CN115659263B - Ship control behavior risk assessment system and method based on big data - Google Patents
Ship control behavior risk assessment system and method based on big data Download PDFInfo
- Publication number
- CN115659263B CN115659263B CN202211260297.6A CN202211260297A CN115659263B CN 115659263 B CN115659263 B CN 115659263B CN 202211260297 A CN202211260297 A CN 202211260297A CN 115659263 B CN115659263 B CN 115659263B
- Authority
- CN
- China
- Prior art keywords
- ship
- data
- navigation
- module
- control behavior
- 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.)
- Active
Links
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T90/00—Enabling technologies or technologies with a potential or indirect contribution to GHG emissions mitigation
Landscapes
- Traffic Control Systems (AREA)
Abstract
The invention discloses a ship control behavior risk assessment system and method based on big data, which are used for assessing ship control behaviors; based on the Internet of things and big data technology, ship azimuth data and cabin monitoring data in the whole process of ship starting, sailing and stopping are collected, a ship control rule knowledge base is fused by utilizing a machine learning algorithm to form a ship control behavior standard model base, a ship control behavior standard assessment result and an assessment report can be automatically and continuously output, informationized support is provided for monitoring and early warning of ship control behaviors, comprehensiveness, economy and timeliness are achieved, and timeliness and accuracy of ship management are improved.
Description
Technical Field
The invention relates to a ship control behavior risk assessment system and method based on big data, which belong to the technical field of ship navigation safety and are used for assessing ship control behaviors.
Background
The ship control behavior is very important for safe navigation of the ship, and some nonstandard control behaviors exist in actual navigation, so that potential safety hazards exist. The ship control behavior refers to various control behaviors of a shipman on a ship in the driving process, and the content comprises navigation starting, navigation stopping, forward and reverse switching and the like. In the actual sailing process, some nonstandard behaviors often exist, and the nonstandard ship control behaviors can damage ship equipment and cause potential safety hazards. For example, if the water temperature does not reach a preset value in the course of starting the ship, the ship starts to run, the lubrication condition of equipment is insufficient, the abrasion of main and auxiliary machine parts is possibly increased, the maintenance cost of the ship is increased if the water temperature is light, the equipment such as a main machine and an auxiliary machine is caused to fail if the water temperature is heavy, the ship navigation safety and the personnel safety on the ship are influenced, and the ship accident is caused. It can be seen whether the ship control behavior is standard or not, and is critical to the safe navigation of the ship. An important means for ensuring the regulation of the ship control behavior is to conduct regulation.
At present, the regulation means for the regulation of the ship operation behavior is single, the automation and data degree is not enough, and the comprehensiveness and timeliness are lacking. The traditional supervision mode is to check the sailing log and the turbine log, and remotely supervise the conditional ship through video monitoring equipment or check and judge whether the ship control behavior accords with the specification through manual patrol. However, because the navigation time of the ship is often long, the manual checking workload is large, the efficiency is low, the accuracy is not high, the influence of subjective judgment is large, various operation parameters of the ship in the navigation process cannot be associated with the video, and comprehensive and deep normative judgment and excavation are difficult to complete. Therefore, it is necessary to use informatization means to assist the manager in automatic supervision, and to find out and prompt out the nonstandard behavior.
Disclosure of Invention
The technical problem to be solved by the invention is to provide a ship control behavior risk assessment system and method based on big data, which can be used for assessing ship control behaviors; based on the Internet of things and big data technology, ship azimuth data and cabin monitoring data in the whole process of ship starting, sailing and stopping are collected, a ship control rule knowledge base is fused by utilizing a machine learning algorithm to form a ship control behavior standard model base, a ship control behavior standard assessment result and an assessment report can be automatically and continuously output, informationized support is provided for monitoring and early warning of ship control behaviors, comprehensiveness, economy and timeliness are achieved, and timeliness and accuracy of ship management are improved.
In order to achieve the technical characteristics, the aim of the invention is realized in the following way: the ship control behavior risk assessment system based on big data comprises a control parameter acquisition module, a data cleaning module, a sample big database module, a decision comparison analysis module, a risk level assessment module and an assessment result feedback module;
the control parameter acquisition module comprises a ship navigation data acquisition module and a cabin monitoring data acquisition module, wherein the ship navigation data acquisition module is used for acquiring ship navigation data, and the cabin monitoring data acquisition module is used for acquiring cabin monitoring data;
the data cleaning module is used for cleaning the data acquired by the control parameter acquisition module;
the sample big database module comprises a static big data module and a dynamic big data module, wherein the static big data module is used for storing initial data of each inherent static parameter of a ship which is input in advance, the dynamic big data module is used for collecting ship navigation data and cabin monitoring data which are collected when the ship is formed through operation and control meeting the standard requirements in normal navigation for a period of time and forming dynamic parameters, and the dynamic parameters and the static parameters are combined to form standard sample parameters;
the decision comparison analysis module is used for extracting the ship navigation data and the cabin monitoring data processed by the data cleaning module, comparing and analyzing the ship navigation data and the cabin monitoring data with the standard sample parameters, judging whether the ship navigation data and the cabin monitoring data are normal, acquiring operation behavior models corresponding to different operation behaviors through a decision tree algorithm, and judging whether the corresponding operation behaviors meet the standard requirements;
the risk level assessment module is used for collecting analysis and judgment results obtained by the decision comparison and analysis module and aiming at the control behavior model, assessing the judgment results to obtain corresponding scores, giving different weights to the different control behavior models, and carrying out risk level assessment after weighting all the scores to obtain the risk level of the current voyage;
and the evaluation result feedback module is used for counting and storing the risk level of the previous voyage within a period of time to form an evaluation report within the period of time.
The ship navigation data comprise ship navigation start-stop time, navigation speed and navigation track; the cabin monitoring data comprises the rotation speed, power, water temperature, oil pressure and oil cabin liquid level of the ship main engine.
The static big data module is used for storing initial data of various inherent static parameters of the ship, including ship type parameters, displacement, main engine parameters, auxiliary engine parameters, cooling system parameters, steering engine parameters, shafting and propulsion system parameters, which are input in advance;
the dynamic big data module is used for acquiring ship navigation data and cabin monitoring data acquired when the ship is in normal navigation through operation and control meeting the specification requirements for a period of time, forming dynamic parameters, and forming standard sample parameters matched with the ship shape through combination of the dynamic parameters and the static parameters.
The different manipulation behavior models include: the system comprises an open-air behavior analysis model, a stopping-air behavior analysis model, a forward and reverse switching analysis model, a course speed analysis model, a full-load operation ratio analysis model, a navigation process guiding equipment analysis model, a night-air process analysis model, a driving process analysis model and a ship stopping correctly-used lamplight analysis model.
The assessment method of the ship control behavior risk assessment system based on big data comprises the following steps:
s100, standard sample data are obtained through a sample big database module, wherein the sample big database module comprises a static big data module and a dynamic big data module, and the static big data module is used for storing initial data of each inherent static parameter of a ship which is input in advance; the dynamic big data module is used for acquiring ship navigation data and cabin monitoring data acquired when the ship is formed through control meeting the specification requirements in normal navigation for a period of time and forming dynamic parameters, and standard sample parameters matched with the ship form are formed through combining the dynamic parameters and the static parameters;
s200, acquiring control parameter data of the navigation through a control parameter acquisition module, wherein the control parameter acquisition module comprises a ship navigation data acquisition module and a cabin monitoring data acquisition module, the ship navigation data acquisition module is used for acquiring ship navigation data, and the cabin monitoring data acquisition module is used for acquiring cabin monitoring data;
s300, data cleaning; the navigation data acquisition module and the cabin monitoring data acquisition module acquired by the control parameter acquisition module are cleaned and noise reduced through the data cleaning module;
s400, extracting the ship navigation data and the cabin monitoring data processed by the data cleaning module through the decision comparison analysis module, carrying out comparison analysis on the ship navigation data and the cabin monitoring data and the standard sample parameters, judging whether the ship navigation data and the cabin monitoring data are normal, acquiring different control behavior models corresponding to the control behaviors through a decision tree algorithm, and judging whether the corresponding control behaviors meet the standard requirements;
s500, collecting analysis and judgment results obtained by a decision comparison analysis module aiming at different operation and control behavior models through a risk level assessment module, assessing the judgment results to obtain scores of the analysis and judgment results, giving weights to the different operation and control behavior models, and carrying out risk level assessment after weighting all the scores to obtain risk levels of the current voyage, wherein the risk levels are correspondingly divided into: high risk, medium risk, low risk, no risk;
and S600, counting and storing the risk level of the previous voyage in a period of time through an evaluation result feedback module to form an evaluation report in the period of time.
The different manipulation behavior models include: the system comprises an open-air behavior analysis model, a stopping-air behavior analysis model, a forward and reverse switching analysis model, a course speed analysis model, a full-load operation ratio analysis model, a navigation process guiding equipment analysis model, a night-air process analysis model, a driving process analysis model and a ship correct use lamplight analysis model.
The ship navigation data comprise ship navigation start-stop time, navigation speed and navigation track; the cabin monitoring data comprise the rotation speed, power, water temperature, oil pressure and oil cabin liquid level of the ship main engine; the data acquisition is realized by adopting a sensor or a measuring instrument;
in the data cleaning process in S300, a part of null values and abnormal values are processed, and because the data volume is large, the part of null values are directly deleted, and the abnormal values are removed after being detected by adopting a quarter bit interval;
the ship navigation data and the cabin monitoring data are mainly time sequence data, and a sliding window method is adopted to extract the average number, the median, the standard deviation, the maximum value, the minimum value, the slope, the skewness and the kurtosis of the time period data as characteristics, so that a proper characteristic data set is finally formed;
for different manipulation behaviors, different characteristic values are extracted.
The static big data module is used for storing initial data of various inherent static parameters of the ship, including ship type parameters, displacement, main engine parameters, auxiliary engine parameters, cooling system parameters, steering engine parameters, shafting and propulsion system parameters, which are input in advance; the dynamic big data module is used for acquiring ship navigation data and cabin monitoring data acquired when the ship is formed through control meeting the specification requirements in normal navigation for a period of time, forming dynamic parameters, and forming standard sample parameters matched with the ship shape through combining the dynamic parameters and the static parameters.
In the decision tree algorithm in S400, the decision tree learns through a training data set to obtain a top-down tree model, and test data is applied to the tree model to obtain a prediction classification result, which is a binary tree model including a root node, an intermediate node and leaf nodes, each node is judged and selected according to the splitting standard of each attribute, and the leaf nodes are the final classification category of the sample;
gini index represents attribute classification uncertainty, the smaller the value of Gini index represents uncertainty, namely the lower the uncertainty is, namely the decision tree based on Gini index as an attribute splitting standard has CART, SLIQ and SPRINT as a standard for selecting optimal characteristics; the Gini formula is as follows, pi represents the probability that a sample belongs to the i category, and the probability that the sample is misclassified is 1-pi:
the CART algorithm is a classification regression binary decision tree, that is, when a certain feature is used for sample division, it can be divided into two sets: equal to the set of given sample features, not equal to the set of given sample features; where S is the sample dataset and S1 and S2 are the sub-sample sets:
generating a model: inputting conditions including training data set S and stopping calculation; the output is CART decision tree, and according to training data set, starting from root node, the following processing is sequentially carried out on each node to construct binary decision tree:
(1) Setting the training data set of the node as S, calculating the Gini index of the existing feature about the data set, in this case, dividing each feature A into two parts S1 and S2 according to the test condition that whether the sample point meets A=a or not for each value a possibly taken by the feature A, and calculating Gini index Gini (S, a) when A=a by using the formula (2);
(2) Selecting the feature with the smallest Gini index and the segmentation point corresponding to the feature as the optimal feature and the optimal segmentation point from all possible features A and all possible segmentation points a of the feature, generating two sub-nodes from the current node according to the optimal feature and the optimal segmentation point, and distributing the training data set into the two sub-nodes according to the feature;
(3) Recursively calling the formula (1) and the formula (2) for the two sub-nodes until a stopping condition is met, namely the number of samples in the node is smaller than a preset threshold value, the Gini index of the sample set is smaller than the threshold value, or no more features exist;
(4) And generating a CART decision tree.
The ship navigation data and the cabin monitoring data generate 9 control behavior models, and main labels and main factors of each control behavior model comprise:
the first control behavior model and the opening behavior analysis model are labeled as follows: whether the operation behavior of the voyage is standard or not, the main factors are as follows: slope of water temperature curve, maximum value of water temperature, slope of host rotation speed curve, phase difference of water temperature-navigational speed curve;
the control behavior model II and the navigation stopping behavior analysis model are labeled as follows: whether the operation and control behavior is standard or not, the main factors are as follows: slope of water temperature curve and slope of navigational speed curve;
the control behavior model III and the forward and reverse switching analysis model are labeled as follows: whether the forward and reverse switching behavior is standard or not, the main factors are as follows: slope of host rotation speed curve, phase difference of rotation speed-navigational speed curve;
controlling a behavior model IV and a course speed analysis model, wherein the labels are as follows: whether the navigational speed course is abnormal or not, the main factors are as follows: course statistics, navigational speed statistics and fuel consumption statistics;
five control behavior models, full load operation ratio analysis models, the labels are: whether the full load operation ratio is abnormal or not, the main factors are as follows: power distribution statistics, maximum power;
controlling a behavior model six, guiding an equipment analysis model in a navigation process, and labeling as follows: the main factors of whether navigation equipment is abnormal or not in the navigation process are as follows: the equipment online time and navigation time;
a control behavior model seven and a night navigation process analysis model are labeled as follows: whether the control behavior is standard in the night navigation process or not, the main factors are as follows: a night voyage number, a night voyage distance, a night voyage host rotation speed curve and a night voyage ship track curve;
a control behavior model eight and a driving process analysis model, wherein the labels are as follows: whether the driving process is normal in operation and control behaviors or not is mainly determined by the following factors: host rotation speed statistics, ship navigational speed statistics and ship track curve;
and a control behavior model nine, a ship correctly uses a lamplight analysis model, and the labels are as follows: whether the lamplight control behavior is standard or not is determined by the following main factors: monitoring a main engine rotating speed curve, a ship navigation stopping track curve and a ship lamplight state;
accordingly, the control behavior model is obtained through a decision tree algorithm, and whether the corresponding control behavior meets the standard requirement is judged.
The invention has the following beneficial effects:
based on the Internet of things and big data technology, the ship navigation system acquires ship azimuth data and cabin monitoring data in the whole process of ship navigation, navigation and stopping navigation, utilizes a machine learning algorithm to fuse a ship control rule knowledge base to form a ship control behavior standard model base, can automatically and continuously output a ship control behavior standard assessment result and an evaluation report, provides informationized support for monitoring and early warning of ship control behaviors, has comprehensiveness, economy and timeliness, and improves timeliness and accuracy of ship management.
Drawings
The invention is further described below with reference to the drawings and examples.
FIG. 1 is a block diagram of a risk assessment system of the present invention.
Fig. 2 is a block diagram of a manipulation parameter acquisition module according to the present invention.
FIG. 3 is a block diagram of a sample large database module of the present invention.
In the figure: the system comprises a control parameter acquisition module 1, a data cleaning module 2, a sample big database module 3, a decision comparison analysis module 4, a risk level evaluation module 5 and an evaluation result feedback module 6;
a ship navigation data acquisition module 101 and a cabin monitoring data acquisition module 102;
a static big data module 301 and a dynamic big data module 302.
Detailed Description
Embodiments of the present invention will be further described with reference to the accompanying drawings.
Example 1:
referring to fig. 1-3, a ship control behavior risk assessment system based on big data comprises a control parameter acquisition module 1, a data cleaning module 2, a sample big database module 3, a decision comparison analysis module 4, a risk level assessment module 5 and an assessment result feedback module 6; the control parameter acquisition module 1 comprises a ship navigation data acquisition module 101 and a cabin monitoring data acquisition module 102, wherein the ship navigation data acquisition module 101 is used for acquiring ship navigation data, and the cabin monitoring data acquisition module 102 is used for acquiring cabin monitoring data; the data cleaning module 2 is used for cleaning the data acquired by the control parameter acquisition module 1; the sample big database module 3 comprises a static big data module 301 and a dynamic big data module 302, wherein the static big data module 301 is used for storing initial data of each inherent static parameter of a ship which is input in advance, the dynamic big data module 302 is used for collecting ship navigation data and cabin monitoring data which are collected when the ship is operated and formed through operation meeting the standard requirements in normal navigation for a period of time and forming dynamic parameters, and the dynamic parameters and the static parameters are combined to form standard sample parameters; the decision comparison analysis module 4 is used for extracting the ship navigation data and the cabin monitoring data processed by the data cleaning module, comparing and analyzing the ship navigation data and the cabin monitoring data with the standard sample parameters, judging whether the ship navigation data and the cabin monitoring data are normal, acquiring operation behavior models corresponding to different operation behaviors through a decision tree algorithm, and judging whether the corresponding operation behaviors meet the standard requirements; the risk level evaluation module 5 is used for collecting the analysis and judgment results obtained by the decision comparison analysis module 4 and aiming at the control behavior model, evaluating the judgment results to obtain corresponding scores, giving different weights to the different control behavior models, and performing risk level evaluation after weighting all the scores to obtain the risk level of the current voyage; the evaluation result feedback module 6 is used for counting and storing the risk level of the past navigation in a period of time to form an evaluation report in the period of time. Through adopting above-mentioned system, it is based on thing networking and big data technology, gathers boats and ships navigation, the whole process boats and ships position data of stopping and cabin monitoring data, utilize the machine learning algorithm, merge the rule knowledge base is controlled to the boats and ships, form the normal model storehouse of behavior is controlled to the boats and ships, can automatic continuously output the normal nature assessment result of behavior is controlled to the boats and ships and evaluation report, provide informationized support for the monitoring early warning of behavior is controlled to the boats and ships, have comprehensiveness, economic nature and timeliness, the timeliness and the accuracy of management have been promoted.
Further, the ship navigation data comprise ship navigation start-stop time, navigation speed and navigation track; the cabin monitoring data comprises the rotation speed, power, water temperature, oil pressure and oil cabin liquid level of the ship main engine. The static big data module 301 is configured to store initial data of various inherent static parameters of the ship, including ship type parameters, displacement, main engine parameters, auxiliary engine parameters, cooling system parameters, steering engine parameters, shafting and propulsion system parameters, which are input in advance; the dynamic big data module 302 is configured to collect ship navigation data and cabin monitoring data collected when the ship is operating according with the specification requirements in normal navigation for a period of time, and form dynamic parameters, and form standard sample parameters matched with the ship shape by combining the dynamic parameters and the static parameters. The subsequent integration of the static big data module 301 and the dynamic big data module 302 to form standard sample parameters is facilitated by the adoption of the ship navigation data and the cabin monitoring data.
Further, the different manipulation behavior models include: the system comprises an open-air behavior analysis model, a stopping-air behavior analysis model, a forward and reverse switching analysis model, a course speed analysis model, a full-load operation ratio analysis model, a navigation process guiding equipment analysis model, a night-air process analysis model, a driving process analysis model and a ship stopping correctly-used lamplight analysis model. The ship control behavior standard model library is formed by fusing the ship control rule knowledge base through the different control behavior models, so that a ship control behavior standard evaluation result and an evaluation report can be automatically and continuously output, and informationized support is provided for monitoring and early warning of ship control behaviors.
Example 2:
the assessment method of the ship control behavior risk assessment system based on big data comprises the following steps:
s100, standard sample data are obtained through a sample big database module 3, wherein the sample big database module comprises a static big data module and a dynamic big data module, and the static big data module is used for storing initial data of each inherent static parameter of a ship which is input in advance; the dynamic big data module is used for acquiring ship navigation data and cabin monitoring data acquired when the ship is formed through control meeting the specification requirements in normal navigation for a period of time and forming dynamic parameters, and standard sample parameters matched with the ship form are formed through combining the dynamic parameters and the static parameters;
s200, acquiring control parameter data of the navigation through a control parameter acquisition module 1, wherein the control parameter acquisition module comprises a ship navigation data acquisition module and a cabin monitoring data acquisition module, the ship navigation data acquisition module is used for acquiring ship navigation data, and the cabin monitoring data acquisition module is used for acquiring cabin monitoring data;
s300, data cleaning; the navigation data acquisition module and the cabin monitoring data acquisition module acquired by the control parameter acquisition module 1 are cleaned and noise reduced through the data cleaning module 2;
s400, extracting the ship navigation data and the cabin monitoring data processed by the data cleaning module 2 through the decision comparison analysis module 4, carrying out comparison analysis on the ship navigation data and the cabin monitoring data and the standard sample parameters, judging whether the ship navigation data and the cabin monitoring data are normal, acquiring different control behavior models corresponding to the control behaviors through a decision tree algorithm, and judging whether the corresponding control behaviors meet the standard requirements;
s500, collecting analysis and judgment results obtained by the decision comparison analysis module 4 aiming at different operation and control behavior models through the risk level assessment module 5, assessing the judgment results to obtain scores of the analysis and judgment results, giving weights to the different operation and control behavior models, and carrying out risk level assessment after weighting all the scores to obtain risk levels of the current voyage, wherein the risk level correspondence is divided into: high risk, medium risk, low risk, no risk;
s600, the evaluation report in a period of time is formed by the evaluation result feedback module 6 for counting and storing the risk level of the previous voyage in the period of time.
Further, the different manipulation behavior models include: the system comprises an open-air behavior analysis model, a stopping-air behavior analysis model, a forward and reverse switching analysis model, a course speed analysis model, a full-load operation ratio analysis model, a navigation process guiding equipment analysis model, a night-air process analysis model, a driving process analysis model and a ship correct use lamplight analysis model.
Further, the ship navigation data comprise ship navigation start-stop time, navigation speed and navigation track; the cabin monitoring data comprise the rotation speed, power, water temperature, oil pressure and oil cabin liquid level of the ship main engine; the data acquisition is realized by adopting a sensor or a measuring instrument;
further, in the data cleaning process in S300, a partial null value and an abnormal value are processed, and because the data volume is large, the partial null value is directly deleted, and the abnormal value is removed after the detection of the quarter bit interval;
further, the ship navigation data and the cabin monitoring data are mainly time sequence data, and a sliding window method is adopted to extract the average number, the median, the standard deviation, the maximum value, the minimum value, the slope, the skewness and the kurtosis of the time period data as characteristics, so that a proper characteristic data set is finally formed; for different manipulation behaviors, different characteristic values are extracted.
Further, the static big data module is used for storing initial data of various inherent static parameters of the ship, including ship type parameters, displacement, main engine parameters, auxiliary machine parameters, cooling system parameters, steering engine parameters, shafting and propulsion system parameters, which are input in advance; the dynamic big data module is used for acquiring ship navigation data and cabin monitoring data acquired when the ship is formed through control meeting the specification requirements in normal navigation for a period of time, forming dynamic parameters, and forming standard sample parameters matched with the ship shape through combining the dynamic parameters and the static parameters.
Example 3:
in the decision tree algorithm in S400, the decision tree learns through a training data set to obtain a top-down tree model, and test data is applied to the tree model to obtain a prediction classification result, which is a binary tree model including a root node, an intermediate node and leaf nodes, each node is judged and selected according to the splitting standard of each attribute, and the leaf nodes are the final classification category of the sample;
gini index represents attribute classification uncertainty, the smaller the value of Gini index represents uncertainty, namely the lower the uncertainty is, namely the decision tree based on Gini index as an attribute splitting standard has CART, SLIQ and SPRINT as a standard for selecting optimal characteristics; the Gini formula is as follows, pi represents the probability that a sample belongs to the i category, and the probability that the sample is misclassified is 1-pi:
the CART algorithm is a classification regression binary decision tree, that is, when a certain feature is used for sample division, it can be divided into two sets: equal to the set of given sample features, not equal to the set of given sample features; where S is the sample dataset and S1 and S2 are the sub-sample sets:
generating a model: inputting conditions including training data set S and stopping calculation; the output is CART decision tree, and according to training data set, starting from root node, the following processing is sequentially carried out on each node to construct binary decision tree:
(1) Setting the training data set of the node as S, calculating the Gini index of the existing feature about the data set, in this case, dividing each feature A into two parts S1 and S2 according to the test condition that whether the sample point meets A=a or not for each value a possibly taken by the feature A, and calculating Gini index Gini (S, a) when A=a by using the formula (2);
(2) Selecting the feature with the smallest Gini index and the segmentation point corresponding to the feature as the optimal feature and the optimal segmentation point from all possible features A and all possible segmentation points a of the feature, generating two sub-nodes from the current node according to the optimal feature and the optimal segmentation point, and distributing the training data set into the two sub-nodes according to the feature;
(3) Recursively calling the formula (1) and the formula (2) for the two sub-nodes until a stopping condition is met, namely the number of samples in the node is smaller than a preset threshold value, the Gini index of the sample set is smaller than the threshold value, or no more features exist;
(4) And generating a CART decision tree.
Example 4:
in the specific implementation process, 9 control behavior models are generated from ship navigation data and cabin monitoring data, and main labels and main factors of each control behavior model comprise:
the first control behavior model and the opening behavior analysis model are labeled as follows: whether the operation behavior of the voyage is standard or not, the main factors are as follows: slope of water temperature curve, maximum value of water temperature, slope of host rotation speed curve, phase difference of water temperature-navigational speed curve;
the control behavior model II and the navigation stopping behavior analysis model are labeled as follows: whether the operation and control behavior is standard or not, the main factors are as follows: slope of water temperature curve and slope of navigational speed curve;
the control behavior model III and the forward and reverse switching analysis model are labeled as follows: whether the forward and reverse switching behavior is standard or not, the main factors are as follows: slope of host rotation speed curve, phase difference of rotation speed-navigational speed curve;
controlling a behavior model IV and a course speed analysis model, wherein the labels are as follows: whether the navigational speed course is abnormal or not, the main factors are as follows: course statistics, navigational speed statistics and fuel consumption statistics;
five control behavior models, full load operation ratio analysis models, the labels are: whether the full load operation ratio is abnormal or not, the main factors are as follows: power distribution statistics, maximum power;
controlling a behavior model six, guiding an equipment analysis model in a navigation process, and labeling as follows: the main factors of whether navigation equipment is abnormal or not in the navigation process are as follows: the equipment online time and navigation time;
a control behavior model seven and a night navigation process analysis model are labeled as follows: whether the control behavior is standard in the night navigation process or not, the main factors are as follows: a night voyage number, a night voyage distance, a night voyage host rotation speed curve and a night voyage ship track curve;
a control behavior model eight and a driving process analysis model, wherein the labels are as follows: whether the driving process is normal in operation and control behaviors or not is mainly determined by the following factors: host rotation speed statistics, ship navigational speed statistics and ship track curve;
and a control behavior model nine, a ship correctly uses a lamplight analysis model, and the labels are as follows: whether the lamplight control behavior is standard or not is determined by the following main factors: monitoring a main engine rotating speed curve, a ship navigation stopping track curve and a ship lamplight state;
accordingly, the control behavior model is obtained through a decision tree algorithm, and whether the corresponding control behavior meets the standard requirement is judged.
Claims (4)
1. The ship control behavior risk assessment system based on big data is characterized by comprising a control parameter acquisition module (1), a data cleaning module (2), a sample big database module (3), a decision comparison analysis module (4), a risk level assessment module (5) and an assessment result feedback module (6);
the control parameter acquisition module (1) comprises a ship navigation data acquisition module (101) and a cabin monitoring data acquisition module (102), wherein the ship navigation data acquisition module (101) is used for acquiring ship navigation data, and the cabin monitoring data acquisition module (102) is used for acquiring cabin monitoring data;
the data cleaning module (2) is used for cleaning the data acquired by the control parameter acquisition module (1);
the sample big database module (3) comprises a static big data module (301) and a dynamic big data module (302), wherein the static big data module (301) is used for storing initial data of each inherent static parameter of a ship which is input in advance, and the dynamic big data module (302) is used for collecting ship navigation data and cabin monitoring data which are collected when the ship passes through a control stroke meeting the standard requirement in normal navigation for a period of time and forming dynamic parameters, and the dynamic parameters and the static parameters are combined to form standard sample parameters;
the decision comparison analysis module (4) is used for extracting the ship navigation data and the cabin monitoring data processed by the data cleaning module, comparing and analyzing the ship navigation data and the cabin monitoring data with the standard sample parameters, judging whether the ship navigation data and the cabin monitoring data are normal or not, acquiring operation behavior models corresponding to different operation behaviors through a decision tree algorithm, and judging whether the corresponding operation behaviors meet the standard requirements or not;
the risk level assessment module (5) is used for collecting analysis and judgment results obtained by the decision comparison analysis module (4) aiming at the control behavior model, assessing the judgment results to obtain corresponding scores, giving different weights to different control behavior models, and carrying out risk level assessment after weighting all the scores to obtain the risk level of the current voyage;
the evaluation result feedback module (6) is used for counting and storing the risk level of the previous voyage within a period of time to form an evaluation report within the period of time;
the ship navigation data comprise ship navigation start-stop time, navigation speed and navigation track; the cabin monitoring data comprise the rotation speed, power, water temperature, oil pressure and oil cabin liquid level of the ship main engine;
the static big data module (301) is used for storing initial data of various inherent static parameters of the ship, including ship type parameters, displacement, main engine parameters, auxiliary engine parameters, cooling system parameters, steering engine parameters, shafting and propulsion system parameters, which are input in advance;
the dynamic big data module (302) is used for acquiring ship navigation data and cabin monitoring data acquired when the ship passes through a control stroke meeting the specification requirements in normal navigation for a period of time and forming dynamic parameters, and standard sample parameters matched with the ship are formed by combining the dynamic parameters and the static parameters;
the different manipulation behavior models include: the system comprises an open-air behavior analysis model, a stopping-air behavior analysis model, a forward and reverse switching analysis model, a course navigational speed analysis model, a full-load operation ratio analysis model, a navigation process navigation equipment analysis model, a night navigation process analysis model, a driving process analysis model and a ship stopping correctly-used lamplight analysis model;
the ship navigation data and the cabin monitoring data generate 9 control behavior models, and the labels and factors of each control behavior model comprise:
the first control behavior model and the opening behavior analysis model are labeled as follows: whether the operation behavior of the voyage is standard or not, the factors are as follows: slope of water temperature curve, maximum value of water temperature, slope of host rotation speed curve, phase difference of water temperature-navigational speed curve;
the control behavior model II and the navigation stopping behavior analysis model are labeled as follows: whether the operation and control behavior is standard or not, the factors are as follows: slope of water temperature curve and slope of navigational speed curve;
the control behavior model III and the forward and reverse switching analysis model are labeled as follows: whether the forward and reverse switching behavior is standard or not, the factor is: slope of host rotation speed curve, phase difference of rotation speed-navigational speed curve;
controlling a behavior model IV and a course speed analysis model, wherein the labels are as follows: whether the navigational speed course is abnormal or not, the factors are as follows: course statistics, navigational speed statistics and fuel consumption statistics;
five control behavior models, full load operation ratio analysis models, the labels are: whether the full load operation ratio is abnormal or not, the factor is: power distribution statistics, maximum power;
controlling a behavior model six, guiding an equipment analysis model in a navigation process, and labeling as follows: whether navigation process leads to equipment unusual, the factor is: the equipment online time and navigation time;
a control behavior model seven and a night navigation process analysis model are labeled as follows: whether the control behavior is standard in the night navigation process or not, the factors are as follows: a night voyage number, a night voyage distance, a night voyage host rotation speed curve and a night voyage ship track curve;
a control behavior model eight and a driving process analysis model, wherein the labels are as follows: whether the driving process control behavior is standard or not, the factors are as follows: host rotation speed statistics, ship navigational speed statistics and ship track curve;
and a control behavior model nine, a ship correctly uses a lamplight analysis model, and the labels are as follows: whether the lamplight control behavior is standard or not is determined by the following factors: monitoring a main engine rotating speed curve, a ship navigation stopping track curve and a ship lamplight state;
accordingly, the control behavior model is obtained through a decision tree algorithm, and whether the corresponding control behavior meets the standard requirement is judged.
2. An evaluation method adopting the ship control behavior risk evaluation system based on big data as set forth in claim 1, characterized in that: the method comprises the following steps:
s100, standard sample data are obtained through a sample big database module (3), wherein the sample big database module comprises a static big data module and a dynamic big data module, and the static big data module is used for storing initial data of each inherent static parameter of a ship which is input in advance; the dynamic big data module is used for acquiring ship navigation data and cabin monitoring data acquired when the ship is formed through control meeting the specification requirements in normal navigation for a period of time and forming dynamic parameters, and standard sample parameters matched with the ship form are formed through combining the dynamic parameters and the static parameters;
s200, acquiring control parameter data of the navigation through a control parameter acquisition module (1), wherein the control parameter acquisition module comprises a ship navigation data acquisition module and a cabin monitoring data acquisition module, the ship navigation data acquisition module is used for acquiring ship navigation data, and the cabin monitoring data acquisition module is used for acquiring cabin monitoring data;
s300, data cleaning; the navigation data acquisition module and the cabin monitoring data acquisition module acquired by the control parameter acquisition module (1) are cleaned and denoised through the data cleaning module (2);
s400, extracting the ship navigation data and the cabin monitoring data processed by the data cleaning module (2) through the decision comparison analysis module (4) for comparison analysis with standard sample parameters, judging whether the ship navigation data and the cabin monitoring data are normal, acquiring different control behavior models corresponding to the control behaviors through a decision tree algorithm, and judging whether the corresponding control behaviors meet the standard requirements;
s500, collecting analysis and judgment results obtained by a decision comparison analysis module (4) aiming at different operation and control behavior models through a risk level assessment module (5), assessing the judgment results to obtain scores of the analysis and judgment results, giving weights to the different operation and control behavior models, and carrying out risk level assessment after weighting all the scores to obtain risk levels of the current voyage, wherein the risk levels are correspondingly divided into: high risk, medium risk, low risk, no risk;
s600, calculating and storing the risk level of the previous voyage in a period of time through an evaluation result feedback module (6) to form an evaluation report in the period of time;
the ship navigation data and the cabin monitoring data generate 9 control behavior models, and the labels and factors of each control behavior model comprise:
the first control behavior model and the opening behavior analysis model are labeled as follows: whether the operation behavior of the voyage is standard or not, the factors are as follows: slope of water temperature curve, maximum value of water temperature, slope of host rotation speed curve, phase difference of water temperature-navigational speed curve;
the control behavior model II and the navigation stopping behavior analysis model are labeled as follows: whether the operation and control behavior is standard or not, the factors are as follows: slope of water temperature curve and slope of navigational speed curve;
the control behavior model III and the forward and reverse switching analysis model are labeled as follows: whether the forward and reverse switching behavior is standard or not, the factor is: slope of host rotation speed curve, phase difference of rotation speed-navigational speed curve;
controlling a behavior model IV and a course speed analysis model, wherein the labels are as follows: whether the navigational speed course is abnormal or not, the factors are as follows: course statistics, navigational speed statistics and fuel consumption statistics;
five control behavior models, full load operation ratio analysis models, the labels are: whether the full load operation ratio is abnormal or not, the factor is: power distribution statistics, maximum power;
controlling a behavior model six, guiding an equipment analysis model in a navigation process, and labeling as follows: whether navigation process leads to equipment unusual, the factor is: the equipment online time and navigation time;
a control behavior model seven and a night navigation process analysis model are labeled as follows: whether the control behavior is standard in the night navigation process or not, the factors are as follows: a night voyage number, a night voyage distance, a night voyage host rotation speed curve and a night voyage ship track curve;
a control behavior model eight and a driving process analysis model, wherein the labels are as follows: whether the driving process control behavior is standard or not, the factors are as follows: host rotation speed statistics, ship navigational speed statistics and ship track curve;
and a control behavior model nine, a ship correctly uses a lamplight analysis model, and the labels are as follows: whether the lamplight control behavior is standard or not is determined by the following factors: monitoring a main engine rotating speed curve, a ship navigation stopping track curve and a ship lamplight state;
accordingly, the control behavior model is obtained through a decision tree algorithm, and whether the corresponding control behavior meets the standard requirement is judged.
3. The method for evaluating the risk evaluation system for the ship handling behavior based on big data according to claim 2, wherein: the ship navigation data comprise ship navigation start-stop time, navigation speed and navigation track; the cabin monitoring data comprise the rotation speed, power, water temperature, oil pressure and oil cabin liquid level of the ship main engine; the data acquisition is realized by adopting a sensor or a measuring instrument;
in the data cleaning process in S300, a part of null values and abnormal values are processed, and because the data volume is large, the part of null values are directly deleted, and the abnormal values are removed after being detected by adopting a quarter bit interval;
the ship navigation data and the cabin monitoring data are time sequence data, and a sliding window method is adopted to extract the average number, the median, the standard deviation, the maximum value, the minimum value, the slope, the skewness and the kurtosis of the time period data as characteristics, so that a proper characteristic data set is finally formed;
for different manipulation behaviors, different characteristic values are extracted.
4. The method for evaluating the risk evaluation system for the ship handling behavior based on big data according to claim 2, wherein: in the decision tree algorithm in S400, the decision tree learns through a training data set to obtain a top-down tree model, and test data is applied to the tree model to obtain a prediction classification result, which is a binary tree model including a root node, an intermediate node and leaf nodes, each node is judged and selected according to the splitting standard of each attribute, and the leaf nodes are the final classification category of the sample;
gini index represents attribute classification uncertainty, the smaller the value of Gini index represents uncertainty, namely the lower the uncertainty is, namely the decision tree based on Gini index as an attribute splitting standard has CART, SLIQ and SPRINT as a standard for selecting optimal characteristics; the Gini formula is as follows, pi represents the probability that a sample belongs to the i category, and the probability that the sample is misclassified is 1-pi:
the CART algorithm is a classification regression binary decision tree, that is, when a certain feature is used for sample division, it can be divided into two sets: equal to the set of given sample features, not equal to the set of given sample features; where S is the sample dataset and S1 and S2 are the sub-sample sets:
generating a model: inputting conditions including training data set S and stopping calculation; the output is CART decision tree, and according to training data set, starting from root node, the following processing is sequentially carried out on each node to construct binary decision tree:
(1) Setting the training data set of the node as S, calculating the Gini index of the existing feature about the data set, in this case, dividing each feature A into two parts S1 and S2 according to the test condition that whether the sample point meets A=a or not for each value a possibly taken by the feature A, and calculating Gini index Gini (S, a) when A=a by using the formula (2);
(2) Selecting the feature with the smallest Gini index and the segmentation point corresponding to the feature as the optimal feature and the optimal segmentation point from all possible features A and all possible segmentation points a of the feature, generating two sub-nodes from the current node according to the optimal feature and the optimal segmentation point, and distributing the training data set into the two sub-nodes according to the feature;
(3) Recursively calling the formula (1) and the formula (2) for the two sub-nodes until a stopping condition is met, namely the number of samples in the node is smaller than a preset threshold value, the Gini index of the sample set is smaller than the threshold value, or no more features exist;
(4) And generating a CART decision tree.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202211260297.6A CN115659263B (en) | 2022-10-14 | 2022-10-14 | Ship control behavior risk assessment system and method based on big data |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202211260297.6A CN115659263B (en) | 2022-10-14 | 2022-10-14 | Ship control behavior risk assessment system and method based on big data |
Publications (2)
Publication Number | Publication Date |
---|---|
CN115659263A CN115659263A (en) | 2023-01-31 |
CN115659263B true CN115659263B (en) | 2023-08-08 |
Family
ID=84988267
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202211260297.6A Active CN115659263B (en) | 2022-10-14 | 2022-10-14 | Ship control behavior risk assessment system and method based on big data |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN115659263B (en) |
Families Citing this family (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115880950B (en) * | 2023-02-27 | 2023-05-05 | 中国船舶集团有限公司第七一九研究所 | Data processing method of automatic ship identification system |
CN117390762B (en) * | 2023-09-15 | 2024-05-14 | 中国人民解放军92942部队 | Full-flow ship safety design analysis method |
CN117037089A (en) * | 2023-10-09 | 2023-11-10 | 亿海蓝(北京)数据技术股份公司 | Method and device for detecting ship unauthorized exit behavior and readable storage medium |
Citations (12)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108009723A (en) * | 2017-12-01 | 2018-05-08 | 大连海事大学 | A kind of ship navigation state judges and energy efficiency state assessment system |
CN109146195A (en) * | 2018-09-06 | 2019-01-04 | 北方爆破科技有限公司 | A kind of blast fragmentation size prediction technique based on cart tree regression algorithm |
CN109657880A (en) * | 2019-01-09 | 2019-04-19 | 中交信息技术国家工程实验室有限公司 | A kind of Collision Accidents of Ships grade prediction technique and system based on Bayesian network |
AU2020101965A4 (en) * | 2020-08-25 | 2020-10-01 | Wuhan University Of Technology | A method of navigational early warming for ship encounter, and system thereof |
CN112208728A (en) * | 2020-09-30 | 2021-01-12 | 震兑工业智能科技有限公司 | Ship fouling monitoring and early warning method |
CN112441196A (en) * | 2020-11-20 | 2021-03-05 | 武汉理工大学 | Processing method, system and storage medium of ship navigation risk assessment information |
CN113032502A (en) * | 2021-02-09 | 2021-06-25 | 北京工业大学 | Ship anomaly detection method based on improved track segment DBSCAN clustering |
CN113128826A (en) * | 2021-03-03 | 2021-07-16 | 武汉理工大学 | Navigation risk prediction method, navigation risk prediction device and storage medium |
CN113212153A (en) * | 2021-04-15 | 2021-08-06 | 东风汽车集团股份有限公司 | Method, device, equipment and storage medium for high-low speed operation of fan |
CN113762394A (en) * | 2021-09-09 | 2021-12-07 | 昆明理工大学 | Blasting block size prediction method |
CN114662575A (en) * | 2022-03-09 | 2022-06-24 | 集美大学 | Wind power water area ship navigation risk estimation method and system and storage medium |
CN115017684A (en) * | 2022-05-25 | 2022-09-06 | 广州星际海洋工程设计有限公司 | Intelligent energy efficiency management system for ship |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111951606B (en) * | 2020-07-29 | 2021-07-30 | 武汉理工大学 | Ship collision risk assessment and early warning method and system |
-
2022
- 2022-10-14 CN CN202211260297.6A patent/CN115659263B/en active Active
Patent Citations (12)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108009723A (en) * | 2017-12-01 | 2018-05-08 | 大连海事大学 | A kind of ship navigation state judges and energy efficiency state assessment system |
CN109146195A (en) * | 2018-09-06 | 2019-01-04 | 北方爆破科技有限公司 | A kind of blast fragmentation size prediction technique based on cart tree regression algorithm |
CN109657880A (en) * | 2019-01-09 | 2019-04-19 | 中交信息技术国家工程实验室有限公司 | A kind of Collision Accidents of Ships grade prediction technique and system based on Bayesian network |
AU2020101965A4 (en) * | 2020-08-25 | 2020-10-01 | Wuhan University Of Technology | A method of navigational early warming for ship encounter, and system thereof |
CN112208728A (en) * | 2020-09-30 | 2021-01-12 | 震兑工业智能科技有限公司 | Ship fouling monitoring and early warning method |
CN112441196A (en) * | 2020-11-20 | 2021-03-05 | 武汉理工大学 | Processing method, system and storage medium of ship navigation risk assessment information |
CN113032502A (en) * | 2021-02-09 | 2021-06-25 | 北京工业大学 | Ship anomaly detection method based on improved track segment DBSCAN clustering |
CN113128826A (en) * | 2021-03-03 | 2021-07-16 | 武汉理工大学 | Navigation risk prediction method, navigation risk prediction device and storage medium |
CN113212153A (en) * | 2021-04-15 | 2021-08-06 | 东风汽车集团股份有限公司 | Method, device, equipment and storage medium for high-low speed operation of fan |
CN113762394A (en) * | 2021-09-09 | 2021-12-07 | 昆明理工大学 | Blasting block size prediction method |
CN114662575A (en) * | 2022-03-09 | 2022-06-24 | 集美大学 | Wind power water area ship navigation risk estimation method and system and storage medium |
CN115017684A (en) * | 2022-05-25 | 2022-09-06 | 广州星际海洋工程设计有限公司 | Intelligent energy efficiency management system for ship |
Non-Patent Citations (1)
Title |
---|
尉浩然 等.基于操舵规律的船舶行为预测的搁浅率动态风险研究.《中国水运》.2019,第74-75页. * |
Also Published As
Publication number | Publication date |
---|---|
CN115659263A (en) | 2023-01-31 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN115659263B (en) | Ship control behavior risk assessment system and method based on big data | |
CN115578015B (en) | Sewage treatment whole process supervision method, system and storage medium based on Internet of things | |
CN112528519A (en) | Method, system, readable medium and electronic device for engine quality early warning service | |
CN111311059A (en) | Knowledge graph-based water mill room fault diagnosis method | |
CN111275335A (en) | Data-driven slurry circulating pump optimization method and system | |
CN114813105A (en) | Gear box fault early warning method and system based on working condition similarity evaluation | |
CN111190349A (en) | Method, system and medium for monitoring state and diagnosing fault of ship engine room equipment | |
CN115330268B (en) | Comprehensive emergency command method and system for coping with mine disasters | |
CN116522088B (en) | Nuclear power plant operation data analysis method and system based on machine learning | |
CN114021932A (en) | Energy efficiency evaluation and diagnosis method, system and medium for wind turbine generator | |
CN111273125A (en) | RST-CNN-based power cable channel fault diagnosis method | |
CN113590396A (en) | Method and system for diagnosing defect of primary device, electronic device and storage medium | |
CN115526515A (en) | Safety monitoring system of gate for water conservancy and hydropower | |
CN116359285A (en) | Oil gas concentration intelligent detection system and method based on big data | |
CN111967003B (en) | Automatic wind control rule generation system and method based on black box model and decision tree | |
CN116714437B (en) | Hydrogen fuel cell automobile safety monitoring system and monitoring method based on big data | |
CN116933656A (en) | Ship power system fault diagnosis method based on deep learning | |
Corrêa et al. | Data-driven approach for labelling process plant event data | |
CN115964883A (en) | State monitoring and maintenance support system of unmanned ship | |
CN116011982A (en) | Online monitoring method and system for breakage of grinding roller of coal mill | |
CN115310517A (en) | Aero-engine online anomaly detection method based on improved piecewise linear representation | |
CN114879632A (en) | Multi-mode fusion vehicle fault diagnosis method and system based on big data | |
Morgan et al. | Predicting future states with $ n $-dimensional Markov chains for fault diagnosis | |
Tao et al. | Research on fault diagnosis method of axial flow induced draft fan of power plant based on machine learning | |
Handayani et al. | Anomaly detection in vessel sensors data with unsupervised learning technique |
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 | ||
GR01 | Patent grant | ||
GR01 | Patent grant |