An intelligent ship power control system

Abstract

The present invention discloses an intelligent ship power control system, comprising a data collection module, a power control strategy adjustment trigger module, a ship navigation status identification model construction module, a ship power strategy generation module, and a power intelligent control module. The present invention relates to the field of ship power control technology, and specifically refers to an intelligent ship power control system. This solution innovatively introduces a power control strategy adjustment trigger method with a dual judgment logic mechanism, effectively realizing the intelligent update of the ship power control strategy; combines the dynamic hyperbox expansion mechanism for overlapping risk avoidance with the membership function construction method for boundary dynamic regulation to construct an identification model, thereby improving the stability of the identification result; designs a multi-stage strategy generation method, and adopts an improved optimization algorithm that introduces a population search direction control factor and an adaptive reverse guidance mechanism to enhance the global optimization capability of the control strategy, thereby realizing refined control of the ship power strategy in a complex navigation environment.

Description

Technical Field

[0001] The present invention relates to the technical field of ship power control, and in particular to an intelligent ship power control system. Background Art

[0002] An intelligent ship power control system refers to an intelligent automatic control system built on a computer platform, which is used for real-time monitoring, state identification and dynamic control of ship power systems. The system obtains operating data, uses computer technology to process data and generate intelligent power control strategies, and implements precise adjustment of various power equipment through actuators, thereby realizing intelligent centralized control of the ship's propulsion system, power system and auxiliary units.

[0003] However, traditional ship power control systems have technical problems of frequent strategy adjustments and delayed responses, which lead to unstable response of the power control system and deterioration of ship navigation performance; existing models suitable for ship navigation state identification have technical problems of hyperbox division rigidity and fuzzy state boundary identification, which make it impossible to accurately adapt to the changeable characteristics of complex navigation states, and thus lead to deterioration of navigation state identification accuracy; existing methods for power control strategy generation have technical problems of lack of physical constraints and poor global optimization capabilities, which lead to the generated power control strategy deviating from the actual ship navigation scenario, and thus resulting in poor strategy adaptability and unstable control effect. Summary of the Invention

[0004] In response to the above situation, in order to overcome the defects of the existing technology, the present invention provides an intelligent ship power control system. It addresses the technical problems of frequent strategy adjustment and delayed response in traditional ship power control systems, which leads to unstable response of the power control system and decreased ship navigation performance. This solution innovatively introduces a power control strategy adjustment triggering method with a dual judgment logic mechanism. By introducing a trigger logic that combines single parameter out-of-bounds judgment with multi-parameter combination deviation judgment, it achieves accurate judgment of the timing of strategy adjustment, effectively avoids false triggering of strategy adjustment, and can effectively realize intelligent update and stable control of ship power control strategy, significantly improving the economy and operational reliability of ship power control. In response to the technical problems of rigid hyperbox division and fuzzy state boundary identification in existing ship navigation state identification models, which leads to the inability to accurately adapt to the variable characteristics of complex navigation states, and thus leads to decreased navigation state identification accuracy, this solution innovatively combines the dynamic hyperbox expansion mechanism for overlapping risk avoidance with the membership function construction method for boundary dynamic regulation to construct an identification model, which can effectively solve the attribution ambiguity faced by traditional models when similar samples are unevenly distributed or category boundaries are adjacent. The proposed method solves the problem of conflict between different classifications, improves the structural flexibility and adaptability of the model, enhances the model's responsiveness to the transition region of the navigation state boundary, and improves the stability and credibility of the recognition results. This method realizes high-precision recognition of the navigation state of the ship navigation state recognition model in dynamic and complex environments. In response to the technical problems of the lack of physical constraints and poor global optimization capabilities in existing methods for power control strategy generation, which cause the generated power control strategies to deviate from the actual ship navigation scenario, resulting in poor strategy adaptability and unstable control effects, this scheme innovatively designs a multi-stage strategy generation method. By constructing a three-layer strategy generation architecture, the first layer completes template strategy selection based on the navigation state recognition results to ensure that the parameter structure has engineering reference. The second layer sets parameter boundaries to limit the search space to a physically feasible range. The third layer introduces an improved intelligent optimization algorithm with a population search direction control factor and an adaptive reverse guidance mechanism to perform dynamic optimization search on the selected parameters under boundary constraints, enhancing the physical consistency and engineering feasibility of the strategy generation, improving the global optimization capability of the control strategy, and significantly improving the accuracy, stability and intelligent adaptability of the ship's power control strategy in complex navigation environments.

[0005] The technical solution adopted by the present invention is as follows: The present invention provides an intelligent ship power control system, including a data collection module, a power control strategy adjustment trigger module, a ship navigation state recognition model construction module, a ship power strategy generation module and a power intelligent control module;

[0006] The data collection module is used to collect the original data required for ship power control, specifically by collecting data from various power subsystems of the ship and the ship's navigation environment to obtain the original data of ship power control;

[0007] The power control strategy adjustment trigger module specifically preprocesses and selects features from the raw data, defines strategy trigger parameters and their threshold ranges, and uses single parameter out-of-bounds and multi-parameter combination deviation judgment logic to determine whether the current power control strategy needs to be adjusted, thereby obtaining ship power control optimization data and strategy adjustment trigger results.

[0008] The ship navigation state recognition model construction module is used to build and train a recognition model for accurately identifying the ship's navigation state under different operating environments and working conditions. Specifically, the model structure parameters are first initialized, and then sample classification processing is performed. The same type of hyperbox matching, the first category hyperbox creation, and the adaptive boundary expansion operations are performed in sequence to form a boundary structure expression of the navigation state. Then, by constructing a navigation state membership function, the membership degree of each input sample to each navigation state category is calculated; finally, the state with the highest membership degree is output as the recognition result, and the model training is completed in combination with historical power control data to obtain a trained ship navigation state recognition model;

[0009] The ship power strategy generation module is used to generate a power optimal control strategy with adaptive performance under the identified current navigation state. Specifically, the module first performs power control template strategy selection based on the navigation state identification result, sets the corresponding power control parameter boundaries, determines the set of parameters to be optimized and their dynamic boundary intervals, and uses an improved intelligent optimization algorithm to dynamically optimize the selected parameters within the constraint range to obtain the optimal control parameter combination that adapts to the current navigation state. Based on the optimal control parameter combination, the module obtains the power optimal control strategy under the current navigation state.

[0010] The intelligent power control module is used to realize adaptive adjustment and intelligent decision-making control of the ship propulsion system, specifically to determine whether the control strategy needs to be updated; if the update is triggered, the current navigation status is identified, and the optimal control strategy is generated based on the status. Finally, the power control system is dynamically adjusted according to the strategy to realize intelligent control of the ship power.

[0011] Furthermore, the data collection module specifically obtains ship power control raw data by collecting the original data required for ship power control from various power subsystems of the ship and the ship's navigation environment; the ship power control raw data includes historical power control data and real-time power control data; the historical power control data and real-time power control data both include ship operation data, ship power data, ship navigation environment data, power control strategy data and power control log data; the historical power control data also includes historical ship navigation status identification results.

[0012] Furthermore, the power control strategy adjustment trigger module includes data preprocessing, navigation state identification feature extraction, strategy adjustment trigger parameter definition, parameter threshold definition and strategy adjustment trigger judgment, specifically including the following steps:

[0013] Data preprocessing, specifically cleaning and standardizing the raw data of ship power control to obtain ship power control optimization data;

[0014] The selection of navigation state identification features is to use the correlation analysis method to screen out the features related to navigation state identification from the ship power control optimization data;

[0015] Strategy adjustment trigger parameter definition, specifically, the strategy adjustment trigger parameter is composed of external environment key parameters and ship status key parameters;

[0016] Parameter threshold definition, specifically, based on the historical data of the strategy adjustment trigger parameters collected during the stable operation of the current power control strategy, average statistics are performed to generate the reference mean of each strategy adjustment trigger parameter, and combined with the fixed deviation range set by the current power control strategy, the strategy adjustment trigger parameter threshold range is obtained;

[0017] The strategy adjustment trigger judgment specifically adopts a dual judgment logic mechanism consisting of single-parameter out-of-bounds judgment logic and multi-parameter combination deviation judgment logic, performs logical judgment in sequence, and determines whether to trigger the control strategy adjustment process based on the judgment result to obtain the strategy adjustment trigger result.

[0018] Furthermore, the ship navigation state recognition model construction module specifically includes the following steps:

[0019] Initialize the structural parameters of the recognition model, specifically initialize the minimum boundary matrix V and the maximum boundary matrix W to all zero matrices, and the dimensions of both are , where M is the number of samples, n is the feature dimension, and the hyperbox set is initialized , and for each category of training samples, construct the corresponding Kd-Tree index structure;

[0020] Sample classification processing, specifically for each sample The following three operations are performed in sequence: similar hyperbox matching determination, first category hyperbox construction, and adaptive hyperbox expansion. If the sample hyperbox update operation is successfully completed in any operation, the sample is considered to have completed the classification process and the classification process of the current sample is terminated. Specifically, the following steps are included:

[0021] Similar super box matching judgment, specifically if the sample Belongs to the existing super box of the same category , then for the hyperbox Perform boundary extension and update the hyperbox The minimum and maximum boundary values ​​of ;

[0022] The first category superbox is constructed, specifically if the sample The category does not appear in the current hyperbox set, that is, no similar hyperbox exists, so a new hyperbox is created directly. , and the sample Join in on it;

[0023] Adaptive hyperbox expansion, specifically if the sample If the category already exists in the superbox set but is not included in any similar superbox, a dynamic superbox expansion operation based on overlap risk avoidance is performed. Specifically, the dynamic superbox expansion operation first performs the same-category nearest superbox selection operation, then performs the same-category boundary overlap check operation, and if there is no similar-category overlap, then performs the heterogeneous boundary overlap risk determination operation.

[0024] The similar nearest hyperbox selection operation is specifically to calculate the sample The Euclidean distance to the geometric center of all existing hyperboxes of the same category, and select the nearest hyperbox as the pre-expanded hyperbox ;

[0025] The same type of boundary overlap check operation is specifically based on the boundary overlap detection condition check Super box with similar categories Is there any boundary overlap? If so, refuse to expand the hyperbox and create a new hyperbox directly. ;

[0026] The heterogeneous boundary overlap risk determination operation is specifically to determine the pre-expanded super box according to the boundary overlap detection condition. Whether it is super box with other categories If there is a boundary overlap, get the overlapping hyperbox. The geometric center of and radius , and use the Kd-Tree structure to search for distances that meet A cross-category sample set If it exists The samples in fall into overlapping hyperboxes If the value is within , it is determined that there is a potential risk of category confusion, and no expansion is performed, and a new hyperbox is created instead. Otherwise, it is considered that there is no obvious overlap risk and the super box is allowed to Perform expansion and update its minimum point With the maximum point ;in, Represents overlapping hyperboxes The minimum boundary vector of Represents overlapping hyperboxes The maximum boundary vector of and Represent overlapping hyperboxes The upper and lower boundaries of the i-th dimension;

[0027] The calculation of the navigation state membership is specifically performed by constructing a navigation state membership function based on boundary dynamic control; specifically, the calculation includes the following steps:

[0028] Dynamic modification of the hyperbox boundary, specifically for each built hyperbox By judging the positional relationship between the dimensional mean of the samples in the hyperbox and the geometric center, the hyperbox boundary is dynamically corrected. , the upper boundary remains unchanged, and the corrected lower boundary , otherwise the lower boundary remains unchanged and the revised upper boundary ;in, represents the j-th hyperbox, represents the original lower boundary of the j-th hyperbox, represents the original upper boundary of the j-th hyperbox, Represents the value of the t-th sample in the i-th dimension;

[0029] Define the navigation status membership function, the formula used is as follows:

[0030] ;

[0031] Where, represents the navigation state membership value of the t-th sample to the j-th hyperbox, n represents the total number of feature dimensions, and i represents the index of the feature dimension;

[0032] The navigation status recognition result is output. Specifically, after completing the membership calculation of the sample to each hyperbox, the hyperbox with the largest membership value is selected. ,That The corresponding navigation status category is the navigation status category to which the current sample belongs;

[0033] Construct and train the model, specifically by initializing the structural parameters of the recognition model, the sample classification processing, the navigation state membership calculation and the navigation state recognition result output, to realize the construction of the ship navigation state recognition model, based on the historical power control data as training input data, complete the training process of the recognition model, and obtain the trained ship navigation state recognition model.

[0034] Furthermore, the ship power strategy generation module specifically includes the following steps:

[0035] The power control template strategy selection is specifically to match the power control strategy data corresponding to the current navigation state in the preset power control template strategy library according to the current navigation state identification result, and extract the corresponding power control parameter values ​​from it as the initial reference values ​​of the optimization parameters, and use the parameter set as the parameter variable set to be optimized;

[0036] Setting the power control parameter boundaries, specifically, automatically retrieving the upper and lower limits corresponding to each control parameter from the state boundary mapping table based on the current navigation state identification result, and generating the dynamic boundary interval of each parameter accordingly;

[0037] The power control parameter optimization is to use an improved optimization algorithm to dynamically optimize the selected power control parameters under the above boundary constraints to obtain the optimal control parameter combination adapted to the current navigation state. The following steps are included:

[0038] The search population is initialized by taking the set of parameter variables to be optimized as the search individual position vector in the optimization algorithm, and generating disturbances based on the initial reference value of the optimization parameter and the dynamic boundary interval of each parameter to form multiple search individual position vectors. , thus completing the initialization of the search population;

[0039] Calculation of search individual fitness value, specifically calculating the fitness value of search individuals in the population ; Using the constructed dynamic control optimization objective function as the fitness evaluation standard, calculate the fitness value of the individual, sort the search individuals of the entire population from best to worst according to the individual fitness value, and obtain the global optimal position of the individual ;

[0040] The power control optimization objective function is used to measure the comprehensive performance of the control parameter combination under the current navigation state, serving as an evaluation criterion for the optimization algorithm. Specifically, the four performance indicators of propulsion efficiency, speed control accuracy, attitude stability, and control smoothness are comprehensively considered, and all performance indicators are normalized to unit, and the power control optimization objective function is constructed in the form of weighted summation.

[0041] The population search direction decision is to introduce a population search direction control factor to regulate the switching of the population between the global search and local optimization stages. The formula used is as follows:

[0042] ;

[0043] Where, Indicates the search direction control factor of the t-th generation population, t represents the current number of iterations, represents the maximum number of iterations, Represents the initialization population event selection factor;

[0044] Search individual multi-stage position update, specifically based on the current iterative population search direction control factor , determine the search direction and perform individual position update, if , then the global search phase is carried out, focusing on large-span search within the solution space. Otherwise, it enters the local optimization phase, focusing on fine-grained optimization around the current optimal solution. According to the above judgment, the individual position is updated to obtain the j-th dimension position of the i-th individual in the t+1 generation population. ;

[0045] Individual position update retention, specifically by comparing the fitness value of the individual's current position with the updated position, to determine the individual to be retained. If the fitness of the updated position is better than the current generation, the updated individual position is retained, otherwise the current individual position is retained;

[0046] Adaptive reverse guidance is used to introduce a probability-driven reverse learning strategy for the individual with the worst fitness in the population. Specifically, an adaptive reverse guidance mechanism is designed to perform reverse perturbation updates on the individual with the worst fitness in the population in a probability-driven manner. The formula used is expressed as follows:

[0047] ;

[0048] ;

[0049] Where, represents the position of the individual after reverse guidance, and Respectively represent the search upper and lower limits of the j-th dimension, Indicates the final updated position of the inferior individual, Both represent random numbers uniformly distributed in the range [0,1]. represents the power control optimization objective function, Indicates the individual with the worst fitness in the population;

[0050] Search for the optimal position of an individual. Specifically, after each round of iteration, the fitness values ​​of all individuals in the current population are evaluated. If there is an individual position with a fitness better than the current global optimal individual position, the individual global optimal position is updated with this individual.

[0051] The search iteration ends, specifically when the fitness value of the search individual When the fitness threshold is higher than the threshold and the maximum number of iterations is reached, the search is terminated and the individual global optimal position is obtained. The individual global optimal position specifically refers to the optimal control parameter combination of the current navigation state;

[0052] The power optimal control strategy output is specifically to generate and output the power optimal control strategy under the current navigation state based on the optimal control parameter combination of the current navigation state.

[0053] Furthermore, the power intelligent control module specifically first determines whether the current power control strategy needs to be updated based on the power control strategy adjustment trigger module; if the strategy adjustment is triggered, the real-time power control data is input into the ship navigation state identification model to generate the current navigation state identification result, and the power strategy generation module outputs the corresponding power optimal control strategy based on the state, and finally completes the dynamic adjustment and control of the ship propulsion system according to the strategy, otherwise maintains the existing control strategy and continues the current power system operation state, thereby realizing the adaptive adjustment and intelligent decision-making execution of the ship power control.

[0054] The beneficial effects achieved by the present invention using the above scheme are as follows:

[0055] (1) Aiming at the technical problems of frequent strategy adjustment and delayed response in traditional ship power control systems, which lead to unstable response of the power control system and deterioration of ship navigation performance, this scheme innovatively introduces a power control strategy adjustment triggering method with a dual judgment logic mechanism. By introducing a trigger logic that combines single parameter out-of-bounds judgment with multi-parameter combination deviation judgment, it can achieve accurate judgment of the timing of strategy adjustment, effectively avoid false triggering of strategy adjustment, and effectively realize intelligent update and stable control of ship power control strategy, significantly improving the economy and operational reliability of ship power control.

[0056] (2) Aiming at the technical problems of rigid hyperbox division and fuzzy state boundary identification in the existing models applicable to ship navigation state recognition, which leads to the inability to accurately adapt to the changeable characteristics of complex navigation states and the decline in navigation state recognition accuracy, this scheme innovatively combines the dynamic hyperbox expansion mechanism of overlapping risk avoidance with the membership function construction method of boundary dynamic regulation to construct a recognition model. It can effectively solve the problems of attribution ambiguity and classification conflict faced by traditional models when the distribution of similar samples is uneven or the category boundaries are adjacent, improve the structural flexibility and adaptability of the model, enhance the model's response ability to the transition area of ​​the navigation state boundary, and improve the stability and credibility of the recognition results. This method realizes the high-precision recognition of the navigation state of the ship navigation state recognition model in a dynamic and complex environment.

[0057] (3) In response to the technical problems of the lack of physical constraints and poor global optimization capabilities in the existing methods for generating power control strategies, which lead to the generated power control strategies deviating from the actual ship navigation scenarios, and thus resulting in poor strategy adaptability and unstable control effects, this scheme innovatively designs a multi-stage strategy generation method. By constructing a three-layer strategy generation architecture, the first layer completes the template strategy selection based on the navigation state recognition results to ensure that the parameter structure has engineering reference; the second layer sets the parameter boundaries to limit the search space to the physically feasible range; the third layer introduces an improved intelligent optimization algorithm with a population search direction control factor and an adaptive reverse guidance mechanism to perform dynamic optimization search on the selected parameters under boundary constraints, thereby enhancing the physical consistency and engineering feasibility of strategy generation, improving the global optimization capability of the control strategy, and significantly improving the accuracy, stability and intelligent adaptability of the ship's power control strategy in complex navigation environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 A schematic diagram of a module of an intelligent ship power control system provided by the present invention;

[0059] Figure 2 A flow chart of the power control strategy adjustment trigger module;

[0060] Figure 3 A flow chart of the module construction for the ship navigation status recognition model;

[0061] Figure 4 A flow chart of the ship power strategy generation module;

[0062] Figure 5 A schematic diagram of the process for optimizing power control parameters in the ship power strategy generation module;

[0063] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. They are used to explain the present invention together with the embodiments of the present invention and do not constitute a limitation of the present invention. DETAILED DESCRIPTION

[0064] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments; based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0065] In the description of the present invention, it should be understood that terms such as "up", "down", "front", "back", "left", "right", "top", "bottom", "inside" and "outside" indicating directions or positional relationships are based on the directions or positional relationships shown in the accompanying drawings. They are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the system or element referred to must have a specific direction, be constructed and operated in a specific direction. Therefore, they should not be understood as limiting the present invention.

[0066] Example 1, see Figure 1 The present invention provides an intelligent ship power control system, which includes a data collection module, a power control strategy adjustment trigger module, a ship navigation state recognition model construction module, a ship power strategy generation module and a power intelligent control module;

[0067] The data collection module is used to collect the raw data required for ship power control, specifically by collecting data from various power subsystems of the ship and the ship's navigation environment, obtaining the raw data of ship power control, and sending the data to the power control strategy adjustment trigger module;

[0068] The power control strategy adjustment trigger module receives the data sent by the data collection module, specifically by preprocessing and feature selecting the raw data, defining the strategy trigger parameters and their threshold ranges, and using the single parameter out-of-bounds and multi-parameter combination deviation judgment logic to determine whether the current power control strategy needs to be adjusted, obtain the ship power control optimization data and strategy adjustment trigger results, and send the data to the ship navigation status identification model construction module;

[0069] The ship navigation state recognition model construction module receives data sent by the power control strategy adjustment trigger module, and is used to construct and train a recognition model for accurately identifying the ship navigation state under different operating environments and working conditions. Specifically, the model structure parameters are first initialized, and then sample classification processing is performed, and similar hyperbox matching, first category hyperbox creation and adaptive boundary expansion operations are performed in sequence to form a boundary structure expression of the navigation state. Then, by constructing a navigation state membership function, the membership degree of each input sample to each navigation state category is calculated; finally, the state with the highest membership degree is output as the recognition result, and the model training is completed in combination with historical power control data to obtain a trained ship navigation state recognition model, and the data is sent to the ship power strategy generation module;

[0070] The ship power strategy generation module receives data sent by the ship navigation state identification model construction module, and is used to generate a power optimal control strategy with adaptive performance under the identified current navigation state. Specifically, the module first performs power control template strategy selection according to the navigation state identification result, sets the corresponding power control parameter boundaries, determines the parameter set to be optimized and its dynamic boundary interval, and uses an improved intelligent optimization algorithm to dynamically optimize the selected parameters within the constraint range to obtain the optimal control parameter combination adapted to the current navigation state. Based on the optimal control parameter combination, the module obtains the power optimal control strategy under the current navigation state, and sends the data to the power intelligent control module;

[0071] The intelligent power control module receives data sent by the ship power strategy generation module, and is used to realize adaptive adjustment and intelligent decision-making control of the ship propulsion system, specifically to determine whether the control strategy needs to be updated; if the update is triggered, the current navigation status is identified, and the optimal control strategy is generated based on the status. Finally, the power control system is dynamically adjusted according to the strategy to realize intelligent control of the ship power.

[0072] Example 2, see Figure 1, this embodiment is based on the above embodiment, the data collection module specifically collects the original data required for ship power control from various power subsystems of the ship and the ship's navigation environment to obtain the ship power control original data; the ship power control original data includes historical power control data and real-time power control data; the historical power control data and real-time power control data both include ship operation data, ship power data, ship navigation environment data, power control strategy data and power control log data; the historical power control data also includes historical ship navigation state identification results; the historical ship navigation state identification results include normal cruising navigation state, wind and wave interference navigation state, load jump navigation state, steady-state high-efficiency navigation state and berthing low-speed navigation state; the ship operation data includes speed, acceleration , deceleration, heading angle, yaw angle, roll angle, pitch angle, draft and cabin pressure; the ship power data includes engine output power, engine load rate, fuel flow, fuel consumption rate, propulsion motor torque and current; the ship navigation environment data includes wave height, wave direction, wind speed, wind direction, water flow direction, water flow speed, air pressure, air temperature, humidity and visibility level; the power control strategy data includes propulsion power setting value, propeller speed, throttle control command, response time constant, attitude adjustment factor and target speed; the throttle control command is used to adjust the output intensity of fuel; the attitude adjustment factor is used to enhance the role weight of attitude stabilization control in the strategy; the power control log data includes control command response time, alarm events, manual intervention records and power control strategy adjustment trigger records.

[0073] Example 3, see Figure 1 and Figure 2 This embodiment is based on the above embodiment. The power control strategy adjustment trigger module includes data preprocessing, navigation state identification feature extraction, strategy adjustment trigger parameter definition, parameter threshold definition and strategy adjustment trigger judgment, and specifically includes the following steps:

[0074] Data preprocessing is used to clean and standardize the original data of ship power control, specifically to clean and standardize the original data of ship power control to obtain ship power control optimization data;

[0075] The data cleaning specifically includes filling missing values, removing outliers and normalizing fields on the original data;

[0076] The standardization process specifically involves normalizing the numerical data in the original data using a minimum-maximum normalization method;

[0077] Navigation state identification feature selection is used to extract key characteristic indicators reflecting the trend of navigation state changes from ship power control optimization data. Specifically, the correlation analysis method is used to screen out features related to navigation state identification from ship power control optimization data;

[0078] The strategy adjustment trigger parameter definition is used to determine the key trigger parameter set for judging the change of the ship's navigation state under the premise of the current control strategy setting; specifically, the strategy adjustment trigger parameter is composed of the external environment key parameters and the ship state key parameters. The external environment key parameters include wave height, wind speed and direction, and water flow speed and direction; the ship state key parameters include ship attitude angle, ship speed, and ship propulsion power;

[0079] Parameter threshold definition, specifically, based on the historical data of the strategy adjustment trigger parameters collected during the stable operation of the current power control strategy, average statistics are performed to generate the reference mean of each strategy adjustment trigger parameter, and combined with the fixed deviation range set by the current power control strategy, the strategy adjustment trigger parameter threshold range is obtained;

[0080] Strategy adjustment trigger judgment, specifically, adopting a dual judgment logic mechanism consisting of single parameter out-of-bounds judgment logic and multi-parameter combination deviation judgment logic, performing logical judgments in sequence, and determining whether to trigger the control strategy adjustment process based on the judgment results to obtain the strategy adjustment trigger result;

[0081] The single parameter out-of-bounds judgment logic is specifically that if the real-time observation value of any strategy adjustment trigger parameter exceeds its corresponding threshold range, it is considered that the current navigation state has deviated from the adaptation range of the current power control strategy, triggering the power control strategy adjustment requirement; if all parameters are within the bounds, the multi-parameter combination deviation judgment logic is continued to be executed;

[0082] The multi-parameter combination deviation judgment logic specifically calculates the deviation of each strategy adjustment trigger parameter separately to obtain the parameter deviation, introduces the corresponding weighting coefficient, and calculates the weighted total deviation value. If the weighted total deviation value exceeds the preset deviation value threshold, it is considered that the current navigation state as a whole deviates from the strategy adaptation range, triggering a power control strategy adjustment request;

[0083] The parameter deviation is calculated by dividing the absolute difference between the real-time observation value of each policy adjustment trigger parameter and its reference mean by the parameter reference mean.

[0084] By performing the above operations, the technical problems of frequent strategy adjustments and delayed response in traditional ship power control systems, which lead to unstable response of the power control system and degradation of ship navigation performance, are addressed. This scheme innovatively introduces a power control strategy adjustment triggering method with a dual judgment logic mechanism. By introducing a trigger logic that combines single-parameter out-of-bounds judgment with multi-parameter combination deviation judgment, it achieves accurate judgment of the timing of strategy adjustment, effectively avoids false triggering of strategy adjustment, and can effectively realize intelligent updating and stable control of ship power control strategies, significantly improving the economy and operational reliability of ship power control.

[0085] Example 4, see Figure 1 and Figure 3 This embodiment is based on the above embodiment, and the ship navigation state recognition model construction module specifically includes the following steps:

[0086] Initialize the structural parameters of the recognition model to define the initial structure and boundary conditions of the recognition model. Specifically, initialize the minimum boundary matrix V and the maximum boundary matrix W to all zero matrices, and the dimensions of both are , where M is the number of samples, n is the feature dimension, and the hyperbox set is initialized , used to store the currently constructed hyperbox structure and build the corresponding Kd-Tree index structure for each category of training samples to accelerate the subsequent similar point search process in space;

[0087] The construction of the corresponding Kd-Tree index structure is specifically to select the feature dimension with the largest variance as the splitting axis in each division, and recursively divide it with the median as the tangent point until no further subset division is possible;

[0088] Sample classification processing is used to achieve refined identification of ships in different navigation states and dynamic construction of state structure to avoid state boundary conflicts. Specifically, for each sample The following three operations are performed in sequence: similar hyperbox matching determination, first category hyperbox construction, and adaptive hyperbox expansion. If the sample hyperbox update operation is successfully completed in any operation, the sample is considered to have completed the classification process and the classification process of the current sample is terminated. Specifically, the following steps are included:

[0089] Similar hyperbox matching judgment is used to determine whether the sample has been included in the hyperbox corresponding to the current category. Specifically, if the sample Belongs to the existing super box of the same category , then for the hyperbox Perform boundary extension and update the hyperbox The minimum and maximum boundary values ​​of ; the formula used is as follows:

[0090] ;

[0091] Where, and Represents the extended front hyperbox The minimum and maximum boundaries of and Represent the expanded hyperbox The minimum and maximum boundaries of represents the minimum function, represents the maximum value function, Indicates a hyperbox The minimum bound in the i-th dimension, Indicates a hyperbox The maximum boundary in the i-th dimension, Represents the value of the t-th sample in the i-th dimension;

[0092] The first category hyperbox construction is used to handle the situation where the category to which the current sample belongs has not yet appeared in the hyperbox set. Specifically, if the sample The category does not appear in the current hyperbox set, that is, no similar hyperbox exists, so a new hyperbox is created directly. , and the sample Join in on it;

[0093] Adaptive hyperbox expansion is used to handle the situation where the current sample is not covered by an existing hyperbox of the same type. Specifically, if the sample If the category already exists in the superbox set but is not included in any similar superbox, a dynamic superbox expansion operation based on overlap risk avoidance is performed. Specifically, the dynamic superbox expansion operation first performs the same-category nearest superbox selection operation, then performs the same-category boundary overlap check operation, and if there is no similar-category overlap, then performs the heterogeneous boundary overlap risk determination operation.

[0094] The similar nearest hyperbox selection operation is specifically to calculate the sample The Euclidean distance to the geometric center of all existing hyperboxes of the same category, and select the nearest hyperbox as the pre-expanded hyperbox ;

[0095] The same type of boundary overlap check operation is specifically based on the boundary overlap detection condition check Super box with similar categories Is there any boundary overlap? If so, refuse to expand the hyperbox and create a new hyperbox directly. ;

[0096] The boundary overlap specifically means that two super boxes have interval intersection in each dimension;

[0097] The boundary overlap detection condition is specifically that if any two super boxes , There is overlap if and only if the boundary intervals of all its dimensions intersect, that is, ,but ; then the two hyperbox boundaries are considered to overlap;

[0098] The heterogeneous boundary overlap risk determination operation is specifically to determine the pre-expanded super box according to the boundary overlap detection condition. Whether it is super box with other categories If there is a boundary overlap, get the overlapping hyperbox. The geometric center of and radius , and use the Kd-Tree structure to search for distances that meet A cross-category sample set If it exists The samples in fall into overlapping hyperboxes If the value is within , it is determined that there is a potential risk of category confusion, and no expansion is performed, and a new hyperbox is created instead. Otherwise, it is considered that there is no obvious overlap risk and the super box is allowed to Perform expansion and update its minimum point With the maximum point ; The formula used is as follows:

[0099] ;

[0100] Where, and Represents the extended front hyperbox The minimum and maximum boundaries of and Represent the expanded hyperbox The minimum and maximum boundaries of Represents overlapping hyperboxes The minimum boundary vector of Represents overlapping hyperboxes The maximum boundary vector of and Represent overlapping hyperboxes The upper and lower boundaries of the i-th dimension;

[0101] The navigation state membership calculation is used to measure the similarity between the input sample and the hyperbox corresponding to each navigation state, thereby reflecting its belonging confidence in each navigation state. Specifically, it is achieved by constructing a navigation state membership function based on boundary dynamic regulation. Specifically, it includes the following steps:

[0102] Dynamic correction of hyperbox boundaries is used to fine-tune the boundaries of each built hyperbox in each characteristic dimension, thereby improving the ability of the membership function to analyze complex state boundaries. Specifically, for each built hyperbox By judging the positional relationship between the dimensional mean of the samples in the hyperbox and the geometric center, the hyperbox boundary is dynamically corrected. , the upper boundary remains unchanged, and the corrected lower boundary , otherwise the lower boundary remains unchanged and the revised upper boundary ; The formula used is as follows:

[0103] ;

[0104] Where, represents the tth sample, represents the j-th hyperbox, represents the original lower boundary of the j-th hyperbox, represents the corrected lower boundary of the j-th hyperbox, represents the original upper boundary of the j-th hyperbox, represents the corrected upper boundary of the j-th hyperbox;

[0105] Define the navigation status membership function, the formula used is as follows:

[0106] ;

[0107] Where, represents the navigation state membership value of the t-th sample to the j-th hyperbox, and n represents the total number of feature dimensions;

[0108] The navigation state recognition result output is used to output the navigation state recognition result corresponding to the input sample; specifically, after completing the membership calculation of the sample to each hyperbox, the hyperbox with the largest membership value is selected. ,That The corresponding navigation status category is the navigation status category to which the current sample belongs;

[0109] Construct and train the model, specifically by initializing the structural parameters of the recognition model, the sample classification processing, the navigation state membership calculation and the navigation state recognition result output, to realize the construction of the ship navigation state recognition model, based on the historical power control data as training input data, complete the training process of the recognition model, and obtain the trained ship navigation state recognition model.

[0110] By performing the above operations, in order to solve the technical problems of rigid hyperbox division and fuzzy state boundary identification in the existing ship navigation state recognition model, which leads to the inability to accurately adapt to the changeable characteristics of complex navigation states and thus leads to a decrease in the accuracy of navigation state recognition, this scheme innovatively combines the dynamic hyperbox expansion mechanism for overlapping risk avoidance with the membership function construction method for dynamic boundary regulation to construct an identification model. It can effectively solve the problems of attribution ambiguity and classification conflict faced by traditional models when the distribution of similar samples is uneven or the category boundaries are adjacent, improve the structural flexibility and adaptability of the model, enhance the model's response ability to the transition area of ​​the navigation state boundary, and improve the stability and credibility of the recognition results. This method realizes high-precision recognition of the navigation state of the ship navigation state recognition model in a dynamic and complex environment.

[0111] Example 5, see Figure 1 、 Figure 4 and Figure 5 This embodiment is based on the above embodiment, and the ship power strategy generation module specifically includes the following steps:

[0112] The power control template strategy selection is specifically to match the power control strategy data corresponding to the current navigation state in the preset power control template strategy library according to the current navigation state identification result, and extract the corresponding power control parameter values ​​from it as the initial reference values ​​of the optimization parameters, and use the parameter set as the parameter variable set to be optimized;

[0113] Each strategy in the power control template strategy library predefines a set of typical parameter values ​​and control logic based on engineering experience for a specific navigation state;

[0114] Setting the power control parameter boundaries is used to set dynamic boundary conditions for the power control parameters to be optimized, ensuring that the parameter search is carried out within the physically reasonable and system stability allowable range. Specifically, based on the current navigation state identification results, the state boundary mapping table automatically retrieves the upper and lower limits corresponding to each control parameter, and generates the dynamic boundary interval for each parameter accordingly;

[0115] The state boundary mapping table predefines the mapping relationship between different navigation states and the upper and lower limits of corresponding control parameters;

[0116] The power control parameter optimization is to use an improved optimization algorithm to dynamically optimize the selected power control parameters under the above boundary constraints to obtain the optimal control parameter combination adapted to the current navigation state. The following steps are included:

[0117] The search population is initialized by taking the set of parameter variables to be optimized as the search individual position vector in the optimization algorithm, and generating disturbances based on the initial reference value of the optimization parameter and the dynamic boundary interval of each parameter to form multiple search individual position vectors. , thus completing the initialization of the search population;

[0118] The fitness value calculation of the search individual is used to evaluate the control performance of each search individual in the population under the current navigation state and guide the convergence direction of the optimization algorithm. Specifically, the fitness value of the search individual in the population is calculated. ; Using the constructed dynamic control optimization objective function as the fitness evaluation standard, calculate the fitness value of the individual, sort the search individuals of the entire population from best to worst according to the individual fitness value, and obtain the global optimal position of the individual ;

[0119] The power control optimization objective function is used to measure the comprehensive performance of the control parameter combination under the current navigation state, and serves as the evaluation standard of the optimization algorithm. Specifically, it comprehensively considers four performance indicators: propulsion efficiency, speed control accuracy, attitude stability, and control smoothness. All performance indicators are normalized to unit, and the power control optimization objective function is constructed in the form of weighted summation. The formula used is as follows:

[0120] ;

[0121] Where, represents the power control optimization objective function, 、 、 and are the weight parameters of propulsion efficiency, speed control accuracy, attitude stability and control smoothness respectively; is the propulsion efficiency function, which represents the unit energy consumption of the propulsion system. is the speed control accuracy function, which represents the error between the actual speed and the target speed. is the attitude stability function, which represents the disturbance amplitude of the hull attitude. is the control smoothness function, which represents the hull vibration measurement;

[0122] The population search direction decision is used to balance the global exploration and local development of the population. Specifically, a population search direction control factor is introduced to regulate the switching of the population between the global search and local optimization stages. The formula used is expressed as follows:

[0123] ;

[0124] Where, Indicates the search direction control factor of the t-th generation population, t represents the current number of iterations, represents the maximum number of iterations, Represents the initialization population event selection factor;

[0125] Search individual multi-stage position update, specifically based on the current iterative population search direction control factor , determine the search direction and perform individual position update, if , then the global search phase is carried out, focusing on large-span search within the solution space. Otherwise, it turns to the local optimization phase, focusing on fine-grained optimization around the current optimal solution. According to the above judgment, the individual position is updated to obtain ; The formula used is as follows:

[0126] ;

[0127] Where, represents the j-th dimension position of the i-th individual in the t+1-th generation population, r represents a random integer uniformly distributed in the range [1, N], represents a random number uniformly distributed in the range [0,1]. represents the random dimensional position of a random individual in the t-th generation population, represents the random dimensional position of the i-th individual in the t-generation population, Represents the remainder when j is divided by 2, j represents the index of the dimension, Represents the randomness parameter used to control the search space, representing a random number uniformly distributed in the range [0,2π], 、 and Both represent random numbers uniformly distributed in the range [0,1]. represents a random dimension, represents the j-th dimension position of a random individual in the t-th generation population, and represents random numbers that follow a normal distribution, Indicates the optimal individual position of the current iteration;

[0128] Individual position update retention, specifically by comparing the fitness value of the individual's current position with the updated position, to determine the individual to be retained. If the fitness of the updated position is better than the current generation, the updated individual position is retained, otherwise the current individual position is retained;

[0129] ;

[0130] Adaptive reverse guidance is used to introduce a probability-driven reverse learning strategy for the individual with the worst fitness in the population. Specifically, an adaptive reverse guidance mechanism is designed to perform reverse perturbation updates on the individual with the worst fitness in the population in a probability-driven manner. The formula used is expressed as follows:

[0131] ;

[0132] ;

[0133] Where, represents the position of the individual after reverse guidance, and Respectively represent the search upper and lower limits of the j-th dimension, Indicates the final updated position of the inferior individual, Both represent random numbers uniformly distributed in the range [0,1]. Indicates the individual with the worst fitness in the population;

[0134] Search for the optimal position of an individual. Specifically, after each round of iteration, the fitness values ​​of all individuals in the current population are evaluated. If there is an individual position with a fitness better than the current global optimal individual position, the individual global optimal position is updated with this individual.

[0135] The search iteration is terminated, specifically when the fitness value fi of the search individual is higher than the fitness threshold and the maximum number of iterations is reached, the search is terminated and the individual global optimal position is obtained. The individual global optimal position specifically refers to the optimal control parameter combination of the current navigation state;

[0136] The power optimal control strategy output is specifically to generate and output the power optimal control strategy under the current navigation state based on the optimal control parameter combination of the current navigation state.

[0137] By performing the above operations, in order to address the technical problems of the lack of physical constraints and poor global optimization capabilities in the existing power control strategy generation methods, which cause the generated power control strategies to deviate from the actual ship navigation scenarios, and thus result in poor strategy adaptability and unstable control effects, this scheme innovatively designs a multi-stage strategy generation method. By constructing a three-layer strategy generation architecture, the first layer completes the template strategy selection based on the navigation state recognition results to ensure that the parameter structure has engineering reference; the second layer sets the parameter boundaries to limit the search space to the physically feasible range; the third layer introduces an improved intelligent optimization algorithm with a population search direction control factor and an adaptive reverse guidance mechanism to dynamically optimize and search the selected parameters under boundary constraints, enhance the physical consistency and engineering feasibility of the strategy generation, improve the global optimization capability of the control strategy, and significantly improve the accuracy, stability and intelligent adaptability of the ship's power control strategy in complex navigation environments.

[0138] Example 6, see Figure 1 This embodiment is based on the above embodiment. Specifically, the power intelligent control module first determines whether the current power control strategy needs to be updated according to the power control strategy adjustment trigger module; if the strategy adjustment is triggered, the real-time power control data is input into the ship navigation state identification model to generate the current navigation state identification result, and the power strategy generation module outputs the corresponding power optimal control strategy based on the state, and finally completes the dynamic adjustment and control of the ship propulsion system according to the strategy; otherwise, the existing control strategy is maintained, the ship power strategy is not generated, and the current power system operation state is continued, thereby realizing the adaptive adjustment and intelligent decision-making execution of the ship power control.

[0139] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus.

[0140] While the embodiments of the present invention have been shown and described, it will be apparent to those skilled in the art that various changes, modifications, substitutions, and alterations can be made to the embodiments without departing from the principles and spirit of the invention.

[0141] The present invention and its embodiments are described above. This description is not restrictive. The drawings show only one embodiment of the present invention, and the actual structure is not limited thereto. In short, if a person skilled in the art is inspired by this and, without departing from the purpose of the present invention, designs structures and embodiments similar to this technical solution without inventiveness, they shall fall within the scope of protection of the present invention.

Claims

1. An intelligent ship power control system, characterized by: It includes data collection module, power control strategy adjustment trigger module, ship navigation status identification model construction module, ship power strategy generation module and power intelligent control module; The data collection module obtains raw data of ship power control by performing data collection operations; The power control strategy adjustment trigger module preprocesses and selects features of the original data, defines the strategy trigger parameters and their threshold ranges, and determines whether the current strategy needs to be adjusted through the judgment logic of single parameter out-of-bounds and multi-parameter combination deviation, thereby obtaining the ship power control optimization data and strategy adjustment trigger results; The ship navigation state recognition model construction module initializes the model structure parameters, sequentially performs similar hyperbox matching, first category hyperbox creation, and adaptive boundary expansion operations combined with overlap risk avoidance to perform sample classification processing, constructs a navigation state membership function based on boundary dynamic regulation, and completes the navigation state membership calculation of each sample; finally, it outputs the state with the highest membership as the recognition result, and combines the historical data training model to obtain the trained ship navigation state recognition model; The ship power strategy generation module specifically determines the set of parameters to be optimized and their dynamic boundary intervals based on the navigation state identification results, and uses an improved optimization algorithm that introduces a population search direction control factor and an adaptive reverse guidance mechanism to optimize the power control parameters within the constraint range to obtain the optimal control parameter combination. Based on this optimal control parameter combination, the optimal power control strategy under the current navigation state is obtained; The power intelligent control module determines whether to update the strategy based on the strategy adjustment trigger result. If the update is triggered, it identifies the current navigation status and generates the optimal control strategy based on the status, and dynamically adjusts the power system accordingly to realize intelligent control of the ship power.

2. The intelligent ship power control system according to claim 1, characterized in that: The power control strategy adjustment trigger module specifically includes the following steps: Data preprocessing, specifically cleaning and standardizing the raw data of ship power control to obtain ship power control optimization data; The selection of navigation state identification features is to use the correlation analysis method to screen out the features related to navigation state identification from the ship power control optimization data; Strategy adjustment trigger parameter definition, specifically, the strategy adjustment trigger parameter is composed of external environment key parameters and ship status key parameters; Parameter threshold definition, specifically, based on the historical data of the strategy adjustment trigger parameters collected during the stable operation of the current power control strategy, average statistics are performed to generate the reference mean of each strategy adjustment trigger parameter, and combined with the fixed deviation range set by the current power control strategy, the strategy adjustment trigger parameter threshold range is obtained; The strategy adjustment trigger judgment specifically adopts a dual judgment logic mechanism consisting of single-parameter out-of-bounds judgment logic and multi-parameter combination deviation judgment logic, performs logical judgment in sequence, and determines whether to trigger the control strategy adjustment process based on the judgment result to obtain the strategy adjustment trigger result.

3. The intelligent ship power control system according to claim 1, characterized in that: The ship navigation state recognition model construction module is used to construct and train a model for accurately identifying the ship navigation state under different operating environments and working conditions, and specifically includes the following steps: Initialize the structural parameters of the recognition model, specifically initialize the minimum boundary matrix V and the maximum boundary matrix W to all zero matrices, and the dimensions of both are , where M is the number of samples, n is the feature dimension, and the hyperbox set is initialized , and for each category of training samples, construct the corresponding Kd-Tree index structure; Sample classification processing, specifically for each sample The following three operations are performed in sequence: similar hyperbox matching judgment, first category hyperbox construction and adaptive hyperbox expansion. If the sample hyperbox update operation is successfully completed in any operation, the sample is considered to have completed the classification process and the classification process of the current sample is terminated; The calculation of the navigation state membership is specifically performed by constructing a navigation state membership function based on boundary dynamic control; specifically, the calculation includes the following steps: Dynamic modification of the hyperbox boundary, specifically for each built hyperbox By judging the positional relationship between the dimensional mean of the samples in the hyperbox and the geometric center, the hyperbox boundary is dynamically corrected. , the upper boundary remains unchanged, and the corrected lower boundary , otherwise the lower boundary remains unchanged and the revised upper boundary ;in, represents the j-th hyperbox, represents the original lower boundary of the j-th hyperbox, represents the original upper boundary of the j-th hyperbox, Represents the value of the t-th sample in the i-th dimension; Define the navigation status membership function, the formula used is as follows: ； Where, represents the navigation state membership value of the t-th sample to the j-th hyperbox, n represents the total number of feature dimensions, and i represents the index of the feature dimension; The navigation status recognition result is output. Specifically, after completing the membership calculation of the sample to each hyperbox, the hyperbox with the largest membership value is selected. ,That The corresponding navigation status category is the navigation status category to which the current sample belongs; Construct and train the model, specifically by initializing the structural parameters of the recognition model, the sample classification processing, the navigation state membership calculation and the navigation state recognition result output, to realize the construction of the ship navigation state recognition model, based on the historical power control data as the training input data, complete the training process of the recognition model, and obtain the trained ship navigation state recognition model.

4. The intelligent ship power control system according to claim 1, characterized in that: The sample classification process specifically includes the following steps: Similar super box matching judgment, specifically if the sample Belongs to the existing super box of the same category , then for the hyperbox Perform boundary extension and update the hyperbox The minimum and maximum boundary values ​​of ; The first category superbox is constructed, specifically if the sample The category does not appear in the current hyperbox set, that is, no similar hyperbox exists, so a new hyperbox is created directly. , and the sample Join in on it; Adaptive hyperbox expansion, specifically if the sample If the category already exists in the superbox set but is not included in any similar superbox, a dynamic superbox expansion operation based on overlap risk avoidance is performed. Specifically, the dynamic superbox expansion operation first performs the same-category nearest superbox selection operation, then performs the same-category boundary overlap check operation, and if there is no similar-category overlap, then performs the heterogeneous boundary overlap risk determination operation. The similar nearest hyperbox selection operation is specifically to calculate the sample The Euclidean distance to the geometric center of all existing hyperboxes of the same category, and select the nearest hyperbox as the pre-expanded hyperbox ; The same type of boundary overlap check operation is specifically based on the boundary overlap detection condition check Super box with similar categories Is there any boundary overlap? If so, refuse to expand the hyperbox and create a new hyperbox directly. ; The heterogeneous boundary overlap risk determination operation is specifically to determine the pre-expanded super box according to the boundary overlap detection condition. Whether it is super box with other categories If there is a boundary overlap, get the overlapping hyperbox. The geometric center of and radius , and use the Kd-Tree structure to search for distances that meet A cross-category sample set If it exists The samples in fall into overlapping hyperboxes If the value is within , it is determined that there is a potential risk of category confusion, and no expansion is performed, and a new hyperbox is created instead. Otherwise, it is considered that there is no obvious overlap risk and the super box is allowed to Perform expansion and update its minimum point With the maximum point ;in, Represents overlapping hyperboxes The minimum boundary vector of Represents overlapping hyperboxes The maximum boundary vector of and Represent overlapping hyperboxes The lower and upper bounds of the i-th dimension.

5. The intelligent ship power control system according to claim 1, characterized in that: The ship power strategy generation module specifically includes the following steps: The power control template strategy selection is specifically to match the power control strategy data corresponding to the current navigation state in the preset power control template strategy library according to the current navigation state identification result, and extract the corresponding power control parameter values ​​from it as the initial reference values ​​of the optimization parameters, and use the parameter set as the parameter variable set to be optimized; Setting the power control parameter boundaries, specifically, automatically retrieving the upper and lower limits corresponding to each control parameter from the state boundary mapping table based on the current navigation state identification result, and generating the dynamic boundary interval of each parameter accordingly; Power control parameter optimization, specifically, using an improved optimization algorithm to dynamically optimize the selected power control parameters under the above-mentioned boundary constraints to obtain the optimal control parameter combination that adapts to the current navigation state; The power optimal control strategy output is specifically to generate and output the power optimal control strategy under the current navigation state based on the optimal control parameter combination of the current navigation state.

6. The intelligent ship power control system according to claim 1, characterized in that: The power control parameter optimization specifically includes the following steps: The search population is initialized by taking the set of parameter variables to be optimized as the search individual position vector in the optimization algorithm, and generating disturbances based on the initial reference value of the optimization parameter and the dynamic boundary interval of each parameter to form multiple search individual position vectors. , thus completing the initialization of the search population; Calculation of search individual fitness value, specifically calculating the fitness value of search individuals in the population ; Using the constructed dynamic control optimization objective function as the fitness evaluation standard, calculate the fitness value of the individual, sort the search individuals of the entire population from best to worst according to the individual fitness value, and obtain the global optimal position of the individual ; The power control optimization objective function is used to measure the comprehensive performance of the control parameter combination under the current navigation state, serving as an evaluation criterion for the optimization algorithm. Specifically, the four performance indicators of propulsion efficiency, speed control accuracy, attitude stability, and control smoothness are comprehensively considered, and all performance indicators are normalized to unit, and the power control optimization objective function is constructed in the form of weighted summation. The population search direction decision is to introduce a population search direction control factor to regulate the switching of the population between the global search and local optimization stages. The formula used is as follows: ； Where, Indicates the search direction control factor of the t-th generation population, t represents the current number of iterations, represents the maximum number of iterations, Represents the initialization population event selection factor; Search individual multi-stage position update, specifically based on the current iterative population search direction control factor , determine the search direction and perform individual position update, if , then the global search phase is carried out, focusing on large-span search within the solution space. Otherwise, it enters the local optimization phase, focusing on fine-grained optimization around the current optimal solution. According to the above judgment, the individual position is updated to obtain the j-th dimension position of the i-th individual in the t+1 generation population. ; Individual position update retention, specifically by comparing the fitness value of the individual's current position with the updated position, to determine the individual to be retained. If the fitness of the updated position is better than the current generation, the updated individual position is retained, otherwise the current individual position is retained; Adaptive reverse guidance is used to introduce a probability-driven reverse learning strategy for the individual with the worst fitness in the population. Specifically, an adaptive reverse guidance mechanism is designed to perform reverse perturbation updates on the individual with the worst fitness in the population in a probability-driven manner. The formula used is expressed as follows: ； ； Where, represents the position of the individual after reverse guidance, and Respectively represent the search upper and lower limits of the j-th dimension, Indicates the final updated position of the inferior individual, Both represent random numbers uniformly distributed in the range [0,1]. represents the power control optimization objective function, Indicates the individual with the worst fitness in the population; Search for the optimal position of an individual. Specifically, after each round of iteration, the fitness values ​​of all individuals in the current population are evaluated. If there is an individual position with a fitness better than the current global optimal individual position, the individual global optimal position is updated with this individual. The search iteration ends, specifically when the fitness value of the search individual When the fitness threshold is higher than the maximum number of iterations, the search is terminated and the individual global optimal position is obtained. The individual global optimal position specifically refers to the optimal control parameter combination of the current navigation state.

7. The intelligent ship power control system according to claim 1, characterized in that: Specifically, the intelligent power control module first determines whether the current power control strategy needs to be updated based on the power control strategy adjustment trigger module; if the strategy adjustment is triggered, the real-time power control data is input into the ship navigation state identification model to generate the current navigation state identification result, and the power strategy generation module outputs the corresponding power optimal control strategy based on the state, and finally completes the dynamic adjustment and control of the ship propulsion system according to the strategy, otherwise maintains the existing control strategy and continues the current power system operation state, thereby realizing the adaptive adjustment and intelligent decision-making execution of the ship power control.

8. The intelligent ship power control system according to claim 1, characterized in that: The data collection module specifically collects the original data required for ship power control from various power subsystems of the ship and the ship's navigation environment to obtain the ship power control original data; the ship power control original data includes historical power control data and real-time power control data; the historical power control data and real-time power control data both include ship operation data, ship power data, ship navigation environment data, power control strategy data and power control log data; the historical power control data also includes historical ship navigation status identification results.

Images