A ship engine-propeller intelligent matching method and system based on multi-source fusion algorithm

Abstract

The present invention relates to the technical field of ship propulsion system design, specifically disclosing a method and system for intelligent ship engine-propeller matching based on a multi-source fusion algorithm. The method comprises: optimizing a preset brake horsepower calculation rule based on ship-related parameters to generate an optimized horsepower calculation rule; then performing nonlinear coupling on the optimized horsepower calculation rule based on an effective horsepower demand calculation rule to generate a coupled horsepower calculation rule; and finally, performing an engine-propeller matching operation based on a propeller map, ship design parameters, the coupled horsepower calculation rule, and a matching horsepower coefficient to obtain a corresponding propeller model. By performing an engine-propeller matching operation based on the propeller map, ship design parameters, the coupled horsepower calculation rule, and the matching horsepower coefficient to obtain a corresponding propeller model, multi-model data fusion is implemented to match propeller models, thereby improving propeller matching efficiency.

Description

Technical Field

[0001] The present invention relates to the technical field of ship propulsion system design, and in particular to a ship engine-propeller intelligent matching method and system based on a multi-source fusion algorithm. Background Art

[0002] A ship's propulsion system is a key component of ship design, and its performance directly affects the ship's speed and economic efficiency. Matching the propeller with the engine is a crucial step in ship propulsion system design. Proper matching can improve propulsion efficiency and reduce energy consumption. However, existing propeller-engine matching technology has the following problems:

[0003] 1. Strong dependence on experience: Relying on a single formula or historical graph, and failing to achieve multi-model data fusion.

[0004] 2. Poor dynamic adaptability: The real-time impact of navigation conditions (such as wind and waves, load) on the matching effect is not considered.

[0005] 3. Low efficiency of graph matching: Manual graph search has large errors and low efficiency; the iterative optimization process does not introduce a machine learning acceleration mechanism. Summary of the Invention

[0006] In order to overcome the above-mentioned technical problems existing in the prior art, an embodiment of the present invention provides a ship engine-propeller intelligent matching method and system based on a multi-source fusion algorithm. By nonlinearly coupling the optimized horsepower calculation rule and the effective horsepower demand calculation rule, a coupled horsepower calculation rule is generated, and a main engine-propeller matching operation is performed based on the propeller map, the ship design parameters, the coupled horsepower calculation rule and the matching horsepower coefficient to obtain the corresponding propeller model, thereby realizing multi-model data fusion matching of propeller models and improving propeller matching efficiency.

[0007] To achieve the above objectives, an embodiment of the present invention provides a ship engine-propeller intelligent matching method based on a multi-source fusion algorithm, the method comprising:

[0008] Obtain ship design parameters and preset brake horsepower calculation rules;

[0009] Optimizing the preset brake horsepower calculation rule based on the ship design parameters to generate an optimized horsepower calculation rule;

[0010] determining an effective horsepower demand calculation rule, and performing nonlinear coupling on the optimized horsepower calculation rule based on the effective horsepower demand calculation rule to generate a coupled horsepower calculation rule;

[0011] Obtaining performance parameters of the ship's main engine, and determining a matching horsepower coefficient based on the performance parameters;

[0012] A propeller map is obtained, and a main engine-propeller matching operation is performed based on the propeller map, the ship design parameters, the coupled horsepower calculation rule, and the matching horsepower coefficient to obtain a corresponding propeller model.

[0013] Preferably, the optimizing the preset brake horsepower calculation rule based on the ship design parameters to generate the optimized horsepower calculation rule includes:

[0014] The preset brake horsepower calculation rule is determined, and the preset brake horsepower calculation rule is characterized as follows:

[0015]

[0016] Determine the dynamic correction function of hull resistance based on hull load and wind and wave environment;

[0017] The preset brake horsepower calculation rule is optimized based on the hull resistance dynamic correction function to generate an optimized horsepower calculation rule, which is characterized as follows:

[0018]

[0019] Among them, K represents the preset horsepower coefficient, Δ represents the ship's design displacement, V represents the ship's design speed, C represents the empirical constant, f(Rt,α) represents the dynamic correction function of the hull resistance, Rt represents the resistance correction amount, and α represents the real-time dynamic factor of the navigation state.

[0020] Preferably, the determining of the effective horsepower requirement calculation rule includes:

[0021] Get the Hankscher formula;

[0022] Determine propeller thrust and torque based on the Hankshall formula and the ship design speed;

[0023] The propeller thrust and the torque are interpolated based on the propeller map to obtain an interpolated thrust T and an interpolated torque Q. The interpolated thrust T and the interpolated torque Q are respectively represented as follows: ,

[0024] Where ρ represents the fluid density, n represents the propeller speed, D represents the propeller diameter, and J represents the advance coefficient. Based on the real-time matching determination of the propeller map, K T Characterized by the propeller thrust coefficient, K Q Characterized as the torque coefficient of the propeller;

[0025] An effective horsepower requirement calculation rule is generated based on the interpolated thrust T and the interpolated torque Q.

[0026] Preferably, performing nonlinear coupling on the optimized horsepower calculation rule based on the effective horsepower demand calculation rule to generate a coupled horsepower calculation rule includes:

[0027] Obtaining the optimized horsepower calculation rule and the effective horsepower requirement calculation rule;

[0028] respectively obtaining dynamic weights of the optimized horsepower calculation rule and the effective horsepower demand calculation rule;

[0029] A coupled horsepower calculation rule is obtained based on the dynamic weight, the optimized horsepower calculation rule, and the effective horsepower demand calculation rule. The coupled horsepower calculation rule is characterized as follows:

[0030]

[0031] in , T is the propeller interpolation thrust, η prop is the propeller efficiency, determined by propeller map matching, 、 is the weight before adjustment.

[0032] Preferably, the performance parameters include main engine speed, main engine power and main engine torque, and determining the matching horsepower coefficient based on the performance parameters includes:

[0033] Generate a corresponding matching horsepower coefficient K based on the main engine speed, the main engine power, and the main engine torque m , the horsepower coefficient K m Characterized by:

[0034]

[0035] Among them, P max is the rated power of the ship's main engine, M max is the maximum torque of the ship's main engine, n rated It is the speed when the main engine of the ship reaches the rated power, and L is the unit conversion constant.

[0036] Preferably, performing the main engine-propeller matching operation to obtain the corresponding propeller model includes:

[0037] Obtain the initial genetic algorithm and initial weights;

[0038] Make real-time predictions of the ship's navigation state and generate navigation state prediction results;

[0039] Adjusting the initial weight in real time based on the flight state prediction result to generate an adjusted weight;

[0040] The initial genetic algorithm is optimized based on the ship design speed, the coupled horsepower calculation rule, and the main engine speed to generate a dynamic matching objective function. The dynamic matching objective function is characterized as follows:

[0041]

[0042] Among them, ω1, ω2 represent the adjusted weights, η prop is the propeller efficiency, BHP engine is the engine brake horsepower, BHP prop is the propeller brake horsepower, BHP max Characterized as the maximum brake horsepower of the ship's main engine;

[0043] Determining an optimal propeller diameter-to-pitch ratio based on the dynamic matching objective function;

[0044] A main engine-propeller matching operation is performed based on the optimal propeller diameter and pitch ratio to obtain a corresponding propeller model.

[0045] Preferably, the real-time prediction of the navigation state of the ship and the generation of the navigation state prediction result include:

[0046] Get the long short-term memory network LSTM;

[0047] The navigation state of the ship is predicted in real time based on the LSTM to generate a navigation state prediction result.

[0048] Preferably, the method further comprises:

[0049] After obtaining the propeller model, obtaining a preset propulsion performance requirement;

[0050] calibrating the propeller model based on the preset propulsion performance requirements and generating a calibration result;

[0051] The propeller model is adjusted based on the verification result to obtain an adjusted propeller model.

[0052] On the other hand, the present invention also provides a ship engine-propeller intelligent matching system based on a multi-source fusion algorithm, the system comprising:

[0053] The first acquisition module is used to obtain ship design parameters and preset brake horsepower calculation rules;

[0054] an optimization module, configured to optimize the preset brake horsepower calculation rule based on the ship design parameters to generate an optimized horsepower calculation rule;

[0055] a coupling module, configured to determine an effective horsepower demand calculation rule, and perform nonlinear coupling on the optimized horsepower calculation rule based on the effective horsepower demand calculation rule to generate a coupled horsepower calculation rule;

[0056] A second acquisition module is used to obtain performance parameters of the ship's main engine and determine a matching horsepower coefficient based on the performance parameters;

[0057] The matching module is used to obtain a propeller map, perform a main engine-propeller matching operation based on the propeller map, the ship design parameters, the coupled horsepower calculation rule and the matching horsepower coefficient, and obtain a corresponding propeller model.

[0058] On the other hand, the present invention further provides a computer-readable storage medium having a computer program stored thereon, which implements the method provided by an embodiment of the present invention when the program is executed by a processor.

[0059] The technical solution provided by the present invention has at least the following technical effects:

[0060] By performing nonlinear coupling on the optimized horsepower calculation rule and the effective horsepower requirement calculation rule, a coupled horsepower calculation rule is generated, and a main engine-propeller matching operation is performed based on the propeller map, the ship design parameters, the coupled horsepower calculation rule, and the matching horsepower coefficient to obtain the corresponding propeller model. This realizes multi-model data fusion matching of propeller models and improves propeller matching efficiency.

[0061] Other features and advantages of the embodiments of the present invention will be described in detail in the subsequent detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] The accompanying drawings are used to provide a further understanding of the embodiments of the present invention and constitute a part of the specification. Together with the following detailed description, they are used to explain the embodiments of the present invention, but do not constitute a limitation of the embodiments of the present invention. In the accompanying drawings:

[0063] Figure 1 1 is a flow chart of a method for intelligently matching ship engine and propeller based on a multi-source fusion algorithm according to an embodiment of the present invention;

[0064] Figure 2 Schematic diagram of the structure of the ship engine-propeller intelligent matching optimization system shown in an embodiment of the present invention. DETAILED DESCRIPTION

[0065] The following describes the specific implementation of the embodiment of the present invention in detail with reference to the accompanying drawings. It should be understood that the specific implementation described herein is only used to illustrate and explain the embodiment of the present invention and is not used to limit the embodiment of the present invention.

[0066] The terms "system" and "network" in the embodiments of the present invention can be used interchangeably. "Multiple" refers to two or more. In view of this, "multiple" can also be understood as "at least two" in the embodiments of the present invention. "And / or" describes the association relationship of associated objects, indicating that three relationships can exist. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / ", unless otherwise specified, generally indicates that the previous and next associated objects are in an "or" relationship. In addition, it should be understood that in the description of the embodiments of the present invention, words such as "first" and "second" are only used to distinguish the purpose of description, and cannot be understood as indicating or implying relative importance, nor can they be understood as indicating or implying order.

[0067] In actual application, existing ship-engine-propeller matching technology usually uses a single formula or historical map for matching, and does not consider the real-time impact of navigation conditions on the matching effect. Therefore, the existing ship-engine-propeller matching technology is prone to poor matching effect and low matching efficiency.

[0068] To solve the above problem, see Figure 1 The embodiment of the present invention provides a ship engine-propeller intelligent matching method based on a multi-source fusion algorithm, the method comprising:

[0069] Obtain ship design parameters and preset brake horsepower calculation rules;

[0070] Optimizing the preset brake horsepower calculation rule based on the ship design parameters to generate an optimized horsepower calculation rule;

[0071] determining an effective horsepower demand calculation rule, and performing nonlinear coupling on the optimized horsepower calculation rule based on the effective horsepower demand calculation rule to generate a coupled horsepower calculation rule;

[0072] Obtaining performance parameters of the ship's main engine, and determining a matching horsepower coefficient based on the performance parameters;

[0073] A propeller map is obtained, and a main engine-propeller matching operation is performed based on the propeller map, the ship design parameters, the coupled horsepower calculation rule, and the matching horsepower coefficient to obtain a corresponding propeller model.

[0074] Among the existing methods for calculating ship brake horsepower, the main method used is the Ohsumi Mitsuhiko formula. However, the existing calculation method relies on static parameters (such as fixed displacement Δ, speed V, empirical coefficient K, etc.), and does not consider dynamic factors of the ship during actual navigation. For example, the wind and wave level (β) will cause resistance fluctuations; cargo distribution (γ) affects the ship's draft and resistance; the resistance of the ship under different operating conditions shows nonlinear changes, etc., which makes the existing calculation method biased, and the error is significant under complex navigation conditions (up to 20% or more).

[0075] In one possible embodiment, the ship design parameters and the preset brake horsepower calculation rules are first optimized, wherein the ship design parameters include speed, hull displacement and size, that is, the preset brake horsepower calculation rules are optimized taking into account various factors in the actual navigation process of the ship, and the optimized horsepower calculation rules are obtained. Then, the optimized horsepower calculation rules and the optimized horsepower calculation rules are nonlinearly coupled to generate coupled horsepower calculation rules. Then, according to the coupled horsepower calculation rules, the horsepower requirement that best meets the actual navigation process of the ship is obtained. Finally, the main engine-propeller matching operation is performed through the propeller map, ship design parameters, coupled horsepower calculation rules and matching horsepower coefficients to obtain the corresponding propeller model.

[0076] By performing nonlinear coupling on the optimized horsepower calculation rule and the effective horsepower requirement calculation rule, a coupled horsepower calculation rule is generated, and a main engine-propeller matching operation is performed based on the propeller map, the ship design parameters, the coupled horsepower calculation rule, and the matching horsepower coefficient to obtain the corresponding propeller model. This realizes multi-model data fusion matching of propeller models and improves propeller matching efficiency.

[0077] However, during the actual navigation of the ship, the resistance of the ship in the fluid is constantly changing, so the brake horsepower of the ship is also changing accordingly. That is, during the actual navigation of the ship, it is necessary to consider the substantial impact of the hull resistance on the hull.

[0078] In order to solve the above problem, further, the preset brake horsepower calculation rule is optimized based on the ship design parameters to generate an optimized horsepower calculation rule, including:

[0079] The preset brake horsepower calculation rule is determined, and the preset brake horsepower calculation rule is characterized as follows:

[0080]

[0081] Determine the dynamic correction function of hull resistance based on hull load and wind and wave environment;

[0082] The preset brake horsepower calculation rule is optimized based on the hull resistance dynamic correction function to generate an optimized horsepower calculation rule, which is characterized as follows:

[0083]

[0084] Among them, K represents the preset horsepower coefficient, Δ represents the ship's design displacement, V represents the ship's design speed, C represents the empirical constant with a value range of 0.9~1.1, and f(Rt,α) represents the dynamic correction function of the hull resistance, which is predicted by the LSTM network and the input is the real-time dynamic factor α. , β is the wind and wave coefficient, γ is the load coefficient, and the output is the resistance correction Rt.

[0085] For example, in an embodiment of the present invention, the dynamic correction function of the hull resistance is represented as: The resistance correction value Rt can be predicted and generated by an LSTM network. For example, the calculation formula is: Rt=LSTM(V,β,γ,Historyt), where the input parameters are: current speed V, wind and wave coefficient β, load coefficient γ, and the historical resistance sequence Historyt; the output parameter is: the resistance correction value Rt for the future time step. The real-time dynamic factor α is set to: α=0.2β+0.8γ, reflecting the correction logic that prioritizes load influence and supplements wind and wave influence. During training, the network weights are updated every 24 hours to ensure a prediction error of ≤5%. CFD simulations also verify the goodness of fit between f(Rt,α) and measured resistance (R²>0.95). The actual navigation data is collected from the ship's black box, which covers a variety of typical navigation states and is sampled at a frequency of 1Hz. The actual navigation data training includes data such as speed (V), wind and wave coefficient (β), load coefficient (γ), resistance (Rt), main engine speed (n), and propeller thrust (T).

[0086] For example, in a specific embodiment, if β = 0.5 (moderate wind and waves), γ = 0.7 (70% load), then: α = 0.2 × 0.5 + 0.8 × 0.7 = 0.66. If the LSTM predicts Rt = 1.2 × 104 N, Δ = 20,000 tons, and V = 10 m / s, then:

[0087]

[0088] This means that the brake horsepower needs to be increased by 0.4% under the current working conditions.

[0089] In the embodiment of the present invention, the displacement and speed V in the existing Ohsumi Mitsuhiko formula are fixed parameters. In order to further improve environmental adaptability and calculation accuracy, they are optimized into dynamic parameters. Specifically, the displacement is corrected in combination with the load coefficient γ: ; Combined with the wind and wave coefficient β, the ship speed V is dynamically corrected: .

[0090] Furthermore, the impact of ship size on the calculation of brake horsepower is explained as follows:

[0091] According to the optimized displacement calculation formula: Δ = ρ·L·B·T·Cb, where L, B, and T are the ship's length, breadth, and draft, respectively, and Cb is the block coefficient, representing the hull's fullness, it can be seen that ship size indirectly influences the brake horsepower value in the optimized horsepower calculation rule through displacement Δ. Based on this, the ship's wetted surface area is directly related to the ship's length, breadth, and draft. The calculation formula is: S = L·(B+2T)·k, where k is the hull lines correction factor, ranging from 1.05 to 1.15. The wave-making resistance coefficient Cw is related to the hull size ratio (such as the length-to-breadth ratio L / B) and ship form characteristics. The viscous resistance coefficient Cv is related to the hull surface roughness and the Reynolds number (which depends on the ship's length L and speed V). Therefore, these parameters, through the resistance decomposition model (such as the viscous resistance Rv and the wave-making resistance Rw), ultimately influence the total resistance correction Rt, thereby correcting the brake horsepower.

[0092] By improving the existing brake horsepower calculation formula, the influence of wind and waves and load on the total resistance is dynamically reflected through the resistance correction; by integrating multiple factors, the combined effect of the environment and load is quantified; by using a neural network to predict the resistance correction, the complex change pattern of resistance over time is captured, thereby effectively solving the dynamic adaptability problem caused by static assumptions in the traditional formula, and greatly improving the accuracy of brake horsepower calculation. For example, under the condition of wind and wave of level 8, the error of the traditional formula is as high as 18%, while after dynamic correction, the error is reduced to within 3%.

[0093] Furthermore, the calculation rule for determining the effective horsepower requirement includes: obtaining the Hankshall formula; determining the propeller thrust and torque based on the Hankshall formula and the ship design speed; and interpolating the propeller thrust and the torque based on the propeller map to obtain an interpolated thrust T and an interpolated torque Q, wherein the interpolated thrust T and the interpolated torque Q are respectively characterized as: ,in, It is represented by the advance coefficient, ρ is represented by the fluid density, n is represented by the propeller speed, D is represented by the propeller diameter, and J is represented by the advance coefficient. Based on the real-time matching determination of the propeller map, K T Characterized by the propeller thrust coefficient, K Q Characterized as a torque coefficient of the propeller; generating an effective horsepower requirement calculation rule based on the interpolated thrust T and the interpolated torque Q.

[0094] In one possible implementation, propeller thrust and torque are first calculated based on the Hankshall formula, and then efficiency interpolation is performed in combination with the propeller map database to obtain the interpolated propeller thrust and torque. Finally, the effective horsepower requirement calculation rules are generated based on the interpolated thrust T and the interpolated torque Q.

[0095] The method for constructing the propeller atlas includes: using the OpenCV image recognition algorithm to perform grayscale processing, binarization, and contour extraction on the paper atlas to generate a digital parameter matrix; extracting the characteristic parameters of the propeller: the characteristic parameters of the propeller include the propeller thrust coefficient , propeller torque coefficient , propeller disc ratio , where Z is the number of propeller blades, is the pitch ratio of the propeller; a digital feature library of the propeller map is established based on MATLAB, and the optimal propeller model is matched in the map neural network according to the characteristic parameters of the propeller that meets the power requirements.

[0096] Among them, the graph neural network adopts the ResNet-18 model, and the training set contains 100,000 sets of propeller graph parameters. The digital feature library of the propeller graph supports fuzzy matching, specifically the Euclidean distance is less than or equal to 0.1. When calculating the Euclidean distance, the propeller disk ratio weight is 0.4, the propeller pitch ratio weight is 0.3, and the propeller blade number weight is 0.3, and the matching accuracy error is controlled within 3%.

[0097] The real-time matching procedure for the propeller map is as follows: input the ship speed V, main engine speed n, and calculate the advance coefficient ; Match the closest one from the feature library through convolutional neural network (CNN) curve; output matching results: propeller model, diameter D, pitch ratio P / D.

[0098] By combining the propeller thrust and torque calculated based on the Hankschel formula with the propeller atlas database for efficiency interpolation, the calculation rules for the effective horsepower requirement generated by the interpolated thrust T and the interpolated torque Q are more in line with the actual navigation conditions, which improves the accuracy of propeller model selection.

[0099] Furthermore, the nonlinear coupling of the optimized horsepower calculation rule based on the effective horsepower demand calculation rule to generate a coupled horsepower calculation rule includes: obtaining the optimized horsepower calculation rule and the effective horsepower demand calculation rule; obtaining dynamic weights of the optimized horsepower calculation rule and the effective horsepower demand calculation rule respectively; and obtaining a coupled horsepower calculation rule based on the dynamic weights, the optimized horsepower calculation rule, and the effective horsepower demand calculation rule, wherein the coupled horsepower calculation rule is characterized as follows:

[0100]

[0101] in , T is the propeller interpolation thrust, η prop is the propeller efficiency, determined by propeller map matching, 、 is the weight before adjustment.

[0102] In the actual data coupling process, BHP is calculated separately 优化 and EHP, normalize EHP through sigmoid function to limit its contribution ratio. 优化 >1 (propeller demand exceeds the host capacity), ω2 will be automatically reduced to avoid overload; if EHP / BHP 优化 <0.8 (host capacity redundancy), then optimize ω2 and optimize energy efficiency.

[0103] For example, in a specific embodiment, BHP 优化 =5000KW, EHP=4800KW, after genetic algorithm optimization, the weights are ω1=0.6, ω2=0.4; calculate the sigmoid value: sigmoid(4800 / 5000)≈0.62, then the horsepower after coupling is:

[0104]

[0105] Through the above optimization, the core contribution of the main engine brake horsepower is retained, and it is flexibly adjusted according to the actual needs of the propeller, avoiding the rigidity of traditional linear weighting.

[0106] The Ohsumi Mitsuhiko formula is based solely on the main engine power requirement and does not incorporate the actual propeller thrust and torque, resulting in deviations from actual results. Therefore, in an embodiment of the present invention, a coupled horsepower calculation rule is derived by nonlinearly coupling the effective horsepower requirement calculation rule with the optimized horsepower calculation rule. This coupled horsepower calculation rule better meets the actual sailing conditions of a vessel, transcending the one-way calculation logic of the traditional formula to achieve a bidirectional match between main engine power and propeller thrust, enabling a more accurate determination of the required propeller model.

[0107] Furthermore, the performance parameters include main engine speed, main engine power, and main engine torque, and determining the matching horsepower coefficient based on the performance parameters includes:

[0108] Generate a corresponding matching horsepower coefficient K based on the main engine speed, the main engine power, and the main engine torque m , the horsepower coefficient K m Characterized by:

[0109]

[0110] Among them, P max is the rated power of the ship's main engine, M max is the maximum torque of the ship's main engine, n rated It is the speed when the main engine of the ship reaches the rated power, and L is the unit conversion constant.

[0111] Furthermore, performing the main engine-propeller matching operation to obtain the corresponding propeller model includes:

[0112] Obtain the initial genetic algorithm and initial weights;

[0113] Make real-time predictions of the ship's navigation state and generate navigation state prediction results;

[0114] Adjusting the initial weight in real time based on the flight state prediction result to generate an adjusted weight;

[0115] The initial genetic algorithm is optimized based on the ship design speed, the coupled horsepower calculation rule, and the main engine speed to generate a dynamic matching objective function. The dynamic matching objective function is characterized as follows:

[0116]

[0117] Among them, ω 11 ,ω 22 Represented as the adjusted weight, η prop is the propeller efficiency, BHP engine is the engine brake horsepower, BHP prop is the propeller brake horsepower, BHP max Characterized as the maximum brake horsepower of the ship's main engine;

[0118] Determining an optimal propeller diameter-to-pitch ratio based on the dynamic matching objective function;

[0119] A main engine-propeller matching operation is performed based on the optimal propeller diameter and pitch ratio to obtain a corresponding propeller model.

[0120] In the embodiment of the present invention, the optimization objective may also be based on a genetic algorithm:

[0121]

[0122] Right now:

[0123] Among them, the constraints are: 11 +ω 22 =1, and 0.3≤ω 11 ,ω 22 ≤0.7, ηprop is the propeller efficiency.

[0124] In a possible embodiment, constructing a dynamic objective function based on a genetic algorithm specifically includes:

[0125] Set the initial population: number of blades, pitch ratio, disk ratio, propeller speed, and diameter;

[0126] Adjust the adaptive crossover rate pc = 0.7 ~ 0.9, and the mutation rate pm = 0.01 ~ 0.05;

[0127] According to the objective function weights ω13 = 0.6 and ω23 = 0.4, the initialized population is input and combined with the NSGA-II algorithm to generate the optimal frontier solution, where ω13 is the propeller fuel economy weight, and ω23 is the propeller propulsion efficiency weight.

[0128] Furthermore, the real-time prediction of the ship's navigation state and generation of the navigation state prediction result includes: obtaining a long short-term memory network LSTM; and real-time prediction of the ship's navigation state based on the LSTM to generate the navigation state prediction result.

[0129] In one possible embodiment, the steps for generating results for LSTM-based state prediction are as follows: acquiring data including ship speed (V), wave coefficient (β), load coefficient (γ), resistance (Rt), main engine speed (n), and propeller thrust (T); and preprocessing the acquired data. This preprocessing includes: denoising, which uses a sliding average filter with a 5-second window size to eliminate high-frequency noise from the collected data; normalization, which uses the Z-score normalization formula to process continuous variables such as ship speed and resistance to uniformly measure and distribute the data; time series alignment, which aligns the aforementioned data from multiple sources based on timestamps and integrates them to generate a synchronized time series matrix to ensure temporal consistency of different parameter data; and sliding window segmentation, which uses a 60-second time window with a 1-second step size to generate input sequences (X) and labels (Y) according to rules. Specifically, the input X is the [V, β, γ, Rt] sequence for the first 59 seconds, and the output Y is the state parameters at the 60th second, such as the resistance correction Rt or the speed change ΔV.

[0130] The model construction and training phase includes the following: Designing the LSTM network architecture: Input layer: Build an input layer capable of receiving 60×4 time series data. The four features correspond to V, β, γ, and Rt, respectively, preparing the data input for subsequent model processing. LSTM layer: Utilizes a two-layer stacked LSTM structure with 128 neurons per layer. The activation function uses tanh. A dropout rate of 0.2 is introduced to prevent overfitting and enable the model to learn temporal dependencies between flight states. Fully connected layer: Maps the LSTM layer output to the prediction target. ReLU is used as the activation function to further transform the features learned by the LSTM layer into representations directly relevant to the prediction task. Output layer: The activation function is selected based on the prediction task type. For regression tasks, a linear activation function is used to output continuous values (such as predicting the resistance correction Rt). For classification tasks, a softmax activation function is used to output classification labels (such as abnormal operating condition classification).

[0131] Configure model parameters, including: Loss function: This function is determined based on the task type. For regression tasks, use mean squared error (MSE), while for classification tasks, use cross-entropy. This function measures the difference between the model's predictions and the true values, providing optimization guidance for model training. Optimizer: Use the Adam optimizer, with an initial learning rate of 0.001 and a 10% reduction in the learning rate every 50 training rounds. This balances rapid convergence in the early stages of training with the need for fine-tuning in the later stages, helping the model find optimal parameters.

[0132] Batch size and training iterations: The batch size is set to 32, which means that each time the model updates the parameters, it is calculated based on 32 sets of data as a batch; the training iterations are 500 rounds (epochs), allowing the model to continuously optimize its own parameters during repeated data learning.

[0133] Data partitioning and dynamic weight updating include: Data partitioning: historical data is divided into 70% training sets, which are used for basic training and learning of the model; the validation set accounts for 15%, which is used for hyperparameter tuning during the training process, such as trying different network structure depths, Dropout ratios, etc., to find the optimal model configuration; the test set accounts for 15%, which is used to evaluate the generalization ability of the model after model training is completed, and to judge the model's prediction accuracy for unseen data.

[0134] Dynamic weight update: Each time the latest data is accessed, an incremental learning strategy is adopted to fine-tune only the weights of the fully connected layer to maintain the model's adaptability to current environmental changes, avoid the model forgetting important features learned previously due to new data, and ensure real-time prediction.

[0135] The actual prediction phase includes an online inference process, including: Data stream access: The system receives real-time sampled data from various ship sensors and caches this data for subsequent processing. Preprocessing: The received real-time data is immediately denoised and normalized, generating a 60×4 input matrix according to pre-defined rules to ensure it meets the model input requirements. Prediction execution: The preprocessed input matrix is fed into a trained LSTM model, which performs calculations and inference based on the learned navigation state patterns, outputting a navigation state prediction result for the next one second, such as specific values such as Rt or speed V, or classification labels for abnormal operating conditions. Result publishing: The predicted results are published to the dynamic correction module and genetic algorithm optimization module via ROS (Robot Operating System), providing data support for real-time decision-making during navigation and realizing intelligent navigation control. Latency control: Throughout the real-time prediction process, the single inference time is strictly controlled to ensure it is less than or equal to 10ms to meet the real-time navigation requirements of the ship and ensure that the prediction results can be used to adjust the ship's operating status in a timely manner.

[0136] Furthermore, the method also includes: after obtaining the propeller model, obtaining a preset propulsion performance requirement; calibrating the propeller model based on the preset propulsion performance requirement to generate a calibration result; adjusting the propeller model based on the calibration result to obtain an adjusted propeller model.

[0137] In one possible embodiment, propeller calibration is a key step in ensuring that it meets preset propulsion performance requirements. In this embodiment of the present invention, the preset propulsion performance requirements include, but are not limited to, propulsion efficiency (the propeller's propulsion efficiency (ηprop) at different speeds must meet a design threshold (ηprop ≥ 0.65)), energy economy (energy consumption per unit of voyage must not exceed a preset value), cavitation performance (no significant cavitation occurs in the propeller at rated speed), vibration and noise (hull vibration acceleration caused by the propeller is ≤ 0.1 m / s², underwater radiated noise is ≤ 110 dB), structural strength and durability (propeller material fatigue life is ≥ 108 cycles, and corrosion resistance meets IMO standards), dynamic response capability (propeller response time is ≤ 10 seconds and thrust fluctuation is ≤ 5% under emergency acceleration / deceleration conditions), matching the power and speed of the main engine (propeller absorbed power must match the rated power of the main engine), and environmental adaptability (propulsion efficiency reduction is ≤ 15% under specific sea conditions). The specific process is as follows:

[0138] Data input: Obtain propeller model parameters, such as diameter D, pitch ratio P / D, disk ratio, number of blades Z, etc., and obtain preset performance thresholds, including specific requirements in terms of propulsion efficiency, energy economy, cavitation performance, vibration and noise, structural strength and durability, dynamic response capability, matching of main engine power and speed, and environmental adaptability.

[0139] Performance calculations: Calculations are performed using CFD (computational fluid dynamics), FEA (finite element analysis), or pattern matching tools. For example, propulsion efficiency can be calculated based on propeller patterns and CFD simulations; cavitation number can be calculated using cavitation tests and empirical formulas; and fuel consumption can be calculated using the main engine power curve.

[0140] Verification and Comparison: Compare the calculated results with the preset requirements and generate a verification report to clarify whether each performance indicator has passed. If not, subsequent adjustments are required. For example, in a container ship case, the MAU5-80 propeller propulsion efficiency was 0.63 < 0.65, and the cavitation number σ = 0.28 < 0.3, which did not meet the preset requirements.

[0141] Dynamic adjustment (if calibration fails)

[0142] Determine optimization goals: Set improvement targets for performance indicators that do not meet the standards, such as improving propulsion efficiency to ηprop ≥ 0.65 and increasing the cavitation number to σ ≥ 0.3.

[0143] Develop adjustment strategies: Optimize propeller parameters by increasing pitch ratio, reducing diameter, increasing number of blades, etc.

[0144] Algorithm optimization: Use a genetic algorithm to determine the optimization variable range and objective function, such as the optimization variables D∈[6.0,6.5]m, P / D∈[1.1,1.2], and Z∈{5,6}. The objective function is MinimizeF=w13×(0.65−ηprop)+w23×(0.3−σ) (weights w13=0.6, w23=0.4), and set constraints, such as fuel consumption ≤200g / kWh and vibration acceleration ≤0.1m / s².

[0145] Optimization results: Get the optimal parameter combination and prediction performance. For example, in a container ship case, the optimal parameter combination is D=6.2m, P / D=1.15, Z=6; the prediction performance is η prop =0.66, σ=0.32, fuel consumption=198g / kWh, vibration acceleration 0.09m / s².

[0146] Recalibration: Recalibrate the various performance indicators of the adjusted propeller, such as CFD verification of propulsion efficiency, experimental measurement of cavitation number, etc., to ensure that all performance indicators meet the standards and determine the final propeller model.

[0147] In an embodiment of the present invention, by combining CFD simulation, cavitation testing and measured data, key performance indicators are quantitatively evaluated, and propeller parameters are optimized through genetic algorithms to balance efficiency, cavitation and vibration, thereby effectively improving the accuracy and qualification of various indicators, ensuring that all performance standards are met, and meeting the actual needs of the enterprise.

[0148] For further information, please refer to Figure 2 This embodiment also provides a ship engine-propeller intelligent matching system based on a multi-source fusion algorithm, the system comprising:

[0149] The first acquisition module is used to obtain ship design parameters and preset brake horsepower calculation rules;

[0150] an optimization module, configured to optimize the preset brake horsepower calculation rule based on the ship design parameters to generate an optimized horsepower calculation rule;

[0151] a coupling module, configured to determine an effective horsepower demand calculation rule, and perform nonlinear coupling on the optimized horsepower calculation rule based on the effective horsepower demand calculation rule to generate a coupled horsepower calculation rule;

[0152] A second acquisition module is used to obtain performance parameters of the ship's main engine and determine a matching horsepower coefficient based on the performance parameters;

[0153] The matching module is used to obtain a propeller map, perform a main engine-propeller matching operation based on the propeller map, the ship design parameters, the coupled horsepower calculation rule and the matching horsepower coefficient, and obtain a corresponding propeller model.

[0154] Furthermore, an embodiment of the present invention also provides a computer-readable storage medium having a computer program stored thereon, which implements the method described in the embodiment of the present invention when the program is executed by a processor.

[0155] The above describes in detail the optional implementation methods of the embodiments of the present invention in conjunction with the accompanying drawings. However, the embodiments of the present invention are not limited to the specific details in the above implementation methods. Within the technical concept of the embodiments of the present invention, various simple modifications can be made to the technical solutions of the embodiments of the present invention, and these simple modifications all fall within the scope of protection of the embodiments of the present invention.

[0156] It should also be noted that the various specific technical features described in the above specific embodiments can be combined in any appropriate manner without contradiction. To avoid unnecessary repetition, the embodiments of the present invention will not further describe various possible combinations.

[0157] Those skilled in the art will understand that all or part of the steps in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a program, which is stored in a storage medium and includes a number of instructions for causing a single-chip microcomputer, chip or processor to execute all or part of the steps of the methods described in each embodiment of the present application. The aforementioned storage medium includes: a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and other media that can store program code.

[0158] In addition, various implementations of the embodiments of the present invention may be arbitrarily combined, and as long as they do not violate the concept of the embodiments of the present invention, they should also be regarded as the contents disclosed in the embodiments of the present invention.

Claims

1. A ship engine-propeller intelligent matching method based on a multi-source fusion algorithm, characterized in that: The method comprises: Obtain ship design parameters and preset brake horsepower calculation rules; Optimizing the preset brake horsepower calculation rule based on the ship design parameters to generate an optimized horsepower calculation rule; determining an effective horsepower demand calculation rule, and performing nonlinear coupling on the optimized horsepower calculation rule based on the effective horsepower demand calculation rule to generate a coupled horsepower calculation rule; Obtaining performance parameters of the ship's main engine, and determining a matching horsepower coefficient based on the performance parameters; Obtaining a propeller map, performing a main engine-propeller matching operation based on the propeller map, the ship design parameters, the coupled horsepower calculation rule, and the matching horsepower coefficient to obtain a corresponding propeller model; The step of optimizing the preset brake horsepower calculation rule based on the ship design parameters to generate an optimized horsepower calculation rule includes: The preset brake horsepower calculation rule is determined, and the preset brake horsepower calculation rule is characterized as follows: ， Determine the dynamic correction function of hull resistance based on hull load and wind and wave environment; The preset brake horsepower calculation rule is optimized based on the hull resistance dynamic correction function to generate an optimized horsepower calculation rule, which is characterized as follows: ， ， Among them, K represents the preset horsepower coefficient, Δ represents the ship's design displacement, V represents the ship's design speed, C represents the empirical constant, f(Rt,α) represents the dynamic correction function of the hull resistance, Rt represents the resistance correction amount, and α represents the real-time dynamic factor of the navigation state.

2. The method according to claim 1, characterized in that The calculation rules for determining the effective horsepower requirement include: Get the Hankscher formula; Determine propeller thrust and torque based on the Hankshall formula and the ship's design speed; The propeller thrust and the torque are interpolated based on the propeller map to obtain an interpolated thrust T and an interpolated torque Q. The interpolated thrust T and the interpolated torque Q are respectively represented as follows: , Where ρ represents the fluid density, n represents the propeller speed, D represents the propeller diameter, and J represents the advance coefficient. and Based on the real-time matching determination of the propeller map, Characterized as the propeller thrust coefficient, Characterized as the torque coefficient of the propeller; An effective horsepower requirement calculation rule is generated based on the interpolated thrust T and the interpolated torque Q.

3. The method according to claim 1, characterized in that The nonlinearly coupling the optimized horsepower calculation rule based on the effective horsepower demand calculation rule to generate a coupled horsepower calculation rule includes: Obtaining the optimized horsepower calculation rule and the effective horsepower requirement calculation rule; respectively obtaining dynamic weights of the optimized horsepower calculation rule and the effective horsepower demand calculation rule; A coupled horsepower calculation rule is obtained based on the dynamic weight, the optimized horsepower calculation rule, and the effective horsepower demand calculation rule. The coupled horsepower calculation rule is characterized as follows: ， in , T is the propeller interpolation thrust, η prop is the propeller efficiency, determined by propeller map matching, 、 is the weight before adjustment.

4. The method according to claim 1, wherein The performance parameters include main engine speed, main engine power, and main engine torque. Determining the matching horsepower coefficient based on the performance parameters includes: Generate a corresponding matching horsepower coefficient K based on the main engine speed, the main engine power, and the main engine torque m , the horsepower coefficient K m Characterized by: ， Among them, P max is the rated power of the ship's main engine, M max is the maximum torque of the ship's main engine, n rated It is the speed when the main engine of the ship reaches the rated power, and L is the unit conversion constant.

5. The method according to claim 4, characterized in that The performing of the main engine-propeller matching operation to obtain the corresponding propeller model includes: Obtain the initial genetic algorithm and initial weights; Make real-time predictions of the ship's navigation state and generate navigation state prediction results; Adjusting the initial weight in real time based on the flight state prediction result to generate an adjusted weight; The initial genetic algorithm is optimized based on the ship design speed, the coupled horsepower calculation rule, and the main engine speed to generate a dynamic matching objective function. The dynamic matching objective function is characterized as follows: ， Among them, ω 11 ,ω 22 Represented as the adjusted weight, η prop is the propeller efficiency, BHP engine is the engine brake horsepower, BHP prop is the propeller brake horsepower, BHP max Characterized as the maximum brake horsepower of the ship's main engine; Determining an optimal propeller diameter-to-pitch ratio based on the dynamic matching objective function; A main engine-propeller matching operation is performed based on the optimal propeller diameter and pitch ratio to obtain a corresponding propeller model.

6. The method according to claim 5, characterized in that The real-time prediction of the ship's navigation state and generation of navigation state prediction results include: Get the long short-term memory network LSTM; The navigation state of the ship is predicted in real time based on the LSTM to generate a navigation state prediction result.

7. The method according to any one of claims 1 to 6, characterized in that The method further comprises: After obtaining the propeller model, obtaining a preset propulsion performance requirement; calibrating the propeller model based on the preset propulsion performance requirements and generating a calibration result; The propeller model is adjusted based on the verification result to obtain an adjusted propeller model.

8. A ship engine-propeller intelligent matching system based on a multi-source fusion algorithm, characterized in that: The system comprises: The first acquisition module is used to obtain ship design parameters and preset brake horsepower calculation rules; an optimization module, configured to optimize the preset brake horsepower calculation rule based on the ship design parameters to generate an optimized horsepower calculation rule; a coupling module, configured to determine an effective horsepower demand calculation rule, and perform nonlinear coupling on the optimized horsepower calculation rule based on the effective horsepower demand calculation rule to generate a coupled horsepower calculation rule; A second acquisition module is used to obtain performance parameters of the ship's main engine and determine a matching horsepower coefficient based on the performance parameters; a matching module, configured to obtain a propeller map, and perform a main engine-propeller matching operation based on the propeller map, the ship design parameters, the coupled horsepower calculation rule, and the matching horsepower coefficient to obtain a corresponding propeller model; The step of optimizing the preset brake horsepower calculation rule based on the ship design parameters to generate an optimized horsepower calculation rule includes: The preset brake horsepower calculation rule is determined, and the preset brake horsepower calculation rule is characterized as follows: ， Determine the dynamic correction function of hull resistance based on hull load and wind and wave environment; The preset brake horsepower calculation rule is optimized based on the hull resistance dynamic correction function to generate an optimized horsepower calculation rule, which is characterized as follows: ， ， Among them, K represents the preset horsepower coefficient, Δ represents the ship's design displacement, V represents the ship's design speed, C represents the empirical constant, f(Rt,α) represents the dynamic correction function of the hull resistance, Rt represents the resistance correction amount, and α represents the real-time dynamic factor of the navigation state.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images