A hydrogen fuel cell ship electric propulsion inertia adaptive optimization control method

Abstract

The application discloses a kind of hydrogen fuel cell ship electric propulsion inertia adaptive optimization control methods, it is related to hydrogen fuel cell ship electric propulsion control technical field, real-time acquisition ship operating state data and ship internal power network fluctuation data, pre-processing and extracting features, calculate virtual inertia coefficient and virtual damping coefficient, generate virtual torque instruction;Using torque speed calculation method, obtain the total power demand of ship and be divided into high frequency, mid-frequency and low frequency ship power distribution instruction;Through virtual synchronous machine model, output ship actual power, real-time monitoring DC bus voltage and hydrogen fuel cell power, and using dynamic compensation power fluctuation method, the operating state of ship is regulated and controlled, and voltage stability index is obtained;Through power regulation algorithm, the inertia of ship propulsion is adaptively optimized, so that the ship can better adapt to different working conditions, and the stability and safety of the ship are enhanced.

Description

Technical Field

[0001] This invention relates to the field of hydrogen fuel cell ship electric propulsion control technology, and in particular to an adaptive optimization control method for the inertial of hydrogen fuel cell ship electric propulsion. Background Technology

[0002] With the increasing demands for ship power performance, energy efficiency, and navigation safety, ships initially relied mainly on traditional diesel engines for direct propulsion. This method was simple in structure but had low energy efficiency and caused pollution. With the development of power electronics and motor control technologies, electric propulsion began to be applied to ships. Hydrogen fuel cell ships convert the electrical energy of the ship's hydrogen fuel cells into the mechanical energy required for ship propulsion, resulting in higher energy efficiency, better maneuverability, and lower emissions. When rapidly changing power demands based on different navigation conditions of hydrogen fuel cell ships, it is impossible to accurately allocate power and dynamically respond. Under high-frequency fluctuations, the control accuracy and response speed are insufficient, making it difficult to maintain stability. Furthermore, traditional methods cannot calculate virtual inertia and damping coefficients in a timely and accurate manner, affecting propulsion dynamic performance and power voltage stability. Summary of the Invention

[0003] In view of the aforementioned existing problems, the present invention is proposed.

[0004] Therefore, this invention provides an adaptive optimization control method for the inertial propulsion of hydrogen fuel cell ships to solve the problem of inaccurate power distribution and dynamic response under complex operating conditions of hydrogen fuel cell ships.

[0005] To solve the above-mentioned technical problems, the present invention provides the following technical solution: In a first aspect, the present invention provides an adaptive optimization control method for the inertial propulsion of hydrogen fuel cell ships, comprising: real-time acquisition of ship operating status data and internal power network fluctuation data, preprocessing and feature extraction, calculating virtual inertia coefficient and virtual damping coefficient, and generating virtual torque commands; using a torque-speed calculation method to obtain the ship's total power demand and classify it into high-frequency, medium-frequency and low-frequency ship power allocation commands; outputting the ship's actual power through a virtual synchronous machine model, monitoring the DC bus voltage and hydrogen fuel cell power in real time, and using a dynamic compensation power fluctuation method to regulate the ship's operating status and obtain a voltage stability index; based on the ship's actual power and voltage stability index, calculating the current optimal virtual inertia coefficient through a reinforcement learning model, and using a power adjustment algorithm to adaptively optimize the ship's propulsion inertia.

[0006] As a preferred embodiment of the adaptive optimization control method for inertial propulsion of hydrogen fuel cell ships described in this invention, the steps of real-time acquisition of ship operating status data and internal power network fluctuation data, preprocessing, and feature extraction are as follows. Real-time data collection of ship acceleration, angular velocity changes, propeller thrust changes, wind speed and wave height data yields ship operating status data; real-time data collection of ship DC bus voltage and hydrogen fuel cell output power yields ship internal power network fluctuation data. Missing values ​​are filled using linear interpolation, and outliers are removed using a threshold detection method. The Z-score method is used to standardize the data to a unified range, and features are extracted using frequency domain analysis to obtain the characteristics of ship operating status and the fluctuation characteristics of the ship's internal power network.

[0007] As a preferred embodiment of the adaptive optimization control method for inertial propulsion of hydrogen fuel cell ships described in this invention, the specific steps for calculating the virtual inertia coefficient and virtual damping coefficient to generate virtual torque commands are as follows: By analyzing data on changes in ship propeller thrust and fluctuations in the ship's internal power network using frequency domain analysis, the dynamic characteristics of electromagnetic torque are obtained. Based on the characteristics of ship operation and the fluctuation characteristics of the ship's internal power network, virtual inertia coefficient and virtual damping coefficient are calculated. Combining the changes in ship angular velocity and the dynamic characteristics of electromagnetic torque, virtual torque commands are generated through virtual torque calculation methods.

[0008] As a preferred embodiment of the hydrogen fuel cell ship electric propulsion inertial adaptive optimization control method described in this invention, the specific steps for obtaining the ship's total power demand using a torque-speed calculation method and dividing it into high-frequency, medium-frequency, and low-frequency ship power allocation commands are as follows. Based on virtual torque commands and ship angular velocity changes, combined with ship operating status characteristics and internal power network fluctuation characteristics, the total power demand of the ship is obtained using torque-speed calculation methods. By using frequency domain analysis, the total power demand of the ship is decomposed in the frequency domain, and the ship power allocation instructions for high frequency, medium frequency and low frequency power are divided.

[0009] As a preferred embodiment of the hydrogen fuel cell ship electric propulsion inertial adaptive optimization control method of the present invention, the step of outputting the actual power of the ship through a virtual synchronous machine model is as follows: Based on historical ship power allocation commands, the dynamic response characteristics of ship power are extracted by decomposition using fast Fourier transform, statistical analysis is performed using a sliding window to extract statistical features, and the parameters of the virtual synchro model are initialized. Based on the dynamic response characteristics and statistical features of ship power, a gradient descent optimization training method is used to train the virtual synchronizing machine model and output the actual power of the ship.

[0010] As a preferred embodiment of the adaptive optimization control method for inertial propulsion of hydrogen fuel cell ships described in this invention, the steps of real-time monitoring of DC bus voltage and hydrogen fuel cell power, and using a dynamic compensation method for power fluctuations to regulate the ship's operating state and obtain voltage stability indicators are as follows. Based on the actual power of the ship, the changes in DC bus voltage, hydrogen fuel cell power and ship power demand are monitored in real time to extract power fluctuation characteristics. By using a dynamic power fluctuation compensation method, the impact of power fluctuation characteristics on ship operation is compensated in real time, the ship's operating status is adjusted, and voltage stability indicators are obtained.

[0011] As a preferred embodiment of the adaptive optimization control method for inertial propulsion of hydrogen fuel cell ships described in this invention, the specific steps for calculating the current optimal virtual inertial coefficient based on the ship's actual power and voltage stability indicators using a reinforcement learning model are as follows: Based on the actual power of historical ships, the state space and action space are defined, and the reward function is defined using historical voltage stability index to construct a reinforcement learning model framework. The reinforcement learning model framework was trained using historical ship power and historical voltage stability indices to obtain the reinforcement learning model; The ship's actual power and voltage stability indices are input into the reinforcement learning model to calculate the current optimal virtual inertia coefficient.

[0012] As a preferred embodiment of the adaptive optimization control method for the inertial propulsion of hydrogen fuel cell ships described in this invention, the specific steps for adaptively optimizing the ship's propulsion inertia using a power regulation algorithm are as follows. Based on the optimal virtual inertia coefficient, and combined with the ship's high-frequency, medium-frequency, and low-frequency power allocation commands, the power regulation algorithm is used to regulate the ship's operating status in real time, thereby optimizing the ship's inertial response characteristics and dynamic response characteristics.

[0013] In a second aspect, the present invention provides a computer device including a memory and a processor, wherein the memory stores a computer program, wherein when the computer program is executed by the processor, it implements any step of the hydrogen fuel cell ship electric propulsion inertial adaptive optimization control method as described in the first aspect of the present invention.

[0014] Thirdly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the hydrogen fuel cell ship electric propulsion inertial adaptive optimization control method as described in the first aspect of the present invention.

[0015] The beneficial effects of this invention are as follows: By calculating the total power demand of a ship using a torque-speed calculation method, the total power demand is divided into high-frequency, mid-frequency, and low-frequency power allocation commands. After frequency domain decomposition, the power output of the ship at different time scales can be controlled. This helps the ship to rationally allocate power according to demand under various navigation conditions, improving energy utilization efficiency and making the ship's operation more stable and efficient. By calculating the current optimal virtual inertia coefficient through a reinforcement learning model, the ship's control strategy can be dynamically adjusted according to the actual operating conditions, exhibiting strong adaptive capabilities. Adaptive optimization of the ship's propulsion inertia using a power regulation algorithm allows the ship to better adapt to different operating conditions, enhancing the ship's stability and safety. Attached Figure Description

[0016] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 A flowchart for an adaptive optimization control method for the inertial propulsion of hydrogen fuel cell ships.

[0018] Figure 2 This is a flowchart illustrating the operation of the virtual synchronous machine model.

[0019] Figure 3 This is a flowchart for power distribution and regulation.

[0020] Figure 4 A flowchart for training and applying reinforcement learning models. Detailed Implementation

[0021] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0022] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0023] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0024] Reference Figures 1-4 This embodiment provides an adaptive optimization control method for the inertial propulsion of hydrogen fuel cell ships, including the following steps: S1. Collect ship operation status data and internal power network fluctuation data in real time, preprocess and extract features.

[0025] S1.1. Real-time acquisition of ship acceleration, angular velocity changes, propeller thrust changes, wind speed and wave height data to obtain ship operating status data; real-time acquisition of ship DC bus voltage and hydrogen fuel cell output power to obtain ship internal power network fluctuation data.

[0026] It should be noted that the acceleration changes of the ship in three-dimensional space are acquired at a fixed frequency using accelerometers, and the acceleration values ​​at different times are recorded; the angular velocity change data of the ship is collected, and the rotational angular velocity of the ship around each axis is measured using gyroscope sensors to obtain the dynamic change value of angular velocity; the propeller thrust change data is collected, and the magnitude of the thrust generated by the propeller is detected by thrust sensors, and the thrust change value is continuously recorded; wind speed data is collected, and the wind speed around the ship is measured using an anemometer installed on the ship's superstructure to obtain the wind speed value at different time points; wave height data is collected, and the wave height is detected using a wave height sensor on the bottom of the ship, and the dynamic change value of wave height is recorded. The collected acceleration, angular velocity change, propeller thrust change, wind speed, and wave height values ​​are arranged in a unified timestamp order to obtain the ship's operating status data; The DC bus voltage data is collected in real time by a voltage sensor, and the voltage fluctuation is recorded. The power sensor deployed at the output end of the hydrogen fuel cell collects the output power data of the hydrogen fuel cell in real time, and the power output change value of the hydrogen fuel cell is continuously recorded to obtain the internal power network fluctuation data of the ship. The collected ship DC bus voltage and hydrogen fuel cell output power are arranged in a unified timestamp order to obtain the internal power network fluctuation data of the ship.

[0027] S1.2. Linear interpolation is used to fill in missing values, and threshold detection is used to remove outliers.

[0028] It should be noted that the expression for imputing missing values ​​using linear interpolation is as follows: ; in, This is the compensated data. It is the value of the known data point preceding the missing value. It is the value of the next known data point after the missing value. It is the time of the previous known data point. It is the time of the next known data point. These are the time points that need interpolation; It should be noted that, for both ship operational status data and internal power network fluctuation data, each variable in both data types is iterated and checked individually. When a missing value is found, the preceding and following normal data points are determined. The difference between the values ​​of the preceding and following normal data points is calculated, and then the ratio of this difference to the time interval between those two points is calculated to obtain the change per unit time. The time interval between the missing value and the preceding normal data point is multiplied by this change per unit time to obtain the increment of the missing value relative to the preceding normal data point. This increment is then added to the value of the preceding normal data point to obtain the value used to fill in the missing value, which is then inserted into the location of the missing value. After completing the missing value imputation, anomaly detection thresholds are set based on the range of ship operating status data and internal power network fluctuation data. Each piece of ship operating status data and internal power network fluctuation data is then evaluated. When a data point of a variable exceeds the anomaly detection threshold range (such as voltage fluctuation ±10%, speed deviation ±5%, or other operating condition-related safety limits), it is marked as an anomaly. Data marked as an anomaly is then removed from the corresponding variable's dataset, ensuring that each dataset retains only ship operating status data and internal power network fluctuation data within the normal range. This completes the imputation of missing values ​​and the removal of anomalies in the ship operating status data and internal power network fluctuation data.

[0029] S1.3. The Z-score method is used to standardize the data to a unified range, and features are extracted using frequency domain analysis to obtain the characteristics of ship operation status and the fluctuation characteristics of the ship's internal power network.

[0030] It should be noted that Z-score standardization is applied to both the ship's operational status data and the internal power network fluctuation data. The sum of all values ​​in both categories is calculated, divided by the total number of data points to obtain the mean. The sum of the squares of the differences between each data point and the mean is then calculated, divided by the total number of data points to obtain the standard deviation. Finally, the standardization process is completed by subtracting the mean from each data point and dividing by the standard deviation. The standardized ship operating status data and internal power network fluctuation data are arranged sequentially into a data sequence, equally divided into even-numbered segments. A butterfly operation is used on the ship operating status data and internal power network fluctuation data. First, the data is divided into odd and even groups, and multiplication, addition, and subtraction are performed on each group to obtain intermediate results. This grouping operation is repeated on the intermediate results until the frequency domain ship operating status data and internal power network fluctuation data are obtained. Then, the spectrum is analyzed to determine the main frequency components and amplitudes. Features related to ship operating status and internal power network fluctuations are extracted, including propeller and structural vibration frequencies, acceleration, and angular velocity amplitudes in ship operating status, and power frequency, harmonic frequency components, voltage, and power amplitudes in internal power network fluctuations. These yield the ship operating status characteristics and internal power network fluctuation characteristics.

[0031] S2. Calculate the virtual inertia coefficient and virtual damping coefficient, and generate virtual torque command.

[0032] S2.1. By analyzing the changes in ship propeller thrust and fluctuations in the ship's internal power network using frequency domain analysis, the dynamic characteristics of electromagnetic torque are obtained.

[0033] It should be noted that Fast Fourier Transform (FFT) was performed on the ship propeller thrust variation data and the ship's internal power network fluctuation data respectively, converting the time-domain ship propeller thrust variation data and the ship's internal power network fluctuation data into frequency-domain representations, thus obtaining the spectra of the ship propeller thrust variation data and the ship's internal power network fluctuation data. Characteristic frequencies related to electromagnetic torque were identified in the spectra. The amplitude and phase information corresponding to the characteristic frequencies related to electromagnetic torque can reflect the characteristics of electromagnetic torque at different frequencies. The amplitude and phase data of the characteristic frequencies in the propeller thrust spectrum and the power grid fluctuation spectrum were processed and analyzed. By cross-comparing the frequency domain peak values, phase differences, and harmonic distributions of the two, the resonant frequency, harmonic coupling strength, and dynamic response delay of the electromagnetic torque were identified. Combined with typical ship operating conditions (such as the correlation between thrust mutation during rapid acceleration and power grid voltage flicker), the frequency-varying characteristics of the electromagnetic torque were determined. S2.2 Based on the characteristics of ship operation status and the fluctuation characteristics of the ship's internal power network, calculate the virtual inertia coefficient and virtual damping coefficient. Combine the changes in ship angular velocity and the dynamic characteristics of electromagnetic torque, generate virtual torque commands through virtual torque calculation methods.

[0034] It should be noted that a large amount of data on ship angular velocity changes, electromagnetic torque dynamic characteristics, and corresponding actual operating effects under different states were collected. The least squares fitting algorithm was used to find a functional expression that can describe the relationship between ship angular velocity changes, electromagnetic torque dynamic characteristics, and corresponding actual operating effects. Thus, the initial virtual inertia coefficient and virtual damping coefficient reference values ​​were calculated based on the ship angular velocity changes and electromagnetic torque dynamic characteristics under the current state mode. Initial virtual inertia and virtual damping coefficient reference values ​​are obtained. Combined with the fluctuation characteristics of the ship's internal power network, the difference in power network fluctuations between adjacent time intervals is calculated to obtain a power change sequence. A fast Fourier transform is performed on the power change sequence to convert the time-domain data into frequency-domain data. In the frequency-domain data, frequency components with cumulative contributions exceeding 80% of the total energy or significant local peaks (e.g., amplitudes exceeding adjacent frequencies by more than 3 dB) are identified. Combined with typical resonant frequency bands of the ship's power system (e.g., 50Hz fundamental wave, 100Hz second harmonic, etc.), the main frequency components of energy concentration are determined, thereby determining the fluctuation frequency of power output in the ship's power network. The standard deviation of power change over a certain time period is calculated as the rate of change of net power fluctuation in the ship's power network. Weighted calculations are performed, assigning different weights based on the importance of the fluctuation frequency of power output and the rate of change of net power fluctuation in the ship's power network. The virtual inertia and virtual damping coefficients of the current actual situation are obtained through weighted summation. It should be noted that the expression for calculating the virtual inertia coefficient is: ; in, It is the virtual inertia coefficient. It is the fluctuation frequency of power output in the ship's power network. It is the frequency weighting coefficient. It refers to the power fluctuation amplitude in the ship's power network. It is the amplitude weighting coefficient; It should be noted that the expression for calculating the virtual damping coefficient is as follows: ; in, It is the virtual damping coefficient. Virtual damping weighting coefficient It is the rate of change of net power fluctuations in the ship's power network; Substitute the adjusted virtual inertia coefficient, virtual damping coefficient, and relevant data on ship angular velocity change and electromagnetic torque dynamic characteristics into the virtual torque calculation method formula. Determine the rate of change of electromagnetic torque based on the electromagnetic torque dynamic characteristics. Multiply the virtual inertia coefficient by the differential value of angular velocity change, and multiply the virtual damping coefficient by the rate of change of electromagnetic torque. Then add these two products to calculate the differential value of ship angular velocity change. Perform the calculations sequentially according to the operation order specified in the virtual torque calculation method formula to finally generate the virtual torque command.

[0035] S3. Using the torque-speed calculation method, the total power demand of the ship is obtained and divided into high-frequency, medium-frequency and low-frequency ship power allocation instructions.

[0036] S3.1 Based on the virtual torque command and the ship's angular velocity change, combined with the ship's operating status characteristics and the fluctuation characteristics of the ship's internal power network, the total power demand of the ship is obtained by using the torque-speed calculation method.

[0037] It should be noted that the virtual torque command and the ship's angular velocity change data are arranged in chronological order. Considering the ship's operational characteristics, such as load and propulsion requirements under different states like stable navigation, acceleration, deceleration, and turning, as well as the fluctuation characteristics of the ship's internal power network, such as voltage stability and power fluctuation range, the virtual torque command and ship angular velocity change data are weighted. Under stable navigation and power network conditions, the virtual torque command is given a larger weight, while the ship's angular velocity change is given a smaller weight. This weighted processing yields a preliminary torque-speed correlation value. This correlation value is then corrected according to the ship's operational characteristics and the fluctuation characteristics of its internal power network. For example, in acceleration mode, the torque-speed correlation value is appropriately adjusted based on the acceleration magnitude and the power network's carrying capacity. The corrected torque-speed correlation value is then substituted into the torque-speed calculation method. First, the power components related to torque and speed are calculated separately. Then, these two power components are added together to obtain the ship's total power requirement.

[0038] S3.2. Using frequency domain analysis, the total power demand of the ship is decomposed in the frequency domain to divide the ship power allocation instructions into high-frequency, medium-frequency and low-frequency power.

[0039] It should be noted that a Fast Fourier Transform (FFT) is performed on the ship's total power demand data to transform the total power demand from the time domain to the frequency domain, obtaining the spectrum of the total power demand. On the obtained spectrum, high-frequency, mid-frequency, and low-frequency boundaries are defined based on the frequency range. According to these boundaries, frequency components belonging to the high-frequency range and their corresponding amplitude information are extracted from the spectrum. These high-frequency components and their corresponding amplitude information constitute the high-frequency power component, which is used to generate the high-frequency power ship power allocation command. Similarly, frequency components belonging to the mid-frequency range and their corresponding amplitude information are extracted from the spectrum to form the mid-frequency power component, which is then used to generate the mid-frequency power ship power allocation command. Finally, frequency components falling into the low-frequency range and their corresponding amplitude information are extracted from the spectrum to form the low-frequency power component, which is used to generate the low-frequency power ship power allocation command.

[0040] S4. Output the actual power of the ship through the virtual synchro model.

[0041] S4.1 Based on historical ship power allocation instructions, the system is decomposed using Fast Fourier Transform to extract dynamic response characteristics of ship power. Statistical analysis is then performed using a sliding window to extract statistical features and initialize the parameters of the virtual synchronizer model.

[0042] It should be noted that a Fast Fourier Transform (FFT) is performed on historical ship power allocation commands to transform them from the time domain to the frequency domain, obtaining the corresponding spectrum. Within this spectrum, the boundaries of high-frequency, mid-frequency, and low-frequency ranges are determined based on the frequency range. Frequency components and corresponding amplitude information of the high-frequency, mid-frequency, and low-frequency components are extracted according to these boundaries. These frequency components and amplitude information constitute the ship's dynamic power response characteristics. A sliding window is set for the extracted dynamic power response characteristics. Starting from the initial position of the dynamic power response characteristic data, the window moves by a fixed step size each time. Statistical analysis is performed on the dynamic power response characteristic data within the window, calculating statistical quantities such as mean, variance, and standard deviation to extract statistical features. Based on the extracted statistical features, the initial parameters of the virtual synchronizer, including moment of inertia, damping coefficient, and synchronization reactance, are quantitatively derived through the mapping relationship between statistical features (such as high-frequency variance reflecting the intensity of power mutation and low-frequency mean representing steady-state deviation) and the stability requirements of the ship's power system (such as voltage fluctuation rate ≤5% and frequency deviation ±0.2Hz). This completes the initialization of the virtual synchronizer model parameters.

[0043] S4.2 Based on the dynamic response characteristics and statistical characteristics of ship power, the virtual synchronous machine model is trained using the gradient descent optimization training method, and the actual ship power is output.

[0044] It should be noted that the dynamic response characteristics and statistical characteristics of ship power are provided as input data to the virtual synchronizing machine model. Initial weight parameters and bias parameters are set for the virtual synchronizing machine model, which then processes the input dynamic response characteristics and statistical characteristics to obtain the output result, which is the ship power predicted by the virtual synchronizing machine model. The error between the currently predicted ship power and the actual ship power is calculated using the mean square error function. It should be noted that the mean square error function expression for training the virtual synchronizer model is: ; in, It is the mean square error function. It is the first The virtual synchronization machine model for the first Predicted ship power values ​​for each sample. It is the first Actual ship power measurements for each sample It is the number of training samples. It is an index variable representing the number of training samples; Based on the obtained mean squared error function, the gradient descent optimization training method is used to update the weight parameters and bias parameters of the virtual synchrotron model. Specifically, the weight parameters and bias parameters are adjusted according to the gradient direction of the loss function with respect to the weight parameters and bias parameters using the learning rate. After adjusting the parameters, the dynamic response characteristics and statistical characteristics of the ship's power are input into the updated virtual synchrotron model again. The previous steps of calculating the output results, calculating the error, and updating the parameters are repeated. This process is iterated continuously until the error has passed the required number of iterations and the virtual synchrotron model outputs the actual power of the ship.

[0045] S5. Real-time monitoring of DC bus voltage and hydrogen fuel cell power, and using dynamic compensation for power fluctuations to regulate the ship's operating status and obtain voltage stability indicators.

[0046] S5.1 Based on the actual power of the ship, monitor the changes in DC bus voltage, hydrogen fuel cell power and ship power demand in real time, and extract power fluctuation characteristics.

[0047] It is important to note the correlation between the amplitude and frequency characteristics of DC bus voltage fluctuations and the trends, severity, and ranges of fluctuations in hydrogen fuel cell power and ship power demand. This involves comparing the fluctuations in hydrogen fuel cell power and ship power demand when DC bus voltage fluctuations are large over different time periods, observing the responses of DC bus voltage and ship power demand when hydrogen fuel cell power rises or falls rapidly, and examining whether there are corresponding patterns in DC bus voltage and hydrogen fuel cell power when ship power demand exhibits periodic changes. Furthermore, a comparison is made of the start time and duration of fluctuations in DC bus voltage, hydrogen fuel cell power, and ship power demand to identify their synchronicity or sequential relationship over time. By integrating these correlations, responses, patterns, and time series relationships, the characteristics of power fluctuations can be summarized.

[0048] S5.2 Utilize the dynamic compensation power fluctuation method to compensate for the impact of power fluctuation characteristics on ship operation in real time, adjust the ship's operating status, and obtain voltage stability indicators.

[0049] It should be noted that, based on the identified power fluctuation characteristics, the compensation parameters required for the dynamic power fluctuation compensation method are determined. For the DC bus voltage fluctuation characteristics, the compensation power required to stabilize the DC bus voltage is calculated based on the fluctuation amplitude and frequency. For the hydrogen fuel cell power fluctuation characteristics, the corresponding compensation power adjustment is determined according to the power change trend and severity. Based on the fluctuation range and cycle of the ship's power demand, a reference value for compensation power to meet the ship's stable operation requirements is calculated. The calculated compensation power values ​​for DC bus voltage, hydrogen fuel cell power, and ship power demand are integrated to form a comprehensive dynamic compensation power scheme. According to the comprehensive compensation power scheme, the operating status of the compensation device is adjusted in real time to compensate for power fluctuations. During the compensation process, the changes in DC bus voltage, hydrogen fuel cell power, and ship power demand are continuously monitored to observe whether they are moving towards stability. The parameters of the compensation device are adjusted in real time to ensure that the compensation effect achieves the expected results. By monitoring the stability of the DC bus voltage, the voltage fluctuation range is calculated to determine whether it has narrowed, and the number of voltage fluctuations within a specified time period is counted to determine whether it has decreased. Based on the monitoring results, voltage stability indicators are obtained.

[0050] S6. Based on the ship's actual power and voltage stability indicators, calculate the current optimal virtual inertia coefficient through a reinforcement learning model.

[0051] S6.1 Based on the actual power of historical ships, define the state space and action space, use the historical voltage stability index to define the reward function, and construct a reinforcement learning model framework.

[0052] It should be noted that historical ship power data was collected, and the maximum, minimum, and distribution density of this data were statistically analyzed. After removing outliers, the 5% to 95th percentile of the power data was used as the value range. Based on the power value range, different power state intervals were defined, with each interval representing a state, thus defining a state space. Operational action data corresponding to the historical ship power was collected, covering various operations such as adjusting ship equipment power output and changing navigation strategies. Based on the characteristics and possible values ​​of the operational actions, they were classified, with each category representing an action, thus defining an action space. Through time-series alignment and correlation analysis (such as Pearson coefficient and Granger causality test), the dynamic impact of changes in ship power and operational actions on voltage stability indicators (such as voltage deviation rate and fluctuation amplitude) was quantified, obtaining the power-action-voltage stability mapping relationship. Based on the mapping relationship between power, action, and voltage stability, a reward rule is set. When an action is taken, if the voltage stability index improves to the expected range, a higher reward is given; if the voltage stability index deteriorates, a lower reward or even a negative reward is given. Based on the defined state space, action space, and set reward rule, a framework for a reinforcement learning model is built using the near-end policy optimization algorithm.

[0053] S6.2. The reinforcement learning model framework is trained using historical ship actual power and historical voltage stability indices to obtain the reinforcement learning model.

[0054] It should be noted that the process involves collecting historical ship power data and historical voltage stability index data, classifying the state space of the historical ship power data to map the historical ship power data and historical voltage stability index data to corresponding states, and associating the historical voltage stability index data with corresponding operational actions and states. Samples are randomly selected from these samples at a certain proportion and divided into multiple batches. For each batch of samples, the state space data is input into the policy network of the reinforcement learning model framework. The policy network selects the probability distribution for each action based on the input state, and samples the action to obtain operational commands (such as adjusting propulsion power) according to this probability distribution. The sampled operational commands (such as adjusting propulsion power) and the corresponding states are input into a neural network structure that has been initialized. This network utilizes the historical ship power and historical voltage stability index states (such as ship power ranges), the executed actions (such as propeller power adjustment commands), and the immediate reward (feedback value calculated based on the voltage stability index). By minimizing the temporal difference error (TD error) or Monte Carlo reward bias, the weight parameters of the value network are iteratively optimized. In the resulting value network, the value function estimate for this action in this state is calculated. Based on the reward function, and considering the current state, the action taken, and historical voltage stability data for the next state, a reward value is calculated. The state, action, reward value, and next state are then input into the policy network and value network to calculate the policy loss function and value loss function. Based on the calculated loss function values, an optimization algorithm is used to update and adjust the parameters of the policy network and value network. This process of sampling, calculating output, calculating loss, and updating parameters is repeated until the set number of training rounds is reached or other stopping conditions are met, resulting in a reinforcement learning model.

[0055] S6.3 Input the ship's actual power and voltage stability indicators into the reinforcement learning model to calculate the current optimal virtual inertia coefficient.

[0056] It should be noted that the expression for calculating the current optimal virtual inertia coefficient is: ; in, It is the optimal virtual inertia coefficient. It is the set of all possible actions. It is an index of actions. It is an instantaneous value function estimate. It is the first reinforcement learning model to adopt One action, It refers to the ship's operational status. It is the cost function of performing the action. It is the change in the instantaneous value function estimate. These are the weighting coefficients of the immediate value function. It is the weighting coefficient of the change in value. These are the weighting coefficients of the cost function; It should be noted that the actual power data of the ship is input into the reinforcement learning model. Based on the input actual power data and the state space, the reinforcement learning model determines the state corresponding to the current actual power of the ship. From the probability distribution of each action selected in the current state output by the reinforcement learning model, actions that may affect the adjustment of the virtual inertia coefficient are initially screened according to the correspondence between actions and virtual inertia coefficients. The ship's voltage stability index data, the information related to the initially screened actions, and the current state information are input into the value network. The value network calculates the value function estimate corresponding to the actions in the current state and evaluates the degree of influence of each action on the voltage stability index. Based on the value function estimate results of each action output by the value network, the action with the highest value function estimate is selected, and the virtual inertia coefficient corresponding to this action is the current optimal virtual inertia coefficient.

[0057] S7. Adaptive optimization of ship propulsion inertia is performed using a power regulation algorithm.

[0058] S7.1 Based on the optimal virtual inertia coefficient, combined with the ship's high-frequency, medium-frequency and low-frequency power allocation commands, the power regulation algorithm is used to regulate the ship's operating status in real time, and optimize the ship's inertial response characteristics and dynamic response characteristics.

[0059] It should be noted that the optimal virtual inertia coefficient, along with the ship's high-frequency, mid-frequency, and low-frequency power allocation commands, is input into the power regulation algorithm. Based on the input optimal virtual inertia coefficient and the ship's high-frequency, mid-frequency, and low-frequency power allocation commands, the power regulation algorithm analyzes the power adjustment requirements corresponding to the high-frequency power allocation commands. It then uses the optimal virtual inertia coefficient to initially regulate the ship's high-frequency operating state, adjusting relevant parameters to optimize high-frequency response characteristics. For mid-frequency power allocation commands, the power regulation algorithm, according to its power adjustment requirements and in conjunction with the optimal virtual inertia coefficient, regulates the ship's mid-frequency operating state, changing various parameters and improving mid-frequency response characteristics. For low-frequency power allocation commands, the power regulation algorithm, according to the power adjustment information in the command and in conjunction with the optimal virtual inertia coefficient, regulates the ship's low-frequency operating state, optimizing parameter settings and improving low-frequency response characteristics. During the regulation process, the changes in the ship's inertial and dynamic response characteristics are continuously monitored. Based on the monitoring results, the regulation parameters of the power regulation algorithm are fine-tuned to ensure that the ship's inertial and dynamic response characteristics are continuously optimized.

[0060] This embodiment also provides a computer device applicable to the adaptive optimization control method for electric propulsion in hydrogen fuel cell ships, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to realize the adaptive optimization control method for electric propulsion in hydrogen fuel cell ships as proposed in the above embodiment.

[0061] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0062] This embodiment also provides a storage medium storing a computer program. When executed by a processor, the program implements the adaptive optimization control method for inertial electric propulsion of hydrogen fuel cell ships as proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0063] In summary, this invention calculates the ship's total power demand using a torque-speed calculation method, dividing the total power demand into high-frequency, mid-frequency, and low-frequency power allocation commands. After frequency domain decomposition, it can control the ship's power output at different time scales. This helps the ship rationally allocate power according to demand under various navigation conditions, improving energy utilization efficiency and making the ship's operation more stable and efficient. By calculating the current optimal virtual inertia coefficient through a reinforcement learning model, the ship's control strategy can be dynamically adjusted according to actual operating conditions, exhibiting strong adaptive capabilities. Adaptive optimization of the ship's propulsion inertia using a power regulation algorithm allows the ship to better adapt to different operating conditions, enhancing the ship's stability and safety.

[0064] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method of inertial adaptive optimal control for hydrogen fuel cell ship electric propulsion, characterized by: include, Real-time data collection of ship operating status and internal power network fluctuations is performed, preprocessed and features are extracted, virtual inertia coefficient and virtual damping coefficient are calculated, and virtual torque commands are generated. Using torque-speed calculation methods, the total power demand of the ship is obtained and divided into high-frequency, medium-frequency, and low-frequency power allocation instructions. The specific steps are as follows. Based on virtual torque commands and ship angular velocity changes, combined with ship operating state characteristics and internal power network fluctuation characteristics, the total power demand of the ship is obtained using torque-speed calculation methods. By using frequency domain analysis, the total power demand of the ship is decomposed in the frequency domain, and the ship power allocation instructions for high frequency, medium frequency and low frequency power are divided. By using a virtual synchronous machine model, the actual power of the ship is output, the DC bus voltage and the power of the hydrogen fuel cell are monitored in real time, and the ship's operating status is regulated by a dynamic compensation power fluctuation method to obtain voltage stability indicators. Based on the ship's actual power and voltage stability indicators, the optimal virtual inertia coefficient is calculated using a reinforcement learning model. Then, an adaptive optimization of the ship's propulsion inertia is performed using a power regulation algorithm. The specific steps are as follows: Based on the historical actual power of ships, a state space and action space are defined, and a reward function is defined using historical voltage stability indices to construct a reinforcement learning model framework; specifically, it includes the following: Collect historical ship actual power data, and statistically analyze the maximum, minimum and distribution density of historical ship actual power data. After removing outliers, use the 5% to 95th percentile of the power data as the value range. Based on the power value range, divide different power state intervals, and define each interval as a state to define the state space. Collect operational data corresponding to the actual power of historical ships. The operational actions cover various operations such as adjusting the power output of ship equipment and changing navigation strategies. Based on the characteristics and possible values ​​of the operational actions, the operational actions are classified, and each category of operational actions is defined as an action, thus defining the action space. By using time-series alignment and correlation analysis, the dynamic impact of actual power changes and operational actions on voltage stability indicators is quantified, and the mapping relationship between power-action-voltage stability is obtained. Based on the mapping relationship between power, action, and voltage stability, a reward rule is set. When an action is taken, if the voltage stability index improves to the expected range, a higher reward is given; if the voltage stability index deteriorates, a lower reward or even a negative reward is given. Based on the well-defined state space, action space, and set reward rule, a framework for a reinforcement learning model is built using the near-end policy optimization algorithm. The reinforcement learning model framework was trained using historical ship power and historical voltage stability indices to obtain the reinforcement learning model; The ship's actual power and voltage stability indices are input into the reinforcement learning model to calculate the current optimal virtual inertia coefficient; the expression for the current optimal virtual inertial system is: ； in, It is the optimal virtual inertia coefficient. It is the set of all possible actions. It is an index of actions. It is an instantaneous value function estimate. It is the first reinforcement learning model to adopt One action, It refers to the ship's operational status. It is the cost function of performing the action. It is the change in the instantaneous value function estimate. These are the weighting coefficients of the immediate value function. It is the weighting coefficient of the change in value. These are the weighting coefficients of the cost function; By inputting the ship's actual power data into a reinforcement learning model, the model determines the current state corresponding to the ship's actual power based on the input data and the state space. From the probability distribution of each action selected in the current state output by the reinforcement learning model, actions that may affect the adjustment of the virtual inertia coefficient are initially screened based on the correspondence between actions and virtual inertia coefficients. The ship's voltage stability index data, the information related to the initially screened actions, and the current state information are input into a value network. The value network calculates the value function estimate corresponding to each action in the current state and evaluates the degree of influence of each action on the voltage stability index. Based on the value function estimate results of each action output by the value network, the action with the highest value function estimate is selected, and the virtual inertia coefficient corresponding to this action is the current optimal virtual inertia coefficient. Based on the optimal virtual inertia coefficient, and combined with the ship's high-frequency, medium-frequency, and low-frequency power allocation commands, the power regulation algorithm is used to regulate the ship's operating status in real time, thereby optimizing the ship's inertial response characteristics and dynamic response characteristics.

2. The adaptive optimization control method for inertial propulsion of hydrogen fuel cell ships as described in claim 1, characterized in that: The real-time acquisition of ship operating status data and internal power network fluctuation data, followed by preprocessing and feature extraction, involves the following steps: Real-time data collection of ship acceleration, angular velocity changes, propeller thrust changes, wind speed and wave height data yields ship operating status data; real-time data collection of ship DC bus voltage and hydrogen fuel cell output power yields ship internal power network fluctuation data. Missing values ​​are filled using linear interpolation, and outliers are removed using a threshold detection method. The Z-score method is used to standardize the data to a unified range, and features are extracted using frequency domain analysis to obtain the characteristics of ship operating status and the fluctuation characteristics of the ship's internal power network.

3. The adaptive optimization control method for inertial propulsion of hydrogen fuel cell ships as described in claim 2, characterized in that: The specific steps for calculating the virtual inertia coefficient and virtual damping coefficient to generate the virtual torque command are as follows: By analyzing data on changes in ship propeller thrust and fluctuations in the ship's internal power network using frequency domain analysis, the dynamic characteristics of electromagnetic torque are obtained. Based on the characteristics of ship operation and the fluctuation characteristics of the ship's internal power network, virtual inertia coefficient and virtual damping coefficient are calculated. Combining the changes in ship angular velocity and the dynamic characteristics of electromagnetic torque, virtual torque commands are generated through virtual torque calculation methods.

4. The adaptive optimization control method for inertial propulsion of hydrogen fuel cell ships as described in claim 3, characterized in that: The process of outputting the ship's actual power using a virtual synchronizer model involves the following steps. Based on historical ship power allocation commands, the dynamic response characteristics of ship power are extracted by decomposition using fast Fourier transform, statistical analysis is performed using a sliding window to extract statistical features, and the parameters of the virtual synchro model are initialized. Based on the dynamic response characteristics and statistical features of ship power, a gradient descent optimization training method is used to train the virtual synchronizing machine model and output the actual power of the ship.

5. The adaptive optimization control method for inertial propulsion of hydrogen fuel cell ships as described in claim 4, characterized in that: The real-time monitoring of DC bus voltage and hydrogen fuel cell power, and the use of dynamic power fluctuation compensation methods to regulate the ship's operating status, yields voltage stability indicators. The specific steps are as follows: Based on the actual power of the ship, the changes in DC bus voltage, hydrogen fuel cell power and ship power demand are monitored in real time to extract power fluctuation characteristics. By using a dynamic power fluctuation compensation method, the impact of power fluctuation characteristics on ship operation is compensated in real time, the ship's operating status is adjusted, and voltage stability indicators are obtained.

6. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the hydrogen fuel cell ship electric propulsion inertial adaptive optimization control method according to any one of claims 1 to 5.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the adaptive optimization control method for inertial electric propulsion of hydrogen fuel cell ships as described in any one of claims 1 to 5.

Images