Method and system for extracting stable navigation data of real ship

Abstract

The invention provides a real ship stable navigation data extraction method and system, and the method comprises the steps: firstly collecting ship data when the working condition of a ship main engine is stable, then calculating a feature value of main engine noise data in the ship data, carrying out the noise cleaning of the main engine noise data through employing a Chebyshev inequality, carrying out the normalization processing of the data after noise cleaning, and carrying out the extraction of the real ship stable navigation data. The method comprises the following steps: extracting data having the maximum correlation with host power by using a recursive feature elimination method as a correlation feature parameter, then calculating an autocorrelation coefficient of the host power by using an autocorrelation function so as to obtain a change trend function, and calculating according to the change trend function to obtain a window period; dividing the ship data after noise cleaning according to a window period to obtain ship data of a plurality of periods, respectively calculating an information entropy value of each related characteristic parameter in each period by adopting an information entropy method, and finally judging whether the related characteristic parameters in the periods are rejected or not. And taking a characteristic parameter data set formed by the periods which are not removed as a stable navigation data set.

Description

Technical Field

[0001] The present invention relates to the technical field of real ship data extraction, and in particular to a method and system for extracting real ship stable navigation data. Background Art

[0002] The added resistance of a ship due to waves is the added resistance caused by waves and motion when the ship is sailing in relatively still water in the actual marine environment. This performance index not only significantly affects the actual energy efficiency (economy) and environmental performance of the ship, but also indirectly determines whether the ship has the ability to effectively maneuver when sailing in harsh marine environment conditions. For a long time, people have generally studied the motion performance of ships by combining numerical calculations with small-scale models in water tank tests. Due to the limitations of the pool wall effect and the authenticity of wind and wave simulation, the current water tank model test technology cannot simulate the actual wave environment, does not have the conditions to complete the research on the environmental adaptability of the ship, and cannot achieve the problem of predicting the wave performance of the actual ship in the actual wave environment, which brings great difficulties to the evaluation and analysis of the hydrodynamic performance of ships.

[0003] At present, constructing a wave resistance increase prediction method based on real ship and real sea area operation data is a new method to study the wave performance of real ships. Although the real ship test is the most realistic test method, the real ship navigation is affected by factors such as weather, driving operation, equipment operation and main engine equipment status. It is costly and difficult to implement, especially for tests in harsh sea conditions. The risk is high and it is destructive to a certain extent. In fact, it is difficult to achieve the expected test goals. In addition, instrument deviations and acquisition equipment failures will also cause errors in real ship data. Therefore, it is necessary to comprehensively consider the influence of multiple related variables and characteristics, and select real ship stable navigation data for wave resistance increase prediction research. Summary of the invention

[0004] In order to solve the problems of many abnormal data points appearing in the actual ship data collection due to the factors affecting the actual ship's navigation, and abnormal main engine power collection data due to the instability of the internal parameters of the main engine equipment, the present invention provides a method for extracting data of an actual ship's stable navigation, which is based on the ship's data when the main engine of the ship is in stable working condition, and uses a specific judgment method and calculation method to screen out the data set of the actual ship's stable navigation, effectively reducing the abnormal data points appearing in the actual ship's data collection and the abnormal main engine power data due to the instability of the internal parameters of the main engine equipment, and determining a stable data interval that can provide wave resistance increase calculation. The present invention also relates to a real ship's stable navigation data extraction system.

[0005] The technical solution of the present invention is as follows:

[0006] A method for extracting stable navigation data of a real ship, characterized by comprising the following steps:

[0007] Data acquisition and extraction steps: Ship data is collected at regular intervals when the operating conditions of the ship's main engine are stable during actual ship navigation. The differences in the forward draft of the main engine, the aft draft of the main engine, and the rotational speed of the main engine in the ship data collected in any adjacent time periods within a certain cycle are compared with the corresponding preset thresholds. If all are less than or equal to the corresponding preset thresholds, the ship data within that cycle is extracted;

[0008] Noise cleaning and correlation extraction steps: Calculate the characteristic values of the main engine noise data in the extracted ship data, and use the Chebyshev inequality to screen out the characteristic data set that meets a certain range from the characteristic values to complete the noise cleaning of the main engine noise data. Then, perform normalization processing on the main engine noise data after noise cleaning. Based on the normalized data, use the recursive feature elimination method to extract the data with the highest correlation with the main engine power in the ship data as the relevant characteristic parameters;

[0009] Window period determination steps: Use the autocorrelation function to calculate the autocorrelation coefficients of the main engine power at different cycle interval lengths, and then obtain the change trend function of the autocorrelation coefficient of the main engine power with respect to the cycle interval length. Calculate the window period according to the change trend function;

[0010] Information entropy value calculation steps: Divide the ship data after noise cleaning according to the window period to obtain ship data for multiple cycles. Calculate the maximum and minimum values of each parameter in the relevant characteristic parameters within the said range, and then divide the maximum and minimum values equally to obtain multiple numerical distribution intervals. Based on the numerical distribution intervals, use the information entropy method to calculate the information entropy values of each relevant characteristic parameter in each cycle;

[0011] Stable navigation data judgment steps: Judge whether the information entropy values of each relevant characteristic parameter in a certain cycle are less than the preset numerical threshold. If the information entropy value of any relevant characteristic parameter is less than the preset numerical threshold, the relevant characteristic parameters of that cycle are excluded, and the characteristic parameter data set composed of the cycles that have not been excluded is retained as the actual ship stable navigation data set.

[0012] Preferably, in the data acquisition and extraction steps, the ship data includes the forward draft of the main engine, the aft draft of the main engine, the rotational speed of the main engine, the main engine power, the speed through water, the outlet temperature of the cooling water of the main engine cylinder head, the outlet temperature of the cooling water of the main engine cylinder liner, the temperature of the fuel side cylinder liner of the main engine cylinder, the temperature of the exhaust gas side cylinder liner of the main engine cylinder, the outlet temperature of the piston cooling oil of the main engine cylinder, the scavenging air box temperature of the main engine cylinder, the exhaust outlet temperature of the main engine cylinder, the exhaust gas outlet pressure of the main engine turbocharger, the exhaust inlet temperature of the main engine turbocharger, and the exhaust outlet temperature of the main engine turbocharger.

[0013] Preferably, in the noise cleaning and correlation extraction step, the main engine noise data includes the cooling water outlet temperature of the main engine cylinder head, the cooling water outlet temperature of the main engine cylinder liner, the cylinder liner temperature on the fuel side of the main engine cylinder, the cylinder liner temperature on the exhaust gas side of the main engine cylinder, the piston cooling oil outlet temperature of the main engine cylinder, the scavenging air box temperature of the main engine cylinder, the exhaust outlet temperature of the main engine cylinder, the exhaust gas outlet pressure of the main engine turbocharger, the exhaust inlet temperature of the main engine turbocharger, and the exhaust outlet temperature of the main engine turbocharger.

[0014] Preferably, in the noise cleaning and correlation extraction step, the eigenvalue includes calculating the average value and standard deviation of ten main engine noise data.

[0015] Preferably, in the noise cleaning and correlation extraction step, the recursive feature elimination method includes training a classifier, calculating the importance measure of permutation, eliminating irrelevant variables, and retraining the classifier using the eliminated features.

[0016] A real ship stable navigation data extraction system, characterized by comprising a data acquisition and extraction module, a noise cleaning and correlation extraction module, a window period determination module, an information entropy value calculation module, and a stable navigation data judgment module, which are connected in sequence.

[0017] The data acquisition and extraction module collects ship data when the main engine of the ship is operating stably during real ship navigation at regular intervals, and compares the differences in the forward draft of the main engine, the differences in the aft draft of the main engine, and the differences in the main engine speed in the ship data collected in any adjacent time periods within a certain cycle with the corresponding preset thresholds. If they are all less than or equal to the corresponding preset thresholds, the ship data within this cycle is extracted.

[0018] The noise cleaning and correlation extraction module calculates the eigenvalues of the main engine noise data in the extracted ship data, uses the Chebyshev inequality to screen out the feature data set that meets a certain interval range from the eigenvalues to complete the noise cleaning of the main engine noise data, normalizes the main engine noise data after noise cleaning, and then uses the recursive feature elimination method to extract the data with the highest correlation with the main engine power in the ship data as the relevant feature parameters.

[0019] The window period determination module calculates the autocorrelation coefficient of the main engine power at different cycle interval lengths using the autocorrelation function, and then obtains the change trend function of the autocorrelation coefficient of the main engine power with respect to the cycle interval length, and calculates the window period according to the change trend function.

[0020] The information entropy value calculation module divides the noise-cleaned ship data according to the window period to obtain ship data of multiple periods, and calculates the maximum and minimum values ​​of each parameter in the relevant characteristic parameters within the interval range, and then divides the maximum and minimum values ​​equally to obtain multiple value distribution intervals, and calculates the information entropy values ​​of each relevant characteristic parameter in each period based on the value distribution interval and using the information entropy method;

[0021] The stable navigation data judgment module determines whether the information entropy value of each relevant characteristic parameter in a certain period is less than the preset numerical threshold. If the information entropy value of any relevant characteristic parameter is less than the preset numerical threshold, the relevant characteristic parameters of the period are eliminated, and the characteristic parameter data set composed of the periods that are not eliminated is retained as the real ship stable navigation data set.

[0022] Preferably, the ship data includes main engine bow draft, main engine stern draft, main engine speed, main engine power, water speed, main engine cylinder head cooling water outlet temperature, main engine cylinder liner cooling water outlet temperature, main engine cylinder fuel side liner temperature, main engine cylinder exhaust side liner temperature, main engine cylinder piston cooling oil outlet temperature, main engine cylinder scavenging box temperature, main engine cylinder exhaust outlet temperature, main engine turbocharger exhaust gas outlet pressure, main engine turbocharger exhaust inlet temperature and main engine turbocharger exhaust outlet temperature.

[0023] Preferably, in the data acquisition and extraction module, the host noise data includes the host cylinder head cooling water outlet temperature, the host cylinder liner cooling water outlet temperature, the host cylinder fuel side liner temperature, the host cylinder exhaust side liner temperature, the host cylinder piston cooling oil outlet temperature, the host cylinder scavenging box temperature, the host cylinder exhaust outlet temperature, the host turbocharger exhaust outlet pressure, the host turbocharger exhaust inlet temperature and the host turbocharger exhaust outlet temperature.

[0024] Preferably, the characteristic value comprises calculating the average value and standard deviation of ten host noise data.

[0025] Preferably, the recursive feature elimination method in the noise cleaning and correlation extraction module includes training a classifier, calculating a permutation importance measure, eliminating irrelevant variables, and retraining the classifier using the eliminated features.

[0026] The beneficial effects of the present invention are:

[0027] A method for extracting actual ship stable navigation data provided by the present invention sequentially sets a data collection and extraction step, a noise cleaning and correlation extraction step, a window period determination step, an information entropy value calculation step, and a stable navigation data judgment step. Each step cooperates with each other and works synergistically. First, ship data is collected at regular intervals when the ship's main engine operating conditions are stable during actual ship navigation, and the differences in the bow draft of the main engine, the differences in the stern draft of the main engine, and the differences in the main engine speed in the ship data collected in any adjacent time periods within a certain cycle are compared with their respective preset thresholds. If they are all less than or equal to the corresponding preset thresholds, the ship data and the ship data of that time period are extracted. Then, the characteristic values of the main engine noise data in the ship data are calculated, and a Chebyshev inequality is used to screen out a characteristic data set that satisfies a certain interval range from the characteristic values to complete the noise cleaning of the main engine noise data, and the main engine noise data after noise cleaning is normalized. Then, based on the normalized data, a recursive feature elimination method is used to extract the data with the greatest correlation with the main engine power in the ship data as the relevant characteristic parameters. Then, the autocorrelation coefficient of the main engine power at different cycle interval lengths is calculated using the autocorrelation function, and then the change trend function of the autocorrelation coefficient of the main engine power with respect to the cycle interval length is obtained. The window period is calculated according to the change trend function, and then the ship data after noise cleaning is divided according to the window period to obtain ship data for multiple cycles. The maximum and minimum values of each parameter in the relevant characteristic parameters within the interval range are calculated, and then the maximum and minimum values are equally divided to obtain multiple numerical distribution intervals. Based on the numerical distribution intervals, the information entropy values of each relevant characteristic parameter in each cycle are calculated using the information entropy method. Finally, it is judged whether the information entropy values of each relevant characteristic parameter in a certain cycle are less than the preset numerical threshold. If the information entropy value of any relevant characteristic parameter is less than the preset numerical threshold, the relevant characteristic parameters of that cycle are excluded, and the characteristic parameter data set composed of the cycles that have not been excluded is retained as the actual ship stable navigation data set. Considering the influence of various relevant characteristic parameters, the actual ship stable navigation data is selected for wave added resistance prediction research, effectively reducing the abnormal data points in actual ship data collection and the abnormal main engine power data caused by unstable internal parameters of the main engine equipment, and determining a stable data interval that can provide wave added resistance calculation.

[0028] The present invention also relates to a system for extracting actual ship stable navigation data. This system corresponds to the above-mentioned method for extracting actual ship stable navigation data and can be understood as a system for implementing the above-mentioned method for extracting actual ship stable navigation data. It includes a data acquisition and extraction module, a noise cleaning and correlation extraction module, a window period determination module, an information entropy value calculation module, and a stable navigation data judgment module that are connected in sequence. Each module works in coordination with each other. Based on the ship data when the ship's main engine conditions are stable during actual ship navigation, and using specific judgment methods and calculation methods, a data set for actual ship stable navigation is screened out, effectively reducing the abnormal data points that occur in actual ship data acquisition and the abnormal main engine power data caused by unstable internal parameters of the main engine equipment, and determining a stable data interval that can provide data for wave added resistance calculation. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 is a flowchart of the method for extracting actual ship stable navigation data of the present invention.

[0030] Figure 2 is a preferred flowchart of the method for extracting actual ship stable navigation data of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0031] The present invention will be described below with reference to the accompanying drawings.

[0032] The present invention relates to a method for extracting actual ship stable navigation data. The flowchart of this method is as Figure 1 shown, and the preferred flowchart is as Figure 2 shown, and successively includes the following steps:

[0033] I. Data acquisition and extraction step, that is, extracting ship data during the stable condition of the main engine: Ship data when the ship's main engine conditions are stable during actual ship navigation is collected at regular intervals, and the differences in the bow draft of the main engine, the differences in the stern draft of the main engine, and the differences in the main engine speed in the ship data collected in any adjacent time periods within a certain cycle are compared with the corresponding preset thresholds respectively. If all are less than or equal to the corresponding preset thresholds, the ship data and the ship data for this time period are extracted;

[0034] Specifically, first, a ship's data is collected every 1 second when the ship's main engine is operating stably, that is, 60 pieces of ship's data are collected in 1 minute. With a 1-minute rolling period, the differences in the bow draft of the main engine, the differences in the stern draft of the main engine, and the differences in the main engine speed in the ship's data collected in any adjacent time periods within a certain cycle are compared with the corresponding preset thresholds respectively, to determine whether the differences in the bow draft of the main engine, the differences in the stern draft of the main engine, and the differences in the main engine speed are all greater than the corresponding preset thresholds, or rather, to determine whether the percentages of the differences in the bow draft of the main engine, the percentages of the differences in the stern draft of the main engine, and the percentages of the differences in the main engine speed are all greater than the corresponding preset thresholds. Preferably, it is determined whether the percentages of each difference are all greater than 5% (① for the bow and stern drafts of the main engine, the deviation between adjacent data does not exceed 5%, ② for the main engine speed, the deviation between adjacent times does not exceed 5%). It is judged whether the data within the judgment cycle meets conditions ① and ②. If the situation of not meeting the conditions occurs, that is, the difference of any data is greater than the preset threshold, the data within that cycle is not extracted. If the conditions are met, that is, the differences are all less than or equal to the preset threshold, the ship's data within that cycle is extracted. Preferably, the ship's data includes the bow draft of the main engine, the stern draft of the main engine, the main engine speed, the main engine power, the speed through water, the outlet temperature of the cooling water of the main engine cylinder head, the outlet temperature of the cooling water of the main engine cylinder liner, the temperature of the fuel side cylinder liner of the main engine cylinder, the temperature of the exhaust side cylinder liner of the main engine cylinder, the outlet temperature of the piston cooling oil of the main engine cylinder, the temperature of the scavenging air box of the main engine cylinder, the outlet temperature of the exhaust of the main engine cylinder, the exhaust gas outlet pressure of the main engine turbocharger, the exhaust inlet temperature of the main engine turbocharger, and the exhaust outlet temperature of the main engine turbocharger, etc.

[0035] II. Noise cleaning and correlation extraction steps: Calculate the eigenvalues of the main engine noise data in the extracted ship's data, and use Chebyshev's inequality to screen out the eigenvalue dataset that meets a certain range to complete the noise cleaning of the main engine noise data, and perform normalization processing on the main engine noise data after noise cleaning. Then, based on the data after normalization processing, use the recursive feature elimination method to extract the data with the greatest correlation with the main engine power in the ship's data as the relevant feature parameters;

[0036] Specifically, first calculate the mean value and standard deviation of the main engine noise data in the ship's data, and use Chebyshev's inequality to screen out the eigenvalue dataset that meets a certain range to complete the noise cleaning of the main engine noise data. Chebyshev's inequality is as follows:

[0037]

[0038] In the above formula, σ is the standard deviation of the main engine noise data, μ is the mean value of the main engine noise data, k is 10, X is the value of each parameter in the main engine noise data, and P is the probability.

[0039] Preferably, the main engine noise data includes the cooling water outlet temperature of the main engine cylinder head, the cooling water outlet temperature of the main engine cylinder liner, the temperature of the main engine cylinder liner on the fuel side, the temperature of the main engine cylinder liner on the exhaust gas side, the cooling oil outlet temperature of the main engine cylinder piston, the scavenging air box temperature of the main engine cylinder, the exhaust outlet temperature of the main engine cylinder, the exhaust gas outlet pressure of the main engine turbocharger, the exhaust inlet temperature of the main engine turbocharger, and the exhaust outlet temperature of the main engine turbocharger. Noise data cleaning is performed on 10 kinds of signals, and the average value and standard deviation of the 10 kinds of signals are calculated.

[0040] Then, the main engine noise data after noise cleaning is normalized (the 10 kinds of signals are normalized separately), and the normalization formula is as follows:

[0041]

[0042] In the above formula, σ is the standard deviation of the main engine noise data, μ is the average value of the main engine noise data, χ is the value of each parameter in the main engine noise data, and Z is the normalized value.

[0043] Based on the data after normalization, the data with the greatest correlation with the main engine power in the ship data is extracted using the recursive feature elimination method as the relevant feature parameters. Preferably, the recursive feature elimination method is used to determine the 5 kinds of signals with the greatest correlation with the main engine power among the 10 kinds of signals. The main steps of the recursive feature elimination method: ① Train the classifier; ② Calculate the importance measure of permutation; ③ Eliminate irrelevant variables; ④ Retrain the classifier with the eliminated features.

[0044] III. Window period determination step, that is, determine the sliding time window period: Calculate the autocorrelation coefficient of the main engine power at different cycle interval lengths using the autocorrelation function, and then obtain the change trend function of the autocorrelation coefficient of the main engine power with respect to the cycle interval length. Calculate the window period according to the change trend function;

[0045] Specifically, first calculate the autocorrelation coefficient value of the main engine power at different cycle interval lengths k using the autocorrelation function, that is, obtain the change trend function f(t) of the autocorrelation coefficient f of the main engine power with respect to k, and calculate according to the following formula:

[0046]

[0047] In the above formula, k is the cycle interval length, N is the length of the sequence (the number of cycles), is the average value of the complete sequence, t is the time point, then x t is the main engine power value at time t.

[0048] Then, take the derivative of the change trend function f(t) to obtain the g(t) function, and then calculate the t value when g(t)=0. This value is the window period.

[0049] IV. Information entropy value calculation steps: Divide the ship data after noise cleaning in the noise cleaning and correlation extraction steps according to the window period to obtain ship data for multiple periods, and calculate the maximum and minimum values of each parameter in the range of the relevant characteristic parameters in the noise cleaning and correlation extraction steps. Then, divide the maximum and minimum values equally to obtain multiple numerical distribution intervals. Based on the numerical distribution intervals, use the information entropy method to calculate the information entropy value of each relevant characteristic parameter in each period;

[0050] Specifically, first divide the ship data after noise cleaning in the noise cleaning and correlation extraction steps according to the window period determined in the window period determination step to obtain the ship data distribution for n periods. Then, calculate the maximum value X of each parameter in the 5 (preferably) relevant characteristic parameters obtained in the noise cleaning and correlation extraction steps within the above range max and the minimum value X min , and then divide the maximum and minimum values equally into 10 parts, so that each relevant characteristic parameter obtains 10 numerical distribution intervals. Then, based on the numerical distribution intervals, use the information entropy method to calculate the information entropy value of each relevant characteristic parameter in each period, that is, calculate the information entropy value of the 5 relevant characteristic parameters in the ship data distribution for n periods. That is, each relevant characteristic parameter will have an information entropy value in each period. The information entropy value H is calculated according to the following formula:

[0051]

[0052] In the above formula, M is the total number of each relevant characteristic parameter, K i is the total number of each relevant characteristic parameter in each numerical distribution interval, H is the result of the information entropy value, and i is the i-th period.

[0053] Therefore, each relevant characteristic parameter has n information entropy value results, that is, the stable values corresponding to n periods.

[0054] V. Stable navigation data judgment steps: Judge whether the information entropy value of each relevant characteristic parameter in a certain period is less than the preset numerical threshold. That is, in a single period, if the information entropy value H of any relevant characteristic parameter is less than 0.2, it is determined that the period is unstable. That is, the relevant characteristic parameters of this period are excluded, and the characteristic parameter data set composed of the periods that are not excluded is retained as the real ship stable navigation data set.

[0055] The present invention also relates to a real ship stable navigation data extraction system, which corresponds to the above-mentioned real ship stable navigation data extraction method and can be understood as a system for implementing the above-mentioned method. The system includes a data acquisition and extraction module, a noise cleaning and correlation extraction module, a window period determination module, an information entropy value calculation module and a stable navigation data judgment module connected in sequence. Specifically,

[0056] The data collection and extraction module collects the ship data when the main engine working condition of the ship is stable during the voyage at regular intervals, and compares the difference of the main engine bow draft, the difference of the main engine stern draft and the difference of the main engine speed in the ship data collected in any adjacent time periods within a certain cycle with the corresponding preset thresholds. If they are all less than or equal to the corresponding preset thresholds, the ship data within the cycle is extracted;

[0057] The noise cleaning and correlation extraction module calculates the characteristic value of the main engine noise data in the extracted ship data, and uses Chebyshev inequality to select a characteristic data set that meets a certain interval range from the characteristic value to complete the noise cleaning of the main engine noise data, and normalizes the main engine noise data after noise cleaning. Based on the normalized data, the recursive feature elimination method is used to extract the data with the greatest correlation with the main engine power in the ship data as the relevant characteristic parameters;

[0058] The window period determination module uses the autocorrelation function to calculate the autocorrelation coefficient of the host power at different period interval lengths, and then obtains the change trend function of the autocorrelation coefficient of the host power with respect to the period interval length, and calculates the window period according to the change trend function;

[0059] The information entropy value calculation module divides the noise-cleaned ship data according to the window period to obtain ship data of multiple periods, and calculates the maximum and minimum values ​​of each parameter in the relevant characteristic parameters within the interval range, and then divides the maximum and minimum values ​​equally to obtain multiple value distribution intervals, and calculates the information entropy values ​​of each relevant characteristic parameter in each period based on the value distribution interval and using the information entropy method;

[0060] The stable navigation data judgment module determines whether the information entropy value of each relevant characteristic parameter in a certain period is less than the preset numerical threshold. If the information entropy value of any relevant characteristic parameter is less than the preset numerical threshold, the relevant characteristic parameters of the period are eliminated, and the characteristic parameter data set composed of the periods that are not eliminated is retained as the real ship stable navigation data set.

[0061] Preferably, the ship data includes the forward draft of the main engine, the aft draft of the main engine, the rotational speed of the main engine, the power of the main engine, the speed through the water, the outlet temperature of the cooling water of the cylinder head of the main engine, the outlet temperature of the cooling water of the cylinder liner of the main engine, the temperature of the cylinder liner on the fuel side of the main engine cylinder, the temperature of the cylinder liner on the exhaust gas side of the main engine cylinder, the outlet temperature of the piston cooling oil of the main engine cylinder, the scavenging air box temperature of the main engine cylinder, the exhaust outlet temperature of the main engine cylinder, the exhaust gas outlet pressure of the main engine turbocharger, the exhaust inlet temperature of the main engine turbocharger, and the exhaust outlet temperature of the main engine turbocharger.

[0062] Preferably, the main engine noise data includes the outlet temperature of the cooling water of the cylinder head of the main engine, the outlet temperature of the cooling water of the cylinder liner of the main engine, the temperature of the cylinder liner on the fuel side of the main engine cylinder, the temperature of the cylinder liner on the exhaust gas side of the main engine cylinder, the outlet temperature of the piston cooling oil of the main engine cylinder, the scavenging air box temperature of the main engine cylinder, the exhaust outlet temperature of the main engine cylinder, the exhaust gas outlet pressure of the main engine turbocharger, the exhaust inlet temperature of the main engine turbocharger, and the exhaust outlet temperature of the main engine turbocharger.

[0063] Preferably, the eigenvalue includes calculating the average value and the standard deviation of ten main engine noise data.

[0064] Preferably, the recursive feature elimination method in the noise cleaning and correlation extraction module includes training a classifier, calculating the importance measure of permutation, eliminating irrelevant variables, and retraining the classifier using the eliminated features.

[0065] The present invention provides an objective and scientific method and system for extracting actual ship stable navigation data. Based on the ship data when the operating conditions of the ship's main engine are stable during actual ship navigation, and using specific judgment methods and calculation methods to screen out the data set of actual ship stable navigation, effectively reducing the abnormal data points in actual ship data acquisition and the abnormal power data of the main engine caused by unstable internal parameters of the main engine equipment, and determining a stable data interval that can provide data for wave resistance calculation.

[0066] It should be noted that the above specific embodiments can enable those skilled in the art to understand the present invention more comprehensively, but do not limit the present invention in any way. Therefore, although this specification has described the present invention in detail with reference to the drawings and embodiments, those skilled in the art should understand that the present invention can still be modified or equivalently replaced. In short, all technical solutions and their improvements that do not depart from the spirit and scope of the present invention should be covered by the protection scope of the patent of the present invention.

Claims

1. A method for extracting stable navigation data of a real ship, characterized in that: The following steps are involved: Data collection and extraction steps: collect ship data when the main engine operating condition of the ship is stable during the voyage of the actual ship at regular intervals, and compare the difference of the main engine bow draft, the difference of the main engine stern draft and the difference of the main engine speed in the ship data collected in any adjacent time periods within a certain period with the corresponding preset thresholds. If they are all less than or equal to the corresponding preset thresholds, extract the ship data within the period; Noise cleaning and correlation extraction steps: Calculate the characteristic value of the main engine noise data in the extracted ship data, use Chebyshev inequality to select a characteristic data set that meets a certain interval range from the characteristic value to complete the noise cleaning of the main engine noise data, and normalize the main engine noise data after noise cleaning. Then, based on the normalized data, use the recursive feature elimination method to extract the data with the greatest correlation with the main engine power in the ship data as the relevant characteristic parameters; Window period determination step: using the autocorrelation function to calculate the autocorrelation coefficient of the host power at different period interval lengths, and then obtaining the change trend function of the autocorrelation coefficient of the host power with respect to the period interval length, and calculating the window period according to the change trend function; Information entropy value calculation steps: divide the noise-cleaned ship data according to the window period to obtain ship data of multiple periods, and calculate the maximum and minimum values ​​of each parameter in the relevant characteristic parameters within the interval range, and then divide the maximum and minimum values ​​equally to obtain multiple value distribution intervals, and calculate the information entropy values ​​of each relevant characteristic parameter in each period based on the value distribution interval and using the information entropy method; Stable navigation data judgment step: determine whether the information entropy value of each relevant characteristic parameter in a certain period is less than the preset numerical threshold. If the information entropy value of any relevant characteristic parameter is less than the preset numerical threshold, the relevant characteristic parameters of the period are eliminated, and the characteristic parameter data set composed of the periods that are not eliminated is retained as the real ship stable navigation data set.

2. The method for extracting stable navigation data of a real ship according to claim 1, characterized in that: In the data collection and extraction step, the ship data includes main engine bow draft, main engine stern draft, main engine speed, main engine power, water speed, main engine cylinder head cooling water outlet temperature, main engine cylinder liner cooling water outlet temperature, main engine cylinder fuel side liner temperature, main engine cylinder exhaust side liner temperature, main engine cylinder piston cooling oil outlet temperature, main engine cylinder scavenging box temperature, main engine cylinder exhaust outlet temperature, main engine turbocharger exhaust gas outlet pressure, main engine turbocharger exhaust inlet temperature and main engine turbocharger exhaust outlet temperature.

3. The method for extracting stable navigation data of a real ship according to claim 1, characterized in that: In the noise cleaning and correlation extraction step, the host noise data includes the host cylinder head cooling water outlet temperature, the host cylinder liner cooling water outlet temperature, the host cylinder fuel side liner temperature, the host cylinder exhaust side liner temperature, the host cylinder piston cooling oil outlet temperature, the host cylinder scavenging box temperature, the host cylinder exhaust outlet temperature, the host turbocharger exhaust outlet pressure, the host turbocharger exhaust inlet temperature and the host turbocharger exhaust outlet temperature.

4. The method for extracting stable navigation data of a real ship according to claim 3, characterized in that: In the noise cleaning and correlation extraction steps, the characteristic value includes calculating the average value and standard deviation of the noise data of ten hosts.

5. The method for extracting stable navigation data of a real ship according to claim 1, characterized in that: In the noise cleaning and correlation extraction steps, the recursive feature elimination method includes training a classifier, calculating a permutation importance measure, eliminating irrelevant variables, and retraining the classifier using the eliminated features.

6. A real ship stable navigation data extraction system, characterized in that: It includes a data acquisition and extraction module, a noise cleaning and correlation extraction module, a window period determination module, an information entropy value calculation module and a stable navigation data judgment module, which are connected in sequence. The data collection and extraction module collects the ship data when the main engine working condition of the ship is stable during the voyage at regular intervals, and compares the difference of the main engine bow draft, the difference of the main engine stern draft and the difference of the main engine speed in the ship data collected in any adjacent time periods within a certain cycle with the corresponding preset thresholds. If they are all less than or equal to the corresponding preset thresholds, the ship data within the cycle is extracted; The noise cleaning and correlation extraction module calculates the characteristic value of the main engine noise data in the extracted ship data, and uses Chebyshev inequality to select a characteristic data set that meets a certain interval range from the characteristic value to complete the noise cleaning of the main engine noise data, and normalizes the main engine noise data after noise cleaning. Based on the normalized data, the recursive feature elimination method is used to extract the data with the greatest correlation with the main engine power in the ship data as the relevant characteristic parameters; The window period determination module uses the autocorrelation function to calculate the autocorrelation coefficient of the host power at different period interval lengths, and then obtains the change trend function of the autocorrelation coefficient of the host power with respect to the period interval length, and calculates the window period according to the change trend function; The information entropy value calculation module divides the noise-cleaned ship data according to the window period to obtain ship data of multiple periods, and calculates the maximum and minimum values ​​of each parameter in the relevant characteristic parameters within the interval range, and then divides the maximum and minimum values ​​equally to obtain multiple value distribution intervals, and calculates the information entropy values ​​of each relevant characteristic parameter in each period based on the value distribution interval and using the information entropy method; The stable navigation data judgment module determines whether the information entropy value of each relevant characteristic parameter in a certain period is less than the preset numerical threshold. If the information entropy value of any relevant characteristic parameter is less than the preset numerical threshold, the relevant characteristic parameters of the period are eliminated, and the characteristic parameter data set composed of the periods that are not eliminated is retained as the real ship stable navigation data set.

7. The real ship stable navigation data extraction system according to claim 6 is characterized in that: The ship data includes main engine bow draft, main engine stern draft, main engine speed, main engine power, water speed, main engine cylinder head cooling water outlet temperature, main engine cylinder liner cooling water outlet temperature, main engine cylinder fuel side liner temperature, main engine cylinder exhaust side liner temperature, main engine cylinder piston cooling oil outlet temperature, main engine cylinder scavenging box temperature, main engine cylinder exhaust outlet temperature, main engine turbocharger exhaust outlet pressure, main engine turbocharger exhaust inlet temperature and main engine turbocharger exhaust outlet temperature.

8. The real ship stable navigation data extraction system according to claim 6 is characterized in that: In the data acquisition and extraction module, the host noise data includes the host cylinder head cooling water outlet temperature, the host cylinder liner cooling water outlet temperature, the host cylinder fuel side liner temperature, the host cylinder exhaust side liner temperature, the host cylinder piston cooling oil outlet temperature, the host cylinder scavenging box temperature, the host cylinder exhaust outlet temperature, the host turbocharger exhaust outlet pressure, the host turbocharger exhaust inlet temperature and the host turbocharger exhaust outlet temperature.

9. The real ship stable navigation data extraction system according to claim 8 is characterized in that: The characteristic value includes calculating the average value and standard deviation of ten host noise data.

10. The real ship stable navigation data extraction system according to claim 6, characterized in that: The recursive feature elimination method in the noise cleaning and correlation extraction module includes training a classifier, calculating a permutation importance measure, eliminating irrelevant variables, and retraining the classifier using the eliminated features.

Images