Method for evaluating influence of hull roughness on ship main engine load through data driving
The impact of hull roughness on the load of ship hosts is evaluated through a data-driven method, and the problem of difficulty in capturing short-term dynamic changes and ignoring the differences in operating environment in the prior art is solved, and the accuracy of hull roughness and energy consumption and emissions are achieved. optimization.
Patent Information
- Application Number
- CN202411151776.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-21
- Publication Date
- 2024-12-06
AI Technical Summary
The prior art is difficult to effectively evaluate the short-term dynamic impact of hull roughness on ship host load, and ignores the differences in operating ships in rivers and oceans, resulting in inaccurate assessment results, affecting ship energy consumption and exhaust emission estimates.
Using a data-driven method, a multi-source data set is obtained through the ship automatic identification system, a host load prediction model is constructed, a control variable experiment is designed, a hull roughness factor is calculated, and the minimum two-mile of seawater temperature, salinity, seawater flow velocity and ship speed is combined. Regression equations to evaluate the effect of hull roughness on host load.
It realizes an accurate assessment of the short-term dynamic impact of hull roughness on host load, provides technical support for formulating a hull cleaning plan and reducing energy consumption, improves the accuracy of exhaust emission estimation, and helps reduce maintenance costs and energy consumption.
Smart Images

Figure CN119089775A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of hull roughness assessment, and in particular to a method for data-driven assessment of the impact of hull roughness on the load of a ship's main engine. Background Art
[0002] Ship exhaust emissions contribute to global warming and affect ambient air quality in port areas. It is reported that global shipping emissions accounted for 3% of anthropogenic carbon dioxide emissions in 2017. Biofouling of hulls and propellers can have a significant impact on shipping emissions. Microorganisms and algae accumulated on the surface of ship hulls immersed in seawater can lead to reduced ship efficiency, thereby increasing fuel consumption and greenhouse gas emissions. According to a report by the International Maritime Organization, biofouling accounts for 9% to 12% of greenhouse gas emissions in the world fleet. The size of the ship's main engine load affects the ship's energy consumption and exhaust emissions, so accurately assessing the impact of hull roughness on the main engine load is crucial to improving the energy efficiency of ship main engines and reducing pollution and carbon emissions.
[0003] In recent years, artificial intelligence technology has been increasingly used in the field of hull roughness assessment. However, when researchers use machine learning technology to evaluate the impact of hull roughness on main engine load, they mainly focus on the changes in hull roughness over a long time scale, resulting in linear changes in the evaluation results. This makes it difficult to capture the dynamic process of short-term biological attachment growth or shedding due to factors such as season, ship operating area or ship speed. In addition, these studies also ignore the differences in hull roughness between ships operating in rivers and oceans. Therefore, it is necessary to use machine learning technology to evaluate the short-term dynamic impact of hull roughness on main engine load, while taking into account the differences in hull roughness between ships operating in rivers and oceans. This can provide technical support for formulating hull cleaning plans, reducing ship energy consumption, reducing maintenance costs, and improving the accuracy of ship exhaust emissions estimates. Summary of the invention
[0004] In view of the shortcomings of the prior art, the present invention proposes a data-driven method for evaluating the impact of hull roughness on the main engine load of a ship, and evaluates the short-term dynamic impact of the hull roughness of an operating ship on the main engine load as the number of hull cleaning days increases, so as to provide technical support for formulating hull cleaning plans, reducing ship energy consumption, reducing maintenance costs and improving the accuracy of ship exhaust emission estimation.
[0005] The technical means adopted by the present invention are as follows:
[0006] A data-driven method for evaluating the effect of hull roughness on ship main engine load comprises the following steps:
[0007] Step 1: According to the recording time or reporting time in the engine log, cargo stowage plan, AIS data and weather / sea condition data set obtained through the ship automatic identification system, the data sets from different sources are matched to obtain a multi-source fusion data set;
[0008] Step 2: construct a data-driven host load prediction model based on the multi-source fusion data set, and evaluate the prediction performance of each model on the host load;
[0009] Step 3: By designing a controlled variable experiment, the hull roughness factor is calculated based on the optimal prediction model for main engine load prediction, and the impact of hull roughness on main engine load as the number of hull cleaning days increases is evaluated;
[0010] Step 4: Evaluate the impact of marine environment and ship operating conditions on hull roughness by constructing the least squares regression equation between the hull roughness factor and seawater temperature, salinity, seawater velocity and ship speed.
[0011] Furthermore, the step 1 includes the following steps in the process of obtaining the fusion:
[0012] Step 11: Extract the cargo stowage diagram data of the ship at the corresponding time according to the voyage number recorded in the engine log, and obtain a fused data set including the ship engine log data and draft data;
[0013] Step 12: Extract weather / sea condition data from the public dataset based on the reporting time and location information in the ship AIS data to obtain a fused dataset containing ship AIS data and weather / sea condition data;
[0014] Step 13: Extract the ship AIS data at the corresponding time according to the recording time in the engine log, and obtain a fused data set including the ship engine log, draft, AIS and weather / sea conditions.
[0015] Furthermore, in step 2, evaluating the prediction performance of each model for the host load includes the following steps:
[0016] Step 21: Divide the multi-source fusion data set into a training set and a test set according to a certain ratio, and build ordinary least squares regression OLS, least absolute shrinkage and selection operator regression LASSO, ridge regression Ridge, decision tree DT, random forest RF and back propagation neural network BPNN models based on the training set data;
[0017] Step 22: Use 10-fold cross validation to train the LASSO, Ridge, DT, RF and BPNN models to determine the optimal hyperparameter combination of the model;
[0018] Step 23: Based on the test set data, the prediction performance of the host load is evaluated to determine the model with the best prediction performance.
[0019] Furthermore, in step 3, evaluating the effect of hull roughness on the main engine load as the number of hull cleaning days increases includes the following steps:
[0020] Step 31: The mode and mean values in the feature data set are used as inputs of the optimal prediction model, namely the ship operating conditions and the external environment encountered;
[0021] Step 32: Use the entire feature data set of the ship as the training set and the mode / mean value of the feature as the test set to predict the main engine load value as the number of hull cleaning days increases;
[0022] Step 33: Compare the predicted values of the main engine load under different hull cleaning days with the predicted values of the main engine load when the ship leaves the dry dock to obtain the hull roughness factor and evaluate the impact of the hull roughness on the main engine load.
[0023] Furthermore, the step 4 includes the following steps:
[0024] Step 41: construct a least square regression equation of the hull roughness factor and seawater temperature, salinity, seawater velocity and ship speed;
[0025] Step 42: Based on the size of the regression coefficient and the significance test results, explain the influence of seawater temperature, salinity, seawater flow rate and ship speed on hull roughness.
[0026] Furthermore, the certain ratio is training set:test set=7:3.
[0027] Compared with the prior art, the present invention has the following advantages:
[0028] Based on the constructed multi-source fusion data set, this application adopts a method combining machine learning technology with single-factor control variables, and the proposed hull roughness correction factor can evaluate the impact of hull roughness changes on the main engine load in the short term.
[0029] This application constructs an evaluation method for the impact of hull roughness on the main engine load of the ship, evaluates the hull roughness characteristics of ships operating in different trade routes and geographical areas (rivers and oceans), provides a scientific basis for formulating ship maintenance and cleaning strategies for specific fleets, and helps reduce ship energy consumption and maintenance costs. BRIEF DESCRIPTION OF THE DRAWINGS
[0030] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.
[0031] Figure 1 The frequency distribution histogram of the characteristic variables of the present invention is shown in Figure 1. Among them, (a) is the main engine speed setting; (b) is the ground speed; (c) is the average draft; (d) is the trim; (e) is the number of days since the last hull cleaning; (f) is the wave height; (g) is the wind speed; (h) is the wind sheer angle; (i) is the stream sheer angle; (j) is the wave sheer angle; and (k) is the main engine load.
[0032] Figure 2 is the Pearson correlation coefficient matrix of the characteristic variables of the present invention. Wherein (a) is the test ship BC1; (b) is the test ship BC2; (c) is the test ship BC3.
[0033] Figure 3 A flow chart for the control variable experimental design of the present invention;
[0034] Figure 4 The change of the hull roughness factor under different hull cleaning days of the present invention. Among them, (a) is the test ship BC1; (b) is the test ship BC2; (c) is the test ship BC3. DETAILED DESCRIPTION
[0035] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.
[0036] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.
[0037] like Figure 1-4 As shown, the present invention provides a method for data-driven evaluation of the impact of hull roughness on ship main engine load, which specifically includes the following steps:
[0038] Step 1: According to the recording time or reporting time in the engine log, cargo stowage plan, AIS data and weather / sea condition datasets obtained through the ship automatic identification system, the datasets from different sources are matched to obtain a multi-source fusion dataset.
[0039] Specifically, in step 1, the fusion process is obtained, including the following steps:
[0040] Step 11: Extract the cargo stowage diagram data of the ship at the corresponding time according to the voyage number recorded in the engine log, and obtain a fused data set including the ship engine log data and draft data;
[0041] Step 12: Extract weather / sea condition data from the public dataset based on the reporting time and location information in the ship AIS data to obtain a fused dataset containing ship AIS data and weather / sea condition data;
[0042] Step 13: Extract the ship AIS data at the corresponding time according to the recording time in the engine log, and obtain a fused data set including the ship engine log, draft, AIS and weather / sea conditions.
[0043] This paper takes three bulk carriers (BC1, BC2 and BC3) operating in the Yangtze River and offshore China as examples, and collects data sets from the time the three ships leave the dry dock after cleaning the hull to the next time they enter the dry dock for cleaning the hull. Specifically, AIS data, engine logs, cargo stowage charts and weather / sea conditions data from May 22, 2019 to April 29, 2022 (BC1), October 10, 2019 to September 22, 2022 (BC2), and July 24, 2020 to July 7, 2023 (BC3) are used to construct a multi-source fusion data set. According to statistics, the number of data points in the three case ship data sets are 4126, 3118 and 2914 respectively.
[0044] Further, step 2: constructing a data-driven host load prediction model based on the multi-source fusion data set, and evaluating the prediction performance of each model on the host load; step 2 includes the following steps:
[0045] Step 21: According to the domain knowledge method, a feature variable dataset for predicting the main engine load is constructed by fusion of multi-source datasets. The feature variables include main engine speed setting (n), ground speed (SOG), ship trim (ts), mean draft (dM), days since last hull cleaning (DSC), wind sheer angle (wrb), stream sheer angle (crb), wave sheer angle (warb), wind speed (Va) and wave height (swh). The data distribution of the input and output variables of the three case ships is shown in the figure below. Figure 1 shown.
[0046] In order to describe the impact of specific weather / sea conditions on the performance of the case ship, the present invention refers to the method of earlier research and divides the wind speed (Va) into breeze and below, gentle breeze, light breeze, strong wind and gusty wind; divides the wave height (swh) into microwave, small wave, light wave, medium wave, big wave and huge wave; divides the wind side angle (wrb) into headwind, port (right) wind, and downwind; divides the current side angle (crb) into headstream, port (right) current and downstream; divides the wave side angle (warb) into front wave, bow wave, transverse wave, oblique wave and back wave. In order to meet the model feature input requirements, the above classifications are converted into digital feature values.
[0047] Each variable was standardized using the minimum-maximum normalization method to scale the input variables to values between 0 and 1; the Pearson correlation coefficient method determined that ts and dM were highly correlated (correlation coefficient>0.8) ( Figure 2 ); after removing ts and retaining dM, the prediction performance of the model is optimal. The final feature data sets include n, SOG, dM, DSC, wrb, crb, warb, Va and swh.
[0048] MEL=F(x),x={n,SOG,DSC,dM,crb,wrb,warb,V a ,swh};
[0049] The multi-source fusion data set is divided into training set and test set in a certain ratio (7:3). Ordinary least squares regression OLS, least absolute shrinkage and selection operator regression LASSO, ridge regression Ridge, decision tree DT, random forest RF and back propagation neural network BPNN models are constructed based on the training set data.
[0050] Step 22: Use 10-fold cross validation to train the LASSO, Ridge, DT, RF and BPNN models to determine the optimal hyperparameter combination of the model;
[0051] Step 23: Based on the test set data, the prediction performance of the main engine load is evaluated to determine the model with the best prediction performance. The root mean square percentage error (RMSPE), mean absolute percentage error (MAPE) and coefficient of determination (R 2 ), and RF was determined to be the optimal model for MEL prediction.
[0052] Table 1 Summary of MEL prediction performance of each model
[0053]
[0054]
[0055] Note: 1) The unit of RMPSE and MAPE is %.
[0056] Furthermore, if Figure 3 As shown, step 3: by designing a control variable experiment, the hull roughness factor is calculated based on the prediction model with the best prediction of the main engine load, and the influence of the hull roughness on the main engine load as the number of hull cleaning days increases is evaluated; specifically, step 3 includes the following steps:
[0057] Step 31: The mode and mean values in the feature data set are used as the input of the optimal prediction model, i.e., the ship operating conditions and the external environment encountered; the mode of n, wrb, crb, warb, Va and swh in the feature data set, and the mean values of the continuous feature variables dM and SOG are selected as the input of the model. These values represent the most typical ship operating conditions and external environment encountered in the training set, which helps the model to obtain better prediction performance under typical conditions.
[0058] Step 32: Use the entire feature data set of the ship as the training set, and the mode / mean value of the feature as the test set to predict the main engine load value with the increase of the number of hull cleaning days; use the entire feature data set of the three case ships as the training set, and the mode / mean value of the feature variables in S31 as the test set to predict the MEL value under different DSC; The MEL predicted value (MEL DSC=i ) and the predicted MEL value when the ship leaves the dry dock after cleaning the hull (MEL DSC=0 ) to obtain the hull roughness factor (HFF) under different DSC conditions. DSC=i ).
[0059]
[0060] Step 33: Compare the predicted values of the main engine load under different hull cleaning days with the predicted values of the main engine load when the ship leaves the dry dock to obtain the hull roughness factor and evaluate the impact of hull roughness on the main engine load. Figure 4 As DSC increases, the change in HFF represents the short-term dynamic increase or decrease of the hull and propeller due to the gradual accumulation or detachment process of biofouling, thereby dynamically evaluating the impact of hull roughness on MEL.
[0061] According to statistics, during the study period, the proportion of BC1 involved in the Yangtze River route was 21%, while the proportions of BC2 and BC3 were 85% and 83%, respectively. It can be found that the HFF of BC1, which mainly operates on the sea route, showed an upward trend with the increase of DSC, while the HFF of BC2 and BC3, which mainly operate on the Yangtze River and sea routes, did not change significantly. The average HFF of BC1, BC2 and BC3 were 1.23, 1.04 and 0.94, respectively. The p values obtained from the nonparametric Wilcoxon rank sum test were 6.30×10 -262 (BC1 and BC2) and 1.76×10 -229 (BC1 vs. BC3). These values are much smaller than the specified significance level (p=0.05). Therefore, the null hypothesis can be rejected, and statistically, there is a significant difference in HFF between BC1, which mainly operates on the sea route, and BC2 and BC3, which mainly operate on the Yangtze River and the sea route. It has been reported that hull organisms submerged in fresh water will be affected by osmotic stress and eventually die. In addition, the water flow velocity significantly affects the number of attached organisms, and generally higher water flow velocities reduce the number and coverage area of attached organisms. Therefore, the fresh water and higher water flow velocities in the Yangtze River result in the MEL of BC2 and BC3 being less affected by hull roughness.
[0062] From the perspective of cargo loading status, the average HFF values of BC1, BC2 and BC3 in full load state were 1.25, 1.11 and 0.96 respectively; the average HFF values in ballast state were 1.21, 0.98 and 0.93 respectively, which were all lower than those in full load state. The p values obtained in the nonparametric Wilcoxon rank sum test were 7.04×10 -89 (BC1), 3.28×10 -33 (BC2) and 2.19×10 -189 (BC3). These values show that the differences in HFF at different cargo load conditions are statistically significant. This is because in a fully loaded state, the wetted area of the hull is larger, and the resistance caused by the hull roughness is higher, which in turn leads to a higher increase in BC1 MEL compared to when the hull is just cleaned.
[0063] The fitting equations of HFF and DSC of the three case ships under ballast and full load conditions are shown in Table 1. 2 The R values of the other equations are 0.373. 2 The values are all greater than 0.735, indicating a good fitting performance. This is mainly due to the large variance and standard deviation of HFF of BC1 under the ballast state (0.006 and 0.075, respectively). The outliers of HFF significantly lead to the R 2The value is lower. Since the HFF is higher under full load, the present invention mainly analyzes the change of hull roughness of BC1 under full load. The results show that when the DSC is 777 days, the HFF value reaches a peak value (1.305), which is the best time to enter the dry dock to clean the hull, 296 days earlier than the actual time of entering the dry dock. If BC1 enters the dry dock to clean the hull when the DSC is 777 days, it can save about US$82,000 in additional fuel consumption (calculated at $497.50 / ton of heavy oil on March 22, 2024) and reduce CO2 emissions by about 514 tons. It is worth noting that after 777 days of DSC, HFF shows a downward trend. This is because during this period, BC1 entered the Yangtze River for navigation five times. The lower fresh water osmotic pressure and higher water flow velocity caused the biological fouling on the hull to die and fall off, thereby reducing the roughness of the hull and resulting in a reduction in HFF. In contrast, the average variance and standard deviation of HFF for BC2 and BC3 in ballast and full load states are 0.001 and 0.024, respectively; both lower than BC1. This is because BC2 and BC3 frequently enter the Yangtze River, resulting in relatively low hull roughness and smaller fluctuations in HFF scatter points. According to the regulations of China Classification Society, ships need to undergo special inspections in dry docks within three years, and special inspections are usually carried out at the same time as hull cleaning. The average hull cleaning cycle of the three case ships is 2.9 years. Taking into account operating costs and environmental impacts, ships that mainly operate on sea routes can carry out hull cleaning in advance, while ships that mainly operate on the Yangtze River and sea routes can appropriately postpone the hull cleaning time.
[0064] Table 1 HFF fitting equation
[0065]
[0066] Note: The fitting equation was constructed by robust covariance polynomial regression using the least angle regression (LAR) method.
[0067] As a preferred embodiment, step 4: by constructing the least squares regression equation of the hull roughness factor with seawater temperature, salinity, seawater velocity and ship speed, the influence of the marine environment and ship operating conditions on the hull roughness is evaluated. In order to evaluate the ability of the proposed HFF in reflecting the influence of hull roughness on MEL, the present invention performs ordinary least squares (OLS) regression analysis (non-robust regression) on the HFF and its corresponding influencing factors (including ship ground speed (SOG), seawater velocity (Vc), sea surface temperature (ST) and salinity (SA)). The R of all three equations is 2 The values were all over 0.750, indicating good fitting performance (Table 2); in addition, most factors had a significant effect on HFFs (p < 0.01) (Table 3). The following steps were included:
[0068] Step 41: construct a least square regression equation of the hull roughness factor and seawater temperature, salinity, seawater velocity and ship speed; specifically,
[0069] Salinity: Salinity has the most significant effect on the HFFs of the three case study ships, with regression coefficients ranging from 0.35 to 0.76. This indicates that HFF increases significantly with increasing salinity (p<0.01). The salinity values of the world's oceans range from 30 to 36 psu, and the salinity tolerance range of hull-bound organisms is also within this range. However, the three case study ships usually sail near the Chinese coast, and there may be local salinity changes near estuaries. Most hull-bound organisms cannot withstand low-salinity environments. Therefore, salinity shows a significant positive effect on hull roughness.
[0070] Speed over ground: Except for the regression coefficient of BC1 SOG which did not reach a significant level (p = 0.702), the regression coefficients of BC2 and BC3 SOG were negative (-0.378 and -0.071). This is mainly because SOG affects the types and number of organisms attached to the hull. When the ship speed is high, the seawater scouring causes most organisms to be unable to continue to attach to the hull. When SOG is low, many organisms can continue to attach to the hull surface. Therefore, higher SOG leads to a decrease in hull roughness.
[0071] Seawater velocity: For BC1, the regression coefficient of Vc is negative (-0.0797), while the regression coefficients for BC2 and BC3 are positive (0.1645 and 0.1458). During the study period, the average Vc encountered by BC1 was about 29% higher than that of BC2 and BC3. This is because biofilms form faster and have higher viscosity under turbulent conditions, while biofilms form slower under lower water flow conditions. It is worth noting that too high Vc can also cause biofilms to detach. Therefore, the higher flow rate encountered by BC1 resulted in a decrease in hull roughness.
[0072] Sea surface temperature: For BC1, the regression coefficient of ST is negative (-0.0374), while the regression coefficients of BC2 and BC3 are positive (0.0074 and 0.2668). During the study period, the average sea surface temperature encountered by BC1 (19.7℃) was higher than that of BC2 and BC3 (15.7℃). Since all three case ships used self-polishing antifouling paint, when the seawater temperature is relatively high, the biocide in the paint is released to prevent biological adhesion on the hull. Therefore, the higher sea temperature encountered by BC1 resulted in a decrease in hull roughness.
[0073] Table 2 HFF fitting equation
[0074]
[0075] Table 3 Ordinary least squares regression results
[0076]
[0077]
[0078] Step 42: Based on the size of the regression coefficient and the significance test results, explain the influence of seawater temperature, salinity, seawater flow rate and ship speed on hull roughness.
[0079] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.
[0080] In the above embodiments of the present invention, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.
[0081] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of the units can be a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.
[0082] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.
[0083] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.
[0084] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, magnetic disk or optical disk and other media that can store program codes.
[0085] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.
Claims
1. A data-driven method for evaluating the effect of hull roughness on ship main engine load, characterized in that: The following steps are involved: Step 1: According to the recording time or reporting time in the engine log, cargo stowage plan, AIS data and weather / sea condition data set obtained through the ship automatic identification system, the data sets from different sources are matched to obtain a multi-source fusion data set; Step 2: construct a data-driven host load prediction model based on the multi-source fusion data set, and evaluate the prediction performance of each model on the host load; Step 3: By designing a controlled variable experiment, the hull roughness factor is calculated based on the optimal prediction model for main engine load prediction, and the impact of hull roughness on main engine load as the number of hull cleaning days increases is evaluated; Step 4: Evaluate the impact of marine environment and ship operating conditions on hull roughness by constructing the least squares regression equation between the hull roughness factor and seawater temperature, salinity, seawater velocity and ship speed.
2. A data-driven method for evaluating the effect of hull roughness on ship main engine load according to claim 1, characterized in that: The step 1 includes the following steps in the process of obtaining the fusion: Step 11: Extract the cargo stowage diagram data of the ship at the corresponding time according to the voyage number recorded in the engine log, and obtain a fused data set including the ship engine log data and draft data; Step 12: Extract weather / sea condition data from the public dataset based on the reporting time and location information in the ship AIS data to obtain a fused dataset containing ship AIS data and weather / sea condition data; Step 13: Extract the ship AIS data at the corresponding time according to the recording time in the engine log, and obtain a fused data set including the ship engine log, draft, AIS and weather / sea conditions.
3. A data-driven method for evaluating the effect of hull roughness on ship main engine load according to claim 1, characterized in that: In step 2, evaluating the prediction performance of each model for host load includes the following steps: Step 21: Divide the multi-source fusion data set into a training set and a test set according to a certain ratio, and build ordinary least squares regression OLS, least absolute shrinkage and selection operator regression LASSO, ridge regression Ridge, decision tree DT, random forest RF and back propagation neural network BPNN models based on the training set data; Step 22: Use 10-fold cross validation to train the LASSO, Ridge, DT, RF and BPNN models to determine the optimal hyperparameter combination of the model; Step 23: Based on the test set data, the prediction performance of the host load is evaluated to determine the model with the best prediction performance.
4. A data-driven method for evaluating the effect of hull roughness on ship main engine load according to claim 1, characterized in that: In step 3, evaluating the effect of hull roughness on the main engine load as the number of days of hull cleaning increases includes the following steps: Step 31: The mode and mean values in the feature data set are used as inputs of the optimal prediction model, namely the ship operating conditions and the external environment encountered; Step 32: Use the entire feature data set of the ship as the training set and the mode / mean value of the feature as the test set to predict the main engine load value as the number of hull cleaning days increases; Step 33: Compare the predicted values of the main engine load under different hull cleaning days with the predicted values of the main engine load when the ship leaves the dry dock to obtain the hull roughness factor and evaluate the impact of the hull roughness on the main engine load.
5. A data-driven method for evaluating the effect of hull roughness on ship main engine load according to claim 1, characterized in that: The step 4 includes the following steps: Step 41: construct a least square regression equation of the hull roughness factor and seawater temperature, salinity, seawater velocity and ship speed; Step 42: Based on the size of the regression coefficient and the significance test results, explain the influence of seawater temperature, salinity, seawater flow rate and ship speed on hull roughness.
6. A data-driven method for evaluating the effect of hull roughness on ship main engine load according to claim 3, characterized in that: The certain ratio is training set:test set=7:3.
