CN117474343A - Petrochemical harbor danger source safety risk early warning method, petrochemical harbor danger source safety risk early warning device, petrochemical harbor danger source safety risk early warning equipment and storage medium - Google Patents
Petrochemical harbor danger source safety risk early warning method, petrochemical harbor danger source safety risk early warning device, petrochemical harbor danger source safety risk early warning equipment and storage medium Download PDFInfo
- Publication number
- CN117474343A CN117474343A CN202311810090.6A CN202311810090A CN117474343A CN 117474343 A CN117474343 A CN 117474343A CN 202311810090 A CN202311810090 A CN 202311810090A CN 117474343 A CN117474343 A CN 117474343A
- Authority
- CN
- China
- Prior art keywords
- early warning
- petrochemical
- data
- harbor
- neural network
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 47
- 238000003860 storage Methods 0.000 title claims abstract description 14
- 238000012549 training Methods 0.000 claims abstract description 60
- 238000012544 monitoring process Methods 0.000 claims abstract description 58
- 238000003062 neural network model Methods 0.000 claims abstract description 45
- 125000004122 cyclic group Chemical group 0.000 claims abstract description 34
- 230000008569 process Effects 0.000 claims abstract description 10
- 239000007788 liquid Substances 0.000 claims description 16
- 238000002790 cross-validation Methods 0.000 claims description 15
- 238000004519 manufacturing process Methods 0.000 claims description 12
- 238000012545 processing Methods 0.000 claims description 9
- 238000001514 detection method Methods 0.000 claims description 6
- 238000010200 validation analysis Methods 0.000 claims description 5
- 238000012795 verification Methods 0.000 claims description 5
- 230000006870 function Effects 0.000 claims description 4
- 238000011156 evaluation Methods 0.000 claims description 3
- 238000000605 extraction Methods 0.000 claims description 3
- 238000005457 optimization Methods 0.000 claims description 3
- 238000005286 illumination Methods 0.000 claims description 2
- 238000012502 risk assessment Methods 0.000 description 8
- 230000000306 recurrent effect Effects 0.000 description 7
- 238000010606 normalization Methods 0.000 description 5
- 230000002159 abnormal effect Effects 0.000 description 4
- 238000013528 artificial neural network Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 2
- 239000012530 fluid Substances 0.000 description 2
- 238000007689 inspection Methods 0.000 description 2
- 238000002372 labelling Methods 0.000 description 2
- 230000007246 mechanism Effects 0.000 description 2
- 238000013058 risk prediction model Methods 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 230000009471 action Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000003542 behavioural effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000012217 deletion Methods 0.000 description 1
- 230000037430 deletion Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000004880 explosion Methods 0.000 description 1
- 239000000383 hazardous chemical Substances 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000013021 overheating Methods 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 230000000717 retained effect Effects 0.000 description 1
- 238000012954 risk control Methods 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0635—Risk analysis of enterprise or organisation activities
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01D—MEASURING NOT SPECIALLY ADAPTED FOR A SPECIFIC VARIABLE; ARRANGEMENTS FOR MEASURING TWO OR MORE VARIABLES NOT COVERED IN A SINGLE OTHER SUBCLASS; TARIFF METERING APPARATUS; MEASURING OR TESTING NOT OTHERWISE PROVIDED FOR
- G01D21/00—Measuring or testing not otherwise provided for
- G01D21/02—Measuring two or more variables by means not covered by a single other subclass
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/02—Agriculture; Fishing; Forestry; Mining
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Business, Economics & Management (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Human Resources & Organizations (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Economics (AREA)
- Artificial Intelligence (AREA)
- Strategic Management (AREA)
- General Health & Medical Sciences (AREA)
- General Business, Economics & Management (AREA)
- Marketing (AREA)
- Tourism & Hospitality (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Entrepreneurship & Innovation (AREA)
- Health & Medical Sciences (AREA)
- Operations Research (AREA)
- Primary Health Care (AREA)
- Mining & Mineral Resources (AREA)
- Marine Sciences & Fisheries (AREA)
- Animal Husbandry (AREA)
- Agronomy & Crop Science (AREA)
- Quality & Reliability (AREA)
- Game Theory and Decision Science (AREA)
- Educational Administration (AREA)
- Development Economics (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
Abstract
The application provides a petrochemical harbor dangerous source safety risk early warning method, device, equipment and storage medium, and belongs to the technical field of petrochemical harbor safety. The method comprises the following steps: monitoring dangerous sources in real time, calling historical accident data, extracting features, setting early warning standards, establishing an early warning sample library, training a circulating neural network model by using the monitoring data and the historical accident data, verifying the performance of the model by K-fold cross, and finally carrying out risk early warning based on the trained model. The method combines real-time monitoring and historical data, processes time sequence information by using a cyclic neural network model, can effectively reduce early warning difficulty and realize timely risk early warning.
Description
Technical Field
The application relates to the field of petrochemical harbor safety, in particular to a petrochemical harbor hazard source safety risk early warning method, device and equipment and a storage medium.
Background
Petrochemical harbors are involved in numerous dangerous goods and high-risk operations, so that personnel safety risk management and control situation is extremely serious, and the petrochemical harbors are an important problem to be solved urgently. As petrochemical storage and transportation enterprises accumulate in coastal areas along the river, the risk of regional harbor significant sources of danger continues to rise. However, the data related to the petrochemical harbor district is complex and complex, including monitoring data, historical accident data and the like, so that the early warning difficulty aiming at important dangerous sources is high, and risks are easy to occur.
Disclosure of Invention
The application aims to overcome the defects in the prior art and provide a petrochemical harbor danger source safety risk early warning method, device, equipment and storage medium.
The application provides a petrochemical harbor district dangerous source safety risk early warning method, which comprises the following steps:
monitoring data including the liquid level of the oil tank, the pressure and the temperature of the pipeline, and the type and the concentration of the combustible gas on a dangerous source of the petrochemical harbor area in real time;
invoking historical accident data of a data center, and extracting dangerous source characteristics in the historical accident data;
presetting a grading early warning standard according to the dangerous source characteristics, marking samples of the historical accident data, and establishing a petrochemical harbor safety production risk classification grading early warning sample library;
forming a model training set by the monitoring data and the historical accident data in the early warning sample library, selecting a cyclic neural network model for training, and learning the representation and prediction of time sequence information;
evaluating the performance of the trained cyclic neural network model through K-fold cross validation;
and based on the evaluated cyclic neural network model, early warning is carried out on the petrochemical harbor dangerous source safety risk.
Optionally, early warning is carried out to petrochemical harbor district dangerous source safety risk, including:
and comparing the output result of the evaluated cyclic neural network model with a preset early warning threshold value, and carrying out early warning on sound, illumination, short messages and/or mails according to the comparison result.
Optionally, the method further comprises:
and displaying a dashboard, a chart and/or a map for visualizing the comparison result and the early warning information.
Optionally, before the monitoring data and the historical accident data in the early warning sample library are formed into a model training set, the method includes:
performing a deduplication operation according to specific fields and identifiers in the monitoring data and the historical accident data;
selectively deleting records containing missing values according to the proportion of the missing values of the monitoring data and the historical accident data;
selectively deleting outliers based on the sources of outliers in the monitored data and the historical incident data;
normalizing the monitoring data and the historical accident data, wherein the formula is as follows:
where Z represents normalized monitoring data, x represents raw monitoring data,representing the minimum value of the original monitoring data, +.>Representing the maximum value of the raw monitoring data.
Optionally, the recurrent neural network model includes:
an input layer, configured to process and convert data in the training set, where an expression of the processing and converting is as follows:
wherein the method comprises the steps ofRepresenting normalized data, +.>Representing raw training set data, +.>Mean value representing the original training set data, +.>Representing the standard deviation of the original training set data.
Optionally, the recurrent neural network model includes:
and the full connection layer is used for carrying out gradient descent processing on the data in the training set, and the expression is as follows:
wherein the method comprises the steps ofRepresentation model optimization weight parameters,/->Output representing current loop layer, +.>Indicates learning rate (I/O)>Representing the gradient of the loss function for the output w of the recycle layer.
Optionally, evaluating the trained recurrent neural network model performance by K-fold cross-validation includes:
dividing the training set into K subsets with equal size by using K-fold cross validation, selecting one subset as a validation set each time, and using the rest K-1 subsets as training sets;
and sequentially using different subsets as verification sets, performing K times of training and verification, and finally obtaining the average value of K model performance evaluation results for verifying and evaluating the model data performance.
The application also provides a petrochemical industry harbor district danger source safety risk early warning device, include:
the detection module is used for monitoring data comprising the liquid level of the oil tank, the pressure and the temperature of the pipeline and the type and the concentration of the combustible gas on a dangerous source of the petrochemical harbor area in real time;
the extraction module is used for calling historical accident data of the data center and extracting dangerous source characteristics in the historical accident data;
the marking module is used for presetting a grading early warning standard according to the dangerous source characteristics, marking samples of the historical accident data and establishing a grading early warning sample library for petrochemical harbor safety production risk classification;
the training module is used for forming a model training set by the monitoring data and the historical accident data in the early warning sample library, selecting a circulating neural network model for training, and learning the representation and prediction of time sequence information;
the estimation module is used for estimating the performance of the trained cyclic neural network model through K-fold cross validation;
and the early warning module is used for carrying out early warning on the safety risk of the petrochemical harbor dangerous source based on the evaluated cyclic neural network model.
The application also provides a petrochemical industry harbor district danger source safety risk early warning equipment, include:
the storage is used for storing a computer executable program of the petrochemical harbor hazard source safety risk early warning method;
a processor for retrieving the computer executable program from the memory and executing: monitoring data including the liquid level of the oil tank, the pressure and the temperature of the pipeline, and the type and the concentration of the combustible gas on a dangerous source of the petrochemical harbor area in real time; invoking historical accident data of a data center, and extracting dangerous source characteristics in the historical accident data; presetting a grading early warning standard according to the dangerous source characteristics, marking samples of the historical accident data, and establishing a petrochemical harbor safety production risk classification grading early warning sample library; forming a model training set by the monitoring data and the historical accident data in the early warning sample library, selecting a cyclic neural network model for training, and learning the representation and prediction of time sequence information; evaluating the performance of the trained cyclic neural network model through K-fold cross validation; and based on the evaluated cyclic neural network model, early warning is carried out on the petrochemical harbor dangerous source safety risk.
The application also provides a storage medium, which stores a computer executable program, wherein the computer executable program is used for being called by a processor to execute the steps of the petrochemical harbor hazard source safety risk early warning method.
The application has the advantages and beneficial effects that:
the application provides a petrochemical harbor district dangerous source safety risk early warning method, which comprises the following steps: monitoring data including the liquid level of the oil tank, the pressure and the temperature of the pipeline, and the type and the concentration of the combustible gas on a dangerous source of the petrochemical harbor area in real time; invoking historical accident data of a data center, and extracting dangerous source characteristics in the historical accident data; presetting a grading early warning standard according to the dangerous source characteristics, marking samples of the historical accident data, and establishing a petrochemical harbor safety production risk classification grading early warning sample library; forming a model training set by the monitoring data and the historical accident data in the early warning sample library, selecting a cyclic neural network model for training, and learning the representation and prediction of time sequence information; evaluating the performance of the trained cyclic neural network model through K-fold cross validation; and based on the evaluated cyclic neural network model, early warning is carried out on the petrochemical harbor dangerous source safety risk. According to the method and the device, the circulating neural network model is trained to perform early warning based on the detection data and the historical accident data, so that risk early warning can be performed in time before the real-time monitoring data, and the early warning difficulty is reduced.
Drawings
FIG. 1 is a schematic diagram of a petrochemical harbor hazard source safety risk early warning process in the present application;
fig. 2 is a schematic diagram of a petrochemical harbor danger source safety risk early warning device in the present application.
Detailed Description
The present application is further described in conjunction with the drawings and detailed embodiments so that those skilled in the art may better understand the present application and practice it.
The following are examples of specific implementation provided for the purpose of illustrating the technical solutions to be protected in this application in detail, but this application may also be implemented in other ways than described herein, and one skilled in the art may implement this application by using different technical means under the guidance of the conception of this application, so this application is not limited by the following specific embodiments.
The application provides a petrochemical harbor district dangerous source safety risk early warning method, which comprises the following steps: monitoring data including the liquid level of the oil tank, the pressure and the temperature of the pipeline, and the type and the concentration of the combustible gas on a dangerous source of the petrochemical harbor area in real time; invoking historical accident data of a data center, and extracting dangerous source characteristics in the historical accident data; presetting a grading early warning standard according to the dangerous source characteristics, marking samples of the historical accident data, and establishing a petrochemical harbor safety production risk classification grading early warning sample library; forming a model training set by the monitoring data and the historical accident data in the early warning sample library, selecting a cyclic neural network model for training, and learning the representation and prediction of time sequence information; evaluating the performance of the trained cyclic neural network model through K-fold cross validation; and based on the evaluated cyclic neural network model, early warning is carried out on the petrochemical harbor dangerous source safety risk. According to the method and the device, the circulating neural network model is trained to perform early warning based on the detection data and the historical accident data, so that risk early warning can be performed in time before the real-time monitoring data, and the early warning difficulty is reduced.
Referring to fig. 1, the method for early warning the security risk of the dangerous source in the petrochemical harbor area includes the following steps:
s101, monitoring data comprising the liquid level of an oil tank, the pressure and the temperature of a pipeline and the type and the concentration of combustible gas on a dangerous source of a petrochemical harbor in real time.
In major sources of risk in petrochemical harbors, various monitoring sensors are installed in order to ensure safety and monitor potential risks in real time. The following are details of the sensors employed in the present application:
a level sensor for monitoring the real-time level of liquid in an oil tank or other liquid storage container. The sensor can acquire liquid level data of the oil tank or other liquid containers in real time so as to know the liquid amount in the containers. Monitoring the level of the tank is critical to preventing spillage, leakage or other related accidents.
Pressure sensors monitor the real-time pressure of various lines, valves, or other fluid transfer systems within the petrochemical dock. The sensor is capable of acquiring pressure data of the pipeline in real time. Abnormal pressure changes are indicative of leaks, blockages, or other system problems.
And the temperature sensor is used for monitoring the real-time temperature of the pipeline in the petrochemical harbor area. The sensor can acquire temperature data of the pipeline in real time. The change in temperature is indicative of an abnormal reaction of the fluid, frictional overheating, or other potential problems.
A gas sensor for monitoring the combustible gas of the leakage source. The sensor is capable of detecting and acquiring type and concentration data of the combustible gas. The leakage of combustible gas causes a fire or explosion, so it is important to detect and take measures in time.
In order to more effectively utilize the data acquired by the sensor, a cloud database interface can be connected. The interface has the main functions of:
historical incident data is obtained, including incident type, cause, and outcome information. Through analysis of historical accidents, common risks and potential problems of harbors can be known.
In the application, a database is created in a data center of a petrochemical harbor area and is used for recording real-time monitoring data acquired from sensors and historical accident data.
Further, preprocessing the detection data and the historical accident data includes:
the deduplication operation ensures that each record in the dataset is unique and has no duplicate information.
Specifically, the inspection is performed based on specific fields and identifiers in the monitoring data and the historical incident data. If two or more records are found to have exactly the same value in these fields, then duplicate records are considered. Duplicate records are deleted, only one of which is retained. This ensures the accuracy of the data and the reliability of the analysis.
The missing values are processed to check if there are some field value deletions in the dataset.
In particular, the dataset is checked piece by piece, and if some fields are found to be empty or unfilled, they are marked as missing values. How to process is determined according to the characteristics of the monitoring data and the historical accident data and the proportion of the missing values. If the missing values are less, the record containing the missing values can be selected to be deleted directly; if there are more missing values, other methods such as mean filling, interpolation, etc. are needed to process.
Abnormal values are processed to identify those in the dataset that deviate significantly from normal ranges, due to measurement errors or other reasons.
Specifically, the inspection is performed based on the characteristics of the monitoring data and the historical accident data, and the source of the outlier. For example, if the reading of a certain sensor suddenly jumps to a value that is not, then the value is an outlier. Abnormal values are selectively deleted or replaced with other suitable values. This needs to be judged according to the actual situation, because some outliers contain useful information.
Data normalization converts the values of the monitored data to a specified range, typically to facilitate subsequent data processing or model training. A min-max normalization method is used. The basic idea of this approach is to linearly transform the raw data to the range of [0,1 ].
The formula:
wherein;
wherein Z represents normalized monitoring data, x represents original monitoring data,representing the minimum value of the original monitoring data, +.>Representing the maximum value of the raw monitoring data.
After normalization, all the monitored data is mapped to the same scale, which helps some algorithms to process the data more efficiently.
S102, calling historical accident data of a data center, and extracting dangerous source features in the historical accident data.
First, historical accident data for petrochemical harbors is obtained from a data center.
The data includes information about the time, place, cause, influence factor, etc. of the accident.
And then analyzing and extracting main reasons of accident occurrence according to the importance weight of the information in the historical accident data. At the same time, various factors affecting the occurrence of the accident and the characteristics of the involved dangerous sources are identified.
And acquiring domain expert knowledge related to the petrochemical harbor district by using the data center. Such knowledge includes hazardous chemical storage methods, safety standards for equipment and facilities, risk points for process flows, and the like.
And (3) combining expert knowledge and historical accident data to arrange various dangerous source characteristics existing in the petrochemical harbor region.
The hazard characteristics are divided into four major categories: chemical storage characteristics, equipment and facility characteristics, process flow characteristics, and personnel behavioral characteristics.
And S103, presetting a grading early warning standard according to the dangerous source characteristics, marking samples of the historical accident data, and establishing a petrochemical harbor safety production risk classification grading early warning sample library.
And taking the sorted dangerous source characteristics as risk category indexes.
Setting three-level early warning standards for each risk category index:
primary early warning: representing a critical situation, meaning that the risk is imminent, and immediate emergency action is required.
Second-level early warning: representing a high risk state, representing a major hidden danger, and needing to take measures as soon as possible for precaution.
Three-stage early warning: representing a medium risk status, indicating that the current risk status requires further observation.
And marking the historical accident data according to the dangerous source characteristics, so as to ensure that each sample represents a specific risk category and grading early warning condition.
Such labeling helps the model to learn and identify the different risk patterns and pre-warning levels more accurately.
And establishing a classification and grading early warning sample library of the safety production risk of the petrochemical harbor area by using the marked samples.
The sample library can be used as a basis for training and testing a machine learning model, and helps the model to more accurately predict and identify various risks and early warning levels of petrochemical harbors.
S104, the monitoring data and the historical accident data in the early warning sample library form a model training set, a circulating neural network model is selected for training, and the representation and the prediction of time sequence information are learned.
The present application employs a Recurrent Neural Network (RNN) model as a risk prediction model.
Specifically, the monitoring data and the historical incident data are combined into a model training set.
This model dataset is divided into three parts: 70% as training set, 10% as validation set, 20% as test set.
A Recurrent Neural Network (RNN) was chosen as a training model, which is particularly suitable for processing sequence data, capable of capturing time dependencies in the data.
The recurrent neural network includes:
the data preprocessing-input layer is used for processing and converting the characteristics of the training set data by the input layer.
The data normalization method is adopted, and the formula is as follows:
；
wherein,representing normalized data, +.>Representing raw training set data, +.>Mean value representing the original training set data, +.>Representing the standard deviation of the original training set data.
The data normalization helps the model to learn the characteristics of the data more effectively, and can improve the stability and convergence speed of training.
And the loop layer is used for being responsible for processing the time sequence information of the sequence data in the training set.
The loop layer creates a time window to define time steps and combines training sets to form time series data.
Each time step receives an input and a hidden state during processing and outputs a hidden state and training results. This facilitates the model to learn the representation and prediction of the timing information.
And the full connection layer is used for mapping the output of the loop layer to the prediction target space.
Each node is connected to all nodes of the previous layer, each connected with a weight.
Training is performed by using a gradient descent algorithm, and weights are optimized. The gradient descent algorithm is formulated as:
；
wherein the method comprises the steps ofRepresentation model optimization weight parameters,/->Output representing current loop layer, +.>Indicates learning rate (I/O)>Representing the gradient of the loss function for the output w of the recycle layer.
By adding multiple fully connected layers, the non-linearity capability of the model can be increased.
And the output layer outputs a prediction result of the model according to the risk early warning.
S105, evaluating the trained performance of the circulating neural network model through K-fold cross validation.
In order to verify and evaluate the performance of the model, the method adopts a K-fold cross verification method.
The training set is divided into K equal-sized subsets. One subset at a time is selected as the validation set, and the remaining K-1 subsets are selected as training sets. This process is repeated K times, each time a different subset is selected as the validation set.
And finally, obtaining the average value of the K model performance evaluation results, wherein the average value is used for evaluating the data performance of the model, and can reflect the generalization capability of the model more comprehensively.
S106, early warning is carried out on the petrochemical harbor dangerous source safety risk based on the evaluated cyclic neural network model.
Firstly, according to the specific safety management requirements of the petrochemical harbor district, a proper risk assessment index is formulated. The index may be derived from actual experience.
The design of these metrics is based on the outcome of the risk prediction model, combined with the relevant safety specifications and standards.
The main risk assessment indexes comprise a safety index and a risk grade, and are used for intuitively reflecting the safety risk degree of the petrochemical harbor district.
After the risk assessment index is determined, a corresponding early warning index is designed. The warning index can be designed according to actual conditions.
And setting a specific threshold value for each early warning index according to the safety management requirements of the petrochemical harbor district.
When a certain risk assessment index exceeds an early warning threshold, the system triggers a corresponding early warning mechanism.
Upon triggering the early warning mechanism, the system generates an early warning signal corresponding to the risk assessment result and the early warning indicator.
These alert signals may be sent in a variety of ways including, but not limited to, sound, light, text messages, and mail to ensure that the relevant personnel are able to receive the alert in a timely manner.
The purpose of the early warning signal is to enable related personnel to know the risk condition at the first time and take emergency measures for risk control and emergency treatment.
In addition to sending the early warning signal, the system also visually displays the risk assessment result and the early warning signal.
The visual display mode can be an instrument panel, a chart or a map, and the like, so that the risk condition of the petrochemical harbor area can be known more intuitively.
And finally, the system also generates a corresponding report according to the risk assessment and early warning results.
These reports include risk assessment reports and early warning reports, which may provide references and decision bases for relevant personnel.
As shown in fig. 2, the present application further provides a petrochemical harbor district hazard source security risk early warning device, including:
a detection module 201 for monitoring data including the liquid level of the oil tank, the pressure and temperature of the pipeline, and the type and concentration of the combustible gas on a dangerous source of the petrochemical harbor in real time;
the extraction module 202 is used for calling historical accident data of the data center and extracting dangerous source features in the historical accident data;
the labeling module 203 is configured to preset a classification early warning standard according to the hazard characteristics, label the historical accident data with samples, and establish a petrochemical harbor safety production risk classification early warning sample library;
the training module 204 is configured to form a model training set from the monitoring data and the historical accident data in the early warning sample library, select a cyclic neural network model for training, and learn the representation and prediction of time sequence information;
an estimation module 205, configured to evaluate the performance of the trained recurrent neural network model through K-fold cross validation;
and the early warning module 206 is used for carrying out early warning on the petrochemical harbor danger source safety risk based on the evaluated cyclic neural network model.
The application also provides a petrochemical industry harbor district danger source safety risk early warning equipment, include:
the storage is used for storing a computer executable program of the petrochemical harbor hazard source safety risk early warning method;
a processor for retrieving the computer executable program from the memory and executing: monitoring data including the liquid level of the oil tank, the pressure and the temperature of the pipeline, and the type and the concentration of the combustible gas on a dangerous source of the petrochemical harbor area in real time; invoking historical accident data of a data center, and extracting dangerous source characteristics in the historical accident data; presetting a grading early warning standard according to the dangerous source characteristics, marking samples of the historical accident data, and establishing a petrochemical harbor safety production risk classification grading early warning sample library; forming a model training set by the monitoring data and the historical accident data in the early warning sample library, selecting a cyclic neural network model for training, and learning the representation and prediction of time sequence information; evaluating the performance of the trained cyclic neural network model through K-fold cross validation; and based on the evaluated cyclic neural network model, early warning is carried out on the petrochemical harbor dangerous source safety risk.
The application also provides a storage medium, which stores a computer executable program, wherein the computer executable program is used for being called by a processor to execute the steps of the petrochemical harbor hazard source safety risk early warning method.
Claims (10)
1. The petrochemical harbor danger source safety risk early warning method is characterized by comprising the following steps of:
monitoring data including the liquid level of the oil tank, the pressure and the temperature of the pipeline, and the type and the concentration of the combustible gas on a dangerous source of the petrochemical harbor area in real time;
invoking historical accident data of a data center, and extracting dangerous source characteristics in the historical accident data;
presetting a grading early warning standard according to the dangerous source characteristics, marking samples of the historical accident data, and establishing a petrochemical harbor safety production risk classification grading early warning sample library;
forming a model training set by the monitoring data and the historical accident data in the early warning sample library, selecting a cyclic neural network model for training, and learning the representation and prediction of time sequence information;
evaluating the performance of the trained cyclic neural network model through K-fold cross validation;
and based on the evaluated cyclic neural network model, early warning is carried out on the petrochemical harbor dangerous source safety risk.
2. The petrochemical harbor hazard safety risk early warning method according to claim 1, wherein the early warning of the petrochemical harbor hazard safety risk comprises:
and comparing the output result of the evaluated cyclic neural network model with a preset early warning threshold value, and carrying out early warning on sound, illumination, short messages and/or mails according to the comparison result.
3. The petrochemical harbor hazard source security risk early warning method according to claim 2, further comprising:
and displaying a dashboard, a chart and/or a map for visualizing the comparison result and the early warning information.
4. The petrochemical harbor hazard source security risk early warning method according to claim 1, wherein before the monitoring data and the historical accident data in the early warning sample library are formed into a model training set, comprising:
performing a deduplication operation according to specific fields and identifiers in the monitoring data and the historical accident data;
selectively deleting records containing missing values according to the proportion of the missing values of the monitoring data and the historical accident data;
selectively deleting outliers based on the sources of outliers in the monitored data and the historical incident data;
normalizing the monitoring data and the historical accident data, wherein the formula is as follows:
；
where Z represents normalized monitoring data, x represents raw monitoring data,representing the minimum value of the original monitoring data, +.>Representing the maximum value of the raw monitoring data.
5. The petrochemical harbor hazard source security risk early warning method according to claim 1, wherein the cyclic neural network model comprises:
an input layer, configured to process and convert data in the training set, where an expression of the processing and converting is as follows:
；
wherein the method comprises the steps ofRepresenting normalized data, +.>Representing raw training set data, +.>Mean value representing the original training set data, +.>Representing the standard deviation of the original training set data.
6. The petrochemical harbor hazard source security risk early warning method according to claim 1, wherein the cyclic neural network model comprises:
and the full connection layer is used for carrying out gradient descent processing on the data in the training set, and the expression is as follows:
；
wherein the method comprises the steps ofRepresentation model optimization weight parameters,/->Output representing current loop layer, +.>The learning rate is indicated as being indicative of the learning rate,representing the gradient of the loss function for the output w of the recycle layer.
7. The petrochemical harbor hazard source security risk early warning method according to claim 1, wherein evaluating the trained performance of the cyclic neural network model through K-fold cross-validation comprises:
dividing the training set into K subsets with equal size by using K-fold cross validation, selecting one subset as a validation set each time, and using the rest K-1 subsets as training sets;
and sequentially using different subsets as verification sets, performing K times of training and verification, and finally obtaining the average value of K model performance evaluation results for verifying and evaluating the model data performance.
8. Petrochemical industry harbor district danger source security risk early warning device, characterized in that includes:
the detection module is used for monitoring data comprising the liquid level of the oil tank, the pressure and the temperature of the pipeline and the type and the concentration of the combustible gas on a dangerous source of the petrochemical harbor area in real time;
the extraction module is used for calling historical accident data of the data center and extracting dangerous source characteristics in the historical accident data;
the marking module is used for presetting a grading early warning standard according to the dangerous source characteristics, marking samples of the historical accident data and establishing a grading early warning sample library for petrochemical harbor safety production risk classification;
the training module is used for forming a model training set by the monitoring data and the historical accident data in the early warning sample library, selecting a circulating neural network model for training, and learning the representation and prediction of time sequence information;
the estimation module is used for estimating the performance of the trained cyclic neural network model through K-fold cross validation;
and the early warning module is used for carrying out early warning on the safety risk of the petrochemical harbor dangerous source based on the evaluated cyclic neural network model.
9. Petrochemical industry harbor district danger source safety risk early warning equipment, characterized in that includes:
a memory for storing a computer executable program of the petrochemical harbor hazard source safety risk early warning method according to any one of claims 1 to 7;
a processor for retrieving the computer executable program from the memory and executing: monitoring data including the liquid level of the oil tank, the pressure and the temperature of the pipeline, and the type and the concentration of the combustible gas on a dangerous source of the petrochemical harbor area in real time; invoking historical accident data of a data center, and extracting dangerous source characteristics in the historical accident data; presetting a grading early warning standard according to the dangerous source characteristics, marking samples of the historical accident data, and establishing a petrochemical harbor safety production risk classification grading early warning sample library; forming a model training set by the monitoring data and the historical accident data in the early warning sample library, selecting a cyclic neural network model for training, and learning the representation and prediction of time sequence information; evaluating the performance of the trained cyclic neural network model through K-fold cross validation; and based on the evaluated cyclic neural network model, early warning is carried out on the petrochemical harbor dangerous source safety risk.
10. A storage medium storing a computer executable program for execution by a processor to perform the steps of the petrochemical harbor hazard source security risk warning method of any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311810090.6A CN117474343A (en) | 2023-12-27 | 2023-12-27 | Petrochemical harbor danger source safety risk early warning method, petrochemical harbor danger source safety risk early warning device, petrochemical harbor danger source safety risk early warning equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311810090.6A CN117474343A (en) | 2023-12-27 | 2023-12-27 | Petrochemical harbor danger source safety risk early warning method, petrochemical harbor danger source safety risk early warning device, petrochemical harbor danger source safety risk early warning equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117474343A true CN117474343A (en) | 2024-01-30 |
Family
ID=89636473
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311810090.6A Pending CN117474343A (en) | 2023-12-27 | 2023-12-27 | Petrochemical harbor danger source safety risk early warning method, petrochemical harbor danger source safety risk early warning device, petrochemical harbor danger source safety risk early warning equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117474343A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117786604A (en) * | 2024-02-23 | 2024-03-29 | 交通运输部水运科学研究所 | Fused safety toughness measurement analysis system based on harbor disaster data |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115511234A (en) * | 2021-06-22 | 2022-12-23 | 中国石油化工股份有限公司 | Grading early warning management method and system for dangerous chemical safety production risk |
| CN115600900A (en) * | 2022-10-28 | 2023-01-13 | 交通运输部水运科学研究所（Cn） | Safety risk assessment method, system and storage medium for petrochemical port area |
| CN116681292A (en) * | 2023-08-04 | 2023-09-01 | 交通运输部水运科学研究所 | Petrochemical harbor security risk analysis and responsibility division method based on deep learning |
| CN116720728A (en) * | 2023-04-26 | 2023-09-08 | 广州地铁设计研究院股份有限公司 | Risk assessment method, electronic device and storage medium |
| CN116757534A (en) * | 2023-06-15 | 2023-09-15 | 中国标准化研究院 | Intelligent refrigerator reliability analysis method based on neural training network |
| CN116882735A (en) * | 2023-06-28 | 2023-10-13 | 刘汉波 | Chemical risk correction early warning method based on process safety index |
-
2023
- 2023-12-27 CN CN202311810090.6A patent/CN117474343A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115511234A (en) * | 2021-06-22 | 2022-12-23 | 中国石油化工股份有限公司 | Grading early warning management method and system for dangerous chemical safety production risk |
| CN115600900A (en) * | 2022-10-28 | 2023-01-13 | 交通运输部水运科学研究所（Cn） | Safety risk assessment method, system and storage medium for petrochemical port area |
| CN116720728A (en) * | 2023-04-26 | 2023-09-08 | 广州地铁设计研究院股份有限公司 | Risk assessment method, electronic device and storage medium |
| CN116757534A (en) * | 2023-06-15 | 2023-09-15 | 中国标准化研究院 | Intelligent refrigerator reliability analysis method based on neural training network |
| CN116882735A (en) * | 2023-06-28 | 2023-10-13 | 刘汉波 | Chemical risk correction early warning method based on process safety index |
| CN116681292A (en) * | 2023-08-04 | 2023-09-01 | 交通运输部水运科学研究所 | Petrochemical harbor security risk analysis and responsibility division method based on deep learning |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117786604A (en) * | 2024-02-23 | 2024-03-29 | 交通运输部水运科学研究所 | Fused safety toughness measurement analysis system based on harbor disaster data |
| CN117786604B (en) * | 2024-02-23 | 2024-04-26 | 交通运输部水运科学研究所 | Fused safety toughness measurement analysis system based on harbor disaster data |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Romano et al. | Automated detection of pipe bursts and other events in water distribution systems | |
| CN113614359A (en) | Method and system for predicting risk of observable damage in wind turbine gearbox assembly | |
| CN117474343A (en) | Petrochemical harbor danger source safety risk early warning method, petrochemical harbor danger source safety risk early warning device, petrochemical harbor danger source safety risk early warning equipment and storage medium | |
| CN109186813A (en) | A kind of temperature sensor self-checking unit and method | |
| CN108508863A (en) | A kind of electromechanical equipment method for diagnosing faults based on gray model | |
| CN114841396B (en) | Method for predicting metamorphic trend and warning catastrophe risk in petrochemical production process | |
| Selvik et al. | On the use of criteria based on the SMART acronym to assess quality of performance indicators for safety management in process industries | |
| US20220405715A1 (en) | Inspection related systems and methods | |
| CN116994418B (en) | Pipeline safety early warning method and system | |
| CN110337640A (en) | Method and system for problem alert polymerization | |
| JP2011145978A (en) | Plant monitoring control system | |
| KR20180123581A (en) | Computerized fluid analysis to determine if an asset has a fluid issue | |
| Gómez-Camperos et al. | Diagnosis of horizontal pipe leaks using neural networks | |
| Ansaldi et al. | Near misses from the Seveso inspections: use of knowledge based methods for safety improvement | |
| EP2956751B1 (en) | Method and monitoring device for monitoring a structure | |
| Knowles et al. | Condition management of marine lube oil and the role of intelligent sensor systems in diagnostics | |
| de León Hijes et al. | Assessment of functional condition of equipment in industrial plants based on multiple measurements | |
| Yang et al. | Machine Learning in Process Safety and Asset Integrity Management | |
| CN117349724A (en) | Chemical storage tank pressure resistance testing method and system based on artificial intelligence | |
| CN117113259B (en) | Coal mine state data processing method and system for predicting potential safety hazards | |
| Gomez Camperos | Application Of Machine Learning Techniques For Leak Detection In Horizontal Pipelines. | |
| Lee et al. | Dynamic risk analysis from the perspective of life cycle approach in Iec 61508 and Iec 61511 | |
| Susarev et al. | Construction of system of extended objects diagnostic monitoring | |
| Hülsmann et al. | Classification of uncertain data: An application in nondestructive testing | |
| WO2024069217A1 (en) | Method for determining instruction for a corrosion monitoring location |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |