A method for dynamic speed fine-tuning for ship speed optimization

Abstract

The present invention relates to a dynamic speed fine-tuning method for optimizing ship speed, which belongs to the field of ship technology and solves the problems of being unable to update models in real time and having large calculation amount and low efficiency in the prior art. The method comprises: pre-processing ship navigation data according to a fixed time period; using a clustering algorithm to divide the processed navigation data into multiple groups of data corresponding to different working conditions; establishing a functional relationship between nautical mile fuel consumption and main engine speed in each group of data based on a polynomial function and introducing an exponential forgetting factor to obtain multiple fitting models; obtaining current navigation data, clustering to obtain actual working conditions, and obtaining the minimum value point closest to the current main engine speed within the current main engine speed setting range in the current navigation data according to the fitting model corresponding to the actual working conditions. If the nautical mile fuel consumption of the minimum value point is less than the current nautical mile fuel consumption obtained according to the current main engine speed, the current main engine speed is adjusted to the minimum value point, otherwise the current main engine speed is kept unchanged. Dynamic speed fine-tuning is achieved.

Description

Technical Field

[0001] The invention relates to the technical field of ships, and in particular to a dynamic rotation speed fine-tuning method for optimizing ship speed. Background Art

[0002] The speed performance of a ship is one of the important technical performances of an operating ship. The speed of a ship has a significant impact on the safety of the ship's navigation, the operating efficiency of the enterprise, and energy conservation and consumption reduction. As the speed increases, the fuel consumption cost also increases significantly. However, in order to ensure the timely delivery of goods and complete the navigation mission, the speed cannot be reduced indefinitely. Ultimately, under the combined effect of the two, the speed is usually stable within a certain range and cannot be adjusted significantly. Therefore, reducing fuel consumption by fine-tuning the speed has become an important cost control method.

[0003] Existing speed optimization methods all establish a correspondence between speed and nautical mile fuel consumption, and then use this correspondence to find the speed that minimizes fuel consumption. Since the factors that affect nautical mile fuel consumption are very complex and many environmental factors are unpredictable, it is often only possible to analyze based on historical experience data to obtain an approximate empirical formula.

[0004] A common method is to use neural networks to approximate the functional relationship between nautical mile fuel consumption and speed. However, due to the large amount of computation required and the lack of interpretability of the neural network itself, it is not suitable as a model to provide dynamic speed optimization suggestions. Summary of the invention

[0005] In view of the above analysis, an embodiment of the present invention aims to provide a dynamic speed fine-tuning method for ship speed optimization, so as to solve the problems that the existing model cannot be updated in real time and the calculation amount is large and the efficiency is low.

[0006] The embodiment of the present invention provides a dynamic speed fine-tuning method for optimizing ship speed, comprising the following steps:

[0007] Preprocessing the ship navigation data according to a fixed time period to obtain processed navigation data;

[0008] Using a clustering algorithm, the processed navigation data is divided into a plurality of groups of data corresponding to different working conditions;

[0009] Based on the polynomial function and the introduction of exponential forgetting factor, the functional relationship between the fuel consumption in nautical miles and the main engine speed in each set of data is established, and multiple fitting models are obtained;

[0010] The current navigation data is obtained, and the actual working conditions are obtained by clustering. It is identified whether there is a minimum point in the fitting model corresponding to the actual working conditions within the set range of the current main engine speed in the current navigation data. If there is a minimum point, and the nautical mile fuel consumption of the minimum point closest to the current main engine speed is less than the current nautical mile fuel consumption obtained according to the current main engine speed, the main engine speed is adjusted to the minimum point, otherwise the main engine speed is kept unchanged.

[0011] Based on the further improvement of the above method, the navigation data includes: timestamp, wind speed, wind direction angle, stern draft, bow draft, main engine fuel inlet flow, main engine fuel outlet flow, actual speed and main engine speed;

[0012] Preprocessing of ship navigation data, including:

[0013] Eliminate false values, including: eliminating illegal values ​​or outliers;

[0014] Eliminate abnormal working values, including eliminating data with host speed of 0 and negative numbers;

[0015] According to the fixed time period, the mean of the speed change value is calculated and the data in the transition state is eliminated.

[0016] Based on the further improvement of the above method, the mean of the speed change value is calculated according to a fixed time period, and the data in the transition state is eliminated, including:

[0017] According to the order of timestamps in a fixed time period, the absolute value of the deviation of the host speed between every two adjacent timestamps is calculated in turn, and the average value of the speed change value is obtained by summing up and averaging;

[0018] It is determined in sequence whether the absolute value of the deviation value is greater than a preset multiple of the mean value. If so, the previous timestamp of each two adjacent timestamps is marked as a transition state, otherwise, it is marked as a stable state;

[0019] For the timestamp of the stable state, if the difference between the average speeds of the first preset time before and after exceeds the set value, all the timestamps within the first preset time before and after are re-marked as transition states;

[0020] Divide the time into multiple time periods according to the second preset time, identify whether there is a transition state in each time period, and if so, mark all timestamps between the earliest and latest transition states that appear within the second preset time before and after the timestamp of the transition state as transition states;

[0021] According to the timestamp marked as a transition state, the corresponding entire navigation data is deleted to obtain the processed navigation data.

[0022] Based on the further improvement of the above method, the working condition is obtained by combining the relative wind speed along the sailing direction and the average draft depth.

[0023] Based on the further improvement of the above method, the order of the polynomial function is greater than the number of forward gears of the main engine of the ship and less than 10.

[0024] Based on the further improvement of the above method, the exponential forgetting factor is related to time, and the calculation formula is:

[0025]

[0026] Among them, x j represents the main engine speed in the jth processed navigation data, Δt j Represents x j The absolute value of the time difference from the acquisition time to the current time. α represents the preset attenuation factor, 0.01≤α≤0.1.

[0027] Based on the further improvement of the above method, the polynomial function is to calculate the fuel consumption per nautical mile according to the main engine speed, and the coefficient matrix of the fitting model is solved by establishing the inner product matrix and the constant term matrix;

[0028] Based on the exponential forgetting factor and the value of the inner product matrix corresponding to the Nth main engine speed in the processed navigation data Recursively obtain the value in the inner product matrix corresponding to the N+1th host speed The calculation formula is:

[0029]

[0030] Among them, 0≤i≤M,0≤k≤M, M represents the order of the polynomial, and represents the basis function of the polynomial function, x N+1 Indicates the N+1th host speed, t N+1 -t N It represents the acquisition time difference between the N+1th host speed and the Nth host speed, α represents the preset attenuation factor, Represents the value of the i+1th row and k+1th column in the inner product matrix.

[0031] Based on the further improvement of the above method, based on the exponential forgetting factor and the value in the constant term matrix corresponding to the N-th nautical mile fuel consumption in the processed voyage data Recursively obtain the value in the constant term matrix corresponding to the N+1th nautical mile fuel consumption The calculation formula is:

[0032]

[0033] Where f(x) represents the fuel consumption in nautical miles, f(x N+1 ) represents the fuel consumption at the N+1th nautical mile, Represents the value of the k+1th row in the constant term matrix.

[0034] Based on a further improvement of the above method, the setting range of the current main engine speed is obtained by floating the current main engine speed up and down according to a preset speed floating value.

[0035] Based on the further improvement of the above method, in each fixed time period, the fitting model corresponding to each working condition is refitted according to the navigation data processed in the current fixed time period, covering the fitting model of the corresponding working condition in the previous fixed time period.

[0036] Compared with the prior art, the present invention can achieve at least one of the following beneficial effects:

[0037] 1. Fitting multiple models according to different working conditions makes the analysis of fuel consumption curve more refined and obtains more accurate fitting results;

[0038] 2. For the first time, a polynomial function is used to fit the relationship curve between the main engine speed and the nautical mile fuel consumption. Considering that the nautical mile fuel consumption shows different characteristics under different forward gears, the order of the polynomial function is set to be greater than the number of main engine forward gears of the ship. The exponential forgetting factor is introduced into the inner product definition, and a recursive expression is obtained, which avoids repeated calculations and greatly improves the efficiency of the traditional fitting algorithm.

[0039] 3. At fixed intervals, a new model is fitted based on the new data collected, so that the model is updated in real time, providing accurate speed fine-tuning suggestions, effectively reducing fuel consumption costs;

[0040] 4. Combined with the actual law of ship operation, the slow change of speed and the unstable state of insufficient duration are taken into account, and the real-time collected ship navigation data is processed for stability judgment to obtain more reasonable and accurate basic data.

[0041] In the present invention, the above-mentioned technical solutions can also be combined with each other to achieve more preferred combination solutions. Other features and advantages of the present invention will be described in the subsequent description, and some advantages can become obvious from the description, or can be understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained through the contents particularly pointed out in the description and the drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] The drawings are only for the purpose of illustrating particular embodiments and are not to be considered limiting of the present invention. Like reference symbols denote like components throughout the drawings.

[0043] Figure 1 is a flow chart of the dynamic speed fine-tuning method in Embodiment 1 of the present invention;

[0044] Figure 2This is a diagram of the model fitting effects under 6 working conditions in Example 2 of the present invention. DETAILED DESCRIPTION

[0045] The preferred embodiments of the present invention are described in detail below in conjunction with the accompanying drawings, wherein the accompanying drawings constitute a part of this application and are used together with the embodiments of the present invention to illustrate the principles of the present invention, but are not used to limit the scope of the present invention.

[0046] Embodiment 1,

[0047] A specific embodiment of the present invention discloses a dynamic speed fine-tuning method for optimizing ship speed, such as Figure 1 As shown, the following steps are included:

[0048] S11: pre-processing the ship navigation data according to a fixed time period to obtain processed navigation data;

[0049] Specifically, the ship's navigation data is obtained through sensors installed on the ship in advance, including: timestamp, wind speed, wind direction angle, stern draft, bow draft, main engine fuel inlet flow, main engine fuel outlet flow, actual speed and main engine speed.

[0050] Preferably, the fixed time period is the average time for crossing a sea area. Exemplarily, the fixed time period is set to 6 hours.

[0051] The preprocessing process of ship navigation data includes:

[0052] S111: Eliminate false values, including: eliminate illegal values ​​or outliers;

[0053] It should be noted that, in actual collection, some illegal values ​​often appear, such as values ​​recorded as NAN, or a few outliers, such as values ​​that exceed the normal range by 3 to 5 times. These are mainly false values ​​recorded by the sensor due to hardware failure, and are directly discarded. The normal range of various data in the above-mentioned navigation data belongs to the common knowledge of those skilled in the art, and will not be described in detail in this embodiment.

[0054] S112: Eliminating abnormal working values, including eliminating data of host speed being 0 and negative numbers;

[0055] It should be noted that a rotation speed of 0 indicates a stop, and a negative rotation speed indicates data recorded when the ship is reversing.

[0056] S113: Calculate the average of the speed change values ​​according to the fixed time period, and remove the data in the transition state.

[0057] Specifically, the process includes: preliminary stability judgment, identification of slowly changing false stable states, and identification of false stable states that do not last long enough. Assume that after processing of step S111 and step S112, there are still n navigation data in the current fixed time period.

[0058] ① In the initial stability judgment process, according to the order of timestamps in a fixed time period, the absolute value of the deviation of the host speed between each two adjacent timestamps is calculated in sequence, and the average value of the speed change is obtained by summing up and averaging. The calculation formula is:

[0059]

[0060] Among them, v i is the time t i The host speed, t1,t n are the first and last timestamps of the segment, and the time difference is the length of the fixed time period. i+1 -v i |Indicates the deviation value of the host speed between two adjacent timestamps;

[0061] It is determined in sequence whether the absolute value of the deviation value is greater than a preset multiple of the mean value. If so, the previous timestamp of each two adjacent timestamps is marked as a transition state, otherwise, it is marked as a stable state;

[0062] It should be noted that if the preset multiple is set too large, not enough transition state points will be screened out. If it is set too small, a large number of stable states will also be judged as transition states. The preset multiple needs to be debugged based on engineering experience.

[0063] For example, the preset multiples are tested from large to small. As the multiple decreases, the number of points in the stable state that are retained gradually increases. When the parameter decreases to around 4.267, obvious speed fluctuations begin to appear in the stable state, indicating that further reducing the multiple will only lead to more misjudgments. Therefore, in this embodiment, 4.267 is selected as the empirical parameter value. When |v i+1 -v i |>4.267*mean(Δv), t i Labeled as transition state.

[0064] ② In the process of identifying the slowly changing pseudo-stable state, for the timestamp of the stable state, if the difference between the average speed of the first preset time before and after exceeds the set value, all the timestamps within the first preset time before and after are re-marked as transition states;

[0065] For example, the first preset time is set to 5 minutes, and the setting value of the mean difference is 1 rpm (Revolutions Per Minute), which means: if the difference between the mean speeds 5 minutes before and after the stable state timestamp exceeds 1 rpm, all timestamps within these 10 minutes are re-marked as transition states;

[0066] ③ In the process of identifying a pseudo stable state that does not last long enough, the time is divided into multiple time periods according to the second preset time, and it is identified whether there is a transition state in each time period. If there is, all timestamps between the earliest and latest transition states that appear within the second preset time before and after the timestamp of the transition state are marked as transition states;

[0067] Exemplarily, based on the normal situation that the stable state lasts for at least 30 minutes, the second preset time is set to 20 minutes, and then judging 20 minutes before and after each transition state timestamp is equivalent to identifying a false stable state timestamp within 40 minutes.

[0068] According to the timestamp marked as a transition state, the corresponding entire navigation data is deleted to obtain the processed navigation data.

[0069] S12: using a clustering algorithm, dividing the processed navigation data into multiple groups of data corresponding to different working conditions;

[0070] It should be noted that during the navigation of a ship, there are two main external factors that have a key impact on the navigation status: the relative wind speed along the navigation direction and the average draft depth. The operating conditions are obtained by combining these two factors.

[0071] Relative wind speed v along the sailing direction r The calculation formula is:

[0072] v r =v wind cos(θ wind ) Formula (2)

[0073] The calculation formula for the average draft d is:

[0074]

[0075] Among them, v wind represents wind speed, θ wind Indicates wind direction angle, d0 indicates stern draft, and d1 indicates bow draft.

[0076] For example, the relative wind speed along the sailing direction is divided into 6 sections, and the average draft corresponds to the two states of no-load and full-load, so a total of 12 different working conditions can be combined. The K-means clustering algorithm is used to divide the two-dimensional data composed of relative wind speed and average draft to obtain working conditions 1 to 12.

[0077] S13: Based on the polynomial function and introducing the exponential forgetting factor, the functional relationship between the nautical mile fuel consumption and the main engine speed in each set of data is established to obtain multiple fitting models;

[0078] The main engine fuel inlet flow rate Q in the voyage data in , Main engine fuel outlet flow Q out and actual speed v ship , get the nautical mile fuel consumption P, the calculation formula is:

[0079] P=(Q in -Q out ) / v ship Formula (4)

[0080] Considering that the same main engine speed corresponds to completely different nautical fuel consumption under different working conditions, for example, the fuel consumption is higher when against the wind and lower when idling, different models need to be fitted for data under different working conditions, but the calculation process is exactly the same.

[0081] The numerical methods represented by neural networks have the problems of large computational complexity and low efficiency in fitting. Considering that the fuel consumption per nautical mile and the main engine speed are generally positively correlated and not periodic, and that the fuel consumption per nautical mile will not increase rapidly and explosively with the increase of the main engine speed, the use of polynomial functions is a more appropriate approach.

[0082] For the polynomial fitting model, it is necessary to select a suitable order. Considering that the fuel consumption in different nautical miles exhibits different characteristics under different forward gears, in order to ensure that there is a minimum point in each forward gear for speed adjustment, the order of the polynomial function must be greater than the number of forward gears of the ship's main engine. In addition, combined with the mathematical theory of polynomial functions: an n-order polynomial has at most n-1 minimum points; if the highest-order coefficient of an even-order polynomial is positive, there must be at least one minimum point; the higher the order of the polynomial, the easier it is to overfit. The order of the polynomial function is less than 10. Preferably, it is set to an even number within the range of greater than the number of forward gears of the ship's main engine and less than 10.

[0083] Exemplarily, the forward gears of the ship's main engine include: slow forward, forward one, forward two, forward three and high-speed cruising, and the 8th-order polynomial is preferentially selected for fitting the model. However, since the characteristics of high-order polynomials cannot be fully characterized, when all minimum points of the 8th-order polynomial are outside the actual speed range, it is adjusted to a 7th-order or 9th-order polynomial.

[0084] The polynomial function calculates the fuel consumption per nautical mile f(x) based on the main engine speed x. Let the actual fuel consumption per nautical mile be f. When the error between the polynomial function f(x) and f is the smallest, the fitting effect is the best. The polynomial function that minimizes the sum of squared errors is the best square approximation polynomial function of f. Let M be the polynomial order, and the set of all M-order polynomial functions is S, where any polynomial function is s, and the best square approximation polynomial function is s * ,but:

[0085]

[0086] Among them, ||*||2 represents the 2-norm, and the calculation formula is:

[0087]

[0088] Among them, c represents the lower limit of the integral, and b represents the upper limit of the integral.

[0089] Suppose a set of basis functions of the polynomial function is Then any M-order polynomial function s can be expressed as:

[0090]

[0091] Among them, a represents the coefficient of the polynomial, a i represents the i-th basis function The previous coefficient.

[0092] Substituting formula (7) into formula (5), we need to make the left side of (5) s * is the best square approximation, which is equivalent to minimizing the following multivariate function:

[0093]

[0094] According to the necessary conditions for multivariate functions to take extreme values: the partial derivatives of each variable are zero, we can get:

[0095]

[0096] Where k also represents the order, formula (9) can be simplified as:

[0097]

[0098] Convert formula (10) to the matrix form in formula (11), and calculate the (M+1)*1 coefficient matrix [a] of the fitting model through the (M+1)*(M+1)-dimensional inner product matrix and the (M+1)*1-dimensional constant term matrix:

[0099]

[0100] in, The value corresponding to the i+1th row and k+1th column in the inner product matrix is ​​abbreviated in the matrix as In the constant term matrix, it is abbreviated as d k , corresponding to the value of the k+1th row. Since the inner product operation satisfies the commutative law, the inner product matrix is ​​a symmetric matrix, satisfying

[0101] In this embodiment, in the processed navigation data, the main engine speed and the calculated nautical mile fuel consumption are discrete data points, so the integral is discretized into a summation form. Assuming that there are N processed navigation data at present, the inner product in formula (11) is defined as:

[0102]

[0103] The constant term in formula (11) is defined as:

[0104]

[0105] Where j represents the navigation data after the jth processing, x j represents the main engine speed in the jth processed navigation data, f(x j ) represents the fuel consumption in nautical miles obtained based on the main engine fuel inlet flow, main engine fuel outlet flow and actual speed in the voyage data processed in Article j.

[0106] Considering that the external environment is constantly changing as the ship sails, the closer the point is to the current moment, the less its working environment has changed compared to the current moment. Conversely, the farther away from the current moment, the lower its reference value. In order to highlight the different effects of different data points on the model, the time-related exponential forgetting factor is introduced. The calculation formula is:

[0107]

[0108] Among them, x j represents the main engine speed in the jth processed navigation data, Δt j Represents x j The absolute value of the time difference from the acquisition time to the current time, α represents a preset attenuation factor, 0.01≤α≤0.1, the larger α is, the faster the attenuation is, and α=0 represents no attenuation. Preferably, α=0.1 is set.

[0109] After introducing formula (14) into formula (12) and formula (13), the calculation formula is:

[0110]

[0111]

[0112] It can be seen from formula (15) and formula (16) that each data in the inner product matrix and the constant term matrix needs to be calculated starting from j = 1. Considering that when collecting the N+1th point, the collection time increases by t relative to the Nth point N+1 -t N , and so on, the acquisition time of each point increases by t relative to the previous point N+1 -t N ,Therefore, in order to avoid repeated calculation and improve ,computational efficiency, a recursive relationship is established to ,calculate the relevant data of the current point based on the previous point and ,the exponential forgetting factor.

[0113] Specifically, based on the exponential forgetting factor and the value of the inner product matrix corresponding to the Nth main engine speed in the processed navigation data Recursively obtain the value in the inner product matrix corresponding to the N+1th host speed Update formula (15) to formula (17):

[0114]

[0115] Among them, x N+1 It indicates the N+1th main engine speed, that is, the main engine speed in the N+1th processed navigation data.

[0116] Similarly, based on the exponential forgetting factor and the value in the constant term matrix corresponding to the N-th nautical mile fuel consumption in the processed navigation data Recursively obtain the value in the constant term matrix corresponding to the N+1th nautical mile fuel consumption Update formula (16) to formula (18):

[0117]

[0118] Among them, f(x N+1 ) represents the fuel consumption at the N+1th nautical mile, i.e., the fuel consumption at the nautical mile is obtained based on the main engine fuel inlet flow, main engine fuel outlet flow and actual speed of the voyage data processed for the N+1th item.

[0119] Basis Functions It can be in various forms, such as the simplest x to the power of k, Legendre polynomials; in this embodiment, considering that selecting Legendre polynomials as basis functions can make the inner product matrix condition number better than monomials, and the fitting result is more accurate, Legendre polynomials are used as basis functions, and are defined as follows:

[0120]

[0121] Based on formula (19), the updated formula (17) and formula (18) are used to calculate the inner product matrix and constant term matrix in formula (11), and the coefficients in the coefficient matrix are obtained to obtain the fitting model.

[0122] For each set of data corresponding to different working conditions, multiple fitting models are obtained according to the above method.

[0123] S14: Acquire current navigation data, cluster to obtain actual working conditions, and obtain the minimum point closest to the current main engine speed within the set range of the current main engine speed in the current navigation data according to the fitting model corresponding to the actual working conditions. If the nautical mile fuel consumption at the minimum point is less than the current nautical mile fuel consumption obtained according to the current main engine speed, adjust the current main engine speed to the minimum point, otherwise keep the current main engine speed unchanged.

[0124] Specifically, the fitting model establishes a polynomial function relationship between nautical mile fuel consumption and main engine speed. By taking the derivative, it is easy to obtain the minimum point of the function and the main engine speed value that minimizes the nautical mile fuel consumption.

[0125] The minimum point of each fitting model is obtained according to the following derivative definition:

[0126]

[0127] The setting range of the current host speed is obtained by floating the current host speed up and down according to the preset speed floating value. If there are multiple minimum points in this range, the minimum point with the closest distance is selected. If there is no minimum point in this range, the current host speed remains unchanged.

[0128] Exemplarily, the speed floating value is set to 5rpm, that is, to identify whether there is a minimum point within the current host speed range of ±5rpm. If there are multiple minimum points, calculate the Euclidean distance from the current speed to each minimum point, and select the minimum point with the closest distance.

[0129] Preferably, if there is no minimum point of the fitting model within the setting range of the current rotation speed for a continuous period of time, such as 6 hours, the order of the polynomial function is adjusted, illustratively, from the 8th order to the 7th order.

[0130] The current main engine speed is passed into the fitting model to obtain the function value of the nautical mile fuel consumption, and then compared with the nautical mile fuel consumption corresponding to the minimum point closest to the current main engine speed. If the nautical mile fuel consumption corresponding to the minimum point is smaller, the current main engine speed will be adjusted to the minimum point, otherwise the current main engine speed will remain unchanged.

[0131] It should be noted that there is a time difference between the adjustment time of the host speed and the current time. It can be set according to the actual situation and a recommended adjustment time can be given.

[0132] Preferably, the current slip rate is calculated based on the current main engine speed, the actual speed and the pitch of the ship, and the real-time speed is calculated from the current slip rate and the adjusted main engine speed, and compared with the actual speed. If the difference is within a threshold range, a main engine speed adjustment suggestion is issued, including: recommended adjustment time, recommended speed, expected speed, expected slip rate and speed change, wherein the speed change indicates speed increase or speed decrease; if the difference exceeds the threshold range, the main engine speed is kept unchanged and a warning message is issued to ensure the correctness of the adjustment.

[0133] In each fixed time period, the fitting model corresponding to each working condition is refitted according to the navigation data processed in the current fixed time period, covering the fitting model of the corresponding working condition in the previous fixed time period.

[0134] During implementation, the collected ship navigation data in the new time period is preprocessed, and a new fitting model is established. At the same time, the current real-time main engine speed is passed into the corresponding existing latest fitting model to determine whether the main engine speed needs to be adjusted according to the minimum point. The fitting model is iteratively updated with the real-time collected data, and the speed adjustment suggestion is also determined according to the latest fitting model at that time.

[0135] Compared with the prior art, the present invention provides a dynamic speed fine-tuning method for ship speed optimization, which combines the actual law of ship operation, takes into account the non-stable state of slow speed change and insufficient duration, and performs stability judgment processing on the ship navigation data collected in real time to obtain more reasonable and accurate basic data; for the first time, a polynomial function is used to fit the relationship curve between the main engine speed and the nautical mile fuel consumption, and considering that the nautical mile fuel consumption shows different characteristics under different forward gears, the order of the polynomial function is set to be greater than the number of main engine forward gears of the ship, the exponential forgetting factor is introduced into the inner product definition, and a recursive expression is obtained, which avoids repeated calculations and greatly improves the efficiency of traditional fitting algorithms; according to the relative wind speed and average draft depth, different working conditions are combined to fit multiple models, so that the analysis of the fuel consumption curve is more precise and a more accurate fitting result can be obtained; at fixed time intervals, a new model is fitted according to the new data collected, so that the model is iteratively updated in real time, and accurate speed fine-tuning suggestions are provided, which effectively reduces the fuel consumption cost.

[0136] Embodiment 2,

[0137] Taking a cargo ship sailing on a route from Corjur Christi (UTC-5) in the United States to Kaohsiung (UTC+8) in China with full load as an example, the specific method of Example 1 is used to fine-tune the rotation speed.

[0138] Since there is only one load state, when dividing the working conditions, only the relative wind speed along the sailing direction is considered, which is divided into 6 different working conditions. The 8th-order polynomial is used, and the 6 fitting models obtained are as follows: Figure 2 As shown, the hollow circles mark the minimum points.

[0139] Based on actual navigation data, a total of 25 speed adjustments were made during the entire flight segment. Compared with the originally planned fuel consumption, the final fuel consumption after the speed adjustment reduced carbon dioxide emissions by 342.83 tons, saved 110.08 tons of fuel, and saved 4.11% of the total fuel consumption.

[0140] Those skilled in the art will appreciate that all or part of the processes of the above-mentioned embodiments can be implemented by controlling related hardware through a computer program, and the program can be stored in a computer-readable storage medium, wherein the computer-readable storage medium is a disk, an optical disk, a read-only storage memory, or a random access memory, etc.

[0141] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by any technician familiar with the technical field within the technical scope disclosed by the present invention should be covered within the protection scope of the present invention.

Claims

1. A dynamic speed fine-tuning method for optimizing ship speed, characterized in that: The steps include: Preprocessing the ship navigation data according to a fixed time period to obtain processed navigation data; Using a clustering algorithm, the processed navigation data is divided into a plurality of groups of data corresponding to different working conditions; Based on the polynomial function and the introduction of exponential forgetting factor, the functional relationship between the fuel consumption in nautical miles and the main engine speed in each set of data is established, and multiple fitting models are obtained; The current navigation data is obtained, and the actual working condition is obtained by clustering. According to the fitting model corresponding to the actual working condition, the minimum value point closest to the current main engine speed is obtained within the setting range of the current main engine speed in the current navigation data. If the nautical mile fuel consumption at the minimum value point is less than the current nautical mile fuel consumption obtained according to the current main engine speed, the current main engine speed is adjusted to the minimum value point, otherwise the current main engine speed is kept unchanged; The exponential forgetting factor is related to time, and the calculation formula is: Among them, x j represents the main engine speed in the jth processed navigation data, Δt j Represents x j The absolute value of the time difference from the acquisition time to the current time, α represents the preset attenuation factor, 0.01≤α≤0.1; By establishing the inner product matrix and the constant term matrix of the polynomial function, solving the coefficient matrix of the fitting model; Based on the exponential forgetting factor and the value of the inner product matrix corresponding to the Nth main engine speed in the processed navigation data Recursively obtain the value in the inner product matrix corresponding to the N+1th host speed The calculation formula is: Among them, 0≤i≤M,0≤k≤M, M represents the order of the polynomial, and represents the basis function of the polynomial function, x N+1 Indicates the N+1th host speed, (t N+1 -t N ) represents the acquisition time difference between the N+1th host speed and the Nth host speed, Represents the value of the i+1th row and k+1th column in the inner product matrix; Based on the exponential forgetting factor and the value in the constant term matrix corresponding to the Nth nautical mile fuel consumption in the processed navigation data Recursively obtain the value in the constant term matrix corresponding to the N+1th nautical mile fuel consumption The calculation formula is: Where f(x) represents the fuel consumption in nautical miles, f(x N+1 ) represents the fuel consumption at the N+1th nautical mile, Represents the value of the k+1th row in the constant term matrix.

2. The dynamic speed fine-tuning method for optimizing ship speed according to claim 1 is characterized in that: Said navigation data includes: timestamp, wind speed, wind direction angle, stern draft, bow draft, main engine fuel inlet flow, main engine fuel outlet flow, actual speed and main engine rotation speed; The pre-processing of ship navigation data comprises: Eliminate false values, including: eliminating illegal values ​​or outliers; Eliminate abnormal working values, including eliminating data with host speed of 0 and negative numbers; According to the fixed time period, the mean of the speed change value is calculated and the data in the transition state is eliminated.

3. The dynamic speed fine-tuning method for ship speed optimization according to claim 2 is characterized in that: The method of calculating the average of the speed change values ​​according to the fixed time period and eliminating the data in the transition state includes: According to the order of timestamps in a fixed time period, the absolute value of the deviation of the host speed between every two adjacent timestamps is calculated in turn, and the average value of the speed change value is obtained by summing up and averaging; It is determined in sequence whether the absolute value of the deviation value is greater than a preset multiple of the mean value, and if so, the previous timestamp of each two adjacent timestamps is marked as a transition state, otherwise, it is marked as a stable state; For the timestamp of the stable state, if the difference between the average speeds of the first preset time before and after exceeds the set value, all the timestamps within the first preset time before and after are re-marked as transition states; Divide the time into multiple time periods according to the second preset time, identify whether there is a transition state in each time period, and if so, mark all timestamps between the earliest and latest transition states that appear within the second preset time before and after the timestamp of the transition state as transition states; According to the timestamp marked as a transition state, the corresponding entire navigation data is deleted to obtain the processed navigation data.

4. The dynamic speed fine-tuning method for optimizing ship speed according to claim 1 or 3, characterized in that: The working condition is obtained by combining the relative wind speed along the sailing direction and the average draft.

5. The dynamic speed fine-tuning method for optimizing ship speed according to claim 4 is characterized in that: The order of the polynomial function is greater than the number of forward gears of the main engine of the ship and is less than 10.

6. The dynamic speed fine-tuning method for optimizing ship speed according to claim 5 is characterized in that: The setting range of the current main engine speed is obtained by floating the current main engine speed up and down according to a preset speed floating value.

7. The dynamic speed fine-tuning method for optimizing ship speed according to claim 6 is characterized in that: In each of the fixed time periods, the fitting models corresponding to the various working conditions are refitted based on the navigation data processed in the current fixed time period, covering the fitting models corresponding to the working conditions in the previous fixed time period.

Images