Process industry total factor risk perception system

Abstract

The invention provides a process industry total factor risk perception system. The system comprises a data acquisition module, a data alignment module, a data fusion module, a feature extraction module, a risk perception module, an intelligent decision module, a data storage module, a communication subscription module and a user interaction and visualization module. According to the method, deep data alignment and fusion are carried out on multivariate heterogeneous data, risk features and rules between deep excavation data are extracted through features, a risk perception model with high precision and high robustness is researched and constructed, the inherent rules and trends of risk evolution are revealed, rapid capture and evaluation of potential risks are realized, and the risk perception model has a good development prospect. And a targeted risk decision scheme is given, and real-time and accurate perception and early warning of the total factor risk of the production operation of the process industry are realized. And meanwhile, based on the distributed micro-service system architecture design, each function is realized by an independent micro-service module, so that the flexibility and expandability of the system are improved.

Description

Technical Field

[0001] The present disclosure relates to the technical field of information processing and analysis, and in particular to a full-factor risk perception system for process industries. Background Art

[0002] Due to its unique industry characteristics, process industries (such as chemical, oil and gas, metallurgy, etc.) face many challenges in terms of scientific production and operation management and accuracy of risk analysis, resulting in a severe and complex production safety situation. Once a production safety accident occurs, the consequences are often very serious, not only causing property losses and casualties, but also causing long-term adverse effects on society. The process industry often involves high-risk environments such as high temperature, high pressure, flammable and explosive, which leads to complex and changeable risk factors in production and operation (especially the potential risks in personnel operation behavior), resulting in the difficulty of real-time updating and dynamic adjustment of multi-dimensional factors such as time, space, environment, and personnel behavior. The transformation of risk assessment results into actionable intelligent decision-making recommendations is a huge technical challenge. Therefore, how to build a comprehensive intelligent risk perception system to achieve the integration of risk monitoring, prediction, evaluation and decision-making is a key issue in improving the overall safety performance of the process industry system. Summary of the invention

[0003] The purpose of the embodiments of the present disclosure is to provide a full-factor risk perception system for the process industry to solve the problems existing in the prior art.

[0004] The embodiment of the present disclosure adopts the following technical scheme: a full-factor risk perception system for process industry, comprising: a data acquisition module, used to collect raw data from various data sources in real time, and to clean and denoise the raw data, wherein the raw data is multivariate heterogeneous data at various process nodes in the process industry; a data alignment module, used to align the cleaned and denoised raw data according to time series; a data fusion module, used to fuse the aligned data to form a comprehensive data set based on time series, and to update the comprehensive data set in real time; a feature extraction module, used to extract risk features from the comprehensive data set according to time nodes using signal processing and machine learning algorithms; a risk sensing module, used to collect raw data from various data sources in real time, and to clean and denoise the raw data ... fusion module, used to fusion the aligned data to form a comprehensive data set based on time series, and to update the comprehensive data set in real time; a risk sensing module, used to collect raw data from various data sources in real time, and to clean and denoise the raw data; a data fusion module, used to fusion the aligned data to form a comprehensive data set based on time series, and to update the comprehensive data set in real time; a risk sensing module, used to collect raw data from various data sources in real time, and to clean and denoise the raw data; a data fusion module, used to fusion the aligned data to form a comprehensive data set based on time series; a risk sensing module, used to fusion the raw data from various data sources in real time; a risk sensing module, used to collect raw data from various data sources in real time; a risk sensing module, used to collect raw data from various data sources in real time; a An awareness module is used to call the risk perception model to perform risk perception according to the risk characteristics; an intelligent decision-making module is used to generate a risk response strategy according to the risk perception results, and push the risk response strategy to the user; a data storage module is used to store the original data, the comprehensive data set, the risk characteristics, the risk perception results and the risk response strategy in the corresponding database, and update each database in real time; a communication subscription module is used to push the updated subscription data to the corresponding user when the database is updated according to the permissions and data subscription status of different users; and a user interaction and visualization module is used to provide an operation interface to the user and display the pushed data.

[0005] In some embodiments, the data alignment module includes at least: a time synchronization unit, which uses a time synchronization algorithm and an interpolation algorithm to unify the time reference of all the original data; a semantic synchronization unit, which is used to unify the units and formats of data of the same type in the original data; a frequency matching unit, which is used to resample the data in the original data whose sampling frequency does not meet the target sampling frequency according to the target sampling frequency to unify the sampling interval; and a compensation unit, which is used to compensate the original data according to external environmental parameters and the upstream and downstream relationships between each process node.

[0006] In some embodiments, the data fusion module includes at least: a data dimension reduction unit, used to simplify the data structure of the aligned data through a principal component analysis algorithm or a t-distribution random neighbor embedding algorithm; a fusion unit, used to input the reduced dimension data into a preset fusion algorithm to form a comprehensive data set based on a time series, and the preset fusion algorithm is at least any one of the following algorithms: weighted average algorithm, Bayesian network, and integrated learning.

[0007] In some embodiments, the feature extraction module includes at least: a key parameter screening unit, which is used to screen out key parameters from the comprehensive data set through any one of the analysis techniques of data correlation analysis, principal component analysis, and cluster analysis, combined with process risk theory; a feature extraction unit, which is used to identify the coupling relationship between the key parameters through any one of the algorithms of neural network, support vector machine, decision tree and random forest, and extract risk characteristics reflecting changes in risk status at each time node by learning from historical data of long-term operation of the process industry.

[0008] In some embodiments, the risk perception module includes at least: a knowledge base construction unit, which is used to construct a historical risk database based on expert knowledge and the historical data, and the historical risk database at least stores various risk failure information, risk propagation path information, root cause analysis information and risk handling information in the operation of the process industry; a model training unit, which is used to extract historical risk failure information and historical risk propagation paths from the historical risk database as output data, extract historical risk features as input data, train the risk perception model until the model converges, so as to obtain a risk perception model for risk perception; a model calling unit, which is used to call the risk perception model, input the risk features output by the feature extraction unit, and output the risk perception results.

[0009] In some embodiments, the intelligent decision-making module includes at least: a risk level determination unit, used to call the historical risk database and determine the risk level of the risk perception result according to the risk assessment method; a strategy output unit, used to call the historical risk database, output the risk response strategy according to the risk level, and push the risk response strategy to the user.

[0010] In some embodiments, the risk assessment method includes at least: risk matrix analysis, failure mode and effect analysis, and hazard and operability analysis.

[0011] In some embodiments, the user interaction and visualization module includes at least: a display unit, used to provide an operation interface to the user, and display the subscription data, risk perception results and risk response strategies pushed to the user; an adjustment unit, used to call the adjustment interface of the processing node according to the processing node involved in the risk response strategy, and receive the adjustment parameters entered by the user in the adjustment interface, and send them to the corresponding processing node.

[0012] In some embodiments, the user interaction and visualization module also includes: a knowledge network unit, which is used to call the historical risk database and build a risk knowledge network using knowledge graph technology; the display unit is also used to display the risk knowledge network through any one of the graphical methods of node link diagram, tree diagram, and heat map.

[0013] In some embodiments, the data source includes at least one of the following functional systems in the process industry: distributed control system, manufacturing execution system, equipment maintenance management system, inspection information system, environmental detection system, network security system, laboratory detection system, and personnel management system.

[0014] The beneficial effects of the disclosed embodiments are as follows: the risk perception system of the embodiment performs deep data alignment and fusion of multivariate heterogeneous data, and digs deep into the risk characteristics and laws between data through feature extraction, studies and constructs a risk perception model with high precision and high robustness, reveals the inherent laws and trends of risk evolution, realizes the rapid capture and quantitative assessment of potential risks, and gives targeted risk decision-making solutions, realizing real-time, accurate perception and early warning of all-factor risks in the production and operation of the process industry. At the same time, the risk perception system of the embodiment is based on the distributed microservice architecture design, and each function is implemented as an independent microservice module, which improves the flexibility and scalability of the system and further strengthens the safety protection capabilities in the process industry production and operation process. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate one or more embodiments of this specification or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this specification. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0016] Figure 1 It is a schematic diagram of the structure of the full-factor risk perception system for the process industry in the embodiment of the present disclosure;

[0017] Figure 2 It is a structural diagram of a data alignment module in an embodiment of the present disclosure;

[0018] Figure 3 This is a structural diagram of a data fusion module in an embodiment of the present disclosure;

[0019] Figure 4 is a structural diagram of a feature extraction module in an embodiment of the present disclosure;

[0020] Figure 5 This is a schematic diagram of the structure of the risk perception module in the embodiment of the present disclosure;

[0021] Figure 6 This is a schematic diagram of the structure of the intelligent decision-making module in the embodiment of the present disclosure;

[0022] Figure 7 It is a schematic diagram of the actual implementation of the risk assessment method and the response strategy in the embodiment of the present disclosure;

[0023] Figure 8 It is a structural diagram of a user interaction and visualization module in an embodiment of the present disclosure;

[0024] Fig. 9 Schematic diagram of a dynamic risk intelligent management and control platform for the process industry in an embodiment of the present disclosure. DETAILED DESCRIPTION

[0025] In order to enable those skilled in the art to better understand the technical solutions in one or more embodiments of this specification, the following will be combined with the drawings in one or more embodiments of this specification to clearly and completely describe the technical solutions in one or more embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this specification, not all of the embodiments. Based on one or more embodiments of this specification, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of this document.

[0026] Due to its unique industry characteristics, process industries (such as chemical, oil and gas, metallurgy, etc.) face many challenges in terms of scientific production and operation management and accuracy of risk analysis, resulting in a severe and complex production safety situation. Once a production safety accident occurs, the consequences are often very serious, not only causing property losses and casualties, but also causing long-term adverse effects on society. The process industry often involves high-risk environments such as high temperature, high pressure, flammable and explosive, which leads to complex and changeable risk factors in production and operation (especially the potential risks in personnel operation behavior), resulting in the difficulty of real-time updating and dynamic adjustment of multi-dimensional factors such as time, space, environment, and personnel behavior. The transformation of risk assessment results into actionable intelligent decision-making recommendations is a huge technical challenge. Therefore, how to build a comprehensive intelligent risk perception system to achieve the integration of risk monitoring, prediction, evaluation and decision-making is a key issue in improving the overall safety performance of the process industry system.

[0027] The embodiment of the present disclosure provides a process industry full-factor risk perception system, and its structural diagram is as follows: Figure 1As shown, it mainly includes a data acquisition module 10, a data alignment module 20, a data fusion module 30, a feature extraction module 40, a risk perception module 50, an intelligent decision module 60, a data storage module 70, a communication subscription module 80 and a user interaction and visualization module 90. Figures 2 to 9 The construction and functional implementation of the full-factor risk perception system for the process industry provided in this embodiment are described in detail.

[0028] The data acquisition module 10 is mainly used to collect raw data from various data sources in real time, and to clean and denoise the raw data. The raw data is multivariate heterogeneous data at each process node in the process industry. Specifically, the data source of this embodiment includes at least one of the following functional systems in the process industry: distributed control system, manufacturing execution system, equipment maintenance management system, inspection information system, environmental detection system, network security system, laboratory detection system, personnel management system, etc. The data source design can actually be carried out according to the process node configuration and system requirements of the specific industry, and this embodiment does not make specific restrictions. It can actually include but is not limited to the following data according to different process industries: real-time images from surveillance cameras, radar imaging data from material flow monitoring, maintenance logs and accident reports filled in by operators, general values ​​such as temperature and flow rate read by sensors, etc. The data acquisition module 10 can realize data communication with each data source through a data interface to obtain raw data in real time, and pre-process the raw data based on conventional data cleaning and denoising methods to maximize the effectiveness of the raw data.

[0029] Based on the multi-source heterogeneous characteristics of the original data, there is a lack of unified standards between various types of data, resulting in serious information islands and low efficiency in data integration and sharing. In addition, the differences in data formats of different modalities also increase the difficulty of integration. The data alignment module 20 of this embodiment is mainly used to align the cleaned and denoised original data according to the time series, so that the multivariate heterogeneous data has a unified standard, eliminates the inconsistency between the data, and facilitates comparison and correlation analysis on the same time axis. In actual implementation, the structural diagram of the data alignment module 20 of this embodiment is as follows Figure 2 As shown, it at least includes a time synchronization unit 21, a semantic synchronization unit 22, a frequency matching unit 23 and a compensation unit 24, wherein the time synchronization unit 21 adopts a time synchronization algorithm and an interpolation algorithm to unify the time reference of all original data; the semantic synchronization unit 22 is used to unify the units and formats of data of the same type in the original data; the frequency matching unit 23 is used to resample the data in the original data whose sampling frequency does not meet the target sampling frequency according to the target sampling frequency to unify the sampling interval; the compensation unit 24 is used to compensate the original data according to external environmental parameters and the upstream and downstream relationships between each process node.

[0030] Specifically, the time synchronization unit 21 adjusts the data sampled at different frequencies to the same sampling rate through interpolation or other time series analysis methods to ensure the temporal continuity of all data points, and can correct the time offset caused by the system clock difference through the Network Time Protocol (NTP) and other methods to ensure the uniformity of the time reference between the data. The semantic synchronization unit 22 can map the different terms used in different data sets for the same concept to the same standard expression by defining a unified ontology or vocabulary, and standardize different physical units, convert the same type of data into a unified unit and format to facilitate subsequent processing. The frequency matching unit 23 can use upsampling or downsampling technology to make data of different frequencies have the same sampling interval, so that they can be directly compared, reducing the complexity of data comparison, and for data records triggered by certain specific events, data can be collected by setting a fixed period to ensure that all related data are in the same time period. The compensation unit 24 can consider the impact of changes in actual environmental conditions (such as temperature, humidity, etc.) on sensor readings or other functional systems, align and make corresponding compensations, and at the same time combine the upstream and downstream relationships between each process node to uniformly compensate and align the affected data in the process.

[0031] In the actual implementation process, in addition to aligning the original data collected from each data source, the data alignment module 20 can also expand the data volume by interpolation or data migration for small sample data. This can not only ensure the consistency of the data volume between different time points in the time series, but also avoid the problems of model bias and insufficient generalization ability caused by small sample data during the subsequent risk assessment model training and use.

[0032] Data fusion is the process of combining data from multiple different sources, formats, or timestamps into a new comprehensive dataset to obtain a more comprehensive and accurate representation of information. Figure 3 FIG. 3 shows a schematic diagram of the structure of the data fusion module 30 in this embodiment, which is mainly used to fuse the aligned data to form a comprehensive data set based on time series, and to update the comprehensive data set in real time. Figure 3As shown, the data fusion module 30 of this embodiment includes at least: a data dimension reduction unit 31, which is used to simplify the data structure of the aligned data through a principal component analysis algorithm (PCA) or a t-distributed stochastic neighbor embedding algorithm (t-SNE), so as to reduce the data dimension while maintaining important information; a fusion unit 32, which is used to input the reduced-dimensional data into a preset fusion algorithm to form a comprehensive data set based on a time series, and the preset fusion algorithm is at least any one of the following algorithms: a weighted average algorithm, a Bayesian network, and an integrated learning.

[0033] In some embodiments, the data dimension reduction unit 31 can also select the most valuable data to reduce redundant information through domain knowledge and statistical analysis. The fusion unit 32 can combine the characteristics of the original data and select one of the preset fusion algorithms to extract and fuse the relationship between the data. In addition, the fusion unit 32 can further combine graph neural networks or deep learning, etc., and generate a model for realizing data fusion through learning historical data, and complete the fusion processing of the current real-time collected data through the trained model.

[0034] The full-factor data of production and operation in process industries contain the deep-seated laws and potential risks of system operation. However, these data not only span multiple levels such as physics, information, and management, but also have complex characteristics such as nonlinearity, dynamics, multimodality, and non-stationarity, which bring challenges to the full extraction of key process information contained in the data. The feature extraction module 40 of this embodiment extracts risk features from the comprehensive data set according to time nodes based on signal processing and machine learning algorithms to accurately characterize the key features of the system status and risk trends, and provide accurate input data support for subsequent risk assessment.

[0035] Figure 4 The structural schematic diagram of the feature extraction module 40 in this embodiment is shown, which mainly includes a key parameter screening unit 41 and a feature extraction unit 42, wherein the key parameter screening unit 41 is used to screen out key parameters from the comprehensive data set through any analysis technology of data correlation analysis, principal component analysis, and cluster analysis, combined with process risk theory. These key parameters are mainly parameters that have a greater impact on process risks. Through the screening of key parameters, the amount of data for subsequent processing can be effectively reduced, and the attention of subsequent models to risks can be improved; the feature extraction unit 42 identifies the coupling relationship between key parameters through any algorithm of neural networks, support vector machines, decision trees and random forests, and extracts risk characteristics reflecting changes in risk status at each time node by learning historical data of long-term operation of the process industry.

[0036] The risk features extracted in this embodiment mainly refer to features that are sensitive, representative, stable, relevant, dynamic, and explainable for risk identification or prediction. Among them, sensitivity means that the risk feature is highly sensitive to changes in the system state (especially abnormal situations), and can detect potential risks at an early stage, ensuring that the model can respond to small changes in a timely manner, thereby achieving the ability to quickly capture risks; representativeness refers to the main factors that affect the behavior of the system, providing a complete perspective to understand the system operation mechanism and ensure that the model can accurately assess risks; stability refers to the characteristics that are relatively stable under normal operating conditions, but have significant changes when risk events occur, so as to enhance the robustness of the model, avoid false positives or negative positives, and improve the reliability of prediction results; correlation means that there is a clear correlation between the risk feature and other features, which can reveal complex causal chains to help the model identify the inherent connection between different features and support deeper risk analysis; dynamicity means that the risk feature has the ability to capture trends that evolve over time and reflect the development process of risks, so that the model not only pays attention to the current state, but also predicts future risk evolution trends and provides a basis for preventive measures; interpretability means that the meaning of the feature is clear and easy to understand, easy to explain and verify, thereby increasing the transparency of the model, allowing users to understand and trust the results of the model output and take appropriate actions.

[0037] Furthermore, this embodiment constructs a risk perception model, and uses the risk perception module 50 to call it to perceive the currently extracted risk features. The output risk perception results mainly characterize whether there are risks or will be risks in each node of the current process industry, so as to serve as data support for subsequent intelligent decision-making. Specifically, the risk perception module 50 mainly includes a knowledge base construction unit 51, a model training unit 52 and a model calling unit 53. Figure 5 As shown, the knowledge base construction unit 51 is used to construct a historical risk database based on expert knowledge and historical data. The historical risk database at least stores various risk failure information, risk propagation path information, root cause analysis information and risk handling information in the operation of the process industry; the model training unit 52 is used to extract historical risk failure information and historical risk propagation paths from the historical risk database as output data, extract historical risk features as input data, train the risk perception model until the model converges, so as to obtain a risk perception model for risk perception; the model calling unit 53 is used to call the risk perception model, input the risk features output by the feature extraction unit, and output the risk perception results.

[0038] During the operation of the process industry, a large amount of historical data will be generated, including data during normal operation and data when risk failures occur. The knowledge base construction unit 51 builds a comprehensive historical risk database based on the above historical data combined with expert knowledge, covering key data information such as various types of fault information, fault propagation path tracing, and root cause analysis. On this basis, the historical risk database can be called and the input and output data related to risk perception in the database can be extracted to realize the training of the risk perception model, so that the trained risk perception model has risk monitoring and prediction functions, and can accurately analyze the risk state of the current system according to the input risk characteristics and make predictions on whether risks will occur in a specific time period in the future. The risk perception model of this embodiment can be implemented based on a supervised learning model (such as a support vector machine, a random forest, a gradient boosting machine, etc.), or based on a deep learning model (such as a convolutional neural network, a recurrent neural network and its variants, etc.), or based on a combination of different models for training, and this embodiment does not make specific restrictions. For the risk characteristics obtained after real-time acquisition and processing, the model calling unit 53 can call the trained risk perception model to process them and obtain real-time risk perception results.

[0039] The training and calling of the risk perception model reveals the inherent laws and trends of risk evolution and enables the rapid capture and assessment of potential risks. The intelligent decision-making module 60 can generate corresponding risk response strategies based on the risk perception results and push the risk response strategies to users to achieve rapid risk response or avoidance of possible risks. Figure 6 A structural schematic diagram of the intelligent decision-making module 60 is shown, which at least includes: a risk level determination unit 61, which is mainly used to call the historical risk database and determine the risk level of the risk perception result according to the risk assessment method; a strategy output unit 62, which is mainly used to call the historical risk database, output the risk response strategy according to the risk level, and push the risk response strategy to the user.

[0040] Specifically, the risk assessment method adopted in this embodiment includes but is not limited to risk matrix analysis (RF), failure mode and effects analysis (FMEA), and hazard and operability analysis (HAZOP). Among them, the risk matrix analysis can divide the risk perception results into different risk levels according to the two dimensions of possibility and severity, and can be used to quickly locate high-risk data and nodes; the failure mode and effects analysis is used to identify potential failure modes and their causes, and comprehensively consider factors such as occurrence frequency, severity and detection difficulty, quantify the risk level of each failure mode, and form a risk priority number (RPN) calculation; the hazard and operability analysis decomposes the entire production process into several operation nodes, conducts risk assessment one by one, and for each node, identifies situations that may deviate from normal operation and analyzes their consequences; finally, based on the output results of the above three analysis methods, a risk level output is formed. It should be noted that the correspondence between the final division method of the risk level and the analysis result can be set according to the actual process industry needs and user needs, and this embodiment does not limit its specific limitations.

[0041] When the strategy output unit 62 outputs the risk response strategy, it can combine the risk level and the output results of the above three analysis methods, cooperate with the expert knowledge and fault tracing analysis of fault history data and the fault forward-looking analysis based on OTS and other strategies to output the risk response strategy suitable for the current execution, so as to respond to the possible risks or avoid the possible risks in the future, such as Figure 7 The diagram shows the actual implementation of the risk assessment method and response strategy.

[0042] Corresponding to the various types of data information output by the above modules, the risk perception system of this embodiment stores them in the corresponding database through the data storage module 70. Specifically, the above stored information includes at least: original data, comprehensive data sets, risk characteristics, risk perception results and risk response strategies, etc. Each type of data is stored in the database described therein, and the corresponding database is updated when new data is generated. The communication subscription module 80 can communicate with any module in the system or realize communication between any modules to realize data transmission between modules. In addition, according to the permissions of different users and data subscription conditions, the communication subscription module 80 pushes the updated subscription data to the corresponding users when the database is updated for users to review, especially the risk perception results and risk response strategies, which need to be pushed to users in time for attention and processing.

[0043] Furthermore, the system of this embodiment also includes a user interaction and visualization module 90, which is used to provide an operation interface to the user and display the pushed data. Specifically, the user interaction and visualization module 90 at least includes a display unit 91 for providing an operation interface to the user and displaying the subscription data, risk perception results and risk response strategies pushed to the user, and an adjustment unit 92 for calling the adjustment interface of the processing node according to the processing node involved in the risk response strategy, receiving the adjustment parameters input by the user in the adjustment interface, and sending them to the corresponding processing node, such as Figure 8 As shown. In addition, in some embodiments, the user interaction and visualization module 90 may also include a knowledge network unit 93, which constructs a risk knowledge network by calling a historical risk database and using knowledge graph technology. The display unit 91 may display the risk knowledge network through any graphical method such as a node link diagram, a tree diagram, or a heat map, thereby realizing visual display and intelligent retrieval of risk information, providing data support for intelligent risk assessment and decision-making, integrating an intelligent decision support system with risk matrix, failure mode and effect analysis, and hazard and operability analysis tools, realizing integrated management of risk monitoring, prediction, assessment, and decision-making, automatically performing risk analysis based on real-time production data, and providing targeted risk decisions and optimization solutions.

[0044] In fact, in the risk perception system constructed by this embodiment, each functional module is an independent microservice module. The data storage module 70 can be used as a data transfer center, combined with the data push service provided by the communication subscription module 80, to realize the risk perception system based on the distributed microservice system. The data storage module 70 is used as the data middle platform, relying on the efficient distributed message queue to provide convenient data subscription and push functions. Based on the risk perception system of this embodiment, a dynamic risk intelligent management and control platform for the process industry can be further constructed, such as Fig. 9As shown in the figure, the dynamic data generated by the production process of the process industry often has very complex structures and situations such as multi-source, heterogeneous, heterogeneous, multi-modal, nonlinear, and non-stationary. The full-factor data acquisition module, data alignment module, and data fusion module built by the platform can perform pre-processing such as cleaning, correction, alignment, and fusion on the collected complex data to form standardized data that is easy to manage and use later, and then store it in the platform's database; the data storage module relies on the distributed message queue of the communication subscription module to provide convenient data subscription and push functions; the platform is supported by the data middle platform and realizes a multi-security collaborative protection architecture based on a distributed microservice system, which can further distribute personnel safety, process monitoring, equipment monitoring, environmental monitoring, and network monitoring to relatively independent microservices for subsequent modeling and analysis, and at the same time establish a risk collaborative microservice to exchange information, cross-validate, and coordinate management with the above five microservices; each microservice subscribes to relevant data from the data middle platform according to its own needs, realizes data-driven analysis logic, and interacts with the verification data set, security rule set, and algorithm tool set to jointly realize dynamic risk intelligent management and control of the process industry.

[0045] The risk perception system of this embodiment performs deep data alignment and fusion of multivariate heterogeneous data, and deeply explores the risk characteristics and laws between data through feature extraction, studies and constructs a risk perception model with high precision and high robustness, reveals the inherent laws and trends of risk evolution, realizes the rapid capture and quantitative assessment of potential risks, and provides targeted risk decision-making solutions, realizing real-time, accurate perception and early warning of all-factor risks in the production and operation of the process industry. At the same time, the risk perception system of this embodiment is based on the distributed microservice architecture design, and each function is implemented as an independent microservice module, which improves the flexibility and scalability of the system and further strengthens the safety protection capabilities in the process industry production and operation process.

[0046] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present disclosure, rather than to limit them. Although the present disclosure has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein by equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present disclosure.

Claims

1. A full-factor risk perception system for process industries, characterized in that: include: The data acquisition module is used to collect raw data from various data sources in real time and clean and denoise the raw data. The raw data is multivariate heterogeneous data at various process nodes in the process industry; The data alignment module is used to align the cleaned and denoised raw data according to the time series; A data fusion module is used to fuse the aligned data to form a comprehensive data set based on time series, and to update the comprehensive data set in real time; A feature extraction module, used to extract risk features from the comprehensive data set according to time nodes using signal processing and machine learning algorithms; A risk perception module, used to call a risk perception model to perform risk perception according to the risk characteristics; An intelligent decision-making module is used to generate risk response strategies based on risk perception results and push the risk response strategies to users; A data storage module, used to store the original data, the comprehensive data set, the risk characteristics, the risk perception results and the risk response strategy in corresponding databases, and update each database in real time; A communication subscription module, for pushing updated subscription data to corresponding users when the database is updated according to the permissions and data subscription conditions of different users; and The user interaction and visualization module is used to provide users with an operation interface and display pushed data.

2. The process industry full-factor risk perception system according to claim 1 is characterized in that: The data alignment module at least includes: A time synchronization unit, using a time synchronization algorithm and an interpolation algorithm to unify the time base of all the raw data; A semantic synchronization unit, used to unify the units and formats of the same type of data in the original data; A frequency matching unit, configured to resample the data in the original data whose sampling frequency does not conform to the target sampling frequency according to the target sampling frequency, so as to unify the sampling interval; The compensation unit is used to compensate the original data according to external environmental parameters and the upstream and downstream relationships between various process nodes.

3. The process industry full-factor risk perception system according to claim 2 is characterized in that: The data fusion module at least includes: A data dimension reduction unit, used to simplify the data structure of the aligned data by using a principal component analysis algorithm or a t-distribution random neighbor embedding algorithm; The fusion unit is used to input the dimension-reduced data into a preset fusion algorithm to form a comprehensive data set based on a time series, and the preset fusion algorithm is at least any one of the following algorithms: weighted average algorithm, Bayesian network, and integrated learning.

4. The process industry full-factor risk perception system according to claim 3 is characterized in that: The feature extraction module at least includes: A key parameter screening unit, used to screen out key parameters from the comprehensive data set by using any one of data correlation analysis, principal component analysis, and cluster analysis combined with process risk theory; The feature extraction unit is used to identify the coupling relationship between the key parameters through any one of the algorithms including neural network, support vector machine, decision tree and random forest, and extract the risk characteristics reflecting the changes in risk status at each time node by learning the historical data of long-term operation of the process industry.

5. The process industry full-factor risk perception system according to claim 4 is characterized in that: The risk perception module at least includes: A knowledge base construction unit, used to construct a historical risk database based on expert knowledge and the historical data, wherein the historical risk database at least stores various risk failure information, risk propagation path information, root cause analysis information and risk handling information in the operation of the process industry; A model training unit, used to extract historical risk failure information and historical risk propagation paths from the historical risk database as output data, extract historical risk features as input data, and train the risk perception model until the model converges to obtain a risk perception model for risk perception; The model calling unit is used to call the risk perception model, input the risk features output by the feature extraction unit, and output the risk perception results.

6. The process industry full-factor risk perception system according to claim 5 is characterized in that: The intelligent decision-making module at least includes: A risk level determination unit, used to call the historical risk database and determine the risk level of the risk perception result according to a risk assessment method; The strategy output unit is used to call the historical risk database, output the risk response strategy according to the risk level, and push the risk response strategy to the user.

7. The process industry full-factor risk perception system according to claim 6 is characterized in that: The risk assessment method at least includes: risk matrix analysis, failure mode and effect analysis, and hazard and operability analysis.

8. The process industry full-factor risk perception system according to claim 6 is characterized in that: The user interaction and visualization module at least includes: The display unit is used to provide an operation interface to the user and display the subscription data, risk perception results and risk response strategies pushed to the user; The adjustment unit is used to call the adjustment interface of the processing node according to the processing node involved in the risk response strategy, receive the adjustment parameters input by the user in the adjustment interface, and send them to the corresponding processing node.

9. The process industry full-factor risk perception system according to claim 8 is characterized in that: The user interaction and visualization module also includes: A knowledge network unit, used to call the historical risk database and construct a risk knowledge network using knowledge graph technology; The display unit is also used to display the risk knowledge network through any one of the graphical representation methods such as node link diagram, tree diagram and heat map.

10. The process industry full-factor risk perception system according to any one of claims 1 to 9, characterized in that: The data source includes at least one of the following functional systems in the process industry: Distributed control systems, manufacturing execution systems, equipment maintenance management systems, inspection information systems, environmental monitoring systems, network security systems, laboratory testing systems, and personnel management systems.

Images