A method for writing risk assessment reports for high consequence areas

Abstract

The present invention relates to the technical field of data processing, and discloses a method for compiling a risk assessment report for a high consequence area, comprising: collecting a number of risk assessment reports, extracting keywords from each risk assessment report, calculating report feature values, and dividing the number of risk assessment reports into template levels; collecting the operating data of the pipeline to be detected, identifying abnormal data in the operating data, and obtaining the risk prediction value of the pipeline to be detected; determining the statistical range of the geographical area according to the risk prediction value to determine whether to adjust the risk prediction value, and obtaining the final risk prediction value; determining the template level corresponding to the evaluation report writing template according to the final risk prediction value, obtaining the matching degree between the abnormal data and each template in the corresponding template level to determine the risk assessment report writing template; and filling the abnormal data into the risk assessment report writing template. The present invention realizes the automation and intelligence of pipeline risk assessment report writing, and improves the efficiency and uniformity of report writing.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a method for compiling a high consequence area risk assessment report. Background Art

[0002] With the continuous advancement of industrialization, pipeline systems have been widely used in many fields such as oil, natural gas, and water resources management. However, due to changes in various external environments and internal pressures during pipeline operation, abnormal conditions are prone to occur, such as abnormal fluctuations in temperature, pressure, flow, or vibration. In order to ensure the safe operation of pipelines and deal with abnormal conditions in a timely manner, risk assessments are required for high-consequence areas of oil and gas pipelines every year to identify risks around buried oil and gas pipelines.

[0003] However, the current risk assessment process is usually conducted manually, requiring professionals to conduct assessments based on surrounding environmental factors, relevant standards and regulations, and then prepare an assessment report. This manual assessment process is time-consuming and labor-intensive, prone to human errors, and unable to reflect the latest changes in the surrounding environment in a timely manner. There is a risk of incomplete information or duplication of work. At the same time, manually preparing an assessment report requires a lot of time and effort, and there may be problems such as insufficient uniformity in report format and content. This inefficient report preparation process may delay the release of assessment results and affect subsequent decision-making and the implementation of response measures.

[0004] Therefore, it is necessary to design a method for writing high consequence area risk assessment reports to solve the problems existing in current technology. Summary of the invention

[0005] In view of this, the present invention proposes a method for compiling a high consequence area risk assessment report, aiming to solve the problems that the current high consequence area risk assessment report is time-consuming and labor-intensive to compile, and the report has low accuracy and is easily affected by human factors.

[0006] The present invention proposes a method for compiling a high consequence area risk assessment report, comprising:

[0007] Collecting a number of risk assessment reports based on big data, extracting keywords from each of the risk assessment reports, calculating report feature values ​​based on the keywords, and classifying the risk assessment reports into template levels based on the report feature values;

[0008] Collecting operation data of the pipeline to be inspected, identifying abnormal data in the operation data, and obtaining a risk prediction value of the pipeline to be inspected based on a random forest model;

[0009] Determine the statistical range of the geographical area according to the risk prediction value, and determine whether to adjust the risk prediction value according to the statistical data within the statistical range of the geographical area to obtain a final risk prediction value;

[0010] Determine the template level corresponding to the evaluation report writing template according to the final risk prediction value, obtain the matching degree between the abnormal data and each template in the corresponding template level after determining the corresponding template level, and determine the risk evaluation report writing template according to the matching degree;

[0011] Filling the abnormal data into the risk assessment report writing template;

[0012] Extracting keywords from each of the risk assessment reports and calculating the report characteristic value based on the keywords includes:

[0013] Calculating the word frequency and inverse document frequency of the keywords in the risk assessment report, and obtaining the TF-IDF value of each keyword according to the word frequency and inverse document frequency;

[0014] Obtaining the report feature value by weighted averaging the TF-IDF values ​​of the keywords;

[0015] When the risk assessment reports are divided into template levels according to the report characteristic values, it includes:

[0016] S1: Initialize K centroids and assign each risk assessment report to the nearest centroid to form K clusters;

[0017] S2: Recalculate the centroid of each cluster;

[0018] S3: Repeat S1 and S2 until the center of mass no longer changes or the preset number of repetitions is completed;

[0019] S4: The centroid of each cluster is taken as the eigenvalue of the cluster;

[0020] The centroid calculation formula is as follows:

[0021] ;

[0022] Among them, Zk represents the centroid of the k-th cluster, represents the number of reports in the kth cluster, Ti represents the feature vector of the i-th report;

[0023] When the risk assessment reports are divided into template levels according to the report characteristic values, it also includes:

[0024] Compare the characteristic value Zk of the cluster with the first preset characteristic value Zk1 and the second preset characteristic value Zk2 respectively, Zk1<Zk2, and classify the risk assessment reports into template levels according to the comparison results;

[0025] When Zk≤Zk1, the risk assessment report in the cluster is classified as the first template level;

[0026] When Zk1<Zk≤Zk2, the risk assessment report in the cluster is classified into the second template level;

[0027] When Zk2<Zk, the risk assessment report in the cluster is classified into the third template level;

[0028] The first template level indicates that the risk value of the risk assessment report is lower than the second template level, and the second template level indicates that the risk value of the risk assessment report is lower than the third template level;

[0029] When determining the risk assessment report writing template based on the matching degree, it includes:

[0030] Obtaining the matching degree between the abnormal data and each template in the corresponding template level, and determining the risk assessment report writing template according to the matching degree, including:

[0031] Obtain the number N of abnormal features according to the abnormal data, and calculate the matching degree according to the following formula;

[0032] ;

[0033] Where P represents the matching degree, Yi represents the feature value of the i-th data in the abnormal data; Mki represents the feature value of the i-th data of template k, N represents the number of abnormal features, and Wi represents the i-th weight;

[0034] The matching degree between the abnormal data and each template in the corresponding template level is arranged from high to low, and the template with the highest matching degree is selected as the template for writing the risk assessment report.

[0035] Furthermore, when obtaining the risk prediction value of the pipeline to be inspected based on the random forest model, it includes:

[0036] Acquire historical operation data from the operation data, and divide the historical operation data into a training set and a test set;

[0037] Use cross validation combined with grid search to find the best parameters of the random forest model and establish a random forest model;

[0038] Fitting the random forest model using the training set;

[0039] Bringing the test set into the random forest model and calculating the accuracy of the risk prediction value;

[0040] When the accuracy reaches a preset accuracy threshold, a risk prediction value of the current pipeline to be detected is obtained according to the abnormal data, and the risk prediction value is recorded as F.

[0041] Furthermore, when determining the statistical scope of the geographical area according to the risk prediction value, it includes:

[0042] The risk prediction value F is compared with a first preset risk prediction value F1 and a second preset risk prediction value F2 respectively, F1<F2, and the statistical scope of the geographical area is determined according to the comparison result;

[0043] When F≤F1, determining the geographic area statistical range to be a first preset range;

[0044] When F1<F≤F2, determining the geographic area statistical range as a second preset range;

[0045] When F2<F, determining the geographic area statistical range as a third preset range;

[0046] The first preset range indicates that the statistical area is smaller than the second preset range, and the second preset range indicates that the statistical area is smaller than the third preset range.

[0047] Furthermore, judging whether to adjust the risk prediction value according to the statistical data within the statistical range of the geographical area includes:

[0048] Collecting population density data R within the statistical range of the geographic area, comparing the population density data R with a population density threshold Rmax, and determining whether to adjust the risk prediction value according to the comparison result;

[0049] When R>Rmax, it is determined that the risk prediction value is adjusted to obtain an adjusted risk prediction value, and the adjusted risk prediction value is used as the final risk prediction value;

[0050] When R≤Rmax, it is determined that the risk prediction value is not to be adjusted, and the risk prediction value is used as the final risk prediction value.

[0051] Further, when determining to adjust the risk prediction value, it includes:

[0052] Obtain a density difference ΔR according to the population density data R and the population density threshold Rmax, ΔR=R-Rmax, compare the density difference ΔR with a first preset density difference ΔR1 and a second preset density difference ΔR2 respectively, ΔR1<ΔR2, and adjust the risk prediction value according to the comparison result;

[0053] When ΔR≤ΔR1, a first preset adjustment coefficient A1 is determined to adjust the risk prediction value to obtain an adjusted risk prediction value F*A1;

[0054] When ΔR1<ΔR≤ΔR2, determine a second preset adjustment coefficient A2 to adjust the risk prediction value to obtain an adjusted risk prediction value F*A2;

[0055] When ΔR2<ΔR, determine a third preset adjustment coefficient A3 to adjust the risk prediction value to obtain an adjusted risk prediction value F*A3;

[0056] Among them, 1<A1<A2<A3<1.2.

[0057] Further, when determining the template level corresponding to the evaluation report writing template according to the final risk prediction value, it includes:

[0058] When the final risk prediction value is in the first interval, the template level corresponding to the evaluation report writing template is determined to be the first template level;

[0059] When the final risk prediction value is in the second interval, the template level corresponding to the evaluation report writing template is determined to be the second template level;

[0060] When the final risk prediction value is in the third interval, the template level corresponding to the evaluation report writing template is determined to be the third template level;

[0061] Among them, the first interval is that the final risk prediction value is less than or equal to the first preset risk prediction value F1, the first interval is that the final risk prediction value is greater than the first preset risk prediction value F1 and less than or equal to the second preset risk prediction value F2, and the third interval is that the final risk prediction value is greater than the second preset risk prediction value F2.

[0062] Compared with the prior art, the beneficial effects of the present invention are: the automation and intelligence of pipeline risk assessment report writing is realized through big data and machine learning technology. By using big data collection and natural language processing technology, keywords in a large number of risk assessment reports are extracted and analyzed, and report feature values ​​are calculated, so that the reports are divided into different template levels, which improves the accuracy of report classification. The pipeline operation data is analyzed and risk predicted in real time through the random forest model, and adjusted in combination with regional statistical data to ensure the accuracy and reliability of the risk prediction value. According to the adjusted risk prediction value, the most matching report writing template is selected to automatically generate a risk assessment report with unified structure and detailed content. It reduces human errors and repetitive work, improves the efficiency and uniformity of report writing, and can also reflect the latest environmental changes in a timely manner, thereby improving the timeliness and effectiveness of pipeline risk management. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Various other advantages and benefits will become apparent to those of ordinary skill in the art by reading the detailed description of the preferred embodiments below. The accompanying drawings are only for the purpose of illustrating the preferred embodiments and are not to be considered as limiting the present invention. Moreover, the same reference symbols are used throughout the accompanying drawings to represent the same components. In the accompanying drawings:

[0064] Figure 1 A flowchart of a method for compiling a high consequence area risk assessment report provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0065] Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments described herein. On the contrary, these embodiments are provided in order to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art. It should be noted that, in the absence of conflict, the embodiments of the present invention and the features described in the embodiments can be combined with each other. The present invention will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0066] With the acceleration of industrialization, pipeline systems are increasingly used in many fields such as oil, natural gas, and water resource management. However, during the operation of pipeline systems, various abnormal conditions often occur due to changes in the external environment and internal pressure, such as abnormal fluctuations in temperature, pressure, flow, or vibration. In order to ensure the safe operation of pipelines and deal with these abnormal conditions in a timely manner, especially in high-consequence areas for oil and gas pipelines, detailed risk assessments are required every year to identify and assess the risks around buried oil and gas pipelines.

[0067] The current risk assessment process mainly relies on manual operations, which has significant limitations and multiple problems. First, the manual risk assessment process is time-consuming and labor-intensive. Professionals need to conduct detailed assessments based on surrounding environmental factors, relevant standards and regulations, which not only requires a large amount of human resources, but also consumes a lot of time. In addition, manual operations are prone to human errors, which greatly reduces the accuracy and reliability of the assessment results. Due to the limitations of manual data processing, manual evaluations often fail to reflect the latest changes in the surrounding environment in a timely manner, which may lead to incomplete information or duplication of work.

[0068] In addition, manually compiling risk assessment reports requires a lot of time and effort. This inefficient report preparation process may delay the release of assessment results, which in turn affects subsequent decision-making and the implementation of countermeasures. Due to the lack of unified standards, manually compiled reports may have inconsistencies in format and content, further reducing the practicality and reference value of the reports. All these problems greatly weaken the efficiency and accuracy of risk assessment and affect the overall effectiveness of pipeline safety management.

[0069] In view of the above problems, it is particularly necessary to design a method for writing risk assessment reports in high consequence areas.

[0070] In some embodiments of the present application, see Figure 1 As shown, a method for compiling a high consequence area risk assessment report includes:

[0071] S100: Collecting a number of risk assessment reports based on big data, extracting keywords from each risk assessment report, calculating report feature values ​​based on the keywords, and classifying the risk assessment reports into template levels based on the report feature values.

[0072] S200: Collecting operation data of the pipeline to be inspected, identifying abnormal data in the operation data, and obtaining a risk prediction value of the pipeline to be inspected based on a random forest model.

[0073] S300: Determine the statistical scope of the geographical area according to the risk prediction value, and determine whether to adjust the risk prediction value according to the statistical data within the statistical scope of the geographical area to obtain a final risk prediction value.

[0074] S400: Determine the template level corresponding to the evaluation report writing template according to the final risk prediction value. After determining the corresponding template level, obtain the matching degree between the abnormal data and each template in the corresponding template level, and determine the risk evaluation report writing template according to the matching degree.

[0075] S500: Fill in the abnormal data into the risk assessment report writing template.

[0076] Specifically, S100 collects a large number of risk assessment reports based on big data, and uses natural language processing (NLP) technology to extract keywords from each risk assessment report. Keywords are important words in the report that reflect risk characteristics. For example, data such as temperature, vibration, and pressure anomalies in pipeline detection, the characteristic value of each report is calculated based on the extracted keywords. The characteristic value is a quantitative representation of the content of the report, including factors such as the frequency and importance of the keyword. According to the characteristic value, the risk assessment report is divided into different template levels, and the template level reflects the importance of the content in the report. S200 collects the operating data of the pipeline to be inspected in real time, including parameters such as temperature, pressure, flow, and vibration, and identifies abnormal conditions in the operating data through preset thresholds or statistical analysis methods. The random forest model is used to analyze the abnormal data and predict the risk value of the pipeline. The random forest model is an integrated learning method that can improve the accuracy and robustness of predictions through the voting results of multiple decision trees. S300 determines the geographical area that needs attention based on the preliminary risk prediction value. In the determined area, relevant statistical data such as population density, key infrastructure, and environmental factors are collected. Based on the collected statistical data, it is determined whether the preliminary risk prediction value needs to be adjusted to improve its accuracy and reliability, and finally the adjusted risk prediction value is obtained. In S400, the appropriate report writing template level is determined based on the final risk prediction value, the abnormal data is matched with each template in the template level, the matching degree is calculated, and the most suitable template is selected based on the matching degree to write the final risk assessment report.

[0077] It is understandable that the automation and accuracy of risk assessment report writing have been improved through big data and machine learning technology. The keywords of the report are extracted and analyzed through big data technology, and the characteristic values ​​are automatically calculated and the template levels are divided. Using the pipeline operation data collected in real time, the random forest model is used for risk prediction, and the predicted value is adjusted in combination with regional statistical data. According to the adjusted predicted value, the appropriate report template is selected to ensure that the report content is accurate and reflects the latest environmental changes, and finally achieve efficient and accurate risk assessment report writing. Through automated keyword extraction and template division, human errors and duplication of work are reduced. Real-time data collection and random forest model prediction ensure the timeliness and accuracy of risk assessment. Adjustment combined with regional statistical data further improves the reliability of the predicted value. The automatically selected and generated risk assessment report has a unified format and detailed content, which significantly improves the decision-making efficiency and response capabilities of risk management.

[0078] In some embodiments of the present application, extracting keywords from each risk assessment report and calculating report feature values ​​according to the keywords includes:

[0079] The word frequency and inverse document frequency of the keywords in the risk assessment report are calculated, and the TF-IDF value of each keyword is obtained based on the word frequency and inverse document frequency.

[0080] The reported feature value is obtained by weighted averaging the TF-IDF values ​​of the keywords.

[0081] Specifically, relevant keywords are extracted from each risk assessment report. The frequency of each keyword in a single risk assessment report is calculated. The word frequency represents the ratio of the number of times a keyword appears in a document to the total number of words in the document. The frequency of each keyword in all risk assessment reports is calculated. The inverse document frequency represents the prevalence of a keyword in all documents. The word frequency and the inverse document frequency are multiplied to obtain the TF-IDF value of each keyword. The TF-IDF value measures the importance of a keyword in a specific document. The TF-IDF values ​​of all keywords in the report are weighted averaged to obtain the characteristic value of the report.

[0082] It is understandable that the use of big data and natural language processing technology has effectively improved the efficiency of risk assessment report analysis and writing. By automatically extracting keywords and calculating TF-IDF values, accurate quantification and objective analysis of report content are ensured. The weighted average feature value provides a reliable basis for template division and reduces human errors.

[0083] In some embodiments of the present application, when a number of risk assessment reports are divided into template levels according to report characteristic values, it includes:

[0084] S1: Initialize K centroids and assign each risk assessment report to the nearest centroid to form K clusters.

[0085] S2: Recalculate the centroid of each cluster.

[0086] S3: Repeat S1 and S2 until the center of mass no longer changes or the preset number of repetitions is completed.

[0087] S4: Take the centroid of each cluster as the eigenvalue of the cluster.

[0088] The centroid calculation formula is as follows:

[0089] ;

[0090] Among them, Zk represents the centroid of the k-th cluster, represents the number of reports in the kth cluster, and Ti represents the feature vector of the i-th report.

[0091] Specifically, K initial centroid points are selected, including using methods such as the elbow method or the silhouette coefficient to determine the optimal number of clusters K. Each risk assessment report is assigned to the centroid closest to it to form K initial clusters. For each cluster, the centroid is recalculated, and steps S1 and S2 are repeated until the centroid position no longer changes (that is, the change of all centroids is less than a certain threshold) or the preset number of iterations is reached. After the iteration, the centroid of each cluster is the eigenvalue of the cluster. This eigenvalue can represent the common characteristics of all reports in the cluster. Through the K-means clustering algorithm, several risk assessment reports can be effectively divided into different template levels according to their eigenvalues. The K-means algorithm gradually optimizes the centroid position of each cluster by initializing the centroid, assigning reports, recalculating the centroid, and iteratively adjusting the centroid position, thereby finding the best report grouping. It can automatically discover the inherent structure of the report, classify reports with similar characteristics into one category, and improve the accuracy and consistency of report classification.

[0092] It can be understood that the use of K-means clustering algorithm to automatically divide risk assessment reports into different template levels significantly improves the efficiency and accuracy of report classification. By iteratively optimizing the centroid position, it ensures that the characteristic values ​​of the cluster can accurately represent the common characteristics of all reports in the cluster. The final classification result can better reflect the similarity of report content and provide a reliable basis for subsequent risk assessment. This automated method reduces human intervention, reduces subjective bias in the classification process, and improves the objectivity and uniformity of report classification.

[0093] In some embodiments of the present application, when dividing a number of risk assessment reports into template levels according to report characteristic values, it also includes: comparing the characteristic value Zk of the cluster with a pre-set first preset characteristic value Zk1 and a second preset characteristic value Zk2, Zk1<Zk2, and dividing the number of risk assessment reports into template levels according to the comparison results.

[0094] Specifically, when Zk≤Zk1, the risk assessment report in the cluster is classified as the first template level. When Zk1<Zk≤Zk2, the risk assessment report in the cluster is classified as the second template level. When Zk2<Zk, the risk assessment report in the cluster is classified as the third template level. The first template level indicates that the risk value of the risk assessment report is lower than the second template level, and the second template level indicates that the risk value of the risk assessment report is lower than the third template level.

[0095] It is understandable that the automated classification method improves efficiency, reduces human errors, and ensures the consistency and accuracy of report classification. By dividing the template into different levels, the risk level of the report can be clearly reflected, providing reliable data support for subsequent risk management and decision-making. It helps to quickly identify and respond to pipeline problems of different risk levels and improve the safety management level of the pipeline system.

[0096] In some embodiments of the present application, when obtaining the risk prediction value of the pipeline to be inspected based on the random forest model, it includes:

[0097] The historical operation data is obtained from the operation data, and the historical operation data is divided into a training set and a test set.

[0098] Cross validation combined with grid search was used to find the optimal parameters of the random forest model and establish a random forest model.

[0099] Fit a random forest model using the training set.

[0100] The test set is fed into the random forest model and the accuracy of the risk prediction value is calculated.

[0101] When the accuracy reaches the preset accuracy threshold, the risk prediction value of the current pipeline to be detected is obtained according to the abnormal data, and the risk prediction value is recorded as F.

[0102] Specifically, key data such as temperature, pressure, flow, vibration, etc. are obtained from the historical operation data of the pipeline. These data record the operation status of the pipeline in different time periods. The historical operation data is divided into a training set and a test set. Usually 70%-80% of the data is used as the training set, and the rest is used as the test set. Ensure that both the training set and the test set contain data of various operating conditions to improve the generalization ability of the model. Use cross-validation combined with grid search to find the best parameters for the random forest model. Cross-validation divides the data into several parts and trains the model multiple times to verify its stability and performance. Grid search exhaustively searches for the best parameter combination in the parameter space, such as the number of trees (n_estimators), the maximum depth of the tree (max_depth), etc. Use the optimized parameters to fit the random forest model with the training set data. Random forest integrates multiple decision trees and averages the prediction results of multiple trees to reduce the risk of overfitting and improve the accuracy and stability of the model. Input the test set data into the trained random forest model and calculate the prediction accuracy of the model. The accuracy reflects the performance of the model on unknown data and is an important indicator for evaluating model performance. After the model reaches the preset accuracy threshold, the current abnormal data is used to input the random forest model to calculate the risk prediction value of the pipeline to be tested, which is recorded as F.

[0103] It is understandable that using historical data to train the random forest model can make full use of the historical operation information of the pipeline and improve the prediction ability of the model. The combination of cross-validation and grid search ensures the optimization of model parameters, making the performance of the model on different data sets more stable and reliable. In the process of model training and verification, by continuously adjusting parameters and verifying model performance, a model with high accuracy is finally obtained. The model is trained using historical data, and the parameters are optimized by combining cross-validation and grid search to ensure the accuracy and stability of the model in practical applications. After reaching the preset accuracy threshold, the model can predict the risk of abnormal data of the pipeline in real time, providing efficient and accurate risk assessment. It not only improves the accuracy of risk prediction, but also can detect potential problems in a timely manner, enhancing the initiative and scientificity of pipeline safety management.

[0104] In some embodiments of the present application, when determining the statistical scope of a geographical area based on a risk prediction value, it includes: comparing the risk prediction value F with a first preset risk prediction value F1 and a second preset risk prediction value F2, respectively, F1<F2, and determining the statistical scope of the geographical area based on the comparison result.

[0105] Specifically, when F≤F1, the geographic area statistical range is determined to be the first preset range. When F1<F≤F2, the geographic area statistical range is determined to be the second preset range. When F2<F, the geographic area statistical range is determined to be the third preset range. The first preset range indicates that the statistical area is smaller than the second preset range, and the second preset range indicates that the statistical area is smaller than the third preset range.

[0106] It is understandable that by comparing the risk prediction value with the preset risk value and determining the statistical scope of different geographical areas, the risk assessment is made more accurate and flexible. By dynamically adjusting the size of the statistical area according to the risk value, potential risks can be more effectively identified and managed to ensure the safety of pipeline operation. Especially when the risk value is high, by expanding the statistical area, the potential risk area can be fully covered, the risk prevention ability is enhanced, and the efficiency and accuracy of risk assessment are improved.

[0107] In some embodiments of the present application, when judging whether to adjust the risk prediction value based on statistical data within the statistical scope of a geographic area, it includes: collecting population density data R within the statistical scope of the geographic area, comparing the population density data R with the population density threshold Rmax, and judging whether to adjust the risk prediction value based on the comparison result.

[0108] Specifically, when R>Rmax, it is determined that the risk prediction value is adjusted, the adjusted risk prediction value is obtained, and the adjusted risk prediction value is used as the final risk prediction value. When R≤Rmax, it is determined that the risk prediction value is not adjusted, and the risk prediction value is used as the final risk prediction value.

[0109] In some embodiments of the present application, when it is determined to adjust the risk prediction value, it includes: obtaining a density difference ΔR based on the population density data R and the population density threshold Rmax, ΔR=R-Rmax, comparing the density difference ΔR with a pre-set first preset density difference ΔR1 and a second preset density difference ΔR2, ΔR1<ΔR2, and adjusting the risk prediction value according to the comparison result.

[0110] Specifically, when ΔR≤ΔR1, the first preset adjustment coefficient A1 is determined to adjust the risk prediction value to obtain the adjusted risk prediction value F*A1. When ΔR1<ΔR≤ΔR2, the second preset adjustment coefficient A2 is determined to adjust the risk prediction value to obtain the adjusted risk prediction value F*A2. When ΔR2<ΔR, the third preset adjustment coefficient A3 is determined to adjust the risk prediction value to obtain the adjusted risk prediction value F*A3. Among them, 1<A1<A2<A3<1.2.

[0111] Specifically, by collecting and analyzing population density data in a region, the potential risk of the region can be assessed more accurately. In areas with high population density, the tolerance for risk is low due to the dense population, so the risk prediction value needs to be adjusted upward. By setting different density differences and corresponding adjustment coefficients, the risk prediction value can be flexibly adjusted according to the specific density situation to ensure the accuracy and rationality of the assessment results.

[0112] It is understandable that by incorporating population density data in the region into the risk assessment system, a more refined and dynamic risk prediction is achieved. The risk prediction value is adjusted according to different population density levels, making the risk assessment more accurate, especially in high population density areas, so that higher risks can be identified and assessed in a timely manner. This improves the accuracy of risk assessment, enhances the flexibility and response capabilities of pipeline management, and ensures the safety of pipeline operation and the safety of residents in the region.

[0113] In some embodiments of the present application, when determining the template level corresponding to the evaluation report writing template according to the final risk prediction value, it includes: when the final risk prediction value is in the first interval, determining that the template level corresponding to the evaluation report writing template is the first template level. When the final risk prediction value is in the second interval, determining that the template level corresponding to the evaluation report writing template is the second template level. When the final risk prediction value is in the third interval, determining that the template level corresponding to the evaluation report writing template is the third template level. Among them, the first interval is that the final risk prediction value is less than or equal to the first preset risk prediction value F1, the first interval is that the final risk prediction value is greater than the first preset risk prediction value F1 and less than or equal to the second preset risk prediction value F2, and the third interval is that the final risk prediction value is greater than the second preset risk prediction value F2.

[0114] In some embodiments of the present application, after determining the corresponding template level, determining the risk assessment report writing template according to the matching degree includes: obtaining the matching degree between the abnormal data and each template in the corresponding template level, and determining the risk assessment report writing template according to the matching degree includes:

[0115] The number N of abnormal features is obtained according to the abnormal data, and the matching degree is calculated according to the following formula.

[0116] ;

[0117] Among them, P represents the matching degree, Yi represents the characteristic value of the i-th data in the abnormal data. Mki represents the characteristic value of the i-th data of template k, N represents the number of abnormal features, and Wi represents the i-th weight. The matching degree of the abnormal data and each template in the corresponding template level is arranged from high to low, and the template with the highest matching degree is selected as the template for writing the risk assessment report.

[0118] Specifically, the template level is determined according to the final risk prediction value, and the final template is determined according to the matching degree between the abnormal data and the template. According to the different degrees of risk and the characteristics of the abnormal data, the applicable evaluation report template can be accurately selected, so as to more accurately describe the risk situation of the pipeline and take corresponding measures.

[0119] It is understandable that by combining the characteristics of risk prediction values ​​and abnormal data, the automatic selection of risk assessment report templates is realized, avoiding manual intervention and subjective judgment bias, and improving the accuracy and consistency of the assessment report. According to different risk prediction value intervals and the template with the highest matching degree, the assessment report is more targeted and practical, providing more effective tools and methods for pipeline safety management.

[0120] In the above embodiments, the automation and intelligence of pipeline risk assessment report writing are realized through big data and machine learning technology. By using big data collection and natural language processing technology, keywords in a large number of risk assessment reports are extracted and analyzed, and report feature values ​​are calculated, so that the reports are divided into different template levels, which improves the accuracy of report classification. The pipeline operation data is analyzed and risk predicted in real time through the random forest model, and adjusted in combination with regional statistical data to ensure the accuracy and reliability of the risk prediction value. According to the adjusted risk prediction value, the most matching report writing template is selected to automatically generate a risk assessment report with unified structure and detailed content. It reduces human errors and repetitive work, improves the efficiency and uniformity of report writing, and can also reflect the latest environmental changes in a timely manner, thereby improving the timeliness and effectiveness of pipeline risk management.

[0121] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems or computer program products. Therefore, the present application may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0122] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems) and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0123] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0124] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process in the computer or other programmable device. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0125] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for compiling a high consequence area risk assessment report, characterized in that: include: Collecting a number of risk assessment reports based on big data, extracting keywords from each of the risk assessment reports, calculating report feature values ​​based on the keywords, and classifying the risk assessment reports into template levels based on the report feature values; Collecting operation data of the pipeline to be inspected, identifying abnormal data in the operation data, and obtaining a risk prediction value of the pipeline to be inspected based on a random forest model; Determine the statistical range of the geographical area according to the risk prediction value, and determine whether to adjust the risk prediction value according to the statistical data within the statistical range of the geographical area to obtain a final risk prediction value; Determine the template level corresponding to the evaluation report writing template according to the final risk prediction value, obtain the matching degree between the abnormal data and each template in the corresponding template level after determining the corresponding template level, and determine the risk evaluation report writing template according to the matching degree; Filling the abnormal data into the risk assessment report writing template; Extracting keywords from each of the risk assessment reports and calculating the report characteristic value based on the keywords includes: Calculating the word frequency and inverse document frequency of the keywords in the risk assessment report, and obtaining the TF-IDF value of each keyword according to the word frequency and inverse document frequency; Obtaining the report feature value by weighted averaging the TF-IDF values ​​of the keywords; When the risk assessment reports are divided into template levels according to the report characteristic values, it includes: S1: Initialize K centroids and assign each risk assessment report to the nearest centroid to form K clusters; S2: Recalculate the centroid of each cluster; S3: Repeat S1 and S2 until the center of mass no longer changes or the preset number of repetitions is completed; S4: The centroid of each cluster is taken as the eigenvalue of the cluster; The centroid calculation formula is as follows: ； Among them, Zk represents the centroid of the k-th cluster, represents the number of reports in the kth cluster, Ti represents the feature vector of the i-th report; When the risk assessment reports are divided into template levels according to the report characteristic values, it also includes: Compare the characteristic value Zk of the cluster with the first preset characteristic value Zk1 and the second preset characteristic value Zk2 respectively, Zk1<Zk2, and divide the risk assessment reports into template levels according to the comparison results; When Zk≤Zk1, the risk assessment report in the cluster is classified as the first template level; When Zk1<Zk≤Zk2, the risk assessment report in the cluster is classified into the second template level; When Zk2<Zk, the risk assessment report in the cluster is classified into the third template level; The first template level indicates that the risk value of the risk assessment report is lower than the second template level, and the second template level indicates that the risk value of the risk assessment report is lower than the third template level; When determining the risk assessment report writing template based on the matching degree, it includes: Obtaining the matching degree between the abnormal data and each template in the corresponding template level, and determining the risk assessment report writing template according to the matching degree, including: Obtain the number N of abnormal features according to the abnormal data, and calculate the matching degree according to the following formula; ； Where P represents the matching degree, Yi represents the feature value of the i-th data in the abnormal data; Mki represents the feature value of the i-th data of template k, N represents the number of abnormal features, and Wi represents the i-th weight; The matching degree between the abnormal data and each template in the corresponding template level is arranged from high to low, and the template with the highest matching degree is selected as the template for writing the risk assessment report.

2. The method for compiling a high consequence area risk assessment report according to claim 1, characterized in that: When obtaining the risk prediction value of the pipeline to be inspected based on the random forest model, it includes: Acquire historical operation data from the operation data, and divide the historical operation data into a training set and a test set; Use cross validation combined with grid search to find the best parameters of the random forest model and establish a random forest model; Fitting the random forest model using the training set; Bringing the test set into the random forest model and calculating the accuracy of the risk prediction value; When the accuracy reaches a preset accuracy threshold, a risk prediction value of the current pipeline to be detected is obtained according to the abnormal data, and the risk prediction value is recorded as F.

3. The method for compiling a high consequence area risk assessment report according to claim 2, characterized in that: When determining the statistical scope of the geographical area based on the risk prediction value, it includes: The risk prediction value F is compared with a first preset risk prediction value F1 and a second preset risk prediction value F2 respectively, F1<F2, and the statistical scope of the geographical area is determined according to the comparison result; When F≤F1, determining the geographic area statistical range to be a first preset range; When F1<F≤F2, determining the geographic area statistical range as a second preset range; When F2<F, determining the geographic area statistical range as a third preset range; The first preset range indicates that the statistical area is smaller than the second preset range, and the second preset range indicates that the statistical area is smaller than the third preset range.

4. The method for compiling a high consequence area risk assessment report according to claim 3, characterized in that: When judging whether to adjust the risk prediction value according to the statistical data within the statistical range of the geographical area, it includes: Collecting population density data R within the statistical range of the geographic area, comparing the population density data R with a population density threshold Rmax, and determining whether to adjust the risk prediction value according to the comparison result; When R>Rmax, it is determined that the risk prediction value is adjusted to obtain an adjusted risk prediction value, and the adjusted risk prediction value is used as the final risk prediction value; When R≤Rmax, it is determined that the risk prediction value is not to be adjusted, and the risk prediction value is used as the final risk prediction value.

5. The method for compiling a high consequence area risk assessment report according to claim 4, characterized in that: When determining to adjust the risk prediction value, it includes: Obtain a density difference ΔR according to the population density data R and the population density threshold Rmax, ΔR=R-Rmax, compare the density difference ΔR with a first preset density difference ΔR1 and a second preset density difference ΔR2 respectively, ΔR1<ΔR2, and adjust the risk prediction value according to the comparison result; When ΔR≤ΔR1, a first preset adjustment coefficient A1 is determined to adjust the risk prediction value to obtain an adjusted risk prediction value F*A1; When ΔR1<ΔR≤ΔR2, determine a second preset adjustment coefficient A2 to adjust the risk prediction value to obtain an adjusted risk prediction value F*A2; When ΔR2<ΔR, determine a third preset adjustment coefficient A3 to adjust the risk prediction value to obtain an adjusted risk prediction value F*A3; Among them, 1<A1<A2<A3<1.

2.

6. The method for compiling a high consequence area risk assessment report according to claim 5, characterized in that: When determining the template level corresponding to the evaluation report writing template according to the final risk prediction value, it includes: When the final risk prediction value is in the first interval, the template level corresponding to the evaluation report writing template is determined to be the first template level; When the final risk prediction value is in the second interval, the template level corresponding to the evaluation report writing template is determined to be the second template level; When the final risk prediction value is in the third interval, the template level corresponding to the evaluation report writing template is determined to be the third template level; Among them, the first interval is that the final risk prediction value is less than or equal to the first preset risk prediction value F1, the first interval is that the final risk prediction value is greater than the first preset risk prediction value F1 and less than or equal to the second preset risk prediction value F2, and the third interval is that the final risk prediction value is greater than the second preset risk prediction value F2.

Images