Water charging station dynamic resource allocation method and system based on multi-objective optimization

Abstract

The invention provides a water charging station dynamic resource allocation method and system based on multi-objective optimization, and relates to the technical field of new energy, and the method comprises the steps: generating environment trend data through a deep learning algorithm, and predicting the impact on the charging efficiency; constructing a multi-objective optimization function of charging efficiency, operation cost and safety risk assessment; a multi-stage adaptive genetic algorithm is adopted for solving, a coding mode is selected according to environmental complexity, parallel rehearsal evolution is executed to obtain robust genes, and finally a charging berth and power distribution scheme is obtained. The method can adapt to a complex hydro-meteorological environment, improves the resource utilization rate of the charging station, and guarantees the charging safety.

Description

Technical Field

[0001] This invention relates to new energy technologies, and in particular to a dynamic resource allocation method and system for water-based charging stations based on multi-objective optimization. Background Technology

[0002] With the increasing electrification of water transportation, the number of electric vessels is gradually increasing, making floating charging stations, as key infrastructure supporting their operation, increasingly important. Floating charging stations need to provide safe and efficient charging services for various types of vessels in the complex and ever-changing aquatic environment, and their operational efficiency directly impacts the sustainable development of water transportation. The aquatic environment possesses unique complexities, including variations in hydrological conditions, fluctuating meteorological factors, and differences in vessel draft, all of which collectively affect charging efficiency, safety, and resource utilization.

[0003] Currently, resource allocation at floating charging stations primarily employs static scheduling methods, allocating charging berths and power based on fixed rules or simple reservation mechanisms. Some advanced systems are also beginning to use single-objective optimization algorithms, such as minimizing waiting time or maximizing charging efficiency, to improve resource utilization. Regarding forecasting technology, some systems incorporate simple meteorological and hydrological forecast data to adjust operational plans in advance, but lack in-depth modeling and analysis of the relationship between environmental factors and charging efficiency. Existing technologies for resource allocation at floating charging stations suffer from the following deficiencies and shortcomings: The lack of accurate modeling of the impact of complex hydrological and meteorological environments on charging efficiency means that resource allocation schemes often fail to achieve the expected results under harsh or drastically changing environmental conditions, leading to low charging efficiency and resource waste. Most existing optimization methods employ single-objective or statically weighted multi-objective optimization, failing to dynamically adjust the priority of each objective according to environmental changes, making it difficult to achieve an optimal balance between efficiency, cost, and safety. Existing algorithms are insufficiently adaptable to environmental fluctuations and lack robustness guarantees in volatile environments. When environmental conditions change significantly, the original resource allocation scheme may completely fail, increasing safety risks and operating costs. Summary of the Invention

[0004] The present invention provides a method and system for dynamic resource allocation of water charging stations based on multi-objective optimization, which can solve the problems in the prior art.

[0005] A first aspect of the present invention provides a method for dynamic resource allocation of floating charging stations based on multi-objective optimization, comprising: The hydrological and meteorological parameters of the floating charging station are obtained, environmental trend data are generated using deep learning algorithms, and the prediction results of the impact of the environment on charging efficiency are output. A multi-objective optimization function is constructed, which includes charging efficiency, operating cost and safety risk assessment based on environmental prediction. The safety risk assessment calculates the impact of the environment on the ship's charging status based on environmental trend data. The constraints of the multi-objective optimization function are dynamically generated according to the environmental trend data. The weights of each item in the multi-objective optimization function are determined through an adaptive weighting mechanism driven by environmental prediction. A multi-level adaptive genetic algorithm is used to solve the multi-objective optimization function. The environmental complexity level is determined based on the environmental trend data, and the corresponding chromosome encoding method is selected. Parallel pre-evolution is performed under the predicted environmental conditions at multiple time points, and chromosome segments are selected as robust genes based on fitness. Based on the robust genes, the genetic operator parameters are adjusted to perform population evolution to obtain the optimal individual that meets the convergence condition. The optimal individual is decoded to obtain the charging berth allocation scheme and the charging power allocation scheme.

[0006] In one alternative implementation, The steps for acquiring hydrological and meteorological parameters of the floating charging station, generating environmental trend data using deep learning algorithms, and outputting predictions of the environmental impact on charging efficiency include: A smoothed hydrological parameter sequence is obtained by using a sliding window to reduce noise in the hydrological parameters, and a comprehensive meteorological index is calculated based on the meteorological parameters. The smoothed hydrological parameter sequence, the comprehensive meteorological index, and the corresponding parameter change rate are used to construct a time-series feature matrix. The time-series feature matrix is ​​then input into a pre-trained long short-term memory network to generate environmental trend data. Based on environmental trend prediction data, the vertical disturbance force, horizontal disturbance force, and rotational torque acting on the charging interface are obtained; the contact pressure value and displacement deviation value of the charging interface under the action of the vertical disturbance force, horizontal disturbance force, and rotational torque are detected; the reference resistance value of the charging interface under standard operating conditions is obtained; the contact area ratio is calculated based on the contact pressure value; the displacement coefficient is calculated based on the displacement deviation value; the real-time contact resistance of the charging interface is obtained by multiplying the reference resistance value, the contact area ratio, and the displacement coefficient; the charging efficiency is calculated based on the change in the real-time contact resistance and the change in temperature.

[0007] In one alternative implementation, The steps for constructing a multi-objective optimization function that includes charging efficiency, operating costs, and safety risk assessment based on environmental predictions, wherein the safety risk assessment calculates the impact of the environment on the ship's charging status based on environmental trend data, the constraints of the multi-objective optimization function are dynamically generated based on environmental trend data, and the weights of each item in the multi-objective optimization function are determined through an adaptive weighting mechanism driven by environmental predictions include: The operating costs include charging time costs and energy consumption costs; The ship's displacement is calculated based on the environmental trend data, and the position risk value is calculated based on the difference between the ship's displacement and the preset safe distance threshold. The vertical disturbance force, horizontal disturbance force, and rotational torque acting on the charging interface are calculated based on the environmental trend data, and the mechanical risk value is calculated. The electrical risk value is calculated based on the real-time contact resistance of the charging interface. The weighted sum of the position risk value, the mechanical risk value, and the electrical risk value is used as the safety risk assessment result. An environmental sensitivity index is obtained by calculating the impact of the environmental trend data on charging efficiency, operating costs, and safety risk assessment results. Adaptive weight coefficients of each term in the multi-objective optimization function are then calculated based on the environmental sensitivity index. The multi-objective optimization function is obtained by multiplying the adaptive weight coefficients with the corresponding terms. Based on the environmental trend data, mechanical safety constraint thresholds and electrical safety constraint thresholds are set as dynamic constraints for the multi-objective optimization function.

[0008] In one alternative implementation, The steps for calculating the impact of the environmental trend data on charging efficiency, operating costs, and safety risk assessment results to obtain the environmental sensitivity index include: The environmental trend data is preprocessed to obtain a benchmark reference value. Based on the benchmark reference value, the basic assessment results of charging efficiency, operating cost and safety risk are calculated. The fluctuation frequency and fluctuation amplitude of environmental factors in the environmental trend data are extracted to construct an environmental disturbance vector. Based on the environmental disturbance vector, the steady-state offset of charging efficiency, operating cost and safety risk is calculated. The steady-state offset is compared with the preset fault tolerance range to obtain the stability margin. The environmental trend data is decomposed into a time series to obtain a trend term, a periodic term, and a random term; the rate of change of the trend term is used to obtain the environmental evolution speed, the dominant period of the periodic term is used to obtain the environmental change pattern, and the variance of the random term is used to obtain the environmental uncertainty. The environmental evolution speed, the environmental change pattern, and the environmental uncertainty are then used to construct an environmental evolution feature matrix. Based on the environmental evolution feature matrix, the dominant and secondary modes of environmental factors are identified, and the response characteristics of charging efficiency, operating costs and safety risks under the dominant and secondary modes are calculated. Based on the response characteristics, an environment-charging process response mapping relationship is constructed and the environmental combined effect coefficient is calculated. The correction coefficient is calculated based on the stability margin and the environmental combined effect coefficient; a benchmark value for the environmental sensitivity index is constructed based on the basic assessment results, and the environmental sensitivity index is obtained by multiplying the benchmark value by the correction coefficient.

[0009] In one alternative implementation, The multi-objective optimization function is solved using a multi-level adaptive genetic algorithm. The environmental complexity level is determined based on the environmental trend data, and the corresponding chromosome encoding method is selected. Parallel pre-evolutionary simulations are performed under predicted environmental conditions at multiple time points, and chromosome segments are selected as robust genes based on fitness. Based on the robust genes, the genetic operator parameters are adjusted to perform population evolution, obtaining the optimal individual that meets the convergence condition. The steps of decoding the optimal individual to obtain the charging berth allocation scheme and the charging power allocation scheme include: The environmental complexity level is determined based on the environmental trend data, and the chromosome encoding method is selected based on the environmental complexity level. The fitness value of individuals in the current population under the multi-objective optimization function is calculated, and representative individuals are selected according to the fitness value. The fitness value is obtained by weighted summation of service efficiency and energy efficiency. Parallel pre-evolution is performed under predicted environmental conditions at multiple time points, and the rate of change of the fitness value of the chromosome segments of the representative individuals under different environments is recorded. Chromosome segments with a fitness value change rate lower than the change rate threshold are extracted as robust genes. Based on the robust gene, calculate the dynamic adjustment coefficients of crossover probability and mutation probability, and perform crossover and mutation operations; select individuals containing the robust gene from the parent population and offspring population to form a new population; iterate the evolution of the new population until convergence to obtain the optimal individual; decode the optimal individual to obtain the charging berth allocation scheme and the charging power allocation scheme. When real-time environmental monitoring data deviates from the predicted trend, the dominant chromosome segments are preserved and the population is reconstructed, and the robust genes are updated based on the latest environmental data.

[0010] In one alternative implementation, The steps of performing parallel pre-evolutionary simulations under predicted environmental conditions at multiple time points, recording the rate of change of fitness values ​​of chromosome segments representing individuals under different environments, and extracting chromosome segments with fitness value change rates below a threshold as robust genes include: The environmental trend data is subjected to Fourier transform to obtain the intensity of environmental disturbance. An environmental element coupling matrix is ​​constructed based on the intensity of environmental disturbance. A multi-time-scale predicted environmental sequence is generated based on the environmental element coupling matrix. The chromosome is divided into functional genes and regulatory genes. The gene expression levels of the functional genes and the regulatory genes under the predicted environmental sequence are calculated. The gene environmental response is calculated based on the gene expression levels. The predicted environment sequence is assigned to a parallel evaluation unit, where population evolution is performed and the fitness values ​​of individuals in each predicted environment are recorded. The rate of change of fitness values ​​in adjacent predicted environments is calculated, and the mean of the rate of change is used as the environmental fitness index of the individual. The change rate threshold is determined based on the intensity of the environmental disturbance. Gene fragments are scored based on the gene environmental responsiveness and the environmental adaptability index. Gene fragments with scores lower than the change rate threshold are extracted to form a robust gene library. The complementarity characteristics of gene fragments in the robust gene library are analyzed, and the combination of gene fragments with the best complementarity is selected for population evolution.

[0011] In one alternative implementation, When real-time environmental monitoring data deviates from the predicted trend, the steps of preserving dominant chromosome segments and reconstructing the population, and updating the robust genes based on the latest environmental data include: The deviation between real-time environmental monitoring data and environmental trend data is calculated. When the deviation exceeds a preset threshold, it is determined that the environment has changed. The environmental complexity level is recalculated based on the real-time environmental monitoring data. Chromosomal segments containing the robust gene in the current population are identified and marked as dominant chromosomal segments. Based on the environmental complexity level, a chromosome encoding method is selected to construct a population template. The dominant chromosome segment is mapped to the corresponding position of the population template. Gene values ​​are generated for the remaining positions of the population template through random initialization to form a reconstructed population. Based on the real-time environmental monitoring data, a predicted environmental sequence of a preset time length is constructed. The reconstructed population is subjected to pre-evolution in the predicted environmental sequence of the preset time length to identify robust genes under the new environment. The similarity of the robust genes before and after the environmental change is calculated. Robust genes with similarity higher than a preset similarity threshold are retained, and newly identified robust genes are merged to update the robust gene library. Genetic operations are performed on the reconstructed population, and the selection probability of individuals containing the updated robust genes is set to a preset selection probability. When the improvement rate of the optimal fitness of the population is lower than the convergence threshold, the current optimal individual is taken as the optimal solution in the new environment.

[0012] A second aspect of the present invention provides a dynamic resource allocation system for floating charging stations based on multi-objective optimization, comprising: The first unit is used to acquire hydrological and meteorological parameters of the water charging station, use deep learning algorithms to generate environmental trend data, and output the prediction results of the impact of the environment on charging efficiency. The second unit is used to construct a multi-objective optimization function that includes charging efficiency, operating cost and safety risk assessment based on environmental prediction. The safety risk assessment calculates the impact of the environment on the ship's charging status based on environmental trend data. The constraints of the multi-objective optimization function are dynamically generated according to the environmental trend data. The weights of each item in the multi-objective optimization function are determined through an adaptive weighting mechanism driven by environmental prediction. The third unit is used to solve the multi-objective optimization function using a multi-level adaptive genetic algorithm, determine the environmental complexity level based on the environmental trend data and select the corresponding chromosome encoding method; perform parallel pre-evolution under the predicted environmental conditions at multiple time points, select chromosome segments as robust genes based on fitness; perform population evolution based on the robust genes by adjusting the genetic operator parameters to obtain the optimal individual that meets the convergence condition; and decode the optimal individual to obtain the charging berth allocation scheme and the charging power allocation scheme.

[0013] A third aspect of the present invention provides an electronic device, comprising: processor; Memory used to store processor-executable instructions; The processor is configured to invoke instructions stored in the memory to execute the aforementioned method.

[0014] A fourth aspect of the present invention provides a computer-readable storage medium having stored thereon computer program instructions that, when executed by a processor, implement the aforementioned method.

[0015] This invention acquires hydrological and meteorological parameters and uses deep learning to predict the impact of the environment on charging efficiency, enabling accurate perception and trend prediction of the charging station's operating environment, thereby improving the adaptability and efficiency of the charging process.

[0016] This invention constructs a multi-objective optimization function that includes charging efficiency, operating costs, and safety risks, and adopts an adaptive weighting mechanism driven by environmental prediction, so that resource allocation decisions can dynamically adjust the focus of optimization objectives according to environmental changes, ensuring safety while taking into account economy and efficiency.

[0017] This invention employs a multi-level adaptive genetic algorithm to solve optimization problems. By using environment complexity-driven encoding selection and parallel pre-evolution to extract robust genes, it significantly improves the algorithm's solution efficiency and stability in complex and variable aquatic environments, enabling charging stations to continuously provide reliable charging services under various environmental conditions. Attached Figure Description

[0018] Figure 1 This is a flowchart illustrating the dynamic resource allocation method for water-based charging stations based on multi-objective optimization, as described in an embodiment of the present invention. Figure 2 This is a flowchart illustrating the charging berth allocation scheme and charging power allocation scheme obtained by the present invention based on a genetic algorithm. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] The technical solution of the present invention will be described in detail below with reference to specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments.

[0021] Figure 1 This is a flowchart illustrating the dynamic resource allocation method for floating charging stations based on multi-objective optimization according to an embodiment of the present invention. Figure 1 As shown, the method includes: The hydrological and meteorological parameters of the floating charging station are obtained, environmental trend data are generated using deep learning algorithms, and the prediction results of the impact of the environment on charging efficiency are output. A multi-objective optimization function is constructed, which includes charging efficiency, operating cost and safety risk assessment based on environmental prediction. The safety risk assessment calculates the impact of the environment on the ship's charging status based on environmental trend data. The constraints of the multi-objective optimization function are dynamically generated according to the environmental trend data. The weights of each item in the multi-objective optimization function are determined through an adaptive weighting mechanism driven by environmental prediction. A multi-level adaptive genetic algorithm is used to solve the multi-objective optimization function. The environmental complexity level is determined based on the environmental trend data, and the corresponding chromosome encoding method is selected. Parallel pre-evolution is performed under the predicted environmental conditions at multiple time points, and chromosome segments are selected as robust genes based on fitness. Based on the robust genes, the genetic operator parameters are adjusted to perform population evolution to obtain the optimal individual that meets the convergence condition. The optimal individual is decoded to obtain the charging berth allocation scheme and the charging power allocation scheme.

[0022] In one optional implementation, the steps of acquiring hydrological and meteorological parameters of the floating charging station, generating environmental trend data using a deep learning algorithm, and outputting a prediction of the impact of the environment on charging efficiency include: A smoothed hydrological parameter sequence is obtained by using a sliding window to reduce noise in the hydrological parameters, and a comprehensive meteorological index is calculated based on the meteorological parameters. The smoothed hydrological parameter sequence, the comprehensive meteorological index, and the corresponding parameter change rate are used to construct a time-series feature matrix. The time-series feature matrix is ​​then input into a pre-trained long short-term memory network to generate environmental trend data. Based on environmental trend prediction data, the vertical disturbance force, horizontal disturbance force, and rotational torque acting on the charging interface are obtained; the contact pressure value and displacement deviation value of the charging interface under the action of the vertical disturbance force, horizontal disturbance force, and rotational torque are detected; the reference resistance value of the charging interface under standard operating conditions is obtained; the contact area ratio is calculated based on the contact pressure value; the displacement coefficient is calculated based on the displacement deviation value; the real-time contact resistance of the charging interface is obtained by multiplying the reference resistance value, the contact area ratio, and the displacement coefficient; the charging efficiency is calculated based on the change in the real-time contact resistance and the change in temperature.

[0023] For example, hydrological parameters (such as water flow velocity, wave height, and water depth variation) and meteorological parameters (such as wind speed, wind direction, rainfall, and temperature) are acquired through a sensor network. For noise reduction of the hydrological parameters, a sliding window technique is used. A window of length 60 is used to smooth the water flow velocity data. For each time t, 60 sampling points before and after it are taken, and the weighted average of these points is calculated as the smoothed result for that time. The weights are distributed using a Gaussian distribution, with a weight of 0.15 at the center point, gradually decreasing towards both sides. For example, for the original measured water flow velocity sequence [1.2, 1.5, 1.3, 1.8, 1.4...] m / s, the smoothed sequence [1.3, 1.4, 1.4, 1.5, 1.4...] m / s is obtained after sliding window processing. The same method is also applied to the processing of other hydrological parameters such as wave height and water depth variation.

[0024] The calculation of the comprehensive meteorological index integrates the influence of multiple meteorological parameters. The wind speed value (e.g., 8 m / s) is multiplied by the wind direction influence coefficient (0.8 for tailwind, 1.2 for crosswind, 1.5 for headwind), then multiplied by the rainfall intensity coefficient (e.g., 1.4 for heavy rain, 1.1 for light rain, 1.0 for no rain), and finally combined with the temperature influence factor (1.0 at 25℃, changing by 0.1 for every 5℃ increase or decrease) to obtain the final comprehensive meteorological index. In a specific scenario with a wind speed of 8 m / s, crosswind conditions, light rain, and a temperature of 20℃, the calculated comprehensive meteorological index is 8 × 1.2 × 1.1 × 0.9 = 9.5.

[0025] The construction of the time-series feature matrix includes three types of features: smoothed hydrological parameter sequences, comprehensive meteorological indices, and corresponding parameter change rates. The parameter change rate is obtained by dividing the difference between adjacent time points by the time interval. For example, if the water flow velocity changes from 1.4 m / s to 1.6 m / s over a time interval of 10 minutes, the change rate is 0.02 m / s / min. Each row of the constructed feature matrix represents a time point, and each column represents a different feature, including water flow velocity, wave height, water depth changes, comprehensive meteorological indices, and their respective change rates, forming a multi-dimensional time-series dataset.

[0026] The generation of environmental trend data relies on a pre-trained Long Short-Term Memory (LSTM) network. This network consists of an input layer, two LSTM hidden layers (128 nodes each), and a fully connected output layer. The network is trained using environmental data collected over the past year, including 50,000 time-series samples, trained via a stochastic gradient descent optimizer with a learning rate of 0.001, a batch size of 64, and 200 training epochs. After inputting the time-series feature matrix, the network can predict the changing trends of environmental parameters over the next four hours, including predicted values ​​for water flow velocity, wave height, and comprehensive meteorological conditions.

[0027] In the process of converting environmental trend forecast data into force analysis of the charging interface, a fluid dynamics model is used to calculate the applied forces. The vertical disturbance force is proportional to the square of the water flow velocity. When the water flow velocity is 1.5 m / s, the force acting on the standard charging interface (cross-sectional area 0.02 m²) is... 2 The vertical disturbance force is approximately 22.5 N. The horizontal disturbance force is affected by wave height and frequency; with a wave height of 0.5 m and a period of 4 s, the peak horizontal disturbance force can reach 18 N. The rotational torque is generated by the uneven distribution of horizontal and vertical forces, and under the above conditions, the typical value is 5.4 N·m.

[0028] The charging interface contact status is detected using a pressure sensor array and a displacement sensor. The standard contact pressure during normal operation is 200N, and the contact area is 500mm². 2 When the vertical disturbance force causes a change in pressure, the contact area will change accordingly. For example, when the vertical disturbance force is 22.5 N, the contact pressure decreases to 177.5 N, and the contact area decreases to 450 mm². 2 The contact area ratio is 0.9. The displacement deviation is measured by a three-dimensional displacement sensor. The horizontal disturbance force of 18N results in a horizontal displacement of 2mm and an angular deviation of 0.5°. The calculated displacement coefficient is 0.85.

[0029] Charging efficiency is calculated based on changes in contact resistance. Under standard operating conditions, the interface reference resistance is 5 mΩ. Combining the aforementioned contact area ratio and displacement coefficient, the real-time contact resistance is calculated as 5 mΩ ÷ 0.9 ÷ 0.85 = 6.54 mΩ. The effect of temperature is also considered; for every 10°C increase in temperature, the resistance increases by approximately 3.9%. The final charging efficiency is calculated based on the resistance change. For example, when the resistance increases from 5 mΩ to 6.54 mΩ, the charging efficiency decreases by approximately 23.5% (from 98% to 74.5%).

[0030] This invention employs a sliding window to denoise hydrological parameters and combines them with comprehensive meteorological indices to construct a time-series feature matrix, which is then input into a long short-term memory network, enabling accurate prediction of complex and variable aquatic environments. In particular, it calculates the three-dimensional disturbance force acting on the charging interface based on environmental trend prediction data, then quantitatively analyzes changes in charging interface performance using contact pressure and displacement deviation values, and finally calculates charging efficiency using real-time contact resistance and temperature changes, establishing a direct mapping relationship between environmental changes and charging efficiency.

[0031] In one optional implementation, a multi-objective optimization function is constructed, comprising charging efficiency, operating cost, and safety risk assessment based on environmental prediction. The safety risk assessment calculates the impact of the environment on the ship's charging status based on environmental trend data. The constraints of the multi-objective optimization function are dynamically generated based on the environmental trend data. The steps for determining the weights of each item in the multi-objective optimization function through an adaptive weighting mechanism driven by environmental prediction include: The operating costs include charging time costs and energy consumption costs; The ship's displacement is calculated based on the environmental trend data, and the position risk value is calculated based on the difference between the ship's displacement and the preset safe distance threshold. The vertical disturbance force, horizontal disturbance force, and rotational torque acting on the charging interface are calculated based on the environmental trend data, and the mechanical risk value is calculated. The electrical risk value is calculated based on the real-time contact resistance of the charging interface. The weighted sum of the position risk value, the mechanical risk value, and the electrical risk value is used as the safety risk assessment result. An environmental sensitivity index is obtained by calculating the impact of the environmental trend data on charging efficiency, operating costs, and safety risk assessment results. Adaptive weight coefficients of each term in the multi-objective optimization function are then calculated based on the environmental sensitivity index. The multi-objective optimization function is obtained by multiplying the adaptive weight coefficients with the corresponding terms. Based on the environmental trend data, mechanical safety constraint thresholds and electrical safety constraint thresholds are set as dynamic constraints for the multi-objective optimization function.

[0032] For example, operating costs include charging time costs and energy consumption costs. Charging time costs are calculated by multiplying the vessel's dwell time at the charging station by the unit time cost. For instance, if the unit time cost at the charging station is 200 yuan / hour, and the vessel stays at the station for 3 hours, the charging time cost is 600 yuan. Energy consumption costs are calculated by multiplying the charging power, charging time, and electricity price. For example, when the charging power is 350kW, the charging time is 2.5 hours, and the electricity price is 0.8 yuan / kWh, the energy consumption cost is 700 yuan. The total operating cost is obtained by weighted summing of these two costs. Considering the differences between different types of vessels, different weights are assigned to charging time costs and energy consumption costs. For example, for passenger vessels, the weight for charging time costs is set to 0.7, and the weight for energy consumption costs is set to 0.3; while for cargo vessels, these weights are 0.4 and 0.6, respectively.

[0033] Safety risk assessment comprises three dimensions: location risk, mechanical risk, and electrical risk. Location risk assessment is achieved by calculating ship displacement. Based on environmental trend data, the combined forces of water flow, waves, and wind on the ship are predicted, and the ship displacement at different time points is calculated. For example, in an environment with a water flow speed of 1.2 m / s, wave height of 0.4 m, and wind speed of 6 m / s, the predicted maximum displacement of the ship within 30 minutes is 0.8 m. The preset safe distance threshold is 1.5 m, therefore the safety margin is 0.7 m. The location risk value is calculated as the ratio of the safety margin to the safe distance threshold; in this example, the location risk value is 0.47. When the wind speed increases to 10 m / s, the predicted displacement increases to 1.2 m, the safety margin decreases to 0.3 m, and the location risk value increases to 0.8.

[0034] Mechanical risk assessment is achieved by analyzing the stress state of the charging interface. Based on environmental trend data, the vertical disturbance force, horizontal disturbance force, and rotational torque acting on the charging interface are calculated. The vertical disturbance force is mainly affected by water flow velocity, the horizontal disturbance force is mainly affected by waves, and the rotational torque is related to the resultant force of both and the point of application. Under typical environmental conditions, the vertical disturbance force on the charging interface is 15N, the horizontal disturbance force is 12N, and the rotational torque is 4N·m. These forces are compared with preset safety thresholds, and the risk coefficients for each force are calculated. A weighted sum is then used to obtain the mechanical risk value. The safety thresholds for vertical disturbance force are 30N, horizontal disturbance force is 25N, and rotational torque is 8N·m, with corresponding risk coefficients of 0.5, 0.48, and 0.5, respectively. The weights for these three factors are 0.3, 0.4, and 0.3, respectively, resulting in a calculated mechanical risk value of 0.494.

[0035] Electrical risk assessment is achieved by analyzing the real-time contact resistance of the charging interface. Higher contact resistance indicates poorer electrical connection quality and greater electrical risk. Changes in contact resistance during charging are monitored and compared to the standard operating resistance. For example, if the standard contact resistance is 5 mΩ and the real-time detected contact resistance is 6.2 mΩ, the resistance increase rate is 24%. The electrical risk coefficient is calculated based on the ratio of the resistance increase rate to a safety threshold. The electrical safety threshold is set at a 50% resistance increase, resulting in an electrical risk value of 0.48. When environmental degradation causes the contact resistance to rise to 7.1 mΩ, the resistance increase rate reaches 42%, and the electrical risk value rises to 0.84.

[0036] The location risk value, mechanical risk value, and electrical risk value are weighted and summed with weights of 0.3, 0.35, and 0.35 respectively to obtain the comprehensive safety risk assessment result. In the example above, the comprehensive safety risk value is 0.497. As environmental conditions change, each risk value is dynamically updated, and the comprehensive safety risk value changes accordingly. When the comprehensive safety risk value exceeds 0.8, a safety warning will be triggered; when it exceeds 0.9, it will be recommended to suspend charging operations.

[0037] Baseline values ​​for charging efficiency, operating cost, and safety risk were determined under baseline environmental conditions, such as a baseline charging efficiency of 95%, a baseline operating cost of 1000 yuan, and a baseline safety risk value of 0.3. The impact of changes in different parameters in environmental trend data on these three indicators was then analyzed. For example, when the water flow velocity increases from 0.8 m / s to 1.2 m / s, the charging efficiency decreases by 5 percentage points, the operating cost increases by 8%, and the safety risk value increases by 0.15. An environmental sensitivity matrix was constructed using these rates of change, calculating the environmental sensitivity for charging efficiency to 0.68, operating cost to 0.45, and safety risk to 0.75. A higher sensitivity value indicates a greater susceptibility of that indicator to environmental influences.

[0038] Based on environmental sensitivity indicators, adaptive weight coefficients are calculated for each term of the multi-objective optimization function. These weight coefficients are directly proportional to environmental sensitivity; indicators with higher sensitivity receive higher weights, ensuring the optimization process focuses on objectives significantly impacted by environmental changes. Under normal conditions, the initial weights for charging efficiency, operating cost, and safety risk in the baseline environment are 0.35, 0.35, and 0.3, respectively. As environmental conditions worsen, the weights are adjusted according to environmental sensitivity, resulting in weights of 0.32, 0.28, and 0.4, respectively. The adaptive weight coefficients are multiplied by the corresponding charging efficiency, operating cost, and safety risk indicators to construct the multi-objective optimization function. Using the weights in the example above, the constructed multi-objective optimization function is: charging efficiency multiplied by 0.32, operating cost multiplied by 0.28, and safety risk multiplied by 0.4. The objective of this optimization function is to maximize charging efficiency while minimizing operating cost and safety risk.

[0039] The constraints in the multi-objective optimization function include mechanical safety constraints that consider the impact of environmental factors on the stress state of the charging interface. For example, when the predicted wind speed exceeds 12 m / s, the mechanical safety constraint thresholds are adjusted, reducing the maximum permissible vertical disturbance force from 30 N to 25 N and the horizontal disturbance force from 25 N to 20 N. Electrical safety constraints consider the impact of environmental factors on the electrical performance of the charging interface. For example, when the predicted rainfall intensity increases or humidity rises, the electrical safety constraint thresholds are adjusted, reducing the maximum permissible increase rate of contact resistance from 50% to 35%. These dynamic constraints ensure that the charging interface always operates within safe limits under different environmental conditions.

[0040] This invention constructs a multi-objective optimization function that organically combines three key dimensions: charging efficiency, operating cost, and safety risk assessment, achieving comprehensive optimization of the allocation of resources for water-based charging stations. In particular, by separately calculating location risk values, mechanical risk values, and electrical risk values, and then weighting and fusing them, a comprehensive safety assessment system is constructed. Simultaneously, an adaptive weighting mechanism driven by environmental sensitivity indicators is introduced, enabling the optimization objectives to dynamically adjust their weights according to environmental changes, avoiding the one-sidedness of traditional fixed-weight methods in complex environments.

[0041] In one optional implementation, the step of calculating the impact of the environmental trend data on charging efficiency, operating costs, and safety risk assessment results to obtain an environmental sensitivity index includes: The environmental trend data is preprocessed to obtain a benchmark reference value. Based on the benchmark reference value, the basic assessment results of charging efficiency, operating cost and safety risk are calculated. The fluctuation frequency and fluctuation amplitude of environmental factors in the environmental trend data are extracted to construct an environmental disturbance vector. Based on the environmental disturbance vector, the steady-state offset of charging efficiency, operating cost and safety risk is calculated. The steady-state offset is compared with the preset fault tolerance range to obtain the stability margin. The environmental trend data is decomposed into a time series to obtain a trend term, a periodic term, and a random term; the rate of change of the trend term is used to obtain the environmental evolution speed, the dominant period of the periodic term is used to obtain the environmental change pattern, and the variance of the random term is used to obtain the environmental uncertainty. The environmental evolution speed, the environmental change pattern, and the environmental uncertainty are then used to construct an environmental evolution feature matrix. Based on the environmental evolution feature matrix, the dominant and secondary modes of environmental factors are identified, and the response characteristics of charging efficiency, operating costs and safety risks under the dominant and secondary modes are calculated. Based on the response characteristics, an environment-charging process response mapping relationship is constructed and the environmental combined effect coefficient is calculated. The correction coefficient is calculated based on the stability margin and the environmental combined effect coefficient; a benchmark value for the environmental sensitivity index is constructed based on the basic assessment results, and the environmental sensitivity index is obtained by multiplying the benchmark value by the correction coefficient.

[0042] For example, the raw environmental trend data is normalized to eliminate dimensional differences between different environmental factors. Typical values ​​for water flow velocity are 0.5-2.5 m / s, wave height is 0.1-1.5 m, and wind speed is 2-15 m / s. The environmental trend data is smoothed using a moving average method, and the average value is calculated based on data within a 30-minute time window. For water flow velocity, the original collected data, such as [1.2, 1.5, 1.3, 1.8, 1.4...] m / s, is processed by moving average to obtain [1.4, 1.5, 1.6, 1.5, 1.4...] m / s. The median of these smoothed data is selected as a benchmark reference value, such as a benchmark reference value of 1.5 m / s for water flow velocity, 0.4 m for wave height, and 6 m / s for wind speed. Based on these benchmark reference values, the basic assessment results for charging efficiency, operating costs, and safety risks are calculated. Under the above-mentioned baseline environmental conditions, the basic assessment result for charging efficiency is 92%, the basic assessment result for operating cost is 1200 yuan, and the basic assessment result for safety risk is 0.35.

[0043] Periodic analysis of environmental trend data was performed to extract the fluctuation frequency and amplitude of environmental factors. Taking water flow velocity as an example, analysis of nearly 4 hours of water flow velocity data identified the main fluctuation period as 40 minutes, with a fluctuation amplitude of ±0.3 m / s. Similarly, the main fluctuation period for wave height was 30 minutes, with a fluctuation amplitude of ±0.15 m, and the main fluctuation period for wind speed was 60 minutes, with a fluctuation amplitude of ±2 m / s. These fluctuation frequencies and amplitudes constituted the environmental disturbance vector. Based on the environmental disturbance vector, the impact of environmental fluctuations on charging efficiency, operating costs, and safety risks was simulated, and steady-state offsets were calculated. The steady-state offset for charging efficiency was ±3.5%, for operating costs it was ±120 yuan, and for safety risks it was ±0.08. These steady-state offsets were compared with preset fault tolerance ranges to obtain stability margins. The preset fault tolerance ranges were ±5% for charging efficiency, ±200 yuan for operating costs, and ±0.15 for safety risks. The calculated stability margins are 0.7, 0.6 and 0.53, respectively, all of which are greater than the safety threshold of 0.5, indicating that the system can maintain stable operation under the current environmental fluctuations.

[0044] Environmental trend data is decomposed into three components: trend, periodic, and random. Taking water flow velocity as an example, 8 hours of continuous monitoring data were collected, and the trend, periodic, and random components were obtained through a decomposition algorithm. The trend component shows that the water flow velocity generally exhibits a trend of first increasing and then decreasing, rising from 1.2 m / s to 1.8 m / s and then decreasing to 1.4 m / s within 8 hours. The periodic component shows that the water flow velocity exhibits multiple periodic fluctuations, with a dominant period of 120 minutes and a secondary period of 40 minutes. The random component shows irregular fluctuations with a variance of 0.04. The rate of change of the trend component is used to calculate the environmental evolution rate. The rate of change of the water flow velocity during the rising phase is 0.15 m / s / h, and the rate of change during the falling phase is -0.2 m / s / h, with an overall environmental evolution rate of 0.175 m / s / h. The environmental change patterns are obtained based on the dominant periods of the periodic component, such as the dominant period of water flow velocity being 120 minutes, wave height being 90 minutes, and wind speed being 180 minutes. Environmental uncertainties were obtained based on the variance of the random terms. The environmental uncertainty for water flow velocity was 0.04, for wave height it was 0.02, and for wind speed it was 0.15. The environmental evolution rate, environmental change patterns, and environmental uncertainties were then used to construct an environmental evolution characteristic matrix.

[0045] The dominant and secondary patterns of environmental factors are identified by analyzing the environmental evolution feature matrix. Cluster analysis is used to identify the change patterns of environmental factors. In the application scenario, three main environmental patterns are identified: a stable pattern, a fluctuating pattern, and a rapidly changing pattern. The stable pattern is characterized by an environmental evolution rate of less than 0.1 m / s / h and an environmental uncertainty of less than 0.05; the fluctuating pattern is characterized by an environmental evolution rate between 0.1 and 0.3 m / s / h and an environmental uncertainty between 0.05 and 0.1; and the rapidly changing pattern is characterized by an environmental evolution rate greater than 0.3 m / s / h and an environmental uncertainty greater than 0.1. Based on the current environmental evolution feature matrix, the water flow velocity is determined to be in a fluctuating pattern, the wave height in a stable pattern, and the wind speed in a rapidly changing pattern. The fluctuating pattern of water flow velocity is the dominant pattern, and the rapidly changing pattern of wind speed is the secondary pattern. The response characteristics of charging efficiency, operating costs, and safety risks under the dominant and secondary patterns are calculated. Under the water flow velocity fluctuation mode, the charging efficiency fluctuates within ±4%, operating cost within ±150 yuan, and safety risk within ±0.1. Under the wind speed fluctuation mode, the charging efficiency fluctuates within ±7%, operating cost within ±250 yuan, and safety risk within ±0.18. Based on these response characteristics, an environment-charging process response mapping relationship is constructed, and the environmental combined effect coefficient is calculated. The combined effect coefficient reflects the degree of influence of multiple environmental factors working together. The calculated environmental combined effect coefficient is 1.25, and a value greater than 1 indicates that there is an amplification effect between environmental factors.

[0046] The correction coefficient calculation integrates the effects of stability margin and environmental combined effect coefficient. The correction coefficient is obtained by multiplying the weighted average of the stability margin by the environmental combined effect coefficient. The weighted average of the stability margin is 0.63, multiplied by the environmental combined effect coefficient of 1.25, resulting in a correction coefficient of 0.79. The correction coefficient reflects the comprehensive impact of environmental changes on performance. Based on the basic assessment results, benchmark values ​​for environmental sensitivity indicators are constructed. The benchmark value for charging efficiency is 0.55, for operating costs it is 0.48, and for safety risk it is 0.65. These benchmark values ​​are multiplied by the correction coefficient 0.79 to obtain the environmental sensitivity indicators. The environmental sensitivity indicator for charging efficiency is 0.43, for operating costs it is 0.38, and for safety risk it is 0.51. These environmental sensitivity indicators are used for the dynamic adjustment of the weights in the subsequent multi-objective optimization function; indicators with higher sensitivity receive higher weights.

[0047] The environmental sensitivity index is calculated using a rolling update mechanism, recalculating every 15 minutes to ensure the optimization process responds promptly to environmental changes. When a sudden environmental shift is detected, such as a wind speed increase exceeding 5 m / s, an immediate update calculation of the environmental sensitivity index is triggered. Furthermore, the preset tolerance range and baseline reference value are adjusted according to different seasons and climatic conditions to adapt to long-term environmental changes. For example, the tolerance range is expanded during periods of high winds and waves in summer, and the baseline reference value is recalibrated during the winter icing period.

[0048] This invention establishes a precise calculation method for environmental sensitivity indicators through multi-dimensional analysis of environmental trend data. Specifically, it decomposes environmental trend data into trend, periodic, and random components, and extracts key features such as the rate of environmental evolution, the patterns of environmental change, and environmental uncertainty, constructing an environmental evolution feature matrix. Based on this, it identifies the dominant and secondary patterns of environmental factors, and then constructs an environment-charging process response mapping relationship, achieving a precise correlation between environmental changes and charging response.

[0049] In one optional implementation, a multi-level adaptive genetic algorithm is used to solve the multi-objective optimization function. The environmental complexity level is determined based on the environmental trend data, and a corresponding chromosome encoding method is selected. Parallel pre-evolutionary regression is performed under predicted environmental conditions at multiple time points, and chromosome segments are selected as robust genes based on fitness. Based on the robust genes, genetic operator parameters are adjusted to perform population evolution, obtaining the optimal individual that meets the convergence condition. The steps of decoding the optimal individual to obtain the charging berth allocation scheme and the charging power allocation scheme include: The environmental complexity level is determined based on the environmental trend data, and the chromosome encoding method is selected based on the environmental complexity level. The fitness value of individuals in the current population under the multi-objective optimization function is calculated, and representative individuals are selected according to the fitness value. The fitness value is obtained by weighted summation of service efficiency and energy efficiency. Parallel pre-evolution is performed under predicted environmental conditions at multiple time points, and the rate of change of the fitness value of the chromosome segments of the representative individuals under different environments is recorded. Chromosome segments with a fitness value change rate lower than the change rate threshold are extracted as robust genes. Based on the robust gene, calculate the dynamic adjustment coefficients of crossover probability and mutation probability, and perform crossover and mutation operations; select individuals containing the robust gene from the parent population and offspring population to form a new population; iterate the evolution of the new population until convergence to obtain the optimal individual; decode the optimal individual to obtain the charging berth allocation scheme and the charging power allocation scheme. When real-time environmental monitoring data deviates from the predicted trend, the dominant chromosome segments are preserved and the population is reconstructed, and the robust genes are updated based on the latest environmental data.

[0050] Combination Figure 2 The flowcharts illustrating the charging berth allocation and charging power allocation schemes derived using a genetic algorithm are presented below. For example, the environmental complexity level is comprehensively assessed by calculating the volatility and prediction uncertainty of key parameters such as water flow velocity, wave height, and wind speed in environmental trend data. The rate of change, fluctuation amplitude, and prediction error of each parameter in the environmental trend data are extracted and weighted to obtain the environmental complexity score. For instance, when the average rate of change of water flow velocity is 0.1 m / s / h, the fluctuation amplitude is ±0.2 m / s, and the prediction error is ±0.15 m / s, the complexity score for the water flow factor is 0.35. Similarly, the complexity scores for wave height and wind speed are calculated to be 0.28 and 0.42, respectively. The weighted summation of these scores yields a comprehensive environmental complexity score of 0.36. The environmental complexity level is divided into four levels: stable (0-0.25), slightly fluctuating (0.25-0.5), significantly fluctuating (0.5-0.75), and drastically changing (0.75-1). In this example, the environmental complexity level fluctuates slightly.

[0051] Different chromosome encoding methods are selected for different levels of environmental complexity. In stable environments, binary encoding is used, with each gene bit representing the allocation status of a charging berth or the discrete level of charging power. In environments with slight fluctuations, a hybrid method of real-number encoding and binary encoding is used, with binary encoding used for berth allocation and real-number encoding used for charging power. In environments with significant fluctuations, all-real-number encoding is used, with redundant encoding bits added to represent adjustment margins. In environments with drastic changes, an adaptive-length hierarchical encoding is used, consisting of a core decision-making layer and an environmental adaptation layer. For the hybrid encoding in environments with slight fluctuations, the chromosome structure consists of two parts: the first part is a binary string of length equal to the number of charging berths, representing the berth allocation status (1 for allocation, 0 for no allocation); the second part is a real-number string of equal length, representing the charging power allocation ratio of each berth. For example, a certain water-based charging station has 5 charging berths, with the chromosome code [1,0,1,1,0,0.3,0,0.4,0.5,0], indicating that berths 1, 3, and 4 are allocated with charging power allocation ratios of 30%, 40%, and 50%, respectively.

[0052] After population initialization, the fitness value of individuals in the current population under the multi-objective optimization function is calculated. The fitness value is obtained by weighted summation of service efficiency and energy efficiency. Service efficiency reflects the rationality of berth allocation, considering factors such as ship waiting time and charging completion rate; energy efficiency reflects the optimization degree of power allocation, considering factors such as energy utilization rate and peak-valley balance. For example, if an individual's berth allocation scheme results in an average waiting time of 15 minutes for 5 ships and a charging completion rate of 92%, the calculated service efficiency is 0.78; if the individual's power allocation scheme results in an energy utilization rate of 85% and a peak-valley difference of 15%, the calculated energy efficiency is 0.82. Under the condition of slight fluctuation in environmental complexity, the service efficiency weight is 0.6 and the energy efficiency weight is 0.4, and the individual's fitness value is 0.78 × 0.6 + 0.82 × 0.4 = 0.796. The population is sorted according to fitness value, and the top 20% of individuals are selected as representative individuals for subsequent parallel pre-evolutionary iterations.

[0053] Based on environmental trend data, predicted environmental conditions are generated for multiple future time points (e.g., 1 hour, 2 hours, 3 hours), and population evolution is performed simultaneously under these predicted environments. Each predicted environment corresponds to a parallel evaluation unit, representing an individual's fitness assessment within each unit. The rate of change of fitness values ​​for chromosome segments representing an individual under different environments is recorded. For example, the fitness value of gene segments 1-5 (representing berth allocation) of a representative individual is 0.8 in the current environment; 0.78 in the predicted environment after 1 hour (a change rate of 2.5%); 0.77 in the predicted environment after 2 hours (a change rate of 3.75%); and 0.75 in the predicted environment after 3 hours (a change rate of 6.25%). Similarly, the fitness change rates of gene segments 6-10 (representing power allocation) of the same individual under different predicted environments are calculated to be 8.5%, 12.3%, and 15.8%, respectively.

[0054] A threshold for the rate of change was determined based on the level of environmental complexity, with a threshold of 10% set for environments with slight fluctuations. Chromosomal segments with a rate of change in fitness values ​​below this threshold were extracted as robust genes. In the example above, the maximum rate of change for gene segments 1-5 was 6.25%, which was below the 10% threshold, and therefore extracted as robust genes; while the rate of change for gene segments 6-10 exceeded the threshold and was not considered robust genes. Through analysis of all representative individuals, a set of robust gene segments was finally extracted, which exhibited high stability under environmental changes.

[0055] The distribution characteristics and stability indices of robust genes were analyzed, and dynamic adjustment coefficients for crossover and mutation probabilities were calculated. For example, when robust genes account for 40% of the total chromosome length, the baseline value for the crossover probability was 0.8, the dynamic adjustment coefficient was 0.85, and the final crossover probability was set to 0.8 × 0.85 = 0.68; the baseline value for the mutation probability was 0.1, the dynamic adjustment coefficient was 1.2, and the final mutation probability was set to 0.1 × 1.2 = 0.12. Different crossover and mutation strategies were adopted for robust gene locations and non-robust gene locations: a conservative strategy was adopted for robust gene locations to reduce the probability of crossover breakpoints occurring within robust genes and to reduce the mutation probability at robust gene locations; an exploratory strategy was adopted for non-robust gene locations to increase mutation intensity to enhance search capabilities.

[0056] After performing crossover and mutation operations, individuals containing robust genes are selected from both the parent and offspring populations to form a new population. The selection process considers two factors: individual fitness value and the robust gene content rate. A base selection probability is set to be proportional to the fitness value, and then adjusted based on the proportion of robust genes contained in the individuals. For example, if an individual has a fitness value of 0.75 and a base selection probability of 0.06; and this individual contains 80% robust genes, with a selection probability adjustment factor of 1.4, the final selection probability is 0.06 × 1.4 = 0.084. In this way, individuals with both high fitness and a large number of robust genes are preferentially selected for the new population.

[0057] After each iteration, the convergence condition is checked: the optimal fitness of the population increases by less than 0.5% for 10 consecutive generations, or the maximum number of iterations (100 generations) is reached. During the iteration process, the crossover and mutation parameters are dynamically adjusted, gradually reducing the crossover and mutation probabilities to enhance the algorithm's local search capability. For example, the initial crossover probability is 0.68, decreasing by 5% every 10 generations, with a minimum of 0.5; the initial mutation probability is 0.12, decreasing by 8% every 10 generations, with a minimum of 0.05. When the convergence condition is met, the optimal individual is output, and it is decoded to obtain the charging berth allocation scheme and the charging power allocation scheme. Taking the aforementioned encoding method as an example, the optimal individual [1,0,1,1,0,0.3,0,0.45,0.48,0] is decoded as follows: berths 1, 3, and 4 are allocated to vessels waiting to be charged, with charging power allocation ratios of 30%, 45%, and 48%, respectively.

[0058] When real-time environmental monitoring data deviates from the predicted trend, a population reconstruction mechanism is initiated. The deviation between real-time and predicted environmental data is calculated. When the deviation exceeds a preset threshold (e.g., a water flow velocity deviation exceeding 0.3 m / s, or a wind speed deviation exceeding 3 m / s), population reconstruction is triggered. The reconstruction process retains chromosome segments containing robust genes in the current population, referred to as dominant chromosome segments. Reassessing the environmental complexity level requires switching the chromosome encoding method. For example, if the environment escalates from slight fluctuations to significant fluctuations, the encoding method is switched from mixed encoding to all-real-number encoding. A new population template is constructed, and dominant chromosome segments are mapped to their corresponding positions in the new template. Gene values ​​are generated for the remaining positions in the template through random initialization. Pre-evolutionary iteration is performed in the new environment to identify new robust genes and compare them with existing robust genes, calculating their similarity. Robust genes with similarity higher than a preset threshold (e.g., 0.7) are retained, and newly identified robust genes are merged to update the robust gene library. Based on the updated robust gene library and the latest environmental data, the genetic algorithm process is re-executed to obtain the optimal solution in the new environment.

[0059] The multi-level adaptive genetic algorithm employed in this invention significantly improves the algorithm's environmental adaptability and convergence efficiency through environment complexity-driven chromosome encoding selection and robust gene-based genetic operator parameter adjustment. In particular, the introduction of a parallel pre-evolutionary mechanism extracts environmentally stable chromosome segments as robust genes by analyzing the fitness change rate of representative individuals in the predicted environment. This provides a stable genetic basis for population evolution, ensuring continuous optimization of the charging resource allocation scheme in dynamic and complex aquatic environments.

[0060] In one optional implementation, the step of performing parallel pre-evolutionary regression under predicted environmental conditions at multiple time points, recording the rate of change of fitness values ​​of the representative individual's chromosome segments under different environments, and extracting chromosome segments with fitness value change rates lower than a threshold as robust genes includes: The environmental trend data is subjected to Fourier transform to obtain the intensity of environmental disturbance. An environmental element coupling matrix is ​​constructed based on the intensity of environmental disturbance. A multi-time-scale predicted environmental sequence is generated based on the environmental element coupling matrix. The chromosome is divided into functional genes and regulatory genes. The gene expression levels of the functional genes and the regulatory genes under the predicted environmental sequence are calculated. The gene environmental response is calculated based on the gene expression levels. The predicted environment sequence is assigned to a parallel evaluation unit, where population evolution is performed and the fitness values ​​of individuals in each predicted environment are recorded. The rate of change of fitness values ​​in adjacent predicted environments is calculated, and the mean of the rate of change is used as the environmental fitness index of the individual. The change rate threshold is determined based on the intensity of the environmental disturbance. Gene fragments are scored based on the gene environmental responsiveness and the environmental adaptability index. Gene fragments with scores lower than the change rate threshold are extracted to form a robust gene library. The complementarity characteristics of gene fragments in the robust gene library are analyzed, and the combination of gene fragments with the best complementarity is selected for population evolution.

[0061] For example, performing a Fourier transform on environmental trend data is an important means of obtaining the characteristics of environmental disturbances. Taking water flow velocity as an example, data from the past 4 hours was collected, with a sampling interval of 1 minute, totaling 240 sampling points, resulting in a data sequence of [1.2, 1.3, 1.25, 1.4...] m / s. A Fourier transform was performed on this time series, converting the time-domain data into a frequency-domain representation, yielding a spectrum. From the spectrum, the main frequency components and their amplitudes were extracted. For example, the amplitude of the component with a frequency of 0.008 Hz (period approximately 120 minutes) is 0.35 m / s, the amplitude of the component with a frequency of 0.017 Hz (period approximately 60 minutes) is 0.22 m / s, and the amplitude of the component with a frequency of 0.033 Hz (period approximately 30 minutes) is 0.15 m / s. The sum of the amplitudes of these main frequency components is defined as the environmental disturbance intensity; the environmental disturbance intensity for water flow velocity is 0.72 m / s. Similarly, the environmental disturbance intensity for wave height is calculated to be 0.45m, and the environmental disturbance intensity for wind speed is 3.6m / s.

[0062] The environmental element coupling matrix is ​​constructed by analyzing the mutual influence relationships between different environmental elements. Correlation coefficients are calculated for each environmental element to form a correlation matrix. For example, the correlation coefficient between water flow velocity and wave height is 0.65, the correlation coefficient between water flow velocity and wind speed is 0.48, and the correlation coefficient between wave height and wind speed is 0.72. Further analysis of the time lag relationships between environmental elements is conducted; for example, wind speed changes typically lead water flow velocity changes by 15 minutes and wave height changes by 10 minutes. Based on the correlation coefficients and time lag relationships, the environmental element coupling matrix is ​​constructed. The diagonal elements of this matrix represent the autocorrelation strength of each environmental element, while the off-diagonal elements represent the coupling strength between different elements. For example, in the coupling matrix composed of three elements (water flow velocity, wave height, and wind speed), the diagonal elements are 0.8, 0.75, and 0.85, representing the autocorrelation strength of each element; the off-diagonal elements, such as the coupling strength between water flow velocity and wave height, are 0.65, and other coupling strengths are similarly filled in.

[0063] A multi-timescale predicted environmental sequence is generated based on the environmental element coupling matrix. The predicted environmental sequence includes three time scales: short-term (1 hour later), medium-term (2 hours later), and long-term (4 hours later). First, based on the current environmental trend and Fourier transform results, baseline environmental parameter values ​​are predicted for each future time point. For example, if the current water flow velocity is 1.5 m / s, it is predicted to be 1.7 m / s after 1 hour, 1.6 m / s after 2 hours, and 1.4 m / s after 4 hours. Then, considering the coupling relationships between environmental elements, the predicted values ​​are corrected. For example, when the predicted wind speed increases, the predicted values ​​of water flow velocity and wave height are adjusted accordingly based on the time lag relationship of the coupling matrix. The final generated predicted environmental sequence contains a complete predicted dataset of multiple environmental elements across three time scales.

[0064] In the resource allocation problem of floating charging stations, chromosome coding represents the charging berth allocation scheme and the charging power allocation scheme. The gene loci in the chromosome that directly determine berth allocation and power allocation are defined as functional genes, and the gene loci that affect the expression of functional genes are defined as regulatory genes. For example, in a mixed coding scheme, the binary part representing the berth allocation status is the functional gene, and the real part representing the power allocation ratio is the regulatory gene. In the chromosome [1,0,1,1,0,0.3,0,0.4,0.5,0], the first 5 positions are functional genes, and the last 5 positions are regulatory genes.

[0065] Gene expression levels reflect the effectiveness of genes under specific environmental conditions. Chromosomes are decoded into specific resource allocation schemes, and the effectiveness of these schemes is evaluated under various predicted environments to calculate gene expression levels. The expression levels of functional genes are calculated using the service efficiency of the resource allocation scheme, while the expression levels of regulatory genes are calculated using energy efficiency. Taking the aforementioned chromosome as an example, in the current environment, the expression level of functional genes (service efficiency) is 0.82, and the expression level of regulatory genes (energy efficiency) is 0.78; in the predicted environment after 1 hour, the expression level of functional genes is 0.80, and the expression level of regulatory genes is 0.72; in the predicted environment after 2 hours, the expression level of functional genes is 0.79, and the expression level of regulatory genes is 0.68; in the predicted environment after 4 hours, the expression level of functional genes is 0.76, and the expression level of regulatory genes is 0.62.

[0066] The rate of change of expression levels of functional and regulatory genes relative to the current environment under each predicted environment is calculated, and the maximum rate of change is taken as the gene's environmental responsiveness. In the example above, the rate of change of expression levels of the functional gene under the three predicted environments are 2.4%, 3.7%, and 7.3%, respectively, with a maximum rate of change of 7.3%, therefore the environmental responsiveness of the functional gene is 7.3%. The rate of change of expression levels of the regulatory gene under the three predicted environments are 7.7%, 12.8%, and 20.5%, respectively, with a maximum rate of change of 20.5%, therefore the environmental responsiveness of the regulatory gene is 20.5%. The lower the environmental responsiveness, the more stable the gene is under environmental changes.

[0067] The generated multi-timescale predicted environment sequences are distributed to parallel evaluation units, with each unit corresponding to the environmental conditions at a predicted time point. For example, three evaluation units are configured to simulate environmental conditions after 1 hour, 2 hours, and 4 hours, respectively. The top 20% of individuals in terms of fitness in the population are selected as representative individuals, and these representative individuals are simultaneously sent to each evaluation unit for evaluation. Each evaluation unit executes a population evolution algorithm to simulate the fitness performance of the representative individuals under specific environments. For example, a representative individual might have a fitness value of 0.85 in the current environment, 0.82 after 1 hour, 0.80 after 2 hours, and 0.75 after 4 hours.

[0068] Calculate the rate of change of an individual's fitness value between adjacent predicted environments. For example, the rate of change from the current environment to 1 hour later is 3.5%, from 1 hour later to 2 hours later is 2.4%, and from 2 hours later to 4 hours later is 6.3%. Calculate the mean of these rates of change as an indicator of the individual's environmental fitness; in this example, the environmental fitness indicator is 4.1%. The lower the environmental fitness indicator, the stronger the individual's ability to maintain stability during environmental changes.

[0069] The change rate threshold is dynamically set based on the intensity of environmental disturbance. The greater the intensity of the environmental disturbance, the higher the change rate threshold should be. In practical applications, the final change rate threshold is determined by multiplying the baseline change rate threshold by an environmental disturbance intensity adjustment coefficient. The baseline change rate threshold is set to 8%, and the environmental disturbance intensity adjustment coefficient is proportional to the intensity of the environmental disturbance and is determined by looking up a table. For example, when the environmental disturbance intensity of water flow velocity is 0.72 m / s, the corresponding adjustment coefficient is 1.2; when the environmental disturbance intensity of wave height is 0.45 m, the corresponding adjustment coefficient is 1.1; and when the environmental disturbance intensity of wind speed is 3.6 m / s, the corresponding adjustment coefficient is 1.5. The weighted average of these adjustment coefficients is taken as the final adjustment coefficient, with the weights of water flow, waves, and wind speed being 0.4, 0.3, and 0.3, respectively, resulting in a final adjustment coefficient of 1.26. Multiplying the baseline change rate threshold of 8% by the adjustment coefficient of 1.26 yields a final change rate threshold of 10.08%.

[0070] Gene segments of chromosomes were scored based on their environmental responsiveness and environmental adaptability. The score was a weighted sum of the environmental responsiveness and environmental adaptability, with weights of 0.7 and 0.3, respectively. For example, the functional gene of the aforementioned chromosome had an environmental responsiveness of 7.3% and an environmental adaptability of 4.1%, resulting in a score of 7.3% × 0.7 + 4.1% × 0.3 = 6.34%; the regulatory gene had an environmental responsiveness of 20.5% and an environmental adaptability of 4.1%, resulting in a score of 20.5% × 0.7 + 4.1% × 0.3 = 15.52%. Gene segments with scores below a change rate threshold were extracted as robust genes. In this example, the functional gene score of 6.34% was below the change rate threshold of 10.08% and was extracted as a robust gene; while the regulatory gene score of 15.52% was above the change rate threshold and was not extracted. Through score analysis of all representative individuals, a set of robust gene segments was extracted, forming a robust gene library.

[0071] This study analyzes the functional and environmental adaptation characteristics of gene fragments in a robust gene pool to identify complementary relationships. Complementarity relationships include both functional complementarity and environmental adaptation complementarity. Functional complementarity refers to different gene fragments controlling different aspects of resource allocation, such as some fragments optimizing berth allocation while others optimize power allocation. Environmental adaptation complementarity refers to different gene fragments exhibiting high stability under different environmental conditions, such as some fragments being stable under changes in water flow while others are stable under changes in wind speed. The complementarity degree between gene fragments is calculated, and the gene fragment combination with optimal complementarity is selected for population evolution. For example, five gene fragments are selected from the robust gene pool to form a gene combination with optimal complementarity, where two fragments optimize berth allocation and three fragments optimize power allocation; simultaneously, these gene fragments exhibit high stability under changes in water flow, waves, and wind speed, respectively, forming environmental adaptation complementarity.

[0072] Environmental trend data is updated every 30 minutes, and Fourier transform analysis of environmental disturbance characteristics is re-performed. When a significant change in environmental disturbance intensity is detected (e.g., an increase or decrease exceeding 30%), the robust gene re-extraction process is triggered. Simultaneously, a historical robust gene library is maintained, recording robust genes extracted under different environmental conditions over the past 24 hours. When the similarity between the new environment and the historical environment exceeds 80%, historical robust genes are prioritized to improve algorithm response speed. Furthermore, based on actual operational performance, the baseline change rate threshold and environmental disturbance intensity adjustment coefficient are periodically evaluated and adjusted to ensure the accuracy and effectiveness of robust gene extraction.

[0073] This invention accurately extracts the intensity of environmental disturbances from environmental trend data using Fourier transform and constructs an environmental element coupling matrix, enabling the scientific generation of predicted environmental sequences across multiple time scales. By combining the functional and regulatory gene classification of chromosomes, gene environmental responsiveness is calculated, establishing a correspondence between gene expression and environmental changes. Through performing population pre-evolution in parallel evaluation units, the fitness change rate is recorded and analyzed in real time, identifying gene fragments with high environmental stability, thus providing an environmentally adaptive optimization basis for resource allocation decisions at floating charging stations.

[0074] In one alternative implementation, when real-time environmental monitoring data deviates from the predicted trend, the steps of preserving dominant chromosome segments and reconstructing the population, and updating the robust genes based on the latest environmental data, include: The deviation between real-time environmental monitoring data and environmental trend data is calculated. When the deviation exceeds a preset threshold, it is determined that the environment has changed. The environmental complexity level is recalculated based on the real-time environmental monitoring data. Chromosomal segments containing the robust gene in the current population are identified and marked as dominant chromosomal segments. Based on the environmental complexity level, a chromosome encoding method is selected to construct a population template. The dominant chromosome segment is mapped to the corresponding position of the population template. Gene values ​​are generated for the remaining positions of the population template through random initialization to form a reconstructed population. Based on the real-time environmental monitoring data, a predicted environmental sequence of a preset time length is constructed. The reconstructed population is subjected to pre-evolution in the predicted environmental sequence of the preset time length to identify robust genes under the new environment. The similarity of the robust genes before and after the environmental change is calculated. Robust genes with similarity higher than a preset similarity threshold are retained, and newly identified robust genes are merged to update the robust gene library. Genetic operations are performed on the reconstructed population, and the selection probability of individuals containing the updated robust genes is set to a preset selection probability. When the improvement rate of the optimal fitness of the population is lower than the convergence threshold, the current optimal individual is taken as the optimal solution in the new environment.

[0075] For example, environmental parameters around the floating charging station are collected in real time via a sensor network, including key indicators such as water flow velocity, wave height, and wind speed. Real-time environmental monitoring data is collected and processed every 30 minutes. The real-time environmental data is compared with previously predicted environmental trend data to calculate the deviation. The deviation is calculated using a relative deviation method, i.e., the difference between the actual and predicted values ​​divided by the predicted value. For example, if the predicted water flow velocity is 1.6 m / s and the actual monitored value is 1.9 m / s, the relative deviation is 18.75%; the predicted wave height is 0.5 m and the actual monitored value is 0.65 m, the relative deviation is 30%; and the predicted wind speed is 7.5 m / s and the actual monitored value is 10 m / s, the relative deviation is 33.3%. Deviation thresholds are set for different environmental parameters: 15% for water flow velocity, 25% for wave height, and 20% for wind speed. When the deviation of any environmental parameter exceeds the corresponding threshold, a significant environmental change is determined, triggering the population reconstruction process. In the example above, the deviations of all three environmental parameters exceeded their respective thresholds, indicating that the environmental changes were significant.

[0076] The recalculation of environmental complexity level is based on the latest real-time environmental data. Features such as the rate of change, fluctuation amplitude, and prediction error are extracted from the real-time environmental data to calculate a new environmental complexity score. Taking real-time monitored water flow data as an example, the average rate of change of water flow velocity over the past hour was 0.18 m / s / h, with a fluctuation amplitude of ±0.3 m / s and a short-term prediction error of ±0.25 m / s, resulting in a complexity score of 0.52 for the water flow factor. Similarly, the complexity scores for wave height and wind speed are calculated to be 0.45 and 0.65, respectively. These scores are then weighted and summed according to weights of 0.4, 0.3, and 0.3, yielding a comprehensive environmental complexity score of 0.54. According to the preset environmental complexity level classification standard, this score corresponds to a significantly fluctuating environmental complexity level (0.5-0.75). Compared to the previous slightly fluctuating environmental complexity level (0.25-0.5), the environmental complexity level has been upgraded.

[0077] Scan the chromosomes of all individuals in the current population to identify chromosome segments containing robust genes. Taking the aforementioned chromosome encoding method as an example, an individual's chromosome is [1,0,1,1,0,0.35,0,0.42,0.48,0], where positions 1-5 have been identified as robust genes. Extract the gene values ​​at these positions and label them as dominant chromosome segments [1,0,1,1,0]. Perform the same operation on all individuals in the current population, collecting all dominant chromosome segments and sorting them according to their frequency of occurrence in the population. Dominant chromosome segments with high frequency are considered to have higher environmental adaptability and are preferentially retained for population reconstruction. In a population of 100 individuals, the dominant chromosome segment [1,0,1,1,0] has a frequency of 32%, ranking first; the segment [1,1,0,1,0] has a frequency of 25%, ranking second; and the segment [0,1,1,1,0] has a frequency of 18%, ranking third. The top 50% of dominant chromosome segments are selected for subsequent population reconstruction.

[0078] The appropriate chromosome encoding method is selected based on the newly calculated environmental complexity level. Since the environmental complexity level has escalated from slight fluctuation to significant fluctuation, the chromosome encoding method is switched from mixed encoding (binary + real numbers) to all-real-number encoding. Under the all-real-number encoding method, berth allocation status is determined by a real-number threshold: berths are allocated if the threshold is greater than 0.5, and not allocated if it is less than or equal to 0.5. A new population template is constructed with a length of 10, using only real-number encoding. The previously identified dominant chromosome segment [1,0,1,1,0] is mapped to the corresponding position in the new template, converting it to real values ​​[0.9,0.2,0.8,0.9,0.1]. Gene values ​​are generated for the remaining positions (positions 6-10) of the template through random initialization, such as [0.4,0.3,0.6,0.7,0.2]. After mapping and initialization, the new chromosome is [0.9,0.2,0.8,0.9,0.1,0.4,0.3,0.6,0.7,0.2]. Perform the same mapping operation on all retained dominant chromosome segments to generate 100 new chromosomes, forming a reconstructed population.

[0079] The construction of the predicted environmental sequence is based on the latest real-time environmental monitoring data. A predicted environmental sequence for a predetermined time period (e.g., 3 hours) is generated based on this data. The prediction process considers the temporal correlation of environmental parameters and the mutual influence between parameters. For example, if the current real-time monitored water flow velocity is 1.9 m / s, wave height is 0.65 m, and wind speed is 10 m / s, the predicted values ​​after 1 hour are: water flow velocity 2.1 m / s, wave height 0.75 m, and wind speed 9.5 m / s; after 2 hours, water flow velocity 2.0 m / s, wave height 0.8 m, and wind speed 8.5 m / s; and after 3 hours, water flow velocity 1.8 m / s, wave height 0.7 m, and wind speed 7.8 m / s. These predicted values ​​constitute a multi-time-point predicted environmental sequence.

[0080] The reconstructed population was subjected to pre-evolutionary regression in the generated predicted environment sequence, using the same method as the previously described parallel pre-evolutionary regression. The rate of change of fitness values ​​of chromosome segments in individuals within the reconstructed population under different predicted environments was calculated, and chromosome segments with a rate of change below a threshold were identified as robust genes in the new environment. Under the new environmental complexity, the rate of change threshold was calculated to be 12.5%. Through pre-evolutionary regression analysis, new robust gene segments were identified in the reconstructed population, such as [0.85,0.25,0.9,0.95,0.15] and [0.3,0.25,0.7,0.8,0.2].

[0081] The similarity of robust genes before and after environmental changes is calculated, and genes with similarity higher than a preset threshold are retained. The similarity calculation uses Euclidean distance normalization of gene values. For example, the robust gene [1,0,1,1,0] (binary encoding) before the environmental change is converted to real number encoding [1.0,0.0,1.0,1.0,0.0]. The Euclidean distance between this gene and the robust gene [0.85,0.25,0.9,0.95,0.15] identified under the new environment is calculated, and after normalization, the similarity is 0.82. A similarity threshold of 0.7 is set, therefore this robust gene is retained. Another newly identified robust gene [0.3,0.25,0.7,0.8,0.2] has a similarity of 0.63 with the original robust gene, which is below the threshold. It is not directly retained, but is merged into the robust gene database as a newly identified robust gene. The updated robust gene pool contains both retained robust genes and newly identified robust genes, providing a stable genetic basis for subsequent population evolution.

[0082] Genetic operations such as crossover and mutation are performed on the reconstructed population to generate a new generation. In the selection operation, the selection probability of individuals containing the updated robust genes is increased. The base selection probability is set to be proportional to the individual's fitness, and then adjusted according to the proportion of robust genes contained in the individual. For example, if an individual has a fitness value of 0.78 and a base selection probability of 0.05, and this individual contains 90% robust genes, its selection probability is increased to 0.05 × 1.8 = 0.09. In this way, individuals containing robust genes are preferentially selected for the next generation, accelerating the convergence of the population towards high-quality solutions.

[0083] The convergence condition is set as follows: the improvement rate of the optimal fitness of the population is less than 0.8% for five consecutive generations, or the maximum number of iterations (50 generations) is reached. During the iteration process, the change in the optimal fitness of the population is monitored. For example, the optimal fitness of the first generation is 0.82, the second generation is 0.835 (1.83%), the third generation is 0.845 (1.2%), the fourth generation is 0.85 (0.59%), the fifth generation is 0.852 (0.24%), and the sixth generation is 0.854 (0.23%). Since the improvement rate of two consecutive generations is less than the convergence threshold of 0.8%, the convergence condition is met. The current optimal individual [0.95, 0.1, 0.9, 0.85, 0.0, 0.35, 0.0, 0.6, 0.75, 0.0] is decoded into a charging berth allocation scheme and a charging power allocation scheme. The decoded solution allocates berths 1, 3, and 4 with charging power distribution ratios of 35%, 60%, and 75%, respectively.

[0084] To address the continuously changing environment, a sliding window monitoring mechanism was implemented. A sliding window containing environmental data from the past two hours is maintained; when new environmental data enters the window, the oldest data point is removed. Based on the sliding window data, environmental change trends and fluctuation characteristics are periodically calculated to predict future environmental changes. When a significant environmental change is predicted within the next hour, genetic operator parameters are adjusted in advance to increase the mutation probability and improve population diversity, preparing for the upcoming environmental change. A periodic evaluation mechanism for robust genes was also implemented, reassessing the adaptability of each gene fragment in the robust gene pool to the current environment every two hours, removing gene fragments with significantly decreased adaptability to maintain the effectiveness of the robust gene pool.

[0085] This invention identifies environmental changes promptly by calculating the deviation between real-time environmental monitoring data and predicted trends, and reassesses the environmental complexity level based on the latest environmental data. In particular, it proposes a mechanism for identifying and preserving dominant chromosome segments, combined with chromosome coding selection adapted to environmental complexity, to achieve population template construction and mapping, ensuring the adaptability of the genetic algorithm to environmental changes. By identifying robust genes in new environments through pre-evolutionary simulation and calculating gene similarity to achieve the organic integration of old and new robust genes, it ensures the continuous optimization and stable operation of the water-based charging station resource allocation scheme in dynamic environments.

[0086] A second aspect of the present invention provides a dynamic resource allocation system for floating charging stations based on multi-objective optimization, comprising: The first unit is used to acquire hydrological and meteorological parameters of the water charging station, use deep learning algorithms to generate environmental trend data, and output the prediction results of the impact of the environment on charging efficiency. The second unit is used to construct a multi-objective optimization function that includes charging efficiency, operating cost and safety risk assessment based on environmental prediction. The safety risk assessment calculates the impact of the environment on the ship's charging status based on environmental trend data. The constraints of the multi-objective optimization function are dynamically generated according to the environmental trend data. The weights of each item in the multi-objective optimization function are determined through an adaptive weighting mechanism driven by environmental prediction. The third unit is used to solve the multi-objective optimization function using a multi-level adaptive genetic algorithm, determine the environmental complexity level based on the environmental trend data and select the corresponding chromosome encoding method; perform parallel pre-evolution under the predicted environmental conditions at multiple time points, select chromosome segments as robust genes based on fitness; perform population evolution based on the robust genes by adjusting the genetic operator parameters to obtain the optimal individual that meets the convergence condition; and decode the optimal individual to obtain the charging berth allocation scheme and the charging power allocation scheme.

[0087] A third aspect of the present invention provides an electronic device, comprising: processor; Memory used to store processor-executable instructions; The processor is configured to invoke instructions stored in the memory to execute the aforementioned method.

[0088] A fourth aspect of the present invention provides a computer-readable storage medium having stored thereon computer program instructions that, when executed by a processor, implement the aforementioned method.

[0089] This invention can be a method, apparatus, system, and / or computer program product. The computer program product may include a computer-readable storage medium having computer-readable program instructions loaded thereon for performing various aspects of the invention.

[0090] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A dynamic resource allocation method for floating charging stations based on multi-objective optimization, characterized in that, include: The hydrological and meteorological parameters of the floating charging station are obtained, environmental trend data are generated using deep learning algorithms, and the prediction results of the impact of the environment on charging efficiency are output. A multi-objective optimization function is constructed, which includes charging efficiency, operating cost and safety risk assessment based on environmental prediction. The safety risk assessment calculates the impact of the environment on the ship's charging status based on environmental trend data. The constraints of the multi-objective optimization function are dynamically generated according to the environmental trend data. The weights of each item in the multi-objective optimization function are determined through an adaptive weighting mechanism driven by environmental prediction. A multi-level adaptive genetic algorithm is used to solve the multi-objective optimization function. The environmental complexity level is determined based on the environmental trend data, and the corresponding chromosome encoding method is selected. Parallel pre-evolution is performed under the predicted environmental conditions at multiple time points, and chromosome segments are selected as robust genes based on fitness. Based on the robust gene, the genetic operator parameters are adjusted to perform population evolution, and the optimal individual that meets the convergence condition is obtained; the optimal individual is decoded to obtain the charging berth allocation scheme and the charging power allocation scheme.

2. The method according to claim 1, characterized in that, The steps for acquiring hydrological and meteorological parameters of the floating charging station, generating environmental trend data using deep learning algorithms, and outputting predictions of the environmental impact on charging efficiency include: A smoothed hydrological parameter sequence is obtained by using a sliding window to reduce noise in the hydrological parameters, and a comprehensive meteorological index is calculated based on the meteorological parameters. The smoothed hydrological parameter sequence, the comprehensive meteorological index, and the corresponding parameter change rate are used to construct a time-series feature matrix. The time-series feature matrix is ​​then input into a pre-trained long short-term memory network to generate environmental trend data. Based on environmental trend prediction data, the vertical disturbance force, horizontal disturbance force, and rotational torque acting on the charging interface are obtained; the contact pressure value and displacement deviation value of the charging interface under the action of the vertical disturbance force, horizontal disturbance force, and rotational torque are detected; the reference resistance value of the charging interface under standard operating conditions is obtained; the contact area ratio is calculated based on the contact pressure value; the displacement coefficient is calculated based on the displacement deviation value; the real-time contact resistance of the charging interface is obtained by multiplying the reference resistance value, the contact area ratio, and the displacement coefficient; the charging efficiency is calculated based on the change in the real-time contact resistance and the change in temperature.

3. The method according to claim 1, characterized in that, The steps for constructing a multi-objective optimization function that includes charging efficiency, operating costs, and safety risk assessment based on environmental predictions, wherein the safety risk assessment calculates the impact of the environment on the ship's charging status based on environmental trend data, the constraints of the multi-objective optimization function are dynamically generated based on environmental trend data, and the weights of each item in the multi-objective optimization function are determined through an adaptive weighting mechanism driven by environmental predictions include: The operating costs include charging time costs and energy consumption costs; The ship's displacement is calculated based on the environmental trend data, and the position risk value is calculated based on the difference between the ship's displacement and the preset safe distance threshold. The vertical disturbance force, horizontal disturbance force, and rotational torque acting on the charging interface are calculated based on the environmental trend data, and the mechanical risk value is calculated. The electrical risk value is calculated based on the real-time contact resistance of the charging interface. The weighted sum of the position risk value, the mechanical risk value, and the electrical risk value is used as the safety risk assessment result. An environmental sensitivity index is obtained by calculating the impact of the environmental trend data on charging efficiency, operating costs, and safety risk assessment results. Adaptive weight coefficients of each term in the multi-objective optimization function are then calculated based on the environmental sensitivity index. The multi-objective optimization function is obtained by multiplying the adaptive weight coefficients with the corresponding terms. Based on the environmental trend data, mechanical safety constraint thresholds and electrical safety constraint thresholds are set as dynamic constraints for the multi-objective optimization function.

4. The method according to claim 3, characterized in that, The steps for calculating the impact of the environmental trend data on charging efficiency, operating costs, and safety risk assessment results to obtain the environmental sensitivity index include: The environmental trend data is preprocessed to obtain a benchmark reference value. Based on the benchmark reference value, the basic assessment results of charging efficiency, operating cost and safety risk are calculated. The fluctuation frequency and fluctuation amplitude of environmental factors in the environmental trend data are extracted to construct an environmental disturbance vector. Based on the environmental disturbance vector, the steady-state offset of charging efficiency, operating cost and safety risk is calculated. The steady-state offset is compared with the preset fault tolerance range to obtain the stability margin. The environmental trend data is decomposed into a time series to obtain a trend term, a periodic term, and a random term; the rate of change of the trend term is used to obtain the environmental evolution speed, the dominant period of the periodic term is used to obtain the environmental change pattern, and the variance of the random term is used to obtain the environmental uncertainty. The environmental evolution speed, the environmental change pattern, and the environmental uncertainty are then used to construct an environmental evolution feature matrix. Based on the environmental evolution feature matrix, the dominant and secondary modes of environmental factors are identified, and the response characteristics of charging efficiency, operating costs and safety risks under the dominant and secondary modes are calculated. Based on the response characteristics, an environment-charging process response mapping relationship is constructed and the environmental combined effect coefficient is calculated. The correction coefficient is calculated based on the stability margin and the environmental combined effect coefficient; a benchmark value for the environmental sensitivity index is constructed based on the basic assessment results, and the environmental sensitivity index is obtained by multiplying the benchmark value by the correction coefficient.

5. The method according to claim 1, characterized in that, A multi-level adaptive genetic algorithm is used to solve the multi-objective optimization function. The environmental complexity level is determined based on the environmental trend data, and the corresponding chromosome encoding method is selected. Parallel pre-evolution is performed under the predicted environmental conditions at multiple time points, and chromosome segments are selected as robust genes based on fitness. Based on the robust genes, the genetic operator parameters are adjusted to perform population evolution, and the optimal individual that meets the convergence condition is obtained; The steps for decoding the optimal individual to obtain the charging berth allocation scheme and the charging power allocation scheme include: The environmental complexity level is determined based on the environmental trend data, and the chromosome encoding method is selected based on the environmental complexity level. The fitness value of individuals in the current population under the multi-objective optimization function is calculated, and representative individuals are selected according to the fitness value. The fitness value is obtained by weighted summation of service efficiency and energy efficiency. Parallel pre-evolution is performed under predicted environmental conditions at multiple time points, and the rate of change of the fitness value of the chromosome segments of the representative individuals under different environments is recorded. Chromosome segments with a fitness value change rate lower than the change rate threshold are extracted as robust genes. Based on the robust gene, calculate the dynamic adjustment coefficients of crossover probability and mutation probability, and perform crossover and mutation operations; select individuals containing the robust gene from the parent population and offspring population to form a new population; iterate the evolution of the new population until convergence to obtain the optimal individual; decode the optimal individual to obtain the charging berth allocation scheme and the charging power allocation scheme. When real-time environmental monitoring data deviates from the predicted trend, the dominant chromosome segments are preserved and the population is reconstructed, and the robust genes are updated based on the latest environmental data.

6. The method according to claim 5, characterized in that, The steps of performing parallel pre-evolutionary simulations under predicted environmental conditions at multiple time points, recording the rate of change of fitness values ​​of chromosome segments representing individuals under different environments, and extracting chromosome segments with fitness value change rates below a threshold as robust genes include: The environmental trend data is subjected to Fourier transform to obtain the intensity of environmental disturbance. An environmental element coupling matrix is ​​constructed based on the intensity of environmental disturbance. A multi-time-scale predicted environmental sequence is generated based on the environmental element coupling matrix. The chromosome is divided into functional genes and regulatory genes. The gene expression levels of the functional genes and the regulatory genes under the predicted environmental sequence are calculated. The gene environmental response is calculated based on the gene expression levels. The predicted environment sequence is assigned to a parallel evaluation unit, where population evolution is performed and the fitness values ​​of individuals in each predicted environment are recorded. The rate of change of fitness values ​​in adjacent predicted environments is calculated, and the mean of the rate of change is used as the environmental fitness index of the individual. The change rate threshold is determined based on the intensity of the environmental disturbance. Gene fragments are scored based on the gene environmental responsiveness and the environmental adaptability index. Gene fragments with scores lower than the change rate threshold are extracted to form a robust gene library. The complementarity characteristics of gene fragments in the robust gene library are analyzed, and the combination of gene fragments with the best complementarity is selected for population evolution.

7. The method according to claim 5, characterized in that, When real-time environmental monitoring data deviates from the predicted trend, the steps of preserving dominant chromosome segments and reconstructing the population, and updating the robust genes based on the latest environmental data include: The deviation between real-time environmental monitoring data and environmental trend data is calculated. When the deviation exceeds a preset threshold, it is determined that the environment has changed. The environmental complexity level is recalculated based on the real-time environmental monitoring data. Chromosomal segments containing the robust gene in the current population are identified and marked as dominant chromosomal segments. Based on the environmental complexity level, a chromosome encoding method is selected to construct a population template. The dominant chromosome segment is mapped to the corresponding position of the population template. Gene values ​​are generated for the remaining positions of the population template through random initialization to form a reconstructed population. Based on the real-time environmental monitoring data, a predicted environmental sequence of a preset time length is constructed. The reconstructed population is subjected to pre-evolution in the predicted environmental sequence of the preset time length to identify robust genes under the new environment. The similarity of the robust genes before and after the environmental change is calculated. Robust genes with similarity higher than a preset similarity threshold are retained, and newly identified robust genes are merged to update the robust gene library. Genetic operations are performed on the reconstructed population, and the selection probability of individuals containing the updated robust genes is set to a preset selection probability. When the improvement rate of the optimal fitness of the population is lower than the convergence threshold, the current optimal individual is taken as the optimal solution in the new environment.

8. A dynamic resource allocation system for floating charging stations based on multi-objective optimization, used to implement the method of any one of claims 1-7, characterized in that, include: The first unit is used to acquire hydrological and meteorological parameters of the water charging station, use deep learning algorithms to generate environmental trend data, and output the prediction results of the impact of the environment on charging efficiency. The second unit is used to construct a multi-objective optimization function that includes charging efficiency, operating cost and safety risk assessment based on environmental prediction. The safety risk assessment calculates the impact of the environment on the ship's charging status based on environmental trend data. The constraints of the multi-objective optimization function are dynamically generated according to the environmental trend data. The weights of each item in the multi-objective optimization function are determined through an adaptive weighting mechanism driven by environmental prediction. The third unit is used to solve the multi-objective optimization function using a multi-level adaptive genetic algorithm, determine the environmental complexity level based on the environmental trend data and select the corresponding chromosome encoding method; perform parallel pre-evolution under the predicted environmental conditions at multiple time points, and select chromosome segments as robust genes based on fitness; Based on the robust gene, the genetic operator parameters are adjusted to perform population evolution, and the optimal individual that meets the convergence condition is obtained; the optimal individual is decoded to obtain the charging berth allocation scheme and the charging power allocation scheme.

9. An electronic device, characterized in that, include: processor; Memory used to store processor-executable instructions; The processor is configured to invoke instructions stored in the memory to execute the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having computer program instructions stored thereon, characterized in that, When the computer program instructions are executed by the processor, they implement the method described in any one of claims 1 to 7.

Images