A safety assessment method for ship offshore operations

Abstract

The present invention relates to the field of ship data processing technology, and in particular to a safety assessment method for ship offshore operations, comprising: step S1, acquiring ship data, and identifying feature fields and feature field types in the ship data when the acquisition is completed; step S2, training the ship data to output a plurality of ship feature data sets, and a central control module outputs a training model according to the plurality of ship feature data sets; step S3, the central control module adjusts the training quantity ratio to a corresponding training quantity ratio according to the amount of ship data trained in a unit period; step S4, when the adjustment of the training quantity ratio is completed, the central control module determines whether the fitting degree is within an allowable range according to the average value of the difference between the training output value and the target value; step S5, the central control module determines whether the current ship data is weakly correlated category data according to the proportion of the number of feature points in the single ship data.

Description

Technical Field

[0001] The present invention relates to the technical field of ship data processing, and in particular to a safety assessment method for ship offshore operations. Background Art

[0002] The ship's offshore operation process involves human, machine and environmental factors such as crew, ship platform, marine environment and atmospheric environment, and there is a correlation between these factors. In the ship data training process, there is a problem of reduced accuracy due to the inaccurate processing of incomplete data and ship data whose feature points do not meet the requirements, and there is a lack of effective processing of ship data with different feature points.

[0003] Chinese patent publication number CN116452002A discloses a real-time assessment system for ship navigation safety risks, including: a pre-risk assessment module, an in-event risk control module, an incident emergency response module and a post-accident investigation module; the pre-risk assessment module is used to establish a pre-navigation risk real-time assessment model based on a simulation model of microscopic marine traffic flow; the in-event risk control module is used to obtain a preliminary implementation plan using the relevant risk control points obtained in the pre-navigation risk real-time assessment model; the incident emergency response module is used to numerically simulate the consequences caused by the accident, and obtain an emergency response plan set according to the pre-navigation risk real-time assessment model; the post-accident investigation module is used for real-time collection and historical accumulation of multi-source data before and after the accident. The present invention improves the emergency management capabilities of maritime regulatory departments throughout the entire process from pre-risk real-time assessment, in-event risk control, incident emergency command and post-accident investigation. It can be seen that the real-time assessment system for ship navigation safety risks has the problem of insufficient accuracy of ship data training due to incomplete ship data and excessive differences in the number of feature points of ship data. Summary of the invention

[0004] To this end, the present invention provides a safety assessment method for ship offshore operations, so as to overcome the problem of insufficient accuracy of ship data training caused by incomplete ship data and excessive differences in the number of feature points of ship data in the prior art.

[0005] To achieve the above-mentioned purpose, the present invention provides a safety assessment method for ship offshore operations, comprising: step S1, acquiring ship data, and identifying the characteristic fields and characteristic field types in the ship data when the acquisition is completed; step S2, training the ship data to output a number of ship characteristic data sets, and the central control module outputs the training model according to the several ship characteristic data sets; step S3, when the output of the training model is completed, the central control module adjusts the training quantity ratio to the corresponding training quantity ratio according to the amount of ship data trained in a unit period; step S4, when the adjustment of the training quantity ratio is completed, The central control module determines whether the degree of fit is within an allowable range based on the average value of the difference between the training output value and the target value, and adjusts the number of samples to the corresponding number of samples when it is determined that the degree of fit is lower than the allowable range, or adjusts the collection benchmark capacity of the ship data to the corresponding benchmark capacity based on the average number of feature points of the ship data to be trained; step S5, when the adjustment of the collection benchmark capacity of the ship data is completed, the central control module determines whether the current ship data is weakly correlated category data based on the proportion of the number of feature points to the single ship data, and adjusts the ship data acquisition frequency to the corresponding frequency when it is determined that the current ship data is weakly correlated category data.

[0006] Further, in step S3, the central control module determines whether the training speed is within the allowable range according to the amount of ship data trained in a unit period in two types of determination methods, wherein:

[0007] The first determination method is that the central control module determines that the training speed is lower than the allowable range under the preset first training data volume condition, and adjusts the training quantity ratio to the corresponding training quantity ratio by calculating the difference between the preset training data volume and the ship data volume trained in the unit period;

[0008] The second determination method is that the central control module determines that the training speed is within the allowable range under the preset second training data amount condition;

[0009] Among them, the preset first training data amount condition is that the amount of ship data trained in a unit period is less than or equal to the preset training data amount; the preset second training data amount condition is that the amount of ship data trained in a unit period is greater than the preset training data amount.

[0010] Furthermore, the central control module determines three types of adjustment methods for the proportion of training quantity according to the difference between the preset training data quantity and the ship data quantity trained in the unit period under the preset first training data quantity condition, wherein:

[0011] The first type of adjustment method is that the central control module adjusts the training quantity ratio to a preset training quantity ratio under a preset first training data quantity difference condition;

[0012] The second type of adjustment method is that the central control module uses a preset first training quantity ratio adjustment coefficient to increase the training quantity ratio to the first ratio under a preset second training data quantity difference condition;

[0013] The third type of adjustment method is that the central control module uses a preset second training quantity ratio adjustment coefficient to increase the training quantity ratio to the second ratio under a preset third training data quantity difference condition;

[0014] The preset first training quantity ratio adjustment coefficient is smaller than the preset second training quantity ratio adjustment coefficient.

[0015] Furthermore, the preset first training data volume difference condition is that the difference between the preset training data volume and the ship data volume trained in the unit period is less than or equal to the preset first training data volume difference;

[0016] The preset second training data volume difference condition is that the difference between the preset training data volume and the ship data volume trained in the unit period is greater than the preset first training data volume difference and less than or equal to the preset second training data volume difference;

[0017] The preset third training data volume difference condition is that the difference between the preset training data volume and the ship data volume trained in the unit period is greater than the preset second training data volume difference;

[0018] The preset first training data amount difference is smaller than the preset second training data amount difference.

[0019] Furthermore, the central control module determines whether the degree of fit is within the allowable range in three ways according to the average value of the difference between the training output value and the target value, wherein:

[0020] The first type of fitting degree determination method is that the central control module determines that the fitting degree is within an allowable range under a preset first average value condition;

[0021] The second type of fitting degree determination method is that the central control module determines that the fitting degree is lower than the allowable range under the preset second average value condition, and adjusts the number of samples to the corresponding number of samples by calculating the difference between the training output deviation average value and the preset first deviation average value;

[0022] The third type of fitting degree determination method is that the central control module determines that the fitting degree is lower than the allowable range under the preset third average value condition, preliminarily determines that there is incomplete data in the acquired ship data, and counts the average number of feature points of the ship data to be trained to perform a secondary determination on whether there is incomplete data in the acquired ship data;

[0023] Among them, the preset first average value condition is that the training output deviation average value is less than or equal to the preset first deviation average value; the preset second average value condition is that the training output deviation average value is greater than the preset first deviation average value and less than or equal to the preset second deviation average value; the preset third average value condition is that the training output deviation average value is greater than the preset second deviation average value;

[0024] The preset first deviation average value is smaller than the preset second deviation average value, and the training output deviation average value is the average value of the difference between the training output value and the target value.

[0025] Furthermore, the central control module determines three types of adjustment methods for the number of samples according to the difference between the training output deviation average value and the preset first deviation average value under the preset second average value condition, wherein:

[0026] The first type of sample quantity adjustment method is that the central control module adjusts the sample quantity to a preset sample quantity under a preset first deviation average value difference condition;

[0027] The second type of sample quantity adjustment method is that the central control module uses a preset first sample quantity adjustment coefficient to increase the sample quantity to the first quantity under a preset second deviation average difference condition;

[0028] The third type of sample quantity adjustment method is that the central control module uses a preset second sample quantity adjustment coefficient to increase the sample quantity to a second quantity under a preset third deviation average difference condition;

[0029] Among them, the preset first deviation average difference condition is that the difference between the training output deviation average and the preset first deviation average is less than or equal to the preset first deviation average difference; the preset second deviation average difference condition is that the difference between the training output deviation average and the preset first deviation average is greater than the preset first deviation average difference and less than or equal to the preset second deviation average difference; the preset third deviation average difference condition is that the difference between the training output deviation average and the preset first deviation average is greater than the preset second deviation average difference;

[0030] Among them, the preset first deviation average value difference is smaller than the preset second deviation average value difference, and the preset first sample quantity adjustment coefficient is smaller than the preset second sample quantity adjustment coefficient.

[0031] Furthermore, the central control module determines whether there is incomplete data in the acquired ship data according to the average number of feature points of the ship data to be trained under the preset third average value condition in three types of secondary determination methods, wherein:

[0032] The first type of complete data secondary determination method is that the central control module determines that there is an error in the ship data type under the preset first average feature point quantity condition and issues a notification for changing the ship data type;

[0033] The second type of complete data secondary determination method is that the central control module secondary determines that there is incomplete data in the acquired ship data under the preset second average feature point number condition, and calculates the difference between the average feature point number of the ship data to be trained and the preset first average feature point number to adjust the collection reference capacity of the ship data to the corresponding reference capacity;

[0034] The third type of complete data secondary determination method is that the central control module secondary determines that there is no incomplete data in the acquired ship data under the preset third average feature point quantity condition;

[0035] Among them, the preset first average feature point number condition is that the average feature point number of the ship data to be trained is less than or equal to the preset first average feature point number; the preset second average feature point number condition is that the average feature point number of the ship data to be trained is greater than the preset first average feature point number and less than or equal to the preset second average feature point number; the preset third average feature point number condition is that the average feature point number of the ship data to be trained is greater than the preset second average feature point number;

[0036] The preset first average number of feature points is smaller than the preset second average number of feature points.

[0037] Furthermore, the central control module determines three types of adjustment methods for the collection benchmark capacity of ship data under the preset second average feature point number condition according to the difference between the average feature point number of the ship data to be trained and the preset first average feature point number, wherein:

[0038] The first type of capacity adjustment mode is that the central control module determines not to adjust the collection base capacity of the ship data under the preset first average feature point quantity difference condition;

[0039] The second type of capacity adjustment method is that the central control module uses a preset first reference capacity adjustment coefficient to adjust the collection reference capacity of the ship data to the first reference capacity under a preset second average feature point quantity difference condition;

[0040] The third type of capacity adjustment method is that the central control module uses a preset second reference capacity adjustment coefficient to adjust the collection reference capacity of the ship data to the second reference capacity under a preset third average feature point quantity difference condition;

[0041] Among them, the preset first average feature point number difference condition is that the difference between the average number of feature points of the ship data to be trained and the preset first average feature point number is less than or equal to the preset first average feature point number difference; the preset second average feature point number difference condition is that the difference between the average number of feature points of the ship data to be trained and the preset first average feature point number is greater than the preset first average feature point number difference and less than or equal to the preset second average feature point number difference; the preset third average feature point number difference condition is that the difference between the average number of feature points of the ship data to be trained and the preset first average feature point number is greater than the preset second average feature point number difference; wherein, the preset first average feature point number difference is less than the preset second average feature point number difference, and the preset first benchmark capacity adjustment coefficient is less than the preset second benchmark capacity adjustment coefficient.

[0042] Furthermore, when the central control module completes the adjustment of the collection reference capacity of the ship data, it determines whether the current ship data is weakly correlated category data according to the proportion of the number of feature points in the single ship data in three types of determination methods, wherein:

[0043] The first type of weakly correlated data determination method is that the central control module determines that the current ship data is weakly correlated category data under a preset first weight condition, determines that there is a first-level training interference risk, and stores the ship data in the ship database;

[0044] The second type of weakly correlated data determination method is that the central control module determines that the current ship data is weakly correlated category data under a preset second weight condition, and adjusts the ship data acquisition frequency to the corresponding frequency by calculating the difference between the ratio of the number of feature points to the single ship data and the preset first weight;

[0045] The third type of weakly correlated data determination method is that the central control module determines that the current ship data is non-weakly correlated category data under a preset third weight condition and sends the ship data to the ship data training module;

[0046] Among them, the preset first proportion condition is that the proportion of the number of feature points in the data of a single ship is less than or equal to the preset first proportion; the preset second proportion condition is that the proportion of the number of feature points in the data of a single ship is greater than the preset first proportion and less than or equal to the preset second proportion; the preset third proportion condition is that the proportion of the number of feature points in the data of a single ship is greater than the preset second proportion; wherein the preset first proportion is less than the preset second proportion.

[0047] Furthermore, the central control module determines three types of adjustment methods for the frequency of obtaining ship data according to the difference between the proportion of the number of feature points in the single ship data and the preset first proportion under the preset second proportion condition, wherein:

[0048] The first frequency adjustment method is that the central control module adjusts the ship data acquisition frequency to a preset ship data acquisition frequency under a preset first specific gravity difference condition;

[0049] The second frequency adjustment method is that the central control module uses a preset first acquisition frequency adjustment coefficient to adjust the ship data acquisition frequency to the first frequency under a preset second specific gravity difference condition;

[0050] The third frequency adjustment method is that the central control module uses a preset second acquisition frequency adjustment coefficient to adjust the ship data acquisition frequency to the second frequency under a preset third specific gravity difference condition;

[0051] Among them, the preset first proportion difference condition is that the difference between the proportion of the number of feature points in the data of a single ship and the preset first proportion is less than or equal to the preset first proportion difference; the preset second proportion difference condition is that the difference between the proportion of the number of feature points in the data of a single ship and the preset first proportion is greater than the preset first proportion difference and less than or equal to the preset second proportion difference; the preset third proportion difference condition is that the difference between the proportion of the number of feature points in the data of a single ship and the preset first proportion is greater than the preset second proportion difference; wherein, the preset first proportion difference is less than the preset second proportion difference, and the preset first acquisition frequency adjustment coefficient is less than the preset second acquisition frequency adjustment coefficient.

[0052] Compared with the prior art, the beneficial effect of the present invention lies in that, by setting steps S1 to S4, the method of the present invention adjusts the training quantity ratio to the corresponding training quantity ratio according to the amount of ship data trained in a unit period, thereby reducing the impact of the decline in ship data training accuracy caused by the unreasonable training quantity ratio and test ship data ratio reflected by the different amounts of ship data trained in a unit period; by adjusting the number of samples to the corresponding number of samples according to the average value of the training output deviation, the impact of the low sample fit caused by the excessive difference between the sample and the target value reflected by the deviation of the training output from the average value is reduced; by adjusting the collection benchmark capacity of the ship data to the corresponding benchmark capacity according to the average number of feature points of the ship data to be trained, the impact of the excessive amount of incomplete data obtained on the accuracy of ship data training reflected by the average number of feature points of the ship data to be trained is reduced, thereby achieving an improvement in the accuracy of ship data training and the stability of ship safety assessment.

[0053] Furthermore, the method of the present invention reduces the impact of decreased training accuracy due to inaccurate detection of training speed by setting a preset training data amount and using two types of judgment methods to determine whether the training speed is within an allowable range based on the amount of ship data trained within a unit period, thereby further improving the accuracy of ship data training and the stability of ship safety assessment.

[0054] Furthermore, the method of the present invention sets a preset first training quantity ratio adjustment coefficient and a preset second training quantity ratio adjustment coefficient, and determines three types of adjustment methods for the training quantity ratio according to the difference between the preset training data volume and the ship data volume trained within a unit period, thereby reducing the impact of the decrease in ship data training accuracy due to inaccurate adjustment of the training quantity ratio, and further achieves improvement in the accuracy of ship data training and the stability of ship safety assessment.

[0055] Furthermore, the method of the present invention reduces the impact of decreased ship data training accuracy caused by inaccurate setting of preset training data amount differences by setting a preset first training data amount difference and a preset second training data amount difference, and further improves the accuracy of ship data training and the stability of ship safety assessment by accurately segmenting the preset training data amount differences.

[0056] Furthermore, the method of the present invention reduces the impact of unclear training effect caused by inaccurate judgment of whether the degree of fitting is within the allowable range by setting a preset first deviation average value and a preset second deviation average value, and by using three types of judgment methods to determine whether the degree of fitting is within the allowable range based on the average value of the difference between the training output value and the target value, thereby further achieving improvement in the accuracy of ship data training and the stability of ship safety assessment.

[0057] Furthermore, the method of the present invention reduces the impact of poor fitting effect of ship data training results caused by inaccurate adjustment of sample quantity by setting a preset first deviation mean difference, a preset second deviation mean difference, a preset first sample quantity adjustment coefficient and a preset second sample quantity adjustment coefficient, and determines three types of adjustment methods for sample quantity according to the difference between the training output deviation mean and the preset first deviation mean, thereby further achieving improvement in the accuracy of ship data training and the stability of ship safety assessment.

[0058] Furthermore, the method of the present invention reduces the impact of decreased accuracy of the ship data training process due to inaccurate screening of incomplete data by setting a preset first average number of feature points and a preset second average number of feature points, and by using three types of secondary determination methods to determine whether there is incomplete data in the acquired ship data according to the average number of feature points of the ship data to be trained, thereby further achieving improvements in the accuracy of ship data training and the stability of ship safety assessment.

[0059] Furthermore, the method of the present invention sets a preset first average feature point number difference, a preset second average feature point number difference, a preset first benchmark capacity adjustment coefficient, and a preset second benchmark capacity adjustment coefficient, and determines three types of adjustment methods for the collection benchmark capacity of ship data according to the difference between the average number of feature points of the ship data to be trained and the preset first average feature point number. This reduces the impact of the decrease in the accuracy of ship data training caused by the standard amount of ship data acquisition not conforming to the actual situation due to the inaccurate adjustment of the collection benchmark capacity of ship data, thereby further achieving an improvement in the accuracy of ship data training and the stability of ship safety assessment.

[0060] Furthermore, the method of the present invention reduces the impact of reduced accuracy of ship data training caused by the difference in correlation between ship data reflected by the proportion of the number of feature points to single ship data and target training ship data by setting a preset first proportion and a preset second proportion, and by using three types of judgment methods to determine whether the current ship data is weakly correlated category data based on the proportion of the number of feature points to single ship data when completing the adjustment of the collection benchmark capacity of the ship data. This further achieves an improvement in the accuracy of ship data training and the stability of ship safety assessment.

[0061] Furthermore, the device of the present invention sets a preset first proportion difference, a preset second proportion difference, a preset first acquisition frequency adjustment coefficient and a preset second acquisition frequency adjustment coefficient, and determines three types of adjustment methods for the ship data acquisition frequency according to the difference between the proportion of the number of feature points in a single ship data and the preset first proportion, thereby reducing the reduction in the continuity of ship data training caused by inaccurate adjustment of the ship data acquisition frequency, and further achieving an improvement in the accuracy of ship data training and the stability of ship safety assessment. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 This is an overall flow chart of a safety assessment method for ship offshore operations according to an embodiment of the present invention;

[0063] Figure 2 This is a specific flow chart of step S4 of the safety assessment method for ship offshore operations according to an embodiment of the present invention;

[0064] Figure 3 This is a specific flow chart of step S5 of the method for safety assessment of ship offshore operations according to an embodiment of the present invention. DETAILED DESCRIPTION

[0065] In order to make the objects and advantages of the present invention more clearly understood, the present invention is further described below in conjunction with embodiments; it should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0066] The preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood by those skilled in the art that these embodiments are only used to explain the technical principles of the present invention and are not intended to limit the protection scope of the present invention.

[0067] See also Figure 1 , Figure 2 as well as Figure 3 As shown, they are respectively an overall flow chart of a safety assessment method for ship offshore operations according to an embodiment of the present invention, a specific flow chart of step S4, and a specific flow chart of step S5; a safety assessment method for ship offshore operations according to the present invention comprises:

[0068] Step S1, acquiring ship data, and identifying characteristic fields and characteristic field types in the ship data when the acquisition is completed;

[0069] Step S2, training the ship data to output a number of ship feature data sets, and the central control module outputs the training model according to the number of ship feature data sets;

[0070] Step S3, when the output of the training model is completed, the central control module adjusts the training quantity ratio to the corresponding training quantity ratio according to the amount of ship data trained in the unit period;

[0071] Step S4, when the adjustment of the training quantity ratio is completed, the central control module determines whether the degree of fit is within the allowable range according to the average value of the difference between the training output value and the target value, and adjusts the number of samples to the corresponding number of samples when it is determined that the degree of fit is lower than the allowable range, or adjusts the collection base capacity of the ship data to the corresponding base capacity according to the average number of feature points of the ship data to be trained;

[0072] Step S5, when the adjustment of the collection baseline capacity for the ship data is completed, the central control module determines whether the current ship data is weakly correlated category data based on the proportion of the number of feature points to the single ship data, and adjusts the ship data acquisition frequency to the corresponding frequency when it is determined that the current ship data is weakly correlated category data.

[0073] Specifically, step S4 includes:

[0074] Step S41, when the adjustment of the training quantity ratio is completed, the central control module determines whether the fitting degree is within the allowable range according to the average value of the difference between the training output value and the target value;

[0075] Step S42, when the central control module determines that the degree of fit is lower than the allowable range, the central control module adjusts the number of samples to the corresponding number of samples, or adjusts the collection base capacity of the ship data to the corresponding base capacity according to the average number of feature points of the ship data to be trained.

[0076] Specifically, step S5 includes:

[0077] Step S51, when the adjustment of the collection reference capacity of the ship data is completed, the central control module determines whether the current ship data is weakly correlated category data according to the proportion of the number of feature points to the single ship data;

[0078] Step S52, when the central control module determines that the current ship data is weakly correlated category data, the ship data acquisition frequency is adjusted to a corresponding frequency according to the difference between the proportion of the number of feature points in the single ship data and the preset first proportion.

[0079] Specifically, the ship data includes the ship's driving environment parameters, the real-time distance to the target point, the ship's operating parameters and the ship's fault type data.

[0080] Specifically, the fault type data of the ship includes the historical fault type data of the ship and the fault characteristic parameters of the ship. The fault characteristic parameters of the ship include the target operating parameters corresponding to the situation where the corresponding equipment or the corresponding system has no output, the operating parameters corresponding to the output of the equipment or system is greater than or less than the standard parameter indicators, the target operating parameters corresponding to the reverse output process, the operating parameters corresponding to the change in the output of the equipment or system, and the target operating parameters corresponding to the output of the equipment or system is wrong. The target parameters include the driving environment parameters of the ship, the real-time distance from the target point, and the operating parameters of the ship.

[0081] Specifically, the method of the present invention sets steps S1 to S5, and adjusts the specific training parameters in the ship data training process accordingly. When the adjustment is completed, the ship safety training model output by the training process is used to evaluate the safety of the ship's maritime operations. The training model or ship safety training module is a training model output by artificial intelligence technology, which is a conventional evaluation technical means well known to technical personnel in this field. Therefore, the construction of the ship safety training model and its use process will not be repeated here.

[0082] The method of the present invention reduces the influence of the decline in the accuracy of ship data training caused by the unreasonable training quantity ratio and test ship data ratio reflected by the different amounts of ship data trained in the unit period by adjusting the training quantity ratio to the corresponding training quantity ratio according to the set steps S1 to S4; reduces the influence of the low sample fit caused by the excessive difference between the sample and the target value reflected by the deviation of the training output from the average value by adjusting the sample quantity to the corresponding sample quantity according to the average value of the training output deviation; reduces the influence of the ship data collection benchmark capacity to the corresponding benchmark capacity according to the average number of feature points of the ship data to be trained on the accuracy of ship data training due to the excessive amount of incomplete data obtained reflected by the average number of feature points of the ship data to be trained, thereby achieving the improvement of the accuracy of ship data training and the stability of ship safety assessment.

[0083] Please continue reading Figure 1 As shown, the central control module determines whether the training speed is within the allowable range in two ways according to the amount of ship data trained in a unit period, wherein:

[0084] The first determination method is that the central control module determines that the training speed is lower than the allowable range under the preset first training data volume condition, and adjusts the training quantity ratio to the corresponding training quantity ratio by calculating the difference between the preset training data volume and the ship data volume trained in the unit period;

[0085] The second determination method is that the central control module determines that the training speed is within the allowable range under the preset second training data amount condition;

[0086] Among them, the preset first training data amount condition is that the amount of ship data trained in a unit period is less than or equal to the preset training data amount; the preset second training data amount condition is that the amount of ship data trained in a unit period is greater than the preset training data amount.

[0087] The method of the present invention reduces the impact of decreased training accuracy due to inaccurate detection of training speed by setting a preset training data amount and by using two types of judgment methods to determine whether the training speed is within an allowable range based on the amount of ship data trained in a unit period, and further improves the accuracy of ship data training and the stability of ship safety assessment.

[0088] Please continue reading Figure 1 and Figure 2 As shown, the central control module determines three types of adjustment methods for the proportion of training quantity according to the difference between the preset training data volume and the ship data volume trained in the unit period under the preset first training data volume condition, wherein:

[0089] The first type of adjustment method is that the central control module adjusts the training quantity ratio to a preset training quantity ratio under a preset first training data quantity difference condition;

[0090] The second type of adjustment method is that the central control module uses a preset first training quantity ratio adjustment coefficient to increase the training quantity ratio to the first ratio under a preset second training data quantity difference condition;

[0091] The third type of adjustment method is that the central control module uses a preset second training quantity ratio adjustment coefficient to increase the training quantity ratio to the second ratio under a preset third training data quantity difference condition;

[0092] The preset first training quantity ratio adjustment coefficient is smaller than the preset second training quantity ratio adjustment coefficient.

[0093] Specifically, the amount of ship data trained within a unit period is recorded as Q, the preset training data amount is recorded as Q0, the preset first training quantity ratio adjustment coefficient is recorded as α1, the preset second training quantity ratio adjustment coefficient is recorded as α2, the difference between the preset training data amount and the amount of ship data trained within the unit period is recorded as △Q, the preset first training data amount difference is recorded as △Q1, the preset second training data amount difference is recorded as △Q2, the preset training quantity ratio is recorded as A0, wherein 1<α1<α2, △Q1<△Q2, the adjusted training quantity ratio is recorded as A', set A'=A0×αi, wherein αi is the preset i-th training quantity ratio adjustment coefficient, and i=1,2.

[0094] Specifically, the proportion of training quantity means the proportion of the ship data in the sample set used for the ship data training process in the sample set. The ship data in the sample set is divided into two parts, one part is used for the ship data training process, and the other part is used for the ship data testing process.

[0095] The method of the present invention sets a preset first training quantity ratio adjustment coefficient and a preset second training quantity ratio adjustment coefficient, and determines three types of adjustment methods for the training quantity ratio according to the difference between the preset training data volume and the ship data volume trained in a unit period, thereby reducing the impact of the decrease in ship data training accuracy caused by inaccurate adjustment of the training quantity ratio, and further achieves the improvement of the accuracy of ship data training and the stability of ship safety assessment.

[0096] Please continue reading Figure 1 As shown, the preset first training data volume difference condition is that the difference between the preset training data volume and the ship data volume trained in the unit period is less than or equal to the preset first training data volume difference;

[0097] The preset second training data volume difference condition is that the difference between the preset training data volume and the ship data volume trained in the unit period is greater than the preset first training data volume difference and less than or equal to the preset second training data volume difference;

[0098] The preset third training data volume difference condition is that the difference between the preset training data volume and the ship data volume trained in the unit period is greater than the preset second training data volume difference;

[0099] The preset first training data amount difference is smaller than the preset second training data amount difference.

[0100] The method of the present invention reduces the impact of decreased accuracy of ship data training due to inaccurate setting of the preset training data amount difference by setting a preset first training data amount difference and a preset second training data amount difference, and further improves the accuracy of ship data training and the stability of ship safety assessment by accurately segmenting the preset training data amount difference.

[0101] Please continue reading Figure 1 As shown, the central control module determines whether the degree of fit is within the allowable range in three ways according to the average value of the difference between the training output value and the target value, wherein:

[0102] The first type of fitting degree determination method is that the central control module determines that the fitting degree is within an allowable range under a preset first average value condition;

[0103] The second type of fitting degree determination method is that the central control module determines that the fitting degree is lower than the allowable range under the preset second average value condition, and adjusts the number of samples to the corresponding number of samples by calculating the difference between the training output deviation average value and the preset first deviation average value;

[0104] The third type of fitting degree determination method is that the central control module determines that the fitting degree is lower than the allowable range under the preset third average value condition, preliminarily determines that there is incomplete data in the acquired ship data, and counts the average number of feature points of the ship data to be trained to perform a secondary determination on whether there is incomplete data in the acquired ship data;

[0105] Among them, the preset first average value condition is that the training output deviation average value is less than or equal to the preset first deviation average value; the preset second average value condition is that the training output deviation average value is greater than the preset first deviation average value and less than or equal to the preset second deviation average value; the preset third average value condition is that the training output deviation average value is greater than the preset second deviation average value;

[0106] The preset first deviation average value is smaller than the preset second deviation average value, and the training output deviation average value is the average value of the difference between the training output value and the target value.

[0107] The method of the present invention reduces the influence of unclear training effect caused by inaccurate judgment of whether the degree of fitting is within the allowable range by setting a preset first deviation average value and a preset second deviation average value, and by using three types of judgment methods to determine whether the degree of fitting is within the allowable range according to the average value of the difference between the training output value and the target value, and further achieves the improvement of the accuracy of ship data training and the stability of ship safety assessment.

[0108] Please continue reading Figure 1 As shown, the central control module determines three types of adjustment methods for the number of samples according to the difference between the training output deviation average value and the preset first deviation average value under the preset second average value condition, wherein:

[0109] The first type of sample quantity adjustment method is that the central control module adjusts the sample quantity to a preset sample quantity under a preset first deviation average value difference condition;

[0110] The second type of sample quantity adjustment method is that the central control module uses a preset first sample quantity adjustment coefficient to increase the sample quantity to the first quantity under a preset second deviation average difference condition;

[0111] The third type of sample quantity adjustment method is that the central control module uses a preset second sample quantity adjustment coefficient to increase the sample quantity to a second quantity under a preset third deviation average difference condition;

[0112] Among them, the preset first deviation average difference condition is that the difference between the training output deviation average and the preset first deviation average is less than or equal to the preset first deviation average difference; the preset second deviation average difference condition is that the difference between the training output deviation average and the preset first deviation average is greater than the preset first deviation average difference and less than or equal to the preset second deviation average difference; the preset third deviation average difference condition is that the difference between the training output deviation average and the preset first deviation average is greater than the preset second deviation average difference;

[0113] Among them, the preset first deviation average value difference is smaller than the preset second deviation average value difference, and the preset first sample quantity adjustment coefficient is smaller than the preset second sample quantity adjustment coefficient.

[0114] Specifically, the training output deviation average is recorded as W, the preset first deviation average is recorded as W1, the preset second deviation average is recorded as W2, the preset first deviation average difference is recorded as △W1, the preset second deviation average difference is recorded as △W2, the preset number of samples is recorded as E0, the difference between the training output deviation average and the preset first deviation average is recorded as △W, the preset first sample number adjustment coefficient is recorded as β1, and the preset second sample number adjustment coefficient is recorded as β2, wherein 1<β1<β2, W1<W2, △W1<△W2, the adjusted number of samples is recorded as E', and E'=E0×(1+βj) / 2 is set, wherein βj is the preset j-th sample number adjustment coefficient, and j=1,2.

[0115] Specifically, the average training output deviation refers to the average value of the difference between several trained ship data output by the ship data training module and the standard output ship data, wherein the ship data is numerical ship data, and the number of samples is the number of ship data included in the sample set. By setting the adjustment of the number of samples, when the training deviation does not meet the requirements, increasing the number of samples can increase the accuracy of the training model, thereby improving the accuracy of ship data training.

[0116] The method of the present invention sets a preset first deviation mean difference, a preset second deviation mean difference, a preset first sample quantity adjustment coefficient and a preset second sample quantity adjustment coefficient, and determines three types of adjustment methods for the sample quantity according to the difference between the training output deviation mean and the preset first deviation mean. This reduces the impact of poor fitting effect of ship data training results caused by inaccurate adjustment of the sample quantity, and further improves the accuracy of ship data training and the stability of ship safety assessment.

[0117] Please continue reading Figure 1 As shown, the central control module determines whether there is incomplete data in the acquired ship data according to the average number of feature points of the ship data to be trained under the preset third average value condition, in which:

[0118] The first type of complete data secondary determination method is that the central control module determines that there is an error in the ship data type under the preset first average feature point quantity condition and issues a notification for changing the ship data type;

[0119] The second type of complete data secondary determination method is that the central control module secondary determines that there is incomplete data in the acquired ship data under the preset second average feature point number condition, and calculates the difference between the average feature point number of the ship data to be trained and the preset first average feature point number to adjust the collection reference capacity of the ship data to the corresponding reference capacity;

[0120] The third type of complete data secondary determination method is that the central control module secondary determines that there is no incomplete data in the acquired ship data under the preset third average feature point quantity condition;

[0121] Among them, the preset first average feature point number condition is that the average feature point number of the ship data to be trained is less than or equal to the preset first average feature point number; the preset second average feature point number condition is that the average feature point number of the ship data to be trained is greater than the preset first average feature point number and less than or equal to the preset second average feature point number; the preset third average feature point number condition is that the average feature point number of the ship data to be trained is greater than the preset second average feature point number;

[0122] The preset first average number of feature points is smaller than the preset second average number of feature points.

[0123] Specifically, the average number of feature points of the ship data to be trained is the average value of the number of feature points of several unit ship data to be trained, and the unit ship data to be trained is the basic unit for dividing the ship data to be trained. For example, the ship data to be trained is divided into units of 5kb in size to divide the ship data to be trained into several data segments of 5kb in size.

[0124] Specifically, the feature point is a feature field of ship data contained in a segment of data, and the feature field of ship data includes the ship's heading, the ship's operating voltage, the ship's speed, and the ship's positioning coordinates.

[0125] The method of the present invention sets a preset first average number of feature points and a preset second average number of feature points, and uses three types of secondary determination methods to determine whether there is incomplete data in the acquired ship data according to the average number of feature points of the ship data to be trained. This reduces the impact of the decrease in the accuracy of the ship data training process due to inaccurate screening of incomplete data, and further improves the accuracy of ship data training and the stability of ship safety assessment.

[0126] Please continue reading Figure 1 and Figure 3 As shown, the central control module determines three types of adjustment methods for the acquisition benchmark capacity of ship data according to the difference between the average number of feature points of the ship data to be trained and the preset first average number of feature points under the preset second average number of feature points condition, wherein:

[0127] The first type of capacity adjustment mode is that the central control module determines not to adjust the collection base capacity of the ship data under the preset first average feature point quantity difference condition;

[0128] The second type of capacity adjustment method is that the central control module uses a preset first reference capacity adjustment coefficient to adjust the collection reference capacity of the ship data to the first reference capacity under a preset second average feature point quantity difference condition;

[0129] The third type of capacity adjustment method is that the central control module uses a preset second reference capacity adjustment coefficient to adjust the collection reference capacity of the ship data to the second reference capacity under a preset third average feature point quantity difference condition;

[0130] Among them, the preset first average feature point number difference condition is that the difference between the average number of feature points of the ship data to be trained and the preset first average feature point number is less than or equal to the preset first average feature point number difference; the preset second average feature point number difference condition is that the difference between the average number of feature points of the ship data to be trained and the preset first average feature point number is greater than the preset first average feature point number difference and less than or equal to the preset second average feature point number difference; the preset third average feature point number difference condition is that the difference between the average number of feature points of the ship data to be trained and the preset first average feature point number is greater than the preset second average feature point number difference; wherein, the preset first average feature point number difference is less than the preset second average feature point number difference, and the preset first benchmark capacity adjustment coefficient is less than the preset second benchmark capacity adjustment coefficient.

[0131] Specifically, the average number of feature points of the ship data to be trained is recorded as S, the preset first average number of feature points is recorded as S1, the preset second average number of feature points is recorded as S2, the preset first average number of feature points difference is recorded as △S1, the preset second average number of feature points difference is recorded as △S2, the preset first benchmark capacity adjustment coefficient is recorded as γ1, the preset second benchmark capacity adjustment coefficient is recorded as γ2, the collection benchmark capacity of ship data is recorded as R0, wherein 1<γ1<γ2, △S1<△S2, the adjusted collection benchmark capacity of ship data is recorded as R', and R'=R0×γk is set, wherein γk is the preset kth benchmark capacity adjustment coefficient, and k=1,2.

[0132] Specifically, the collection benchmark capacity of ship data is the benchmark ship data capacity of the data acquisition module. Since ship data is incomplete, the benchmark ship data capacity is increased. When acquiring ship data, more ship data is acquired while reducing the proportion of incomplete data acquisition, so as to improve the accuracy of ship data training.

[0133] The method of the present invention sets a preset first average feature point number difference, a preset second average feature point number difference, a preset first benchmark capacity adjustment coefficient and a preset second benchmark capacity adjustment coefficient, and determines three types of adjustment methods for the collection benchmark capacity of ship data according to the difference between the average feature point number of the ship data to be trained and the preset first average feature point number, thereby reducing the impact of the decrease in the accuracy of ship data training caused by the standard amount of ship data acquisition not meeting the actual situation due to the inaccurate adjustment of the collection benchmark capacity of ship data, and further achieves the improvement of the accuracy of ship data training and the stability of ship safety assessment.

[0134] Please continue reading Figure 1 As shown, when the central control module completes the adjustment of the acquisition reference capacity of the ship data, it determines whether the current ship data is weakly correlated category data according to the proportion of the number of feature points in the single ship data in three types of judgment methods, among which:

[0135] The first type of weakly correlated data determination method is that the central control module determines that the current ship data is weakly correlated category data under a preset first weight condition, determines that there is a first-level training interference risk, and stores the ship data in the ship database;

[0136] The second type of weakly correlated data determination method is that the central control module determines that the current ship data is weakly correlated category data under a preset second weight condition, and adjusts the ship data acquisition frequency to the corresponding frequency by calculating the difference between the ratio of the number of feature points to the single ship data and the preset first weight;

[0137] The third type of weakly correlated data determination method is that the central control module determines that the current ship data is non-weakly correlated category data under a preset third weight condition and sends the ship data to the ship data training module;

[0138] Among them, the preset first proportion condition is that the proportion of the number of feature points in the data of a single ship is less than or equal to the preset first proportion; the preset second proportion condition is that the proportion of the number of feature points in the data of a single ship is greater than the preset first proportion and less than or equal to the preset second proportion; the preset third proportion condition is that the proportion of the number of feature points in the data of a single ship is greater than the preset second proportion; wherein the preset first proportion is less than the preset second proportion.

[0139] The method of the present invention reduces the impact of reduced accuracy of ship data training caused by the difference in correlation between ship data reflected by the proportion of the number of feature points to the single ship data and target training ship data, by setting a preset first proportion and a preset second proportion, and by using three types of judgment methods to determine whether the current ship data is weakly correlated category data according to the proportion of the number of feature points to the single ship data when completing the adjustment of the collection benchmark capacity of the ship data, thereby further achieving improved accuracy of ship data training and stability of ship safety assessment.

[0140] Please continue reading Figure 1 As shown, the central control module determines three types of adjustment methods for the frequency of obtaining ship data according to the difference between the proportion of the number of feature points in the single ship data and the preset first proportion under the preset second proportion condition, wherein:

[0141] The first frequency adjustment method is that the central control module adjusts the ship data acquisition frequency to a preset ship data acquisition frequency under a preset first specific gravity difference condition;

[0142] The second frequency adjustment method is that the central control module uses a preset first acquisition frequency adjustment coefficient to adjust the ship data acquisition frequency to the first frequency under a preset second specific gravity difference condition;

[0143] The third frequency adjustment method is that the central control module uses a preset second acquisition frequency adjustment coefficient to adjust the ship data acquisition frequency to the second frequency under a preset third specific gravity difference condition;

[0144] Among them, the preset first proportion difference condition is that the difference between the proportion of the number of feature points in the data of a single ship and the preset first proportion is less than or equal to the preset first proportion difference; the preset second proportion difference condition is that the difference between the proportion of the number of feature points in the data of a single ship and the preset first proportion is greater than the preset first proportion difference and less than or equal to the preset second proportion difference; the preset third proportion difference condition is that the difference between the proportion of the number of feature points in the data of a single ship and the preset first proportion is greater than the preset second proportion difference; wherein, the preset first proportion difference is less than the preset second proportion difference, and the preset first acquisition frequency adjustment coefficient is less than the preset second acquisition frequency adjustment coefficient.

[0145] Specifically, the proportion of the number of feature points to the data of a single ship is recorded as D, the preset first proportion is recorded as D1, the preset second proportion is recorded as D2, the preset first proportion difference is recorded as △D1, the preset second proportion difference is recorded as △D2, the preset first acquisition frequency adjustment coefficient is recorded as g1, the preset second acquisition frequency adjustment coefficient is recorded as g2, and the preset ship data acquisition frequency is recorded as F0, wherein 1<g1<g2, D1<D2, △D1<△D2, the adjusted ship data acquisition frequency is recorded as F', and F'=F0×(1+gc) / 2 is set, wherein gc is the preset cth acquisition frequency adjustment coefficient, and c=1,2.

[0146] Specifically, the ship data acquisition frequency is the frequency at which the data acquisition module acquires ship data. When it is determined to be weakly correlated ship data, the ship data acquisition frequency is increased to improve the processing efficiency of weakly correlated ship data, thereby improving the efficiency and accuracy of ship data training.

[0147] The device of the present invention sets a preset first proportion difference, a preset second proportion difference, a preset first acquisition frequency adjustment coefficient and a preset second acquisition frequency adjustment coefficient, and determines three types of adjustment methods for the ship data acquisition frequency according to the difference between the proportion of the number of feature points in a single ship data and the preset first proportion, thereby reducing the reduction in the continuity of ship data training caused by inaccurate adjustment of the ship data acquisition frequency, and further achieving an improvement in the accuracy of ship data training and the stability of ship safety assessment. Example

[0148] The central control module described in this embodiment 1 determines three types of adjustment methods for the training quantity ratio according to the difference between the preset training data volume and the ship data volume trained in the unit period under the preset second training data volume condition, wherein the preset first training quantity ratio adjustment coefficient is recorded as α1, the preset second training quantity ratio adjustment coefficient is recorded as α2, the difference between the preset training data volume and the ship data volume trained in the unit period is recorded as △Q, the preset first training data volume difference is recorded as △Q1, the preset second training data volume difference is recorded as △Q2, and the preset training quantity ratio is recorded as A0, wherein α1=1.1, α2=1.15, △Q1=240bit, △Q2=480bit, A0=0.8,

[0149] In this embodiment, △Q=300bit is obtained, the central control module determines △Q1<△Q≤△Q2 and uses the preset first training quantity ratio adjustment coefficient α1 to increase the training quantity ratio to the first ratio, the first ratio is recorded as A', and A'=0.8×1.1=0.88 is set.

[0150] The system described in this embodiment 1 sets a preset first training quantity ratio adjustment coefficient and a preset second training quantity ratio adjustment coefficient, and determines three types of adjustment methods for the training quantity ratio according to the amount of ship data trained in a unit period, thereby reducing the impact of the slow ship data training caused by the untimely adjustment of the training quantity ratio and the resulting reduction in accuracy, and further achieves the improvement of the accuracy of ship data training and the stability of ship safety assessment. Example

[0151] The central control module described in this embodiment 2 determines three types of adjustment methods for the training quantity ratio according to the difference between the preset training data volume and the ship data volume trained in the unit period under the preset second training data volume condition, wherein the preset first training quantity ratio adjustment coefficient is recorded as α1, the preset second training quantity ratio adjustment coefficient is recorded as α2, the difference between the preset training data volume and the ship data volume trained in the unit period is recorded as △Q, the preset first training data volume difference is recorded as △Q1, the preset second training data volume difference is recorded as △Q2, and the preset training quantity ratio is recorded as A0, wherein α1=1.1, α2=1.15, △Q1=240bit, △Q2=480bit, A0=0.8,

[0152] In this embodiment, △Q=500bit is obtained, the central control module determines that △Q>△Q2 and uses the preset second training quantity ratio adjustment coefficient α2 to increase the training quantity ratio to the second ratio, the second ratio is recorded as A', and A'=0.8×1.15=0.92 is set.

[0153] The system described in this embodiment sets a preset first training quantity ratio adjustment coefficient and a preset second training quantity ratio adjustment coefficient, and adjusts the training quantity ratio according to the difference between the preset training data volume and the ship data volume trained in the unit period, thereby reducing the impact of the decline in ship data training accuracy due to inaccurate adjustment of the training quantity ratio, and further achieves improvement in the accuracy of ship data training and the stability of ship safety assessment.

[0154] So far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it is easy for those skilled in the art to understand that the protection scope of the present invention is obviously not limited to these specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will fall within the protection scope of the present invention.

[0155] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A safety assessment method for ship offshore operations, characterized in that: include: Step S1, acquiring ship data, and identifying characteristic fields in the ship data respectively when the acquisition is completed; The characteristic point is a characteristic field of the ship data contained in a piece of data. The characteristic field of the ship data includes the ship's heading, the ship's operating voltage, the ship's speed and the ship's positioning coordinates; Step S2, training the ship data to output a number of ship feature data sets, and the central control module outputs the training model according to the number of ship feature data sets; by correspondingly adjusting the specific training parameters in the ship data training process, when the adjustment is completed, using the ship safety training model output by the training process to evaluate the safety of the ship's marine operations; Step S3, when the output of the training model is completed, the central control module adjusts the training quantity ratio to the corresponding training quantity ratio according to the amount of ship data trained in the unit period; The meaning of the training quantity ratio is the quantity ratio of this part of ship data in the sample set used for the ship data training process in the sample set; Step S4, when the adjustment of the training quantity ratio is completed, the central control module determines whether the degree of fit is within the allowable range according to the average value of the difference between the training output value and the target value, and adjusts the number of samples to the corresponding number of samples when it is determined that the degree of fit is lower than the allowable range, or adjusts the collection base capacity of the ship data to the corresponding base capacity according to the average number of feature points of the ship data to be trained; The sample size is the number of ship data included in the sample set; Step S5, when the adjustment of the collection baseline capacity for the ship data is completed, the central control module determines whether the current ship data is weakly correlated category data based on the proportion of the number of feature points to the single ship data, and adjusts the ship data acquisition frequency to the corresponding frequency when it is determined that the current ship data is weakly correlated category data.

2. The safety assessment method for ship offshore operations according to claim 1, characterized in that: In step S3, the central control module determines whether the training speed is within the allowable range according to the amount of ship data trained in a unit period in two ways, wherein: The first determination method is that the central control module determines that the training speed is lower than the allowable range under the preset first training data volume condition, and adjusts the training quantity ratio to the corresponding training quantity ratio by calculating the difference between the preset training data volume and the ship data volume trained in the unit period; The second determination method is that the central control module determines that the training speed is within the allowable range under the preset second training data amount condition; Among them, the preset first training data amount condition is that the amount of ship data trained in a unit period is less than or equal to the preset training data amount; the preset second training data amount condition is that the amount of ship data trained in a unit period is greater than the preset training data amount.

3. The safety assessment method for ship offshore operations according to claim 2, characterized in that: The central control module determines three types of adjustment methods for the proportion of training quantity according to the difference between the preset training data volume and the ship data volume trained in the unit period under the preset first training data volume condition, wherein: The first type of adjustment method is that the central control module adjusts the training quantity ratio to a preset training quantity ratio under a preset first training data quantity difference condition; The second type of adjustment method is that the central control module uses a preset first training quantity ratio adjustment coefficient to increase the training quantity ratio to the first ratio under a preset second training data quantity difference condition; The third type of adjustment method is that the central control module uses a preset second training quantity ratio adjustment coefficient to increase the training quantity ratio to the second ratio under a preset third training data quantity difference condition; The preset first training quantity ratio adjustment coefficient is smaller than the preset second training quantity ratio adjustment coefficient.

4. The safety assessment method for ship offshore operations according to claim 3 is characterized in that: The preset first training data volume difference condition is that the difference between the preset training data volume and the ship data volume trained in the unit period is less than or equal to the preset first training data volume difference; The preset second training data volume difference condition is that the difference between the preset training data volume and the ship data volume trained in the unit period is greater than the preset first training data volume difference and less than or equal to the preset second training data volume difference; The preset third training data volume difference condition is that the difference between the preset training data volume and the ship data volume trained in the unit period is greater than the preset second training data volume difference; The preset first training data amount difference is smaller than the preset second training data amount difference.

5. The safety assessment method for ship offshore operations according to claim 4, characterized in that: The central control module determines whether the degree of fit is within the allowable range in three ways according to the average value of the difference between the training output value and the target value, wherein: The first type of fitting degree determination method is that the central control module determines that the fitting degree is within an allowable range under a preset first average value condition; The second type of fitting degree determination method is that the central control module determines that the fitting degree is lower than the allowable range under the preset second average value condition, and adjusts the number of samples to the corresponding number of samples by calculating the difference between the training output deviation average value and the preset first deviation average value; The third type of fitting degree determination method is that the central control module determines that the fitting degree is lower than the allowable range under the preset third average value condition, preliminarily determines that there is incomplete data in the acquired ship data, and counts the average number of feature points of the ship data to be trained to perform a secondary determination on whether there is incomplete data in the acquired ship data; Among them, the preset first average value condition is that the training output deviation average value is less than or equal to the preset first deviation average value; the preset second average value condition is that the training output deviation average value is greater than the preset first deviation average value and less than or equal to the preset second deviation average value; the preset third average value condition is that the training output deviation average value is greater than the preset second deviation average value; The preset first deviation average value is smaller than the preset second deviation average value, and the training output deviation average value is the average value of the difference between the training output value and the target value.

6. The safety assessment method for ship offshore operations according to claim 5, characterized in that: The central control module determines three types of adjustment methods for the number of samples according to the difference between the training output deviation average value and the preset first deviation average value under the preset second average value condition, wherein: The first type of sample quantity adjustment method is that the central control module adjusts the sample quantity to a preset sample quantity under a preset first deviation average value difference condition; The second type of sample quantity adjustment method is that the central control module uses a preset first sample quantity adjustment coefficient to increase the sample quantity to the first quantity under a preset second deviation average difference condition; The third type of sample quantity adjustment method is that the central control module uses a preset second sample quantity adjustment coefficient to increase the sample quantity to a second quantity under a preset third deviation average difference condition; Among them, the preset first deviation average difference condition is that the difference between the training output deviation average and the preset first deviation average is less than or equal to the preset first deviation average difference; the preset second deviation average difference condition is that the difference between the training output deviation average and the preset first deviation average is greater than the preset first deviation average difference and less than or equal to the preset second deviation average difference; the preset third deviation average difference condition is that the difference between the training output deviation average and the preset first deviation average is greater than the preset second deviation average difference; Among them, the preset first deviation average value difference is smaller than the preset second deviation average value difference, and the preset first sample quantity adjustment coefficient is smaller than the preset second sample quantity adjustment coefficient.

7. The safety assessment method for ship offshore operations according to claim 6, characterized in that: The central control module determines whether there is incomplete data in the acquired ship data according to the average number of feature points of the ship data to be trained under the preset third average value condition, in which: The first type of complete data secondary determination method is that the central control module determines that there is an error in the ship data type under the preset first average feature point quantity condition and issues a notification for changing the ship data type; The second type of complete data secondary determination method is that the central control module secondary determines that there is incomplete data in the acquired ship data under the preset second average feature point number condition, and calculates the difference between the average feature point number of the ship data to be trained and the preset first average feature point number to adjust the collection reference capacity of the ship data to the corresponding reference capacity; The third type of complete data secondary determination method is that the central control module secondary determines that there is no incomplete data in the acquired ship data under the preset third average feature point quantity condition; Among them, the preset first average feature point number condition is that the average feature point number of the ship data to be trained is less than or equal to the preset first average feature point number; the preset second average feature point number condition is that the average feature point number of the ship data to be trained is greater than the preset first average feature point number and less than or equal to the preset second average feature point number; the preset third average feature point number condition is that the average feature point number of the ship data to be trained is greater than the preset second average feature point number; The preset first average number of feature points is smaller than the preset second average number of feature points.

8. The safety assessment method for ship offshore operations according to claim 7, characterized in that: The central control module determines three types of adjustment methods for the collection benchmark capacity of ship data according to the difference between the average number of feature points of the ship data to be trained and the preset first average number of feature points under the preset second average number of feature points condition, wherein: The first type of capacity adjustment mode is that the central control module determines not to adjust the collection base capacity of the ship data under the preset first average feature point quantity difference condition; The second type of capacity adjustment method is that the central control module uses a preset first reference capacity adjustment coefficient to adjust the collection reference capacity of the ship data to the first reference capacity under a preset second average feature point quantity difference condition; The third type of capacity adjustment method is that the central control module uses a preset second reference capacity adjustment coefficient to adjust the collection reference capacity of the ship data to the second reference capacity under a preset third average feature point quantity difference condition; Among them, the preset first average feature point number difference condition is that the difference between the average number of feature points of the ship data to be trained and the preset first average feature point number is less than or equal to the preset first average feature point number difference; the preset second average feature point number difference condition is that the difference between the average number of feature points of the ship data to be trained and the preset first average feature point number is greater than the preset first average feature point number difference and less than or equal to the preset second average feature point number difference; the preset third average feature point number difference condition is that the difference between the average number of feature points of the ship data to be trained and the preset first average feature point number is greater than the preset second average feature point number difference; wherein, the preset first average feature point number difference is less than the preset second average feature point number difference, and the preset first benchmark capacity adjustment coefficient is less than the preset second benchmark capacity adjustment coefficient.

9. The safety assessment method for ship offshore operations according to claim 8, characterized in that: When the central control module completes the adjustment of the collection reference capacity of the ship data, it determines whether the current ship data is weakly correlated category data according to the proportion of the number of feature points in the single ship data in three types of judgment methods, among which: The first type of weakly correlated data determination method is that the central control module determines that the current ship data is weakly correlated category data under a preset first weight condition, determines that there is a first-level training interference risk, and stores the ship data in the ship database; The second type of weakly correlated data determination method is that the central control module determines that the current ship data is weakly correlated category data under a preset second weight condition, and adjusts the ship data acquisition frequency to the corresponding frequency by calculating the difference between the ratio of the number of feature points to the single ship data and the preset first weight; The third type of weakly correlated data determination method is that the central control module determines that the current ship data is non-weakly correlated category data under a preset third weight condition and sends the ship data to the ship data training module; Among them, the preset first proportion condition is that the proportion of the number of feature points in the data of a single ship is less than or equal to the preset first proportion; the preset second proportion condition is that the proportion of the number of feature points in the data of a single ship is greater than the preset first proportion and less than or equal to the preset second proportion; the preset third proportion condition is that the proportion of the number of feature points in the data of a single ship is greater than the preset second proportion; wherein the preset first proportion is less than the preset second proportion.

10. The safety assessment method for ship offshore operations according to claim 9, characterized in that: The central control module determines three types of adjustment methods for the frequency of obtaining ship data according to the difference between the proportion of the number of feature points in the single ship data and the preset first proportion under the preset second proportion condition, wherein: The first frequency adjustment method is that the central control module adjusts the ship data acquisition frequency to a preset ship data acquisition frequency under a preset first specific gravity difference condition; The second frequency adjustment method is that the central control module uses a preset first acquisition frequency adjustment coefficient to adjust the ship data acquisition frequency to the first frequency under a preset second specific gravity difference condition; The third frequency adjustment method is that the central control module uses a preset second acquisition frequency adjustment coefficient to adjust the ship data acquisition frequency to the second frequency under a preset third specific gravity difference condition; Among them, the preset first proportion difference condition is that the difference between the proportion of the number of feature points in the data of a single ship and the preset first proportion is less than or equal to the preset first proportion difference; the preset second proportion difference condition is that the difference between the proportion of the number of feature points in the data of a single ship and the preset first proportion is greater than the preset first proportion difference and less than or equal to the preset second proportion difference; the preset third proportion difference condition is that the difference between the proportion of the number of feature points in the data of a single ship and the preset first proportion is greater than the preset second proportion difference; wherein, the preset first proportion difference is less than the preset second proportion difference, and the preset first acquisition frequency adjustment coefficient is less than the preset second acquisition frequency adjustment coefficient.

Images