the corresponding standard deviation threshold and interquartile range threshold, the data type is determined to be concentrated distribution; When the data is positively skewed and the proportion of data exceeding the upper bound is higher than the concentration threshold, the data type is determined to be positively deviating from concentration; When the data is positively skewed, but the proportion of data exceeding the upper bound is lower than the concentration threshold, the data type is determined to be positive long tail; When the data is negatively skewed and the proportion of data below the lower bound is higher than the concentration threshold, the data type is determined to be negatively skewed; When the data is negatively skewed, but the proportion of data below the lower bound is lower than the concentration threshold, the data type is determined to be negative long tail.
6. The pipe network fault diagnosis system according to claim 1, characterized in that: The working process of the real-time warning module includes: Based on the cumulative change in historical pipeline operation data, the comprehensive score output by the scoring model, and the abnormality level, the fault prediction value is obtained through the fault warning analysis model. Then, the fault warning level is divided according to the comparison result between the fault prediction value and the fault prediction interval. The fault warning levels are blue, yellow, and red from low to high. Differentiated early warning measures are taken for different fault levels, including sound and light alarms, SMS notifications, and automatic shutdown.
7. The pipe network fault diagnosis system according to claim 6, characterized in that: The expression of the fault warning analysis model is: In the formula, σ(S) represents the transformation result of the Sigmoid function on the comprehensive score S, ρ1, ρ2, and ρ3 represent exponential coefficients, C represents the cumulative change in historical pipeline operation data, ta and tb represent the start and end times, respectively, and x i (t) represents the value of the i-th pipeline operating parameter at time t, ψ i represents the weight factor corresponding to the change in the i-th pipeline operating parameter, L represents the abnormality level, with low level being 1, medium level being 2, and high level being 3, Γ represents the intermediate impact index, and ξ j represents the comprehensive score characteristic value obtained based on historical data statistics, ω j represents the weight factor of the jth Sigmoid function variant, m represents the number of Sigmoid function variants, and P represents the fault prediction value; When P exceeds the fault prediction interval, the fault warning level is red; when P falls within the fault prediction interval, the fault warning level is yellow; when P is lower than the fault prediction interval, the fault warning level is blue.
8. The pipe network fault diagnosis system according to claim 1, characterized in that: The models of the model development module include: Residual models are suitable for data with a central distribution or positive deviation from central distribution; Isolation forest, suitable for data with positive long tail distribution; Support vector machines are suitable for data with negative deviation concentration or negative long tail distribution.
Citation Information
Patent Citations
Model training method and system and storage medium
CN114692886A
Intelligent factory management system based on Internet of Things
CN118897514A
Image classification method for long-tail category distribution
CN119206288A
Integrated pipeline management system, method, equipment and medium
CN119244942A
Failure Prediction and Analysis Techniques
US20190316902A1
Cited By
Component operation and maintenance prediction method, system and equipment of semiconductor equipment and medium
CN120975766A
A method and device for detecting abnormality of municipal sewer network liquid level data
CN122505378A
