A marine fan system with data feedback self-adaptive power regulation and a method thereof

Abstract

The present application relates to the field of ship energy saving and emission reduction, and discloses a marine fan system with data feedback self-adaptive power adjustment and a method thereof, comprising: a multi-source data acquisition module, which is used for synchronously acquiring the ship's speed against water, the draft of the bow, the draft of the stern, the nozzle back pressure, the air blower power consumption and the main engine output power in each control cycle, and calculating the average draft; a variable working condition feedforward calculation module, which is used for calculating the theoretical jet flow according to the current effective ship type coefficient, the speed index and the draft index, combining the speed and the average draft, and calculating the feedforward power target value after the back pressure compensation function correction and the combination of the pipeline total pressure loss and the fan characteristic curve. The present application realizes the real-time verification and optimization of the net energy-saving effect of the air blower power adjustment by constructing the net power benefit signal as the closed-loop feedback quantity, and adapts to the long-term drift of the system characteristics through the online self-correction mechanism.

Description

Technical Field

[0001] This invention relates to the field of marine energy conservation and emission reduction technology, specifically to a marine fan system and method with adaptive power adjustment based on data feedback. Background Technology

[0002] Marine air-lubrication drag reduction systems reduce frictional resistance by injecting compressed air into the hull bottom to create an air-water mixture layer. This transforms the solid-liquid friction interface on the hull surface into a solid-air-liquid interface, utilizing the density and viscosity differences between air and water. For large, low-speed vessels, frictional resistance can account for up to 70% of total drag, making air-lubrication drag reduction technology significant for reducing fuel consumption and carbon emissions. The power control of the blower in this system directly determines the balance between drag reduction and the system's own energy consumption. If the blower power is too low, the air layer coverage is insufficient, reducing drag reduction; if the power is too high, the blower's own energy consumption exceeds the propulsion power savings, resulting in a net energy loss. Determining and adjusting the blower's optimal power in real time is a core challenge in this field.

[0003] In existing technologies, the power control schemes for marine air lubrication system blowers mainly fall into three categories. The first is fixed-gear control, where several power levels are pre-defined based on the ship's speed, and operators select the corresponding level according to the speed, or the control system automatically switches between levels. The second is main engine power signal mapping control, which collects the main engine power as a feedforward signal and directly determines the blower speed or power through a preset functional relationship. The third is closed-loop environmental parameter regulation, which uses measurable environmental parameters such as temperature or gas concentration as feedback quantities, and adjusts the blower output through a PID controller. A common feature of these schemes is that the control loop uses a directly measurable physical quantity as a reference or feedback signal, causing the blower output power to track changes in that signal. Although existing technologies have achieved a basic correlation between blower power and ship operating conditions, shortcomings remain. The control objective of existing schemes is to make the blower power follow a reference signal, without directly verifying whether the adjustment action produces a positive net energy-saving effect. For fixed-gear control, the gear division is based on preset design conditions. However, changes in cargo load, draft due to fuel consumption, and sea state during actual ship operation cause deviations from the design conditions. Fixed-gear control cannot respond to these real-time changes, leading to excessive or insufficient air supply under off-design conditions. For main engine power signal mapping control, changes in main engine power may be caused by factors such as deteriorating sea state or hull fouling. In such cases, increasing fan power may not improve drag reduction but could instead lead to a decrease in net benefit due to blindly increasing fan energy consumption. For closed-loop adjustment of environmental parameters, the controlled variable is a single measurable physical quantity. However, the controlled objective in the air lubrication drag reduction system is the net energy saving benefit, a composite indicator that requires indirect calculation, which traditional PID frameworks cannot directly handle. The root cause of these problems lies in the lack of direct closed-loop feedback on the difference between blower power consumption and propulsion power savings in existing technologies. This makes it impossible to determine in real time whether the current power adjustment has exceeded the optimal operating point, and it also fails to adapt to changes in system characteristics caused by hull fouling and fan performance drift during long-term operation. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a marine blower system and method with adaptive power adjustment based on data feedback. This solves the problem in existing marine air lubrication and drag reduction systems where blower power control lacks direct closed-loop feedback on the difference between blower power consumption and propulsion power savings, resulting in the inability to determine in real time whether the adjustment generates net energy savings and the inability to adapt to system characteristic drift during long-term operation.

[0005] To achieve the above objectives, the present invention provides the following technical solution: The first aspect of this invention provides a marine blower system with adaptive power adjustment based on data feedback. This system includes a multi-source data acquisition module, a variable operating condition feedforward calculation module, a net power feedback calculation module, a power adaptive adjustment controller, a constraint protection module, a steady-state data qualification determination module, a sample pool management module, an online coefficient estimation module, and a linkage protection module. Each module is deployed as a software functional module within the same industrial programmable logic controller (PLC). This PLC is connected to a speed sensor, a draft sensor, a back pressure transmitter, a power transmitter, a ship engine room monitoring system data bus, and a blower frequency converter via analog input modules and digital communication interfaces, respectively.

[0006] In marine air lubrication and drag reduction systems, a blower injects compressed air into the hull to form an air-water mixture, reducing hull friction resistance. The difference between the blower's power consumption and the power savings in propulsion represents the net energy saving benefit. Existing control schemes adjust blower power based on single physical quantities such as ship speed or main engine power, lacking direct feedback verification of this net energy saving benefit and making it impossible to determine whether the current power adjustment generates a positive net benefit.

[0007] This invention addresses the aforementioned problems by constructing a net power gain signal as a feedback quantity, establishing a heterogeneous dual-loop control architecture with feedforward and feedback, and using the operating data accumulated by the system under optimal steady state for online self-calibration of the feedforward model. Its working process is as follows.

[0008] The multi-source data acquisition module synchronously acquires the ship's speed in the water, draft at the bow, draft at the stern, nozzle back pressure, blower power consumption, and main engine output power in each control cycle, and calculates the average draft from the draft at the bow and draft at the stern.

[0009] The variable operating condition feedforward calculation module internally stores hull form coefficients, speed index, draft index, back pressure compensation function, and wind turbine characteristic curves. This module calculates the target feedforward power value in the following manner.

[0010] Module calculates theoretical jet flow rate: In the formula, For ship type coefficient, For speed, For speed index, To achieve average draft, Draft index, The width of the boat.

[0011] The module corrects the theoretical jet flow rate using a back pressure compensation function: In the formula, For nozzle back pressure, For design reference back pressure, This is the back pressure compensation function, which represents the flow correction factor required to achieve a drag reduction effect equivalent to the reference back pressure condition under non-reference back pressure.

[0012] The module calculates the target feedforward power value based on the corrected flow rate: In the formula, For the total pressure loss of the pipeline, This is the efficiency value corresponding to the corrected flow rate in the wind turbine characteristic curve. This feedforward power target value is output to the power adaptive control controller.

[0013] The net power feedback calculation module applies sliding window filtering to both the main engine power and the blower power, maintaining a reference time and storing the corresponding sliding average values ​​of the main engine power and blower power. In each control cycle, the module calculates the difference between the current sliding average value of the main engine power and the reference time, using this as the propulsion power change; it also calculates the difference between the current sliding average value of the blower power and the reference time, using this as the blower power change. The module takes the difference between the propulsion power change and the blower power change to obtain the net power gain signal. This net power gain signal serves as the basis for judging whether the current power adjustment produces a positive net energy saving effect.

[0014] The power adaptive control controller receives the feedforward power target value and the net power gain signal, and generates the power command increment using an incremental PI control law with dead time. The control law is executed in three regions, with a dead time width of ε, a net power gain signal of ΔPnet, a power command increment of ΔPcmd, and a control period of Ts. when hour, ; when hour, ; when hour, ; In the formula, and For positive adjustment of proportional gain and integral gain, and The proportional gain and integral gain are used for inverse adjustment, and the gain of inverse adjustment is greater than or equal to the corresponding gain of forward adjustment. The feedforward power target value and the power command increment are added together to form the initial power command.

[0015] In this heterogeneous dual-loop structure, the feedforward channel uses the output of the variable operating condition feedforward calculation module as the basic component to provide a pre-response when the operating condition changes; the feedback channel uses the net power gain signal as the controlled variable to verify and correct the adjustment direction.

[0016] The constraint protection module sequentially executes back pressure-surge boundary limits, minimum power maintenance constraints, and power command change rate limits on the initial power command. Back pressure-surge boundary limits compare the current operating point with a pre-stored turbine surge boundary curve, forcibly limiting the power command when the operating point enters the warning zone. Minimum power maintenance constraints force the power command to the minimum maintenance value when the speed falls below a set threshold. Power command change rate limits saturate the power command increments between adjacent control cycles. The processed final power command is output to the blower frequency converter.

[0017] While the real-time control loop is operating, the steady-state data eligibility determination module checks whether the following conditions are met in each control cycle: the absolute value of the net power gain signal is continuously within the dead zone width; the rate of change of the final power command is lower than the preset steady-state power change rate threshold; and the speed fluctuation range and the average draft fluctuation range do not exceed preset tolerances. When all of the above conditions are met simultaneously and the duration reaches the steady-state duration threshold, the system is determined to have entered a high-confidence steady state.

[0018] After the high-confidence steady-state period ends, the sample pool management module calculates the arithmetic mean of the target values ​​of speed, average draft, back pressure, and feedforward power during that period, combines these values ​​with a timestamp to form sample points, and stores them in a fixed-capacity rolling sample pool. The sample pool adopts a first-in-first-out queue structure.

[0019] The online coefficient estimation module retrieves all sample points from the sample pool when a trigger condition is met. The trigger condition is either that the number of sample points in the sample pool reaches a preset capacity, or that the time since the last successful estimation exceeds a preset self-correction period, whichever comes first. The module transforms the logarithm of the feedforward model into a linear form. In the formula, This represents the efficiency constant of the blower near the reference operating condition. The module employs a weighted least squares method with a forgetting factor to jointly estimate the intercept parameter, speed index, and draft index of the combined constant term, separating the updated hull form coefficients from the intercept parameter. The estimated coefficients undergo a reasonableness boundary check; estimates exceeding the preset range are discarded, and the original coefficients remain unchanged.

[0020] Before writing the verified new coefficients into the variable operating condition feedforward calculation module, the linkage protection module first expands the dead zone width of the power adaptive regulation controller by a preset multiple. After writing the new coefficients, a transition monitoring timer is started to continuously monitor whether the absolute value of the net power gain signal continuously exceeds the expanded dead zone within a preset transition duration. If no continuous exceedance occurs, the transition is determined to be smooth, the dead zone is restored to its original value, and the new coefficients are fixed. If continuous exceedance occurs, the transition is determined to be abnormal, the coefficients are rolled back to their pre-update values, the dead zone is restored to its original value, and a sample clearing command is sent to the sample pool management module to delete all sample points participating in this estimation. After the coefficient update or rollback is completed, the reference time used by the net power feedback calculation module to calculate the power change is reset to the current time.

[0021] A second aspect of the present invention provides a method for adaptively adjusting the power of a marine wind turbine with data feedback, comprising the following steps: During each control cycle, the ship's speed in the water, draft at the bow, draft at the stern, nozzle back pressure, blower power consumption, and main engine output power are collected synchronously to calculate the average draft.

[0022] Based on the ship type coefficient, speed index, and draft index, the theoretical jet flow rate is calculated by combining the speed and average draft. After correction by the back pressure compensation function and combined with the total pipeline pressure loss and the fan characteristic curve, the feedforward power target value is calculated.

[0023] After applying sliding filters to the main unit power and blower power, the changes in propulsion power and blower power are calculated with reference to the base time, and the difference between the two is taken as the net power gain signal.

[0024] Using the net power gain signal as the controlled variable and the feedforward power target value as the basic component, an incremental PI control law with dead zone is used to generate the superimposed initial power command.

[0025] The initial power command is sequentially subjected to back pressure-surge boundary constraints, minimum power maintenance constraints, and power command change rate constraints to obtain the final power command output to the blower frequency converter.

[0026] While performing the above steps in real time, if the net power gain signal is continuously within the dead zone, the rate of change of power command is lower than the preset threshold, and the fluctuations in speed and draft do not exceed the tolerance, and the above conditions continue to reach the steady-state duration threshold, it is determined that the state has entered a high-confidence steady state.

[0027] After the high-confidence steady-state period ends, the average values ​​of the target values ​​of speed, average draft, back pressure and feedforward power during that period are calculated, and timestamps are added to form sample points, which are then stored in the rolling sample pool.

[0028] When the number of sample points reaches the preset capacity or the time since the last successful parameter update exceeds the self-correction cycle, parameter estimation is triggered. After taking the logarithm of the feedforward model, the weighted least squares method with forgetting factor is used to re-estimate the ship type coefficient, speed index, and draft index. After passing the rationality boundary test, the new coefficients are output.

[0029] Before writing the new coefficients that have passed the test into the feedforward calculation stage, the dead zone is expanded. After writing, the net power gain signal is monitored within the preset transition time to see if it continuously exceeds the limit. If the transition is smooth, the dead zone is restored and the new coefficients are fixed. If it exceeds the limit, the coefficients are rolled back, the dead zone is restored, and the corresponding samples are deleted.

[0030] After the coefficient update or rollback is completed, the reference time used to calculate the power change is reset to the current time, and the current moving average of the host power and the moving average of the blower power are latched as the new reference benchmark.

[0031] This invention provides a marine wind turbine system and method with adaptive power adjustment based on data feedback. It offers the following advantages: 1. This invention constructs a net power gain signal as feedback, enabling blower power regulation to directly optimize net energy saving. Existing solutions adjust blower power based on single physical quantities such as engine speed or main engine power, failing to determine whether the adjustment action truly generates net benefit. This invention calculates the difference between the change in main engine power and the change in blower power, using this difference as the controlled variable fed into the regulator. The regulator only operates when the net power gain signal exceeds the dead zone, avoiding ineffective regulation caused by background noise in the power signal. The regulator's reverse gain is greater than its forward gain, rapidly withdrawing power when the net benefit is negative, shortening the time the system remains in the negative benefit region.

[0032] 2. This invention employs a heterogeneous dual-loop control architecture combining feedforward and feedback, balancing both speed and accuracy of adjustment. The feedforward channel calculates and estimates the power value in real time based on ship operating parameters, allowing for pre-adjustment of turbine power without waiting for feedback signals when speed or draft changes, resulting in minimal response lag. The feedback channel verifies the adjustment direction using net power gain signals, correcting feedforward deviations and preventing over- or under-adjustment of turbine power due to feedforward model errors. The outputs of the two channels are superimposed, each undertaking different control functions, thus avoiding the trade-off between response speed and stability found in single-loop control.

[0033] 3. This invention utilizes an online self-calibration mechanism based on high-confidence steady-state data to maintain the accuracy of the feedforward model during long-term ship operation. The system collects samples only during periods when net power gain is stable within the dead zone, power command changes are gradual, and speed and draft fluctuations are minimal, ensuring low noise and representativeness of the sample data. Online estimation employs a weighted least squares method with a forgetting factor, giving higher weight to recent samples, enabling model coefficients to track the slow drift of hull fouling and blower performance. Before updating coefficients, the feedback dead zone is expanded, and net power gain signals are monitored during the transition period. If the limits are exceeded, automatic rollback occurs, ensuring that the self-calibration process does not compromise the stability of real-time control. Attached Figure Description

[0034] Figure 1 This is a schematic diagram of the system functional modules of the present invention; Figure 2 This is a schematic diagram of the heterogeneous dual-loop control structure of the power adaptive regulation controller of the present invention; Figure 3 This is a flowchart of the net power gain signal generation process of the present invention; Figure 4 This is a flowchart of the online self-calibration mechanism of the present invention; Figure 5 This is a flowchart of the method of the present invention; Figure 6 This is a schematic diagram of the fast and slow dual-ring collaborative working timing of the present invention.

[0035] Among them, 101, multi-source data acquisition module; 102, variable operating condition feedforward calculation module; 103, net power feedback calculation module; 104, power adaptive adjustment controller; 105, constraint protection module; 106, steady-state data qualification judgment module; 107, sample pool management module; 108, coefficient online estimation module; 109, linkage protection module. Detailed Implementation

[0036] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0037] It should be noted that, unless otherwise defined, the technical or scientific terms used in this invention should have the ordinary meaning understood by one of ordinary skill in the art to which this invention pertains. The terms "first," "second," and similar terms used in this invention do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the elements or objects listed following the word and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are used only to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0038] Example 1: Reference Figure 1 , Figure 1 This is a schematic diagram of the system functional modules of a marine blower system with adaptive power adjustment and data feedback according to the present invention. The system includes a multi-source data acquisition module 101, a variable operating condition feedforward calculation module 102, a net power feedback calculation module 103, a power adaptive adjustment controller 104, a constraint protection module 105, a steady-state data qualification determination module 106, a sample pool management module 107, an online coefficient estimation module 108, and a linkage protection module 109. All modules are deployed as software functional modules within the same industrial programmable logic controller (PLC). This PLC is connected to a speed sensor, a draft sensor, a back pressure transmitter, a power transmitter, a ship engine room monitoring system data bus, and a blower frequency converter via analog input modules and digital communication interfaces.

[0039] The multi-source data acquisition module 101 synchronously acquires the ship's speed in the water, bow draft and stern draft, back pressure in the air jet nozzle inlet pipe, real-time power consumption of the blower, and output power of the ship's main engine in each control cycle. This module calculates the average draft and heel angle from the bow and stern drafts, and provides the speed, average draft, and back pressure to the variable operating condition feedforward calculation module 102, while providing the main engine power and blower power to the net power feedback calculation module 103.

[0040] The variable operating condition feedforward calculation module 102 stores the currently effective ship type coefficient, speed index, draft index, back pressure compensation function, and wind turbine characteristic curve. Based on the received speed, average draft, and back pressure, this module calculates the theoretically optimal jet flow rate according to a preset model. After back pressure compensation correction, it combines the total pipeline pressure loss and wind turbine efficiency mapping to obtain the feedforward power target value and outputs it to the power adaptive regulation controller 104.

[0041] The net power feedback calculation module 103 performs moving average filtering with a sliding window length of Tw on the received main engine power and turbine power. This module maintains a reference time and its corresponding average power value, and calculates the change in the current average power relative to the reference time in each control cycle, obtaining the change in propulsion power and the change in turbine power. The difference between the two is the net power gain signal. This net power gain signal is sent to the power adaptive regulation controller 104 as feedback.

[0042] The power adaptive control controller 104 receives the feedforward power target value and the net power gain signal. Using the feedforward power target value as the base component and the net power gain as the controlled variable, it executes an incremental PI control law with dead zone to generate a power command increment. The dead zone width ε is set according to the wind turbine's rated power and the background noise of the actual power signal. The feedforward base component and the feedback increment are superimposed to form the initial power command, which is output to the constraint protection module 105.

[0043] The constraint protection module 105 sequentially processes the initial power command through back pressure-surge boundary constraints, minimum power maintenance constraints, and rate of change constraints, generating the final power command output to the blower frequency converter. Back pressure-surge protection is achieved by comparing the current operating point with the pre-stored blower surge boundary curve; the minimum power maintenance constraint forces the power command to a preset minimum maintenance value when the speed is lower than a set threshold; and the rate of change constraint saturates the power command increment between adjacent control cycles.

[0044] While the real-time control loop continues to operate, the system performs online self-calibration in the form of a slow task.

[0045] The steady-state data eligibility determination module 106 receives the net power gain signal, the power command after constraint protection, speed, and average draft in each control cycle. When the absolute value of the net power gain is continuously within the dead zone, the rate of change of the power command is lower than a preset threshold, and the range of speed fluctuation and the range of average draft fluctuation do not exceed preset tolerances, and the duration of the above conditions being met simultaneously reaches the steady-state duration threshold, the system is determined to have entered a high-confidence steady state. This module provides the high-confidence steady-state identifier and related signals to the sample pool management module 107.

[0046] After the high-confidence steady-state period ends, the sample pool management module 107 calculates the arithmetic mean of the speed, average draft, back pressure, and feedforward power during that period, combines them with a timestamp to form a sample point, and stores it in a fixed-capacity rolling sample pool. The sample pool adopts a first-in-first-out queue structure; whenever a new sample arrives and the sample pool is full, the oldest sample is removed.

[0047] The online coefficient estimation module 108 is triggered when the number of samples in the sample pool reaches a preset capacity or when the time since the last successful estimation exceeds a preset self-correction trigger period. Upon triggering, the module retrieves all samples from the sample pool and estimates the parameters in the feedforward model using a weighted least squares method with a forgetting factor. The sample weights decay exponentially with the sample timestamp. After estimation, a reasonableness boundary check is performed on the obtained parameters; estimated values ​​exceeding a preset range are discarded, and the original coefficients are maintained.

[0048] The linkage protection module 109 performs transition protection when the coefficients need to be updated. This module first temporarily expands the feedback dead zone of the power adaptive regulation controller 104 to a preset multiple, and then writes the new coefficients into the variable operating condition feedforward calculation module 102. Within a preset transition monitoring period, it continuously monitors changes in the net power gain signal; if there is no continuous breach of the expanded dead zone, the transition is considered smooth, the dead zone is restored to its original value, and the new coefficients are fixed; if continuous breaches are detected, the coefficients are immediately rolled back to their pre-update values, the original dead zone is restored, the samples participating in this estimation are removed from the sample pool, and a fault notification is issued. After the coefficients are updated or rolled back, the linkage protection module 109 resets the reference time used by the net power feedback calculation module 103.

[0049] Through the coordinated operation of the above modules, the variable operating condition feedforward calculation module 102 and the power adaptive adjustment controller 104 form a fast time scale real-time power optimization closed loop, and the net power feedback calculation module 103 provides feedback signals for the closed loop with net power gain as the object; the steady-state data qualification judgment module 106, the sample pool management module 107, the coefficient online estimation module 108 and the linkage protection module 109 form a slow time scale feedforward model self-correction path, which uses the operating data accumulated by the system in the optimal steady state to adaptively update the feedforward model, and ensures the stability of real-time control through dead zone release and monitoring rollback mechanisms during the update process.

[0050] The multi-source data acquisition module 101 is used to acquire all the real-time parameters required by the system. This module connects to various field devices through hardware wiring and communication interfaces, and completes signal synchronous latching within each control cycle. The control cycle is set according to the response time constant of the blower frequency converter and the power regulation rate allowed by the ship's power grid, with a typical range of 0.5 seconds to 2 seconds. After being converted by the engineering unit, each signal is directly mapped to the input interfaces of the variable operating condition feedforward calculation module 102 and the net power feedback calculation module 103 as global variables of the PLC, without undergoing intermediate data packaging or protocol conversion.

[0051] The ship's speed signal relative to the water is provided by a Doppler log installed on the hull. The Doppler log outputs an NMEA0183 format message, which the PLC receives and parses from the water speed field in the message via a serial communication module to obtain the speed value.

[0052] The bow and stern drafts are obtained by submersible pressure and level transmitters installed in the bow and stern measuring wells, respectively. The transmitters output a 4-20mA current signal, which is connected to the PLC's analog input module. The PLC converts the current signal into a draft value based on the transmitter's range limits and zero offset. The PLC further calculates the average draft and heel angle from the bow and stern drafts. The average draft is the arithmetic mean of the two values, and the heel angle is obtained by the arctangent of the ratio of the draft difference to the horizontal distance between the sensors. The horizontal distance between the sensors is written as a fixed constant into the PLC during system debugging.

[0053] Nozzle back pressure is measured by a pressure transmitter installed on the inlet header of the air jet nozzle. The range of the pressure transmitter is selected based on the rated outlet pressure of the blower, with a preset proportional margin. The transmitter outputs a 4-20mA signal to the PLC analog input module, where the PLC calculates the back pressure value using range conversion.

[0054] There are two methods to obtain the real-time power consumption of the blower. One method involves installing current and voltage transformers on the input side of the blower's frequency converter, with a power transmitter connected to the secondary side of the transformers. The power transmitter outputs a 4-20mA signal proportional to the active power, which is then acquired by the PLC's analog input module. The other method uses the frequency converter's own Modbus RTU communication interface, allowing the PLC to directly read the value of the holding register address corresponding to the power parameter. When using the latter method, the PLC reads the data in a polling manner.

[0055] The ship's main engine output power is obtained through the ship's engine room monitoring system data bus. The PLC is configured as a Modbus RTU slave or master station, and reads the register value corresponding to the main engine power according to the register mapping table specified by the engine room monitoring system, with a refresh cycle of no more than 1 second. If the engine room monitoring system provides an NMEA2000 or CAN bus interface, the PLC accesses and extracts the main engine power parameters through the corresponding communication module.

[0056] All of the above signals are stored as global variables within the PLC and refreshed once per control cycle. Speed, average draft, and back pressure are transmitted to the variable operating condition feedforward calculation module 102, while main engine power and blower power are transmitted to the net power feedback calculation module 103. The pitch angle is used by other modules for conditional judgment or recording.

[0057] The variable operating condition feedforward calculation module 102 receives the speed, average draft, and nozzle back pressure provided by the multi-source data acquisition module 101, and calculates the feedforward power target value based on the preset mathematical model and stored turbine characteristic data. This feedforward power target value serves as the fundamental component of the power adaptive regulation controller 104, providing the system with pre-response capability when operating conditions change.

[0058] The variable operating condition feedforward calculation module 102 internally stores a set of feedforward model coefficients, including hull form coefficients, speed index, and draft index. The initial values ​​of these coefficients are obtained through experimental calibration during the ship's sea trials and are written to the controller upon system commissioning. The calibration method is as follows: under multiple combinations of speed and draft, the blower jet flow rate is adjusted, and the jet flow rate that maximizes the difference between propulsion power savings and blower power consumption is recorded as the actual optimal flow rate under that condition. Using the actual optimal flow rate as the fitting objective and speed and draft as independent variables, a nonlinear regression is performed on the theoretical optimal jet flow rate formula. The initial values ​​of the hull form coefficients, speed index, and draft index are determined by minimizing the sum of squared residuals between the predicted flow rate and the actual optimal flow rate. During subsequent system operation, this set of coefficients can be updated by the online coefficient estimation module 108.

[0059] The specific process of the variable operating condition feedforward calculation module 102 calculating the target value of feedforward power is as follows.

[0060] In marine air lubrication and drag reduction systems, the jet volumetric flow rate required to maintain an effective air layer at the hull bottom is related to ship speed, draft, and beam. This module uses the following empirical correlation model to calculate the theoretically optimal jet flow rate: In the formula, , is the hull form factor, which reflects the influence of the hull geometry and jet hole arrangement on drag reduction, and is dimensionless; For speed index; The draft index; The ship's speed relative to water is provided by the multi-source data acquisition module 101; The average draft provided for the multi-source data acquisition module 101; The width of the ship is written as a fixed parameter during the debugging phase.

[0061] Because the back pressure at the air jet nozzle outlet varies with speed and draft, when the actual back pressure deviates from the design reference back pressure, the work done by the blower under the same jet volume flow rate changes, leading to a shift in the equivalent drag reduction effect. To compensate for this effect, the variable operating condition feedforward calculation module 102 incorporates a back pressure compensation function to correct the theoretical jet flow rate. In the formula, The nozzle back pressure provided to the multi-source data acquisition module 101; The back pressure is taken as the reference back pressure under design conditions, and the back pressure value at the nozzle inlet is taken under the rated operating conditions of the blower. This is the back pressure compensation function. The physical meaning of this compensation function is: under non-reference back pressure conditions, the correction factor of the jet flow rate required to achieve the same drag reduction effect as the reference back pressure condition. It is obtained by: on a blower factory test bench, maintaining a constant jet flow rate under multiple back pressure conditions, measuring the change in blower power, converting it into the flow correction factor required to maintain the equivalent drag reduction effect, and then fitting it as a function with the back pressure ratio as the independent variable. The conversion relationship of the equivalent flow correction factor is: dividing the change in blower power under non-reference back pressure conditions by the change in blower power corresponding to a unit flow rate change under the reference back pressure condition, obtaining the flow correction amount to be compensated, and then calculating the correction factor. In specific implementation, this function is stored in the controller memory in the form of a two-dimensional lookup table. The table records the compensation factor corresponding to different back pressure ratios, which is obtained through linear interpolation based on the current back pressure ratio during runtime. The node density of the lookup table is determined based on the curvature of the compensation function measured on the blower test bench, increasing the number of nodes for back pressure intervals where the function changes drastically. The back pressure compensation function can also be implemented using a piecewise cubic polynomial, with the polynomial coefficients determined by least squares fitting of offline experimental data.

[0062] After back pressure correction is completed, the corresponding turbine power requirement is calculated based on the corrected flow rate. The formula for calculating the feedforward power target value is: In the formula, The estimated total pressure loss of the pipeline from the blower outlet to the nozzle inlet is composed of the friction loss along the pipeline, the local resistance loss of elbows and valves, and the hydrostatic pressure at the nozzle. It is determined by the pipeline hydraulic calculation during the system design phase and written into the controller as a fixed parameter. For the blower at a volumetric flow rate of Equivalent efficiency at that time.

[0063] The relationship between blower efficiency and flow rate is stored in the controller as a characteristic curve table. This table data originates from the blower's factory performance test report and contains several sets of volumetric flow rate-equivalent efficiency data pairs. During operation, the variable operating condition feedforward calculation module 102 determines two adjacent data points in the table based on the currently calculated corrected flow rate value and obtains the corresponding efficiency value through linear interpolation. For marine blowers such as centrifugal and screw blowers, their efficiency curves change smoothly over a wide flow range, and linear interpolation can meet engineering accuracy requirements. If a multi-stage centrifugal blower is used and operates near the inflection point of the efficiency curve, the table data density is increased in the flow range where efficiency changes significantly, or cubic spline interpolation is used instead.

[0064] The variable operating condition feedforward calculation module 102 executes the above calculation process once in each control cycle, outputs the obtained feedforward power target value to the power adaptive adjustment controller 104, and sends the value to the sample pool management module 107 for steady-state data acquisition.

[0065] Reference Figure 3 , Figure 3 This is a flowchart illustrating the generation of a net power gain signal for a marine wind turbine system with adaptive power adjustment and data feedback, according to the present invention. The net power feedback calculation module 103 receives the main engine power and blower power provided by the multi-source data acquisition module 101, performs moving average filtering on each, extracts the power change using a reference time, and constructs a net power gain signal. This net power gain signal serves as the controlled variable of the power adaptive adjustment controller 104, used to evaluate whether the current power adjustment action produces a positive net energy saving effect.

[0066] The net power feedback calculation module 103 internally maintains two fixed-length annular buffers, which store historical sampled values ​​of main engine power and blower power, respectively. In each control cycle, newly sampled main engine power and blower power are written to the current pointer position of the corresponding buffer. The pointers increment and reset to the starting position when they reach the end of the buffer. The buffer length is determined by the ratio of the sliding window length to the controller cycle. The sliding window length is taken as a preset multiple of the typical wave cycle encountered by the ship, and not less than a preset lower limit value, to filter out main engine power fluctuations caused by waves.

[0067] In digital controllers, the equivalent implementation of moving average filtering is to take the arithmetic mean of all elements in the buffer. Let the most recent host power samples stored in the buffer be the current content of the buffer. Then, the moving average of the host power is calculated using the following formula: The sliding average power of the blower is calculated in the same way: In the formula, The current moment; To control the cycle; The number of control cycles contained within the sliding window is obtained by dividing the length of the sliding window by the control cycle and then rounding down. The distance from the current time The host power sample value before each control cycle; The distance from the current time The blower power sampling value before each control cycle.

[0068] Moving average filtering can also be implemented using an exponentially weighted moving average. Its recursive form can reduce buffer storage overhead. The specific form should be selected by those skilled in the art based on the controller's memory resources, and will not be elaborated here. The equivalent cutoff frequencies of the two filtering methods should be close.

[0069] The net power feedback calculation module 103 sets a reference time and stores the corresponding sliding average power of the main unit and the sliding average power of the blower. The initial value of the reference time is recorded after the system is powered on and enters a stable operating state. During system operation, when the incremental output of the power adaptive control controller 104 remains within a preset narrow range for a preset duration, the net power feedback calculation module 103 determines that the system has entered a new steady state, records the current time as the new reference time, and simultaneously latches the sliding average power of the main unit and the sliding average power of the blower at this moment, overwriting the data of the original reference time. The width of the preset narrow range is a preset proportion of the dead zone, and the preset duration is a preset multiple of the sliding window length.

[0070] The update logic for the reference time also includes a freezing mechanism. When the system is in a high-confidence steady-state period or when the online coefficient estimation module 108 is performing parameter estimation, the update of the reference time is paused to avoid changing the reference benchmark for power change during the sampling of steady-state data used for self-calibration, thereby ensuring that the power change corresponding to the steady-state sample is calculated based on the same reference time.

[0071] In each control cycle, the net power feedback calculation module 103 calculates the change in propulsion power at the current moment relative to the reference moment: And the change in blower power: In the formula, This is the current valid reference time; The moving average of the host power at the reference time; The sliding average value of the blower power at the reference time; This represents the current moment's moving average of the host power. This represents the sliding average value of the blower power at the current moment.

[0072] The physical meaning of the change in propulsion power is: the decrease in main engine output power from the reference time to the current time. When the drag reduction effect of air lubrication is enhanced, the required propulsion power at the same speed is reduced, and the change in propulsion power is positive, indicating a saving in main engine power. The physical meaning of the change in blower power is: the increase in blower power consumption from the reference time to the current time. When the blower increases its output, this change is positive, representing the additional energy cost incurred to achieve drag reduction.

[0073] The net power gain signal is calculated from the changes in propulsion power and blower power: When the net power gain signal is positive, it indicates that the power savings from the blower power adjustment since the reference time are greater than the increase in the blower's own power consumption, and the adjustment has resulted in positive net energy savings. When the net power gain signal is negative, it indicates that the increase in blower power has exceeded the power savings, and the adjustment has no net gain or the net gain is negative.

[0074] The net power feedback calculation module 103 outputs the calculated net power gain signal to the power adaptive adjustment controller 104 in each control cycle, and simultaneously outputs the signal to the steady-state data qualification judgment module 106 for high-confidence steady-state judgment. The net power feedback calculation module 103 also provides the currently valid reference time and the corresponding host power sliding average value and blower power sliding average value to the linkage protection module 109 for reference time reset after coefficient update.

[0075] Reference Figure 2 , Figure 2 This is a schematic diagram of the heterogeneous dual-loop control structure of a power adaptive adjustment controller for a marine wind turbine system with data feedback and adaptive power adjustment according to the present invention. The power adaptive adjustment controller 104 receives the feedforward power target value output by the variable operating condition feedforward calculation module 102 and the net power gain signal output by the net power feedback calculation module 103. Using the feedforward power target value as the base component and the net power gain signal as the controlled variable, it generates a power command increment using an incremental PI control law with dead time. The base component and the increment are superimposed to form an initial power command, which is then output to the constraint protection module 105.

[0076] The power adaptive regulation controller 104 adopts a heterogeneous dual-loop structure with feedforward and feedback. The feedforward channel calculates the estimated power value based on the ship's operating parameters, providing a pre-response when the operating conditions change; the feedback channel uses the net power gain signal as the controlled variable to verify and correct the regulation direction. The outputs of the two channels are superimposed to form the power command.

[0077] The target value of the feedforward power is directly taken from the output of the variable operating condition feedforward calculation module 102. The incremental output of the feedback channel is generated by an incremental PI controller, which has a dead zone set, and the dead zone width is denoted as... The purpose of the dead zone is to maintain the current incremental output when the absolute value of the net power gain signal is small, thus avoiding frequent adjustments caused by background noise and small fluctuations in the power signal.

[0078] The control law of the incremental PI controller is executed in three regions. Let the current control cycle number be... .

[0079] when At this point, the system is in the positive adjustment domain, and the power savings from the increase in blower power have exceeded the increase in blower energy consumption; therefore, the blower power should be further increased. The update formula for the incremental output is: when ≤ At this time, the system is in a dead zone, and the incremental output retains the value of the previous cycle: when At this point, the system is in the reverse regulation domain, and the energy cost of the increased blower power exceeds the power saving of the propulsion system; therefore, the blower power should be reduced. The update formula for the incremental output is: Of the above formulas, The controller's incremental output for the current control cycle; This is the incremental output of the controller from the previous control cycle; This is the net power gain signal for the current control cycle; This is the net power gain signal from the previous control cycle; To control the cycle; and These are the proportional gain and integral gain of the positive adjustment domain, respectively; and These are the proportional gain and integral gain of the inverse adjustment domain, respectively.

[0080] In the reverse control domain, the proportional gain and integral gain are greater than or equal to the corresponding gains in the forward control domain. When the net power gain is negative, the system has already passed the optimal operating point. A larger gain is used to withdraw the power command, shortening the time the system spends in the negative gain region. In the forward control domain, a relatively gentle gain is used, allowing the system to gradually approach the optimal operating point with smaller steps when the net gain is positive, reducing overshoot.

[0081] The dead zone width ε is set based on the blower's rated power and the background noise level of the power signal, taking a fixed value within the range of 2% to 5% of the rated power. In actual ship implementation, during the sea trial phase, the background noise level is determined by analyzing the fluctuation amplitude of the moving average signals of the main engine power and blower power during steady-state navigation, and then the specific value of the dead zone is determined accordingly. When the ship is equipped with a motion reference unit, the controller acquires the vertical acceleration signal and divides the sea state into several levels based on the effective value of the vertical acceleration. Each level corresponds to a dead zone expansion coefficient, and the dead zone width is the product of the reference value and the expansion coefficient corresponding to the current sea state level. The mapping relationship between sea state levels and expansion coefficients is preset in the controller, with higher sea state levels corresponding to larger expansion coefficients.

[0082] In each control cycle, the power adaptive regulation controller 104 superimposes the feedforward power target value with the PI incremental output to obtain the initial power command: In the formula, The target value of feedforward power provided to the variable operating condition feedforward calculation module 102. The superimposed value... After internal limiting processing, the power is restricted to between the minimum allowable power and the rated power of the blower, and output as an initial power command to the constraint protection module 105.

[0083] The power adaptive regulation controller 104 also provides the current incremental output value to the net power feedback calculation module 103 for reference time update determination.

[0084] The constraint protection module 105 receives the initial power command output by the power adaptive regulation controller 104, and sequentially executes the back pressure-surge boundary limit, minimum power maintenance constraint and power command change rate limit to generate the final power command output to the blower frequency converter.

[0085] The back pressure-surge boundary limit is used to prevent the blower from entering surge conditions. The constraint protection module 105 internally stores the blower's surge boundary curve in the form of a two-dimensional lookup table. This lookup table consists of several sets of "volume flow rate - surge critical pressure ratio" data pairs, sourced from the blower's factory performance test report. Linear interpolation is used between adjacent data points in the lookup table. During operation, the pressure ratio and flow rate at the current operating point are compared with the surge boundary curve. The pressure ratio at the current operating point is characterized by the ratio of back pressure to reference back pressure. The flow rate at the current operating point is calculated from the feedforward power target value: using the power-flow correspondence stored in the blower characteristic curve, the corresponding volume flow rate is looked up using the current feedforward power target value. If the flow rate at the current operating point is lower than the critical flow rate corresponding to the same pressure ratio on the surge boundary curve, and the difference between the critical flow rate and the current flow rate falls within a preset safety margin range, then the operating point is determined to have entered the surge warning zone. At this time, the constraint protection module 105 forcibly limits the power command to the preset surge safe power upper limit value, and simultaneously issues a warning signal through the PLC output module. The upper limit of the surge safety power is determined during the wind turbine factory test. The method is as follows: under rated back pressure conditions, a specified multiple of the surge critical flow rate is taken as the safe flow rate. Then, the wind turbine shaft power corresponding to the safe flow rate is found from the power-flow correspondence in the wind turbine characteristic curve. This power value is written into the controller as the upper limit of the surge safety power.

[0086] The minimum power sustaining constraint is used to maintain airflow when the ship is operating below the effective drag-reducing speed threshold for air lubrication. The constraint protection module 105 reads the speed value in each control cycle and compares it with a preset speed threshold. The speed threshold is determined based on the ship's design data and the effective drag-reducing speed limit provided by the air lubrication system supplier. When the speed is below this threshold, the constraint protection module 105 directly forces the power command to the preset minimum sustaining power value. This power value corresponds to the output power of the blower operating at the minimum permissible frequency, ensuring a slight positive pressure is maintained at the nozzle outlet. When the speed recovers and exceeds the sum of the speed threshold and the preset hysteresis value, the constraint protection module 105 releases the minimum sustaining power constraint and resumes the power command output by the power adaptive control controller 104.

[0087] The power command rate limit is used to restrict the magnitude of the output power command change between adjacent control cycles. Let the final power command of the previous control cycle be... The current power command to be output is The upper limit of the power change rate is taken as the smaller value between the maximum allowable transient power change rate of the generator set and the rated power of the frequency converter, and the control cycle is... The saturation treatment is then performed according to the following formula: In the formula, For a saturation function, when Time value ,when Time value ,when Time value The upper limit of the power change rate is calculated and determined during the commissioning phase based on the actual configuration parameters of the generator set and the rated power of the frequency converter, and then written into the controller.

[0088] The three protection functions of the constraint protection module 105 are executed in series in the order of back pressure-surge boundary limit, minimum power maintenance constraint, and power command change rate limit. The back pressure-surge boundary limit has the highest priority, and its output covers the processing results of the first two protection functions. The final power command after three-layer protection processing is output to the frequency command channel of the blower inverter and simultaneously fed back to the steady-state data qualification judgment module 106 for steady-state judgment. The constraint protection module 105 also feeds back the final power command to the power adaptive adjustment controller 104. The controller compares the final power command with the difference between the feedforward value and the incremental summation result. When the two are inconsistent, it indicates that the incremental output has been limited by the constraint protection module. The controller synchronously updates the internal register value of the incremental output to the difference between the final power command and the feedforward value, so that the historical state of the incremental output is consistent with the actual execution amount.

[0089] The steady-state data eligibility determination module 106 receives the net power gain signal output by the net power feedback calculation module 103, the final power command output by the constraint protection module 105, and the speed and average draft provided by the multi-source data acquisition module 101. This module performs conditional judgment on the above signals in each control cycle. When all conditions are simultaneously met and continue for a specified duration, the system is determined to have entered a high-confidence steady state, and the high-confidence steady-state identifier and start and end times are output to the sample pool management module 107.

[0090] The steady-state data eligibility module 106 internally maintains three condition discriminators and one steady-state timer. The three condition discriminators operate in parallel. The steady-state data eligibility module 106 shares the dead-zone width parameter with the power adaptive regulation controller 104.

[0091] The condition discriminator determines whether the net power gain signal is within the dead zone. The discriminator reads the net power gain signal for the current control cycle, takes its absolute value, and compares it with the dead zone width ε. The condition is met when the absolute value is less than or equal to ε.

[0092] The second condition discriminator determines whether the rate of change of the power command is lower than a preset threshold. The discriminator performs a first-order backward differential on the final power command output by the constraint protection module 105, calculates the absolute value of the power command change between adjacent control cycles, and divides it by the control cycle to obtain an approximate value of the rate of change. This condition is met when the rate of change is less than a preset steady-state power change rate threshold. The steady-state power change rate threshold is determined based on the following: at this rate of change, the cumulative change of the power command within a sliding window does not exceed a specified proportion of the dead zone width, thus ensuring that the gradual change of the power command does not cause the net power gain signal to drift beyond the dead zone.

[0093] The third conditional discriminator determines whether fluctuations in speed and average draft are within preset tolerances. Internally, the discriminator maintains a ring-shaped buffer, the same length as the sliding window, for both speed and average draft, storing recent speed and average draft sample values ​​respectively. In each control cycle, the discriminator calculates the range of the sample values ​​within the speed buffer (the difference between the maximum and minimum values) and compares it to the preset speed fluctuation tolerance; simultaneously, it calculates the range of the sample values ​​within the average draft buffer and compares it to the preset average draft fluctuation tolerance. The condition is met if both do not exceed their respective tolerances. The speed fluctuation tolerance is determined by taking the partial derivative of the feedforward model with respect to speed at typical speed and draft operating points, multiplying the partial derivative by the speed fluctuation tolerance to obtain the estimated change in feedforward power. This change should not exceed the dead zone width. The average draft fluctuation tolerance is determined similarly by taking the partial derivative of the feedforward model with respect to average draft.

[0094] A steady-state timer records the continuous duration for which all three conditions are met simultaneously. In each control cycle, if all three conditions are met, the steady-state timer increments by one control cycle duration; if any condition is not met, the steady-state timer is reset to zero. When the accumulated duration of the steady-state timer reaches a preset steady-state duration threshold, the steady-state data eligibility module 106 records the current time as the end time of the high-confidence steady-state period and, based on the end time and steady-state duration, calculates the start time to generate a high-confidence steady-state identifier signal. This identifier signal, along with the start and end times, is output to the sample pool management module 107. The steady-state timer continues counting after the identifier is generated until it is reset to zero when any condition is violated. When the high-confidence steady-state duration spans multiple steady-state duration thresholds, the steady-state data eligibility module 106 generates high-confidence steady-state identifiers in segments with each steady-state duration threshold as an interval.

[0095] The steady-state duration threshold is taken as an integer multiple of the sliding window length, and the specific multiple is selected during the system debugging phase based on the noise characteristics of the power signal.

[0096] During the high-confidence steady-state period or when the online coefficient estimation module 108 is performing parameter estimation, the steady-state data qualification determination module 106 outputs a reference time freeze request signal to the net power feedback calculation module 103. The logic and function of this signal have been described in the net power feedback calculation module 103 section and will not be repeated here.

[0097] The sample pool management module 107 receives the high-confidence steady-state identifier and start and end times output by the steady-state data qualification judgment module 106. Simultaneously, it acquires the speed and average draft from the multi-source data acquisition module 101, the feedforward power target value from the variable operating condition feedforward calculation module 102, and the nozzle back pressure from the multi-source data acquisition module 101. After the high-confidence steady-state period ends, this module extracts the average value of each physical quantity within that period and adds a timestamp to form a sample point, which is then stored in a fixed-capacity rolling sample pool for use by the online coefficient estimation module 108.

[0098] The sample pool management module 107 internally maintains a sample queue, the capacity of which is written during system initialization. The queue adopts a first-in, first-out (FIFO) structure. When a new sample point arrives and the queue is full, the sample point with the earliest timestamp is deleted from the queue, and the new sample point is added to the end of the queue.

[0099] The start and end times of the high-confidence steady-state period are carried by the identifier signal output by the steady-state data qualification determination module 106. When the sample pool management module 107 receives the identifier indicating the end of the high-confidence steady-state period, it reads all the sampled data buffered within that period from its internal ring buffer. This ring buffer continuously and cyclically stores the target values ​​of speed, average draft, back pressure, and feedforward power for each control cycle during system operation. In each control cycle, the buffer directly reads the corresponding signal values ​​from the output interfaces of the multi-source data acquisition module 101 and the variable operating condition feedforward calculation module 102. The control cycles of each module are unified, and sampling is synchronized. The buffer length at least covers the number of sampling points corresponding to a steady-state duration threshold, ensuring that all sampled data within the complete high-confidence steady-state period can be back-read.

[0100] The sample pool management module 107 determines the sampling point index range covered by the steady-state period based on the start and end times, extracts all data from the corresponding interval from the buffer, calculates the arithmetic mean of each physical quantity within the period, and generates a sample point. The components of the sample point are calculated using the following formula: In the formula, This represents the number of control cycles contained within this high-confidence steady-state period. For the first time period Speed ​​sampling values ​​for each control cycle; For the first Average draft sampling value for each control cycle; For the first Back pressure sampling value for each control cycle; For the first The target value of feedforward power for each control cycle.

[0101] The sample pool management module 107 combines the above four average values ​​with the current time's timestamp into a single sample point. The data structure includes fields for average speed, average draft, average back pressure, average feedforward power, and timestamp. The timestamp is taken as the end time of the high-confidence steady-state period.

[0102] The sample pool is implemented in the controller's memory using an array with head and tail pointers. The head pointer points to the storage location of the oldest sample point, and the tail pointer points to the next empty location to be written. When a new sample point is added, it is written to the location pointed to by the tail pointer, which is then incremented. If the tail pointer coincides with the head pointer after incrementing, it indicates that the queue is full; the head pointer is then incremented, and the oldest sample point is discarded. Both the head and tail pointers are used in a modulo-based loop. The current number of samples in the queue is calculated from the relative positions of the tail and head pointers.

[0103] The sample pool management module 107 provides sample data to the online coefficient estimation module 108. The online coefficient estimation module 108 can query the number of sample points currently stored in the queue. When the online coefficient estimation module 108 triggers a parameter estimation task, the sample pool management module 107 sorts all sample points in the queue in ascending order by timestamp and outputs them all at once via memory copy. The sorting operation does not change the storage order of the queue.

[0104] When the linkage protection module 109 determines that a coefficient update is abnormal and performs a rollback, it sends a sample clearing instruction to the sample pool management module 107. This instruction includes a cutoff timestamp, which is the latest timestamp among the sample points used in the abnormal estimation batch. Upon receiving this instruction, the sample pool management module 107 iterates backwards from the tail of the queue to the head, deleting all sample points with timestamps greater than or equal to the cutoff timestamp, and then backs the tail pointer to the corresponding position. If the queue is empty after clearing, the trigger condition for the online coefficient estimation module 108 will only be triggered when the time condition since the last estimation is met.

[0105] When the triggering condition is met, the online coefficient estimation module 108 obtains all sample points from the sample pool management module 107, and re-estimates the feedforward model coefficients using the weighted least squares method with a forgetting factor. The estimation results are then output to the linkage protection module 109 after being checked for reasonableness boundaries.

[0106] The online coefficient estimation module 108 has two triggering conditions. One is that the number of sample points in the sample pool reaches the preset sample pool capacity, which is checked after each new sample point is stored in the sample pool management module 107. The other is that the time since the last successful estimation exceeds a preset self-correction triggering cycle, which is checked by an independent timer in each control cycle. The estimation task is triggered by the first of the two conditions.

[0107] Upon triggering, the online coefficient estimation module 108 retrieves all sample points in the queue through the sample data output interface provided by the sample pool management module 107. Each sample point contains the average speed, average draft, average back pressure, average feedforward power, and a timestamp. Let the number of sample points in the current sample pool be... .

[0108] The feedforward power calculation formula defined in the variable operating condition feedforward calculation module 102 is as follows: Near the operating point where the system is in a high-confidence steady state, the equivalent efficiency of the blower changes very little and is taken as a constant. This constant is taken as the efficiency value corresponding to the design flow rate point of the blower. Rearranging the above relationship and taking the natural logarithm, we obtain the linearized estimation model: Define the following variables to simplify the expression. Let Logarithmic power after back pressure compensation: In the formula, This is the back pressure compensation function stored and used in the variable operating condition feedforward calculation module 102. During online estimation, this function is directly called to calculate the compensation factor corresponding to each sample point.

[0109] make The intercept parameter after combining constant terms: Define eigenvectors and parameter vector : For each sample point, the following linear relationship holds: For the sample pool The nth sample point, the th Feature vector of each sample point The target value is constructed by taking the logarithm of the average speed and average draft of the sample points. The result is obtained by taking the logarithm of the average feedforward power and average backpressure at this sample point after the above correction. The timestamp is recorded as follows. The current time is recorded as .

[0110] To accommodate the slow drift of hull fouling and fan performance over time, recent samples are given higher weights during the estimation process. The weighting function employs an exponential forgetting mechanism, where the weighted average is the first digit of the first digit. Weight of each sample point for: In the formula, This is the forgetting factor, a constant with a value greater than zero and not exceeding 1. When the value is 1, all samples are equally weighted.

[0111] Construct the weighted sum of squared residual objective function: Find the parameter vector estimates that minimize the objective function. Setting the gradient of the objective function with respect to the parameter vector to zero, we obtain the canonical equation, whose analytical solution is: In the formula, for The matrix whose first... Travel constitute; for dimensional column vector, the first The elements are ; for diagonal matrix, .

[0112] For matrix inversion operations, when the number of sample points is large, a recursive least squares algorithm can be used to reduce the computational burden. Its recursive form is a well-known technique in this field and will not be elaborated upon here. With the sample pool capacity set to several dozen, the matrix inversion computation load of batch processing is within the controller's tolerance.

[0113] From the estimation results Calculate the updated values ​​of the feedforward model coefficients: In the formula, , , These represent the first, second, and third components of the parameter vector estimate; the constant term. The calculations are performed and stored in the controller during the system debugging phase.

[0114] The online coefficient estimation module 108 performs a reasonable boundary check on the updated values. The controller internally presets allowable ranges for each coefficient: allowable range for hull form coefficients, allowable range for speed index, and allowable range for draft index. These ranges are set based on empirical laws of ship hydrodynamics and the fluctuation range of sea trial calibration results. If any updated value exceeds its corresponding allowable range, the estimation is deemed invalid, all updated values ​​are discarded, the current coefficients in the variable operating condition feedforward calculation module 102 remain unchanged, and an estimation anomaly event is recorded. If all updated values ​​are within the allowable range, the new coefficients, along with the timestamp of estimation completion, are output to the linkage protection module 109.

[0115] During the estimation task, the online coefficient estimation module 108 outputs a reference time freeze request signal to the net power feedback calculation module 103 through the steady-state data eligibility determination module 106. This mechanism has been described in the net power feedback calculation module 103 section. The request signal is released after the estimation task is completed.

[0116] The linkage protection module 109 receives the new coefficients and estimation completion flags output by the online coefficient estimation module 108. When the coefficients need to be updated, it executes the transition protection process, safely writes the new coefficients into the variable operating condition feedforward calculation module 102, and monitors the system response during the transition period. Based on the monitoring results, it decides whether to solidify the coefficients or roll them back to the state before the update.

[0117] The linkage protection module 109 internally maintains two coefficient storage areas. Storage area one stores the currently effective feedforward model coefficients, which are read and used by the variable operating condition feedforward calculation module 102 in each control cycle. Storage area two stores the feedforward model coefficients that have been verified in the previous version, as rollback backup values. When the system is initially powered on, the same set of initial calibration coefficients is written to both storage areas.

[0118] When the online coefficient estimation module 108 completes the estimation and outputs new coefficients, the linkage protection module 109 extracts the current dead zone width of the power adaptive adjustment controller 104, expands the dead zone width, and writes it back. The expansion factor is a preset constant with a value greater than 1. The expansion factor is set as follows: during the debugging phase, based on the maximum step change in feedforward power caused by the variation of the feedforward model coefficients within a reasonable range, the ratio of this change to the dead zone width is rounded up and then incremented by one as the expansion factor, so that the expanded dead zone width can cover the feedforward power deviation caused by the writing of new coefficients. After the dead zone is expanded, the response sensitivity of the power adaptive adjustment controller 104 to the net power gain signal is temporarily reduced, which can tolerate the feedforward power calculation deviation in the short term after the new coefficients are written.

[0119] The linkage protection module 109 then writes the new coefficients into the coefficient storage area of ​​the variable operating condition feedforward calculation module 102, overwriting the currently effective coefficients. The variable operating condition feedforward calculation module 102 uses the new coefficients to calculate the feedforward power target value starting from the next control cycle.

[0120] After the coefficients are written, the linkage protection module 109 starts a transition monitoring timer with a preset duration. The transition monitoring duration is an integer multiple of the sliding window length, ensuring that the net power gain signal completes at least several full sliding window updates during the monitoring period. During the timer's operation, the linkage protection module 109 reads the net power gain signal output by the net power feedback calculation module 103 in each control cycle and compares its absolute value with the expanded dead zone width. Simultaneously, the linkage protection module 109 maintains an out-of-bounds counter to record the number of times the absolute value of the net power gain signal continuously exceeds the expanded dead zone width.

[0121] The threshold for consecutive out-of-bounds events is set as follows: under the assumption that the net power gain signal contains only zero-mean random noise, the probability of consecutive out-of-bounds events occurring is lower than the minimum number of consecutive out-of-bounds events corresponding to the preset probability value.

[0122] If the out-of-bounds counter does not reach the preset threshold for consecutive out-of-bounds occurrences during the entire transition monitoring period, the transition is considered successful. The linkage protection module 109 updates the spare coefficients in storage area two to the newly written coefficient values ​​and restores the dead zone width of the power adaptive regulation controller 104 to its original value before expansion. The transition monitoring timer is cleared.

[0123] If the out-of-bounds counter reaches the preset threshold for consecutive out-of-bounds occurrences within the transition monitoring period, a transition anomaly is determined. The linkage protection module 109 immediately performs the following rollback operations: writes the previous version of coefficients stored in storage area two back to the coefficient storage area of ​​the variable operating condition feedforward calculation module 102, restoring the coefficients before the update; restores the dead zone width of the power adaptive control controller 104 to its original value before expansion; sends a sample clearing command to the sample pool management module 107, carrying a cutoff timestamp, notifying the sample pool management module 107 to remove the sample points involved in this estimation; and sends a self-calibration anomaly notification to the outside world through the PLC output module. The transition monitoring timer and the out-of-bounds counter are reset.

[0124] After a successful transition or rollback operation, the linkage protection module 109 releases the reference time freeze request sent to the net power feedback calculation module 103 and outputs a reference time reset signal to the net power feedback calculation module 103. Upon receiving this signal, the net power feedback calculation module 103 updates the reference time to the current time in the next control cycle, and simultaneously latches the current host power sliding average and blower power sliding average as reference benchmarks for subsequent net power gain calculations.

[0125] The linkage protection module 109 records the timestamp of each coefficient update and the transition result as an update log, which is stored in the controller's non-volatile storage area. The update log includes the update sequence number, update timestamp, coefficient before update, coefficient after update, final effective coefficient, and transition determination result.

[0126] Reference Figure 4 and Figure 6 The system operates on two time scales: the real-time control loop runs at control cycles, while the self-correction loop is triggered at intervals much longer than the control cycles. Both operate in parallel within the same controller, cooperating through data sharing and timing coordination mechanisms.

[0127] The real-time control loop consists of a variable operating condition feedforward calculation module 102, a net power feedback calculation module 103, a power adaptive regulation controller 104, and a constraint protection module 105. Within each control cycle, it completes all calculations including data acquisition, feedforward calculation, net power gain calculation, incremental PI regulation, and constraint protection, and outputs the final power command to the blower inverter. The control cycle is a preset ratio of the larger of the inverter's response time constant and the time constant corresponding to the grid's allowable power regulation rate; this ratio does not exceed one-tenth.

[0128] The self-calibration loop consists of a steady-state data qualification module 106, a sample pool management module 107, an online coefficient estimation module 108, and a linkage protection module 109. It does not participate in the generation of power commands for each control cycle, but only performs re-estimation and updating of the feedforward model coefficients when trigger conditions are met. The triggering period of the online coefficient estimation module 108 is no less than tens of times the control cycle and no less than several times the sliding window length, thus separating the self-calibration action from real-time power regulation in time and avoiding periodic disturbances to the real-time control loop caused by parameter updates.

[0129] Data interaction between the two loops is achieved through the controller's global variable area. The variable operating condition feedforward calculation module 102 writes the calculated feedforward power target value into the global variable area in each control cycle, the net power feedback calculation module 103 writes the net power gain signal and reference time data into the global variable area, and the constraint protection module 105 writes the final power command into the global variable area. The steady-state data qualification determination module 106 and the sample pool management module 107 read the above data from the global variable area without needing to perform a synchronization handshake with the real-time control loop.

[0130] The improvement of the self-calibrating loop on the real-time control loop is reflected in the continuous improvement of the accuracy of the feedforward model. After the online coefficient estimation module 108 completes the coefficient re-estimation based on the high-confidence steady-state sample and safely writes the new coefficients into the variable operating condition feedforward calculation module 102 through the linkage protection module 109, the feedforward power target value output by the feedforward channel is closer to the true optimal operating point. At this time, the correction amount required by the feedback channel in the power adaptive regulation controller 104 is reduced, the fluctuation amplitude of the incremental output is reduced, and the system remains within the dead zone for a longer period of time. Stable operation within the dead zone reduces the frequency of ineffective adjustment of wind turbine power and also creates conditions for the next high-confidence steady-state determination and sample acquisition.

[0131] The coordination between the two loops is also reflected in the linkage protection. When the online coefficient estimation module 108 starts the estimation task, it outputs a reference time freeze request signal to the net power feedback calculation module 103 through the steady-state data qualification judgment module 106. The net power feedback calculation module 103 suspends the reference time update to ensure that the changes in propulsion power and wind turbine power during the entire estimation and transition period are calculated based on the same reference time. Before updating the coefficients, the linkage protection module 109 expands the dead zone width of the power adaptive adjustment controller 104 to reduce the response intensity of the feedback loop to possible short-term deviations in the early stage of model switching. During the transition monitoring, if the linkage protection module 109 detects that the net power gain signal continuously exceeds the limit, it immediately rolls back the coefficients and clears the relevant samples to prevent inaccurate model coefficients from affecting the adjustment direction of the real-time control loop for a long time. After the transition is successful or the rollback is completed, the linkage protection module 109 releases the reference time freeze request and sends a reset signal. The net power feedback calculation module 103 updates the reference time to the current time. Since the parameters of the feedforward model have changed after the coefficient update, the power balance point corresponding to the original reference time is no longer applicable. Resetting the reference time ensures that subsequent net power gain calculations are based on the new steady state, guaranteeing that the calculation benchmarks for propulsion power changes and turbine power changes match the new model. This mechanism ensures that parameter updates in the self-calibrating loop do not disrupt the stability of the real-time control loop.

[0132] Example 2: See attached document Figure 5 , Figure 5 This is a flowchart illustrating a method for adaptive power adjustment of a marine blower with data feedback, according to an embodiment of the present invention. The method is implemented based on the same industrial programmable logic controller (PLC), which is connected to a speed sensor, draft sensor, back pressure transmitter, power transmitter, ship engine room monitoring system data bus, and blower frequency converter via analog input modules and digital communication interfaces, respectively. The specific steps are as follows: S1. In each control cycle, the ship's speed in the water, draft at the bow, draft at the stern, back pressure in the air jet nozzle inlet pipe, real-time power consumption of the blower, and output power of the ship's main engine are collected synchronously. The average draft and trim angle are calculated from the draft at the bow and the draft at the stern.

[0133] S2. Based on the currently effective ship type coefficient, speed index, and draft index, calculate the theoretical optimal jet flow rate by combining the speed and average draft. Then, correct the theoretical optimal jet flow rate using the back pressure compensation function. Combine the corrected flow rate with the total pipeline pressure loss and the correspondence between efficiency and flow rate in the fan characteristic curve to calculate the feedforward power target value.

[0134] S3. Perform sliding window filtering on the main unit power and blower power respectively to obtain the sliding average value of the main unit power and the sliding average value of the blower power; Maintain a baseline time and store the corresponding sliding average power of the main unit and the sliding average power of the blower at that baseline time; In each control cycle, the difference between the current host power sliding average value and the host power sliding average value at the reference time is calculated to obtain the propulsion power change. Calculate the difference between the current blowing power sliding average value and the blowing power sliding average value at the reference time to obtain the change in blowing power; The net power gain signal is obtained by subtracting the change in propulsion power from the change in blower power.

[0135] S4. Using the net power gain signal as the controlled variable and the feedforward power target value as the base component, execute the incremental PI control law with dead zone to generate the power command increment, and superimpose the feedforward power target value and the power command increment to obtain the initial power command. When the absolute value of the net power gain signal does not exceed the dead zone width, the power command increment remains unchanged from the value of the previous control cycle.

[0136] S5. The initial power command is processed sequentially by back pressure-surge boundary constraint, minimum power maintenance constraint and power command change rate constraint to obtain the final power command output to the blower frequency converter.

[0137] S6. While executing steps S1 to S5 in real time, each control cycle determines whether the following conditions are met simultaneously: the absolute value of the net power gain signal is continuously within the dead zone, the rate of change of the final power command is lower than the preset steady-state rate threshold, and the speed fluctuation range and the average draft fluctuation range do not exceed the preset speed tolerance and draft tolerance, respectively. When the duration of all conditions being met simultaneously reaches a preset steady-state duration threshold, the current time period is determined to be a high-confidence steady-state period. The steady-state rate threshold is determined based on the dead zone width and the sliding window length, ensuring that the cumulative change in power command within the next sliding window duration at that rate does not exceed a specified proportion of the dead zone width; The speed tolerance and draft tolerance are determined based on the partial derivatives of the feedforward model with respect to speed and average draft, as well as the dead zone width, so that the change in feedforward power caused by changes in speed or draft within the corresponding tolerance does not exceed the dead zone width.

[0138] S7. After the high-confidence steady-state period ends, calculate the arithmetic mean of the target values ​​of speed, average draft, back pressure and feedforward power during the period, and combine it with the timestamp to form a sample point, which is then stored in a rolling sample pool of fixed capacity. The sample pool adopts a first-in-first-out queue structure. When the sample pool is full and a new sample point arrives, the sample point with the earliest timestamp is removed.

[0139] S8. When the number of sample points in the sample pool reaches the preset capacity, or when the time since the last successful parameter update operation exceeds the preset self-correction period, the parameter estimation task is triggered: take all sample points from the sample pool, take the logarithm of the feedforward model to obtain the linearized model, and use the weighted least squares method with forgetting factor to re-estimate the ship form coefficient, speed index and draft index. The weight of a sample point decays exponentially with the sample timestamp, according to the forgetting factor. The estimated coefficients are subjected to a reasonableness boundary test. Estimated values ​​that exceed the preset range are discarded and the original coefficients are kept unchanged.

[0140] S9. Before writing the new coefficients that have passed the rationality test into the feedforward calculation stage, the dead zone width in step S4 is multiplied by the preset expansion factor. After writing the new coefficients, start the transition monitoring timer to continuously monitor whether the absolute value of the net power gain signal continuously exceeds the expanded dead zone width within the preset transition monitoring duration. If no consecutive boundary crossings occur, the transition is considered successful, the dead zone width is restored to its original value, and the new coefficient is fixed. If consecutive out-of-bounds errors occur, the transition is deemed abnormal. The coefficients are immediately rolled back to their values ​​before the update, the dead zone width is restored to its original value, all sample points participating in this estimation are removed from the sample pool, and a fault notification is issued.

[0141] S10. After step S9 is completed, the reference time used to calculate the power change in step S3 is reset to the current time, and the current sliding average value of the host power and the sliding average value of the blower power are latched as the new reference benchmark.

[0142] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0143] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented in software, the above embodiments can be implemented, in whole or in part, as a computer program product. Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution.

[0144] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0145] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A marine wind turbine system with adaptive power adjustment based on data feedback, characterized in that, include: The multi-source data acquisition module is used to synchronously acquire the ship's speed in the water, draft at the bow, draft at the stern, nozzle back pressure, blower power consumption, and main engine output power in each control cycle, and to calculate the average draft. The variable operating condition feedforward calculation module is used to calculate the theoretical jet flow rate based on the currently effective ship type coefficient, speed index and draft index, combined with the speed and the average draft. After being corrected by the back pressure compensation function, the feedforward power target value is calculated by combining the total pipeline pressure loss and the fan characteristic curve. The net power feedback calculation module is used to perform sliding window filtering on the output power of the host and the power consumed by the blower, respectively, and calculate the change in propulsion power and the change in blower power with reference to the reference time, and take the difference between the two as the net power gain signal. A power adaptive adjustment controller is used to generate a power command increment by using the feedforward power target value as the base component and the net power gain signal as the controlled variable, and to superimpose the base component and the increment to form an initial power command. The constraint protection module is used to sequentially execute back pressure-surge boundary limit, minimum power maintenance constraint and power command change rate limit on the initial power command to obtain the final power command and output it to the blower frequency converter; The steady-state data qualification determination module is used to determine that the system has entered a high-confidence steady state under the following conditions: the absolute value of the net power gain signal is continuously within the dead zone, the rate of change of the final power command is lower than the preset steady-state power change rate threshold, and the speed fluctuation range and the average draft fluctuation range do not exceed the preset tolerance. The sample pool management module is used to calculate the average value of the speed, average draft, back pressure and the feedforward power target value during the high confidence steady state period after the end of the period, and add a timestamp to form sample points and store them in the rolling sample pool. The online coefficient estimation module is used to obtain sample points from the sample pool when the triggering conditions are met, and to re-estimate the ship type coefficient, speed index and draft index using the weighted least squares method with forgetting factor, and output new coefficients after passing the rationality boundary test. The linkage protection module is used to expand the dead zone of the power adaptive adjustment controller before writing the new coefficient into the variable operating condition feedforward calculation module. After writing, it monitors the boundary of the net power gain signal within a preset transition time. If the transition is smooth, the dead zone is restored and the new coefficient is fixed. If the boundary is exceeded, the coefficient is rolled back and the dead zone is restored.

2. The marine wind turbine system with adaptive power adjustment and data feedback according to claim 1, characterized in that, The variable operating condition feedforward calculation module calculates the target value of feedforward power in the following manner: In the formula, Theoretical jet flow rate, For ship type coefficient, For speed, For speed index, To achieve average draft, Draft index, For the width of the boat; This is the jet flow rate after back pressure correction. This is the back pressure compensation function. For nozzle back pressure, Reference back pressure; The target value for feedforward power. For the total pressure loss of the pipeline, The efficiency value corresponding to the corrected jet flow rate in the fan characteristic curve is denoted as .

3. A marine wind turbine system with adaptive power adjustment and data feedback as described in claim 1, characterized in that, In the power adaptive regulation controller, the incremental PI control law with dead zone is: when hour, ; when hour, ; when hour, ; In the formula, This is the net power gain signal. dead zone width, For power command increment, and For positive adjustment of proportional gain and integral gain, and For inverse adjustment of proportional gain and integral gain, To control the cycle, the proportional gain and integral gain of the reverse adjustment are greater than or equal to the corresponding gains of the forward adjustment.

4. A marine wind turbine system with adaptive power adjustment and data feedback as described in claim 1, characterized in that, The net power feedback calculation module includes a reference time update mechanism: When the incremental output of the power adaptive adjustment controller is continuously within a preset narrow range and remains within a preset duration, the current moment is recorded as a new reference moment, and the sliding average value of the host power and the sliding average value of the blower power at that moment are latched, overwriting the original reference data.

5. A marine wind turbine system with adaptive power adjustment and data feedback according to claim 4, characterized in that, The net power feedback calculation module also includes a reference time freezing mechanism: When the system is in the high-confidence steady-state period or when the online coefficient estimation module is performing parameter estimation, the update of the reference time is paused.

6. A marine wind turbine system with adaptive power adjustment and data feedback according to claim 1, characterized in that, In the steady-state data qualification determination module, the preset steady-state power change rate threshold is set based on the dead zone width and the sliding window length, so that the cumulative change of power command within the next sliding window time does not exceed a specified proportion of the dead zone width. The preset speed tolerance and draft tolerance are determined based on the partial derivatives of the feedforward power target value with respect to speed and average draft, and the dead zone width, respectively, so that the amount of feedforward power change caused by changes in speed or draft within the corresponding tolerance does not exceed the dead zone width.

7. A marine wind turbine system with adaptive power adjustment and data feedback according to claim 1, characterized in that, The triggering condition for the online coefficient estimation module is: The sample pool is considered complete when the number of sample points reaches the preset capacity or when the time since the last successful estimation exceeds the preset self-correction period, whichever comes first.

8. A marine wind turbine system with adaptive power adjustment and data feedback according to claim 1, characterized in that, After taking the logarithm of the feedforward model and converting it into a linear form, the online coefficient estimation module assigns weights to each sample point according to the exponential forgetting factor decay. It then uses the weighted least squares method to jointly estimate the intercept parameter, speed index, and draft index of the merged constant term, and separates the updated ship form coefficients from the intercept parameter.

9. A marine wind turbine system with adaptive power adjustment and data feedback according to claim 1, characterized in that, After executing the coefficient rollback, the linkage protection module sends a sample clearing instruction to the sample pool management module, which then deletes all sample points whose timestamps are later than or equal to the latest timestamp of the samples used in this estimation.

10. A method for adaptively adjusting the power of a marine wind turbine with data feedback, characterized in that, A marine wind turbine system with adaptive power adjustment and data feedback as described in any one of claims 1-9, comprising the following steps: S1. Synchronously collect data on the ship's speed in the water, draft at the bow, draft at the stern, nozzle back pressure, blower power consumption, and main engine output power during each control cycle, and calculate the average draft. S2. Based on the ship type coefficient, speed index and draft index, and combined with the speed and average draft, calculate the theoretical jet flow rate, correct it with the back pressure compensation function, and combine it with the total pipeline pressure loss and the fan characteristic curve to calculate the feedforward power target value. S3. After sliding filtering of the main unit power and blower power, calculate the change in propulsion power and the change in blower power with reference to the reference time, and take the difference between the two as the net power gain signal. S4. Using the net power gain signal as the controlled variable and the feedforward power target value as the basic component, an incremental PI control law with dead zone is used to generate the superimposed initial power command. S5. The initial power command is sequentially subjected to back pressure-surge boundary constraint, minimum power maintenance constraint and power command change rate constraint to obtain the final power command output to the blower frequency converter. S6. While executing steps S1 to S5 in real time, if it is determined that the net power gain signal is continuously within the dead zone, the power command change rate is lower than the preset threshold, and the speed and draft fluctuations do not exceed the tolerance, and the above conditions continue to reach the steady state duration threshold, it is determined that the high confidence steady state has been entered. S7. After the high-confidence steady-state period ends, calculate the average values ​​of the target values ​​of speed, average draft, back pressure and feedforward power during that period, add timestamps to form sample points, and store them in the rolling sample pool. S8. When the number of sample points reaches the preset capacity or the time since the last successful parameter update exceeds the self-correction cycle, parameter estimation is triggered. After taking the logarithm of the feedforward model, the weighted least squares method with forgetting factor is used to re-estimate the ship type coefficient, speed index and draft index. After passing the rationality boundary test, the new coefficients are output. S9. Before writing the new coefficients that have passed the test into the feedforward calculation stage, the dead zone is expanded. After writing, the net power gain signal is monitored within the preset transition time to see if it continuously exceeds the limit. If the transition is smooth, the dead zone is restored and the new coefficients are solidified. If it exceeds the limit, the coefficients are rolled back, the dead zone is restored, and the corresponding samples are deleted. S10. After the coefficient update or rollback is completed, the reference time used to calculate the power change is reset to the current time, and the current host power sliding average and blower power sliding average are latched as the new reference benchmark.

Images