Hybrid adaptive vibration damping method and device for ship thrust bearing oil film thermoelectric energy harvesting

Abstract

This invention discloses a hybrid adaptive vibration reduction method and device for oil film thermoelectric energy harvesting in ship thrust bearings. A thermoelectric energy harvesting device is embedded within the thrust bearing. Utilizing the shear temperature rise of the oil film to create a temperature difference, waste heat is converted into electrical energy through the Seebeck effect. After energy storage, this energy supplies power to sensors, control modules, and actuators. The longitudinal vibration signal of the thrust ring is collected and transmitted to the control module. A time-varying dynamic model is established, and oil film parameters are inferred from the generated voltage to provide real-time input for the algorithm. Control signals are generated using online identification of the control channel, reconstruction of internal model reference signals, and a hybrid algorithm combining neural networks and LMS. The actuator outputs equal-amplitude, anti-phase compensation displacement to suppress vibration. Vibration suppression improves thermoelectric conversion efficiency, forming a positive closed loop of energy harvesting, vibration reduction, and power supply. The entire system, including sensors, controllers, and actuators, is energy-self-sufficient, requiring no external power supply. It exhibits good steady-state vibration suppression and anti-interference capabilities, making it suitable for long-period, variable-condition intelligent vibration reduction requirements in ships.

Description

Technical Field

[0001] This invention relates to the field of ship shafting vibration control and energy harvesting technology, specifically to a hybrid adaptive vibration reduction device and method for oil film thermoelectric energy harvesting in ship thrust bearings. Background Technology

[0002] As a core component of energy dissipation in propulsion shafting, the thrust bearing of a ship experiences significant temperature rise in its oil film during high-frequency shearing. Existing vibration reduction technologies mostly focus on addressing a single energy domain: mechanical solutions dissipate vibration energy through structural optimization or damping materials; electromagnetic solutions use induction coils to cut magnetic field lines to obtain electrical energy, but require additional permanent magnet assemblies and air gap structures, increasing axial space occupation and assembly complexity; purely active control methods rely entirely on external power sources, neglecting the potential for vibration energy recovery. None of these technologies achieve a closed-loop coupling between energy flow and information flow, leading to a trade-off between energy harvesting efficiency and vibration reduction effectiveness.

[0003] Longitudinal vibrations caused by propeller pulsating thrust, main engine imbalance, and wave impact, coupled with the high-frequency, high-amplitude shearing of the thrust bearing oil film during propulsion shaft operation, pose significant challenges. While adaptive control methods based on the filter-leaved least mean square (FxLMS) algorithm are widely used in active vibration control, they still face numerous problems for longitudinal vibrations in ship propulsion shafts. Variable operating conditions during ship navigation, such as speed and load changes, alter the stiffness and damping of the thrust bearing oil film, leading to time-varying or even abrupt changes in the control channel characteristics. This forces traditional algorithms to employ small convergence factors to maintain stability, ultimately limiting control performance and convergence speed.

[0004] With the development of artificial intelligence (AI) technology, many new intelligent vibration active control algorithms have emerged. AI methods typically use large amounts of data for offline training. While the training phase can be time-consuming, their forward inference in the control phase requires very little time. Leveraging the powerful redundancy, superior approximation, and memory capabilities of neural networks, supervised training guided by reinforcement learning under various environments can significantly improve the network's adaptability and robustness to frequency changes. On the other hand, in practical vibration control, the longitudinal vibration reference signal of a ship's shafting is not easily obtained, and the reference signal is affected by the canceled vibration signal, leading to inaccurate reference signals measured by sensors. Therefore, conventional adaptive filtering vibration control methods face significant problems in the practical application of ship shafting. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to address the above-mentioned deficiencies of the prior art by providing a hybrid adaptive vibration reduction method and device for oil film thermoelectric energy harvesting in marine thrust bearings, thereby solving at least one of the following technical problems: Existing energy harvesting and vibration reduction devices occupy a large space and have poor compatibility with the thrust bearings of existing ships.

[0006] Changes in ship speed, load, and rotational speed under varying operating conditions cause time-varying oil film stiffness / damping, resulting in abrupt changes in control channel characteristics. Traditional FxLMS algorithms suffer from slow convergence, poor stability, and reduced vibration suppression effectiveness.

[0007] Longitudinal vibration of the shaft system reduces the thermocouple fit. The thermoelectric energy harvesting efficiency and vibration reduction effect are mutually restrictive, and an energy-information-control closed loop is not formed.

[0008] Traditional vibration control reference signals are difficult to obtain accurately, and additional noise can introduce residual vibrations, resulting in insufficient system robustness. The shearing of oil film in ship thrust bearings generates thousands of watts of waste heat that is not recovered, resulting in energy waste. Furthermore, traditional active vibration damping relies on external power supply, which is unsustainable.

[0009] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows: A hybrid adaptive vibration reduction method for oil film thermoelectric energy harvesting in marine thrust bearings includes: S1: A thermoelectric energy harvesting device is embedded in the thrust bearing. It uses the temperature rise of oil film shear to form a temperature difference, and converts waste heat into electrical energy through the Seebeck effect. After energy storage, it provides self-powered power to sensors, control modules and actuators. S2: The vibration sensor collects the longitudinal vibration signal of the thrust ring and transmits it to the control module; S3: The control module establishes a time-varying dynamic model and uses the generator voltage to infer the oil film parameters, providing real-time input for the algorithm; S4: Control signals are generated using online identification of control channels, reconstruction of internal model reference signals, and a hybrid algorithm combining neural networks and LMS. S5: Drives the actuator to output equal amplitude and opposite phase to compensate for displacement and suppress vibration; S6: Vibration suppression improves thermoelectric conversion efficiency, forming a positive closed loop of energy capture-vibration reduction-power supply.

[0010] In the above technical solution, step S1 utilizes the oil film shear temperature rise to form a stable temperature difference between the hot end and the cold end.

[0011] In the above technical solution, S3: The control module establishes a time-varying dynamic model of the interference channel and the control channel based on the time-varying characteristics of the oil film stiffness and damping as a function of rotational speed and thrust. At the same time, it uses the thermoelectric voltage signal to infer the oil film operating parameters and provides real-time time-varying input for the control algorithm.

[0012] In the above technical solution, S4: The control module adopts a hybrid algorithm that integrates online identification of the control channel, reconstruction of the internal model reference signal, neural network reinforcement learning and LMS adaptive filtering to calculate and generate an adaptive control signal.

[0013] In the above technical solution, S5: the control signal drives the actuator to output a compensation displacement that is equal in amplitude and opposite in phase to the vibration displacement, thereby offsetting the longitudinal vibration of the ship's propulsion shaft system.

[0014] In the above technical solution, S6: Vibration suppression improves the fit between the thermocouple and the thrust pad, providing more stable power to the control system and forming a positive closed loop of "energy capture-vibration reduction-power supply".

[0015] In the above technical solution, the control channel filter is updated in real time according to the steepest gradient descent method, and the auxiliary filter is updated synchronously and iteratively, so that the control channel can be identified online without the need for additional random noise. In the above technical solution, in step S4, the reference signal is reconstructed based on the internal model control principle, and the output of the identified control channel filter is superimposed with the residual vibration error signal to obtain the controller reference input signal. In the above technical solution, in step S4, the neural network controller adopts supervised training guided by reinforcement learning, and the network input and output terminals are respectively set with state normalization and denormalization processing. The training adopts policy gradient and experience replay mechanism, and the network parameters are not updated in the real-time control stage. In the above technical solution, in step S4, the controller uses the LMS algorithm to update, taking the reference signal after channel filtering and the error signal output by the analog control channel as input, and iteratively generating the optimal control signal.

[0016] A hybrid adaptive vibration reduction device for oil film thermoelectric energy harvesting in a ship thrust bearing is installed inside the ship thrust bearing, including a thermoelectric energy harvesting device, a vibration sensing module, an adaptive actuation module, and a control module. The vibration sensing module includes a vibration sensor fixed to the outer surface of the thrust ring, and the adaptive actuation module includes actuators symmetrically arranged on opposite sides of the thrust ring and the thrust disk, with axial output force in the same direction as the longitudinal vibration of the bearing; an oil film is formed between the thrust ring and the thrust disk, and thrust pads are evenly distributed circumferentially on one side of the thrust disk. The thermoelectric energy harvesting device is embedded in the thrust bearing. Its hot end is connected to the oil film for thermal conduction, and its cold end is connected to the ship's lubrication circulation system. Its output end supplies power to the sensor, actuator and control module through the energy storage module. The vibration sensor collects the vibration signal of the thrust ring and transmits it to the control module. The control module drives the actuator to generate a compensating displacement that is equal in amplitude but opposite in phase to the vibration.

[0017] In the above technical solution, the axial output direction of the actuator is consistent with the longitudinal vibration direction of the thrust bearing.

[0018] In the above technical solution, the right end of rotor one is fixedly connected to the surface of the thrust ring, the left end of rotor two is fixedly connected to the surface of the thrust disk, and thrust bearings are evenly distributed circumferentially on the side of the thrust disk away from rotor two. An evenly distributed oil film is formed between the opposite end faces of the thrust ring and the thrust disk.

[0019] In the above technical solution, the power output terminal of the thermoelectric energy harvesting device is electrically connected to the energy storage module through a sealed wire.

[0020] In the above technical solution, the energy storage module is connected to the vibration sensor, actuator, and control module for power supply respectively; the signal output terminal of the vibration sensor is connected to the control module for signal acquisition of the real-time vibration displacement signal of the thrust ring and transmission to the control module; the control signal output terminal of the control module is connected to the actuator drive, and drives the actuator to generate a compensation displacement with the same amplitude and opposite phase as the vibration displacement by outputting an adaptive control signal. In the above technical solution, the installation support structure of the thrust pad includes a limiting component, a support transmission component, and a balance bearing component; the limiting component includes a limiting device one and a limiting device two fixed to the surface of the thrust plate, the limiting device one and the limiting device two being symmetrically arranged on both sides of the thrust pad to limit the radial displacement of the thrust pad; the support transmission component includes an upper support block, a pad fulcrum, and a lower support block, the top end of the upper support block being fixedly connected to the bottom end of the thrust pad, the bottom end of the upper support block being hinged to the balance bearing component through the pad fulcrum, the bottom end of the balance bearing component being fixedly connected to the lower support block, and the bottom end of the lower support block being fixedly connected to the surface of the thrust plate; the balance bearing component includes an upper flat... The structure comprises a balance block, a lower balance block fulcrum one, a lower balance block fulcrum two, a balance block fulcrum one, and a balance block fulcrum two; the top of the upper balance block is hinged to the bearing fulcrum, and the bottom of the upper balance block is fixedly connected to the lower support block; the two sides of the lower support block are fixedly connected to the lower balance block fulcrum one and the lower balance block fulcrum two respectively through the lower balance block fulcrum one and the lower balance block fulcrum two; the bottom ends of the balance block fulcrum one and the balance block fulcrum two are fixedly connected to the surface of the thrust disk; the support transmission component and the balance bearing component form an elastic bearing structure to buffer the vibration impact of the thrust bearing, while ensuring the relative positional stability of the thermoelectric energy harvesting device and the oil film, and its stiffness parameters are adapted to the natural vibration frequency of the ship's propulsion shaft system.

[0021] In the above technical solution, the thermoelectric energy harvesting device includes a hot node, a P-type thermocouple, an N-type thermocouple, a cold node, a heat insulation baffle, and a cooling oil cavity; a mounting groove is reserved on the back side of each thrust pad, and the thermoelectric energy harvesting device is integrally embedded in the groove. The heat insulation baffle surrounds the P-type and N-type thermocouples, forming a sealed cooling oil cavity with the inner wall of the groove of the single thrust pad; the hot node is fixed to the top of the P-type and N-type thermocouples, and the distance between the top surface of the hot node and the working surface of the thrust pad is less than 2mm, for directly capturing the shear temperature rise of the oil film; the cold node is fixed to the P-type and N-type thermocouples. At the bottom of the thermocouple, the cold node is completely embedded in the cooling oil chamber. The cooling oil chamber is connected to the ship's lubrication circulation system through a preset oil circuit, forming a continuous hot oil discharge and cold oil replenishment circuit. Conductive wires are connected to both ends of the cold node. A sealable wire hole is provided on the thrust bearing. After the wire passes through the sealable wire hole, it is electrically connected to the voltage sensor. The voltage sensor is electrically connected to the energy storage module to stably input the electrical energy generated by thermoelectric power generation into the energy storage module. The signal output terminal of the voltage sensor is connected to the control module to transmit the real-time collected power generation voltage signal to the control module. The control module uses the voltage signal to infer the shear power, stiffness, and damping variation parameters of the oil film, providing time-varying parameter input for the adaptive control algorithm.

[0022] Compared with the prior art, the beneficial effects of this invention are: This innovative approach deeply integrates oil film shear heat conversion with intelligent vibration control. By utilizing the oil film temperature rise of the thrust bearing to construct a "hot-cold" node, the shear heat of the oil film, amounting to several kilowatts, is continuously converted into electrical energy through the Seebeck effect. This enables the entire system of sensors, controllers, and actuators in active vibration control to be energy self-sufficient, requiring no external power supply.

[0023] In vibration control systems, a time-varying dynamic model of the disturbance channel and the control channel is established by simplifying the model. This model reflects the changes in parameters such as load and speed in real time, providing a basis for accurate time-varying parameter prediction for the control algorithm and significantly improving the matching accuracy between the control strategy and the actual working conditions.

[0024] By deeply integrating the online identification method of control channels based on auxiliary filters with the internal model control reference signal reconstruction technology, and supplementing it with the powerful approximation and memory capabilities of neural networks and the adaptive real-time iteration capability of filters, the steady-state vibration suppression effect and anti-interference performance of the control are significantly improved.

[0025] The vibration control effect and thermoelectric conversion efficiency form a positive feedback loop: vibration suppression improves the thermocouple fit, thereby increasing power generation efficiency and ultimately providing a more stable energy source for the adaptive controller.

[0026] This invention achieves energy recovery through temperature difference energy harvesting, and realizes performance adaptation by relying on adaptive filtering and neural network control. The three form a "energy-information-control" closed loop, which is particularly suitable for the intelligent vibration reduction needs of long-cycle and variable operating conditions of ship propulsion systems. Attached Figure Description

[0027] The present invention will be further described below with reference to the accompanying drawings and embodiments. In the accompanying drawings: Figure 1 This is a schematic diagram of the overall structure of the propulsion shaft system in an embodiment of the present invention; Figure 2 This is a distribution diagram of the thrust bearings in an embodiment of the present invention; Figure 3 This is a structural diagram of the thrust bearing in an embodiment of the present invention; Figure 4 This is a distribution diagram of the energy harvesting modules in an embodiment of the present invention; Figure 5 This is a structural diagram of the energy harvesting module in an embodiment of the present invention; Figure 6 This is a control block diagram of the neural network-feedback FxLMS algorithm in an embodiment of the present invention; Figure 7 This is a supervised learning graph of the neural network under the guidance of deep learning in an embodiment of the present invention; Figure 8 This is a flowchart illustrating the active control of longitudinal vibration of the ship's propulsion shaft system in an embodiment of the present invention.

[0028] The corresponding labels in the attached diagram are as follows: 1-Rotor 1; 2-Rotor 2; 3-Thrust ring; 4-Oil film; 5-Thrust pad; 6-Hot node; 7-P-type thermocouple; 8-N-type thermocouple; 9-Cold node; 10-Sensor; 11-Actuator; 12-Thrust disc; 13-Limit device 1; 14-Limit device 2; 15-Pad fulcrum; 16-Lower balance block fulcrum 1; 17-Balance block fulcrum 1; 18-Upper balance block; 19-Lower balance block fulcrum 2; 20-Balance block fulcrum 2; 21-Upper support block; 22-Lower support block; 23-Baffle; 24-Cooling oil chamber; 25-Wire; 26-Sealable wire hole; 27-Battery; 28-Voltage sensor; 501-Single thrust pad. Detailed Implementation

[0029] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0030] It should be noted that, depending on the implementation needs, the various steps / components described in this application can be broken down into more steps / components, or two or more steps / components or parts of the operation of steps / components can be combined into new steps / components to achieve the purpose of this invention.

[0031] Example 1 like Figures 1-8 As shown, this embodiment provides a hybrid adaptive vibration reduction device and method for oil film thermoelectric energy harvesting in marine thrust bearings, including the following core steps: 1. Integrated Structure A thermoelectric energy harvesting device is embedded inside the thrust pad, with the hot end close to the oil film and the cold end connected to the lubrication circulation system to form a stable temperature difference; a matching elastic support structure ensures stable thermocouple contact. The thrust bearing has a tilting pad structure with eight thrust pads evenly distributed circumferentially. Figure 2 On the back side of each thrust bearing 5, away from the oil film 4, the upper support block 21, bearing fulcrum 15, upper balance block 18, and lower support block 22 are stacked in sequence. The load is then transferred to the thrust disc 12 through the lower balance block fulcrum 116, lower balance block fulcrum 219, balance block fulcrum 117, and balance block fulcrum 220, forming a complete force flow path. Figure 3 Simultaneously, the potential difference generated by the temperature difference between the hot and cold nodes is stored in the battery via wires, serving as the energy source for subsequent sensors and actuators. An oil film 4 is formed between the opposing end faces of the thrust ring 3 and the thrust disk 12. The thickness, stiffness, and damping of the oil film 4 change in real time with variations in rotational speed and thrust, constituting the main source of the time-varying characteristics of the control channel. A ring-shaped actuator 11 is fixedly installed on the outer cylindrical surface of the thrust ring 3. Figure 1 The axial output direction of actuator 11 is consistent with the longitudinal vibration direction of the thrust bearing, and is used to output the main power with the same amplitude but opposite phase as the external excitation. This forms an integrated ship thrust bearing oil film thermoelectric energy harvesting hybrid adaptive vibration reduction device, which is set inside the ship thrust bearing 10 and includes a thermoelectric energy harvesting device, a vibration sensing module and an adaptive actuation module.

[0032] The vibration sensing module includes a vibration sensor 10 fixed to the outer surface of the thrust ring 3, and the adaptive actuation module includes actuators 11 symmetrically arranged on opposite sides of the thrust ring 3 and the thrust disk 12. The axial output direction of the actuators 11 is consistent with the longitudinal vibration direction of the thrust bearing. The right end of the rotor 11 is fixedly connected to the surface of the thrust ring 3, and the left end of the rotor 22 is fixedly connected to the surface of the thrust disk 12. Thrust pads 5 are evenly distributed circumferentially on the side of the thrust disk 12 away from the rotor 22. An evenly distributed oil film 4 is formed between the opposite end faces of the thrust ring 3 and the thrust disk 12. The thermoelectric energy harvesting device is embedded inside the thrust bearing 5. Its hot end forms heat conduction with the oil film 4, and its cold end forms heat communication with the ship's lubrication circulation system. The power output end of the thermoelectric energy harvesting device is electrically connected to the energy storage module through a sealed wire. The energy storage module is connected to the vibration sensor 10, the actuator 11 and the subsequent control module for power supply. The signal output terminal of the vibration sensor 10 is connected to the control module signal, which is used to collect the real-time vibration displacement signal of the thrust ring 3 and transmit it to the control module. The control signal output terminal of the control module is connected to the actuator 11 drive, and drives the actuator 11 to generate a compensation displacement with the same amplitude and opposite phase as the vibration displacement by outputting an adaptive control signal.

[0033] The mounting support structure of the thrust pad 5 includes a limiting component, a support transmission component, and a balance bearing component; the limiting component includes a limiting device 113 and a limiting device 214 fixed to the surface of the thrust plate 12, the limiting device 113 and the limiting device 214 being symmetrically arranged on both sides of the thrust pad 5 to limit the radial displacement of the thrust pad 5. The support and transmission assembly includes an upper support block 21, a tile fulcrum 15 and a lower support block 22. The top of the upper support block 21 is fixedly connected to the bottom of the thrust tile 5. The bottom of the upper support block 21 is hinged to the balance bearing assembly through the tile fulcrum 15. The bottom of the balance bearing assembly is fixedly connected to the lower support block 22. The bottom of the lower support block 22 is fixedly connected to the surface of the thrust plate 12. The balanced load-bearing assembly includes an upper balance block 18, a lower balance block fulcrum 116, a lower balance block fulcrum 219, a balance block fulcrum 117, and a balance block fulcrum 220. The top of the upper balance block 18 is hinged to the tile fulcrum 15, and the bottom of the upper balance block 18 is fixedly connected to the lower support block 22. The two sides of the lower support block 22 are fixedly connected to the balance block fulcrum 117 and the balance block fulcrum 220 respectively through the lower balance block fulcrum 116 and the lower balance block fulcrum 219. The bottom ends of the balance block fulcrum 117 and the balance block fulcrum 220 are fixedly connected to the surface of the thrust plate 12. The support and transmission components and the balance bearing components form an elastic bearing structure to buffer the vibration impact of the thrust bearing 5, while ensuring the relative positional stability of the thermoelectric energy harvesting device and the oil film 4. Its stiffness parameters are adapted to the natural vibration frequency of the ship's propulsion shaft system.

[0034] The thermoelectric energy harvesting device includes a hot node 6, a P-type thermocouple 7, an N-type thermocouple 8, a cold node 9, a heat insulation partition 23, and a cooling oil chamber 24. The back side of the single thrust pad 501 has a pre-reserved mounting groove, and the thermoelectric energy harvesting device is embedded in the groove. The heat insulation plate 23 surrounds the P-type thermocouple 7 and N-type thermocouple 8, and together with the inner wall of the groove of the single thrust pad 501, forms a sealed cooling oil cavity 24. The thermal node 6 is fixed to the top of the P-type thermocouple 7 and the N-type thermocouple 8. The distance between the top surface of the thermal node 6 and the working surface of the thrust bearing 501 is less than 2mm, which is used to directly capture the shear temperature rise of the oil film 4. The cold node 9 is fixed to the bottom end of the P-type thermocouple 7 and the N-type thermocouple 8. The cold node 9 is completely embedded in the cooling oil chamber 24. The cooling oil chamber 24 is connected to the ship's lubrication circulation system through a preset oil circuit to form a continuous hot oil discharge and cold oil replenishment circuit. The cold node 9 is connected to conductive wires 25 at both ends. The thrust pad 501 is provided with a sealable wire hole 26. The wire 25 passes through the sealable wire hole 26 and is electrically connected to the voltage sensor 28. The voltage sensor 28 is electrically connected to the energy storage module 27 and is used to stably input the electrical energy generated by the thermoelectric generator into the energy storage module 27. The signal output terminal of the voltage sensor 28 is connected to the control module signal, which is used to transmit the real-time acquired generator voltage signal to the control module. The control module uses the voltage signal to infer the shear power, stiffness and damping variation parameters of the oil film 4, and provides time-varying parameter input for the adaptive control algorithm.

[0035] The sensor 10 is fixedly connected to the outer surface of the thrust ring 3, and the battery 27 is fixedly connected to the outer surface of the thrust disk 12. The battery 27 is connected to the voltage sensor 28.

[0036] 2. Oil film waste heat thermoelectric conversion The Seebeck effect is used to convert the shear heat of the oil film into electrical energy, which is then stored under stable voltage to provide self-sufficiency for the entire system of sensors, controllers, and actuators. 3. Time-varying dynamics modeling An interference channel and control channel model with time-varying characteristics of oil film stiffness / damping is established, and the oil film parameters are inversely derived using the potential signal to provide real-time input for control. 4 Hybrid Adaptive Control Algorithm The system integrates online identification of control channels, reconstruction of internal model reference signals, neural network reinforcement learning, and LMS adaptive filtering to output equal-amplitude, anti-phase compensation force to suppress vibration. 5. Energy Harvesting and Control Positive Feedback Closed Loop Vibration suppression improves thermocouple fit and power generation efficiency, while efficient power generation provides stable energy to the controller, forming a continuous collaborative working closed loop.

[0037] The thrust bearing is a tilting pad structure with eight thrust pads evenly distributed circumferentially. On the back side of each thrust pad, away from the oil film, an upper support block, a pad fulcrum, an upper balance block, and a lower support block are stacked sequentially. The load is ultimately transferred to the thrust disk through lower balance block fulcrum 1, lower balance block fulcrum 2, balance block fulcrum 1, and balance block fulcrum 2, forming a complete force flow path. An oil film is formed between the opposite end faces of the thrust ring and the thrust disk. The thickness, stiffness, and damping of the oil film change in real time with variations in rotational speed and thrust, constituting the main source of the time-varying characteristics of the control channel. A ring-shaped actuator 11 is fixedly installed on the outer cylindrical surface of the thrust ring. The axial output direction of the actuator 11 is consistent with the longitudinal vibration direction of the thrust bearing, used to output the main power with the same amplitude but opposite phase to the external excitation.

[0038] The actual uncontrolled vibration in the thrust ring serves as a disturbance signal. Error signals during real-time control are measured by sensors and superimposed on the output of the identified control channel model to obtain a reference signal, which serves as the controller's input signal. The identified control channel filter filters the reference signal to obtain a filtered reference signal. This filtered reference signal, along with the reference error signal obtained by subtracting the analog control channel output from the filtered reference signal, jointly controls the iterative updates of the controller. The reference signal is normalized and used as input to a trained neural network, then inversely normalized and added to the adaptive control filter output to obtain the algorithm's output control signal.

[0039] The neural network is trained using supervised learning guided by reinforcement learning strategies. Leveraging the network's high redundancy and combining it with reinforcement learning, supervised learning is performed across multiple operating conditions in a pure simulation environment to achieve efficient control within a narrow target bandwidth. State normalization and denormalization are performed to ensure that the amplitude of the output signal of the proposed neural network control algorithm is linearly correlated with the amplitude of the disturbance signal, effectively mitigating the adverse effects of vibration source amplitude differences on network performance. During training, Gaussian random noise is artificially injected into the state signal to simulate sensor noise and other interference in the real world. Once the network is trained in a pure simulation environment, it can be directly deployed in the real world for real-time control. No parameter updates are performed during real-time control. The neural network consists of three fully connected layers. The hidden layer contains 64 nodes, while the output layer has only one node. The number of nodes in the input layer is determined by the training environment. To ensure the symmetry and boundedness of the algorithm's output, all layers employ the hyperbolic tangent Tanh activation function.

[0040] Before implementing active vibration control, it is necessary to plan the positions of sensors and actuators, set control algorithm parameters, calculate the variable parameter response models of the disturbance and control channels, and train the neural network. Once these preparations are complete, real-time control can begin. Sensors are placed on the thrust ring shell, primarily measuring axial vibration. The measured signals are converted into discrete signals via A / D conversion and used as the algorithm input. Actuators are placed below the thrust ring, in contact with the hull base. The actuator base and hull base are rigidly connected, and the actuator's execution components are connected to the thrust ring. The control signal output from the algorithm is converted into a voltage signal via D / A conversion and a power amplifier to control the actuator's movement, generating a displacement equal in magnitude but opposite in direction to the actual vibration through the control channel.

[0041] The change in vibration model parameters specifically refers to the change in the longitudinal stiffness and damping of the oil film between the thrust ring 3 and the thrust bearing 5 at the thrust bearing, which changes with the main shaft speed and the stable thrust of the propeller. This causes a change in the longitudinal vibration transmission path of the ship's propulsion shaft system, which consists of the propeller, tail shaft, intermediate shaft, thrust bearing, and base. As a result, the response of the interference channel and the control channel has time-varying characteristics. Since the oil film temperature rise is proportional to the shear power, the thermoelectric potential output by the thermoelectric generator implies the modulation information of vibration intensity and frequency.

[0042] The adaptive vibration control method described is an active vibration control method based on a hybrid algorithm of neural network feedback adaptive filtering, including modules such as reference signal reconstruction, neural network controller, and controller update based on LMS algorithm. The hybrid adaptive control algorithm outputs a control signal, which drives actuator 11 through digital-to-analog conversion and power amplifier to generate a displacement with the same amplitude but opposite phase as the vibration at thrust ring 3, thereby canceling the vibration response of the propeller shaft system caused by the propeller interference force.

[0043] The identified control channel filter The identification error signal from the controller output signal and the auxiliary filter The control is updated in real time, and the update criterion is based on the steepest gradient descent method. The update formula is as follows:

[0044] In the formula, To control the step size parameter of the channel filter update, the update formula for the auxiliary filter is:

[0045] In the formula, It is the iteration step size of the auxiliary filter.

[0046] Sensor 10 acquires the residual vibration signal of the thrust bearing position in real time. The real-time control system uses an A / D converter to discretely sample the actual residual vibration signal and transmits it to the real-time input error signal of the algorithm. Based on the internal model control principle, a real-time reconstruction method for the reference signal is established, and the identified control channel filter is used... The output signal is superimposed with the error signal to obtain the controller reference input signal. .

[0047] The neural network controller employs supervised training guided by reinforcement learning. State normalization and denormalization are added before and after the neural network, respectively. Data acquisition uses a reinforcement learning-inspired exploratory mode, and network updates use a policy gradient mode with an experience replay buffer. The network parameters are updated based on sampled data by minimizing the loss function given in the following equation. :

[0048] In the formula, It is the batch size for network training. It is the inverse normalization coefficient, determined by the root mean square (RMS) of the actual reconstructed interference signal and the RMS of the interference signal configured during simulation training. It is the order of the adaptive filter. It is network input. It is an ideal control signal.

[0049] The controller is updated using the LMS algorithm, utilizing the identified control channel filter. For reference signal Filtering is performed to obtain the signal. and the reference signal Output of analog control channel Subtracting them yields the reference error signal. As input to the LMS algorithm, the controller update formula is:

[0050] In the formula, This is the iteration step size for the controller.

[0051] like Figure 6 As shown, thrust ring 3 Figure 1 The actual vibration under no-control conditions is described as an interference signal. The error signal in the real-time control process is measured by sensor 10. The signal is then superimposed on the output of the identified control channel model to obtain the reference signal. As the input signal to the controller. Identified control channel filter. For reference signal Filtering is performed to obtain the signal. and through reference signal Output of analog control channel The reference error signal obtained by subtraction The common control controller undergoes iterative updates. The reference signal is normalized and used as input to the trained neural network, then inversely normalized and added to the output of the adaptive control filter to obtain the algorithm's output control signal.

[0052] like Figure 7 As shown, the neural network is trained using supervised learning guided by a reinforcement learning strategy. Leveraging the network's high redundancy and combining it with the reinforcement learning strategy, supervised learning is performed across multiple operating conditions in a pure simulation environment to achieve efficient control within a narrow target bandwidth. State normalization and denormalization are performed to ensure that the amplitude of the output signal of the proposed neural network control algorithm is linearly correlated with the amplitude of the disturbance signal, effectively mitigating the adverse effects of vibration source amplitude differences on network performance. During training, Gaussian random noise is artificially injected into the state signal to simulate sensor noise and other interferences in the real world. Once the network is trained in a pure simulation environment, it can be directly deployed in the real world for real-time control. No parameter updates are performed during real-time control. The neural network consists of three fully connected layers. The hidden layer contains 64 nodes, while the output layer has only one node. The number of nodes in the input layer is determined by the training environment. To ensure the symmetry and boundedness of the algorithm's output, all layers employ the hyperbolic tangent Tanh activation function.

[0053] like Figure 8 As shown, before implementing active vibration control, it is necessary to position the sensor 10 and actuator 11, set the control algorithm parameters, calculate the variable parameter response models of the interference channel and control channel, and train the neural network. Once these preparations are complete, real-time control can begin. The sensor 10 is placed on the outer shell of the thrust ring 3, primarily measuring axial vibration. The measured signal is converted into a discrete signal by an A / D converter and used as the algorithm input signal. The actuator 11 is placed below the thrust ring 3, in contact with the hull base. The actuator 11 base is fixedly connected to the hull base, and the actuator 11's actuating component is connected to the thrust ring 3.

[0054] Figure 6 Output control signal After D / A conversion and power amplification, the signal becomes a voltage signal that controls the actuator 11 to move, generating a displacement that is equal in magnitude but opposite in direction to the actual vibration through the control channel.

[0055] The principle and workflow of this invention: Energy Harvesting Section: Once the shaft system begins to rotate, the oil film between the thrust ring and the thrust bearing rapidly heats up due to viscous shear. Heat diffuses from the oil film itself to the surrounding area. As the surface temperature of the thrust bearing rises, it is captured by pre-embedded hot nodes. The hot nodes are separated from the oil film by only a layer of thrust bearing surface material, allowing heat to flow in and causing the temperature of the hot nodes to rise closely following the oil film. Subsequently, heat passes axially through tightly welded P-type and N-type thermocouples: in P-type thermocouples, thermal energy drives holes to migrate from the high-temperature end to the low-temperature end; in N-type thermocouples, it drives electrons to diffuse in the same direction. These two types of charge carriers form directional movements within their respective lattices, thus creating a continuous potential difference between the hot and cold nodes. The cold node remains at a lower temperature maintained by the circulating lubricating oil, thus continuously absorbing heat from the P / N-type thermocouples and maintaining the temperature difference. The heat is eventually released into the lubricating oil and carried away by the oil flow, completing a full heat flow cycle from the oil film through the hot nodes, PN-type thermocouples, and back to the cold node. As long as the shaft system is running, the process of oil film shearing heat generation, hot nodes absorbing heat, thermocouple charge carrier migration, and cold nodes dissipating heat will repeat continuously, converting the constantly generated oil film heat into charge and achieving energy capture.

[0056] Control Section: During the operation of the ship's propulsion shafting, hydrodynamic excitation from the propeller, engine rotational imbalance, ship start-up and shutdown speed, propeller thrust stability changes, and external impact excitation that the hull may experience under certain harsh environments are transmitted along the propulsion shaft to the thrust bearing, causing longitudinal vibration, which in turn affects the hull base. Sensors are used to measure the longitudinal vibration of the thrust bearing position in real time and input it into the control algorithm. The control algorithm outputs a control signal based on real-time identified control channels, LMS-based controller updates, and neural network controller iterative calculations. After D / A conversion and power amplification, it is converted into a voltage signal to control the actuator between the thrust bearing and the base, generating a control quantity to counteract the longitudinal vibration displacement of the thrust bearing position.

[0057] Energy Harvesting-Control Coordinated Workflow: During shaft operation, the shear heat of the oil film raises the surface temperature of the thrust bearing. Embedded hot nodes capture heat and conduct it to the cooling oil chamber through a P / N thermocouple array, while cold nodes maintain a relatively low temperature. The continuous heat flow drives carrier migration, generating a potential difference. The output of the thermocouple array, after rectification and voltage regulation, can charge the battery. The charging power is positively correlated with the shear strength of the oil film; that is, the more severe the vibration, the higher the energy harvesting power, which precisely meets the power supply capacity requirements of the control algorithm under strong disturbances. Simultaneously, the fluctuation frequency of the thermoelectric potential is consistent with the dominant frequency of the shaft's longitudinal vibration. This characteristic signal is extracted and used to enhance the input of the neural network controller, improving the phase accuracy of the control output. When the ship slows down or sails under low load, the oil film temperature rise decreases, the energy harvesting power drops, and the system automatically switches to a low-power filtering mode. The neural network enters dormancy, relying solely on adaptive filtering to maintain basic vibration reduction. At this time, the thermoelectric potential signal is only used for operational condition monitoring. When the power recovers, the neural network is reactivated, resuming high-precision control. This dynamic switching mechanism of control strategy based on energy state enables adaptive matching of energy capture and control capabilities, avoiding the energy unsustainability problem caused by continuous high power consumption in traditional active control systems.

[0058] Example 2 This embodiment is based on Embodiment 1 and forms a hybrid adaptive vibration reduction device for oil film thermoelectric energy harvesting in a ship thrust bearing. It is installed inside the ship thrust bearing and includes a thermoelectric energy harvesting device, a vibration sensing module, an adaptive actuation module, and a control module. The vibration sensing module includes a vibration sensor fixed to the outer surface of the thrust ring. The adaptive actuation module includes actuators symmetrically arranged on opposite sides of the thrust ring and thrust disk. The axial output direction of the actuators is consistent with the longitudinal vibration direction of the thrust bearing. The right end of rotor one is fixedly connected to the surface of the thrust ring, and the left end of rotor two is fixedly connected to the surface of the thrust disk. Thrust pads, the thrust ring, and the thrust disk are evenly distributed circumferentially on the side of the thrust disk away from rotor two. A uniformly distributed oil film is formed between the opposite end faces; the thermoelectric energy harvesting device is embedded inside the thrust bearing, with its hot end conducting heat to the oil film and its cold end forming heat communication with the ship's lubrication circulation system; the electrical output end of the thermoelectric energy harvesting device is electrically connected to the energy storage module through a sealed wire; the energy storage module is connected to the vibration sensor, actuator, and control module for power supply; the signal output end of the vibration sensor is connected to the control module for signal acquisition of the real-time vibration displacement signal of the thrust ring and transmission to the control module; the control signal output end of the control module is connected to the actuator drive, and drives the actuator to generate a compensating displacement with the same amplitude but opposite phase as the vibration displacement by outputting an adaptive control signal.

[0059] The thrust bearing mounting support structure includes a limiting component, a support transmission component, and a balance bearing component. The limiting component includes a limiting device one and a limiting device two fixed to the surface of the thrust plate. Limiting device one and limiting device two are symmetrically arranged on both sides of the thrust bearing to limit the radial displacement of the thrust bearing. The support transmission component includes an upper support block, a bearing fulcrum, and a lower support block. The top of the upper support block is fixedly connected to the bottom of the thrust bearing, and the bottom of the upper support block is hinged to the balance bearing component via the bearing fulcrum. The bottom of the balance bearing component is fixedly connected to the lower support block, and the bottom of the lower support block is fixedly connected to the surface of the thrust plate. The balance bearing component includes an upper balance block and a lower balance block. The structure includes a first balance block fulcrum, a second lower balance block fulcrum, a first balance block fulcrum, and a second balance block fulcrum. The top of the upper balance block is hinged to the bearing fulcrum, and the bottom of the upper balance block is fixedly connected to the lower support block. The two sides of the lower support block are fixedly connected to the first and second balance block fulcrums respectively via the first and second lower balance block fulcrums. The bottom ends of the first and second balance block fulcrums are fixedly connected to the surface of the thrust disc. The support transmission assembly and the balance bearing assembly form an elastic bearing structure to buffer the vibration impact of the thrust bearing and ensure the relative positional stability of the thermoelectric energy harvesting device and the oil film. Its stiffness parameters are adapted to the natural vibration frequency of the ship's propulsion shaft system.

[0060] The thermoelectric energy harvesting device includes a hot node, a P-type thermocouple, an N-type thermocouple, a cold node, a heat insulation baffle, and a cooling oil cavity. A mounting groove is pre-reserved on the back side of each thrust pad, and the entire thermoelectric energy harvesting device is embedded in this groove. The heat insulation baffle surrounds the P-type and N-type thermocouples, forming a sealed cooling oil cavity with the inner wall of the groove of each thrust pad. The hot node is fixed to the top of the P-type and N-type thermocouples, with the top surface of the hot node less than 2mm from the working surface of the thrust pad, for directly capturing the shear temperature rise of the oil film. The cold node is fixed to the P-type and N-type thermocouples. At the bottom of the thermocouple, the cold node is completely embedded in the cooling oil chamber. The cooling oil chamber is connected to the ship's lubrication circulation system through a preset oil circuit, forming a continuous hot oil discharge and cold oil replenishment circuit. Conductive wires are connected to both ends of the cold node. A sealable wire hole is provided on the thrust bearing. After the wire passes through the sealable wire hole, it is electrically connected to the voltage sensor. The voltage sensor is electrically connected to the energy storage module to stably input the electrical energy generated by thermoelectric power generation into the energy storage module. The signal output terminal of the voltage sensor is connected to the control module to transmit the real-time collected power generation voltage signal to the control module. The control module uses the voltage signal to infer the shear power, stiffness, and damping variation parameters of the oil film, providing time-varying parameter input for the adaptive control algorithm.

[0061] It should be understood that those skilled in the art can make improvements or modifications based on the above description, and all such improvements and modifications should fall within the protection scope of the appended claims.

Claims

1. A hybrid adaptive vibration reduction method for oil film thermoelectric energy harvesting in ship thrust bearings, characterized in that, include: S1: A thermoelectric energy harvesting device is embedded in the thrust bearing. It uses the temperature rise of oil film shear to form a temperature difference, and converts waste heat into electrical energy through the Seebeck effect. After energy storage, it provides self-powered power to sensors, control modules and actuators. S2: The vibration sensor collects the longitudinal vibration signal of the thrust ring and transmits it to the control module; S3: The control module establishes a time-varying dynamic model and uses the generator voltage to infer the oil film parameters, providing real-time input for the algorithm; S4: Control signals are generated using online identification of control channels, reconstruction of internal model reference signals, and a hybrid algorithm combining neural networks and LMS. S5: Drives the actuator to output equal amplitude and opposite phase to compensate for displacement and suppress vibration; S6: Vibration suppression improves thermoelectric conversion efficiency, forming a positive closed loop of energy capture-vibration reduction-power supply.

2. The hybrid adaptive vibration reduction method for oil film thermoelectric energy harvesting in ship thrust bearings according to claim 1, characterized in that, Step S1 utilizes the shear temperature rise of the oil film to form a stable temperature difference between the hot and cold ends.

3. The hybrid adaptive vibration reduction method for oil film thermoelectric energy harvesting in ship thrust bearings according to claim 1, characterized in that, In S3: The control module establishes a time-varying dynamic model of the disturbance channel and the control channel based on the time-varying characteristics of oil film stiffness and damping as a function of rotational speed and thrust. At the same time, it uses the thermoelectric voltage signal to infer the oil film operating parameters, providing real-time time-varying input for the control algorithm.

4. The hybrid adaptive vibration reduction method for oil film thermoelectric energy harvesting in ship thrust bearings according to claim 1, characterized in that, In step S4, the reference signal is reconstructed based on the internal model control principle. The output of the identified control channel filter is superimposed with the residual vibration error signal to obtain the controller reference input signal.

5. The hybrid adaptive vibration reduction method for oil film thermoelectric energy harvesting in ship thrust bearings according to claim 1, characterized in that, In step S4, the controller updates using the LMS algorithm, taking the filtered reference signal from the identification channel and the error signal output from the analog control channel as inputs, and iteratively generates the optimal control signal.

6. The hybrid adaptive vibration reduction method for oil film thermoelectric energy harvesting in ship thrust bearings according to claim 1, characterized in that, In S5: The control signal drives the actuator to output a compensation displacement that is equal in amplitude and opposite in phase to the vibration displacement, thereby offsetting the longitudinal vibration of the ship's propulsion shaft system.

7. The hybrid adaptive vibration reduction method for oil film thermoelectric energy harvesting in ship thrust bearings according to claim 1, characterized in that, In S6: Vibration suppression improves the fit between the thermocouple and the thrust pad, providing more stable power to the control system.

8. A hybrid adaptive vibration reduction device for oil film thermoelectric energy harvesting in marine thrust bearings, characterized in that... Located within the ship's thrust bearing, it includes a thermoelectric energy harvesting device, a vibration sensing module, an adaptive actuation module, and a control module; The vibration sensing module includes a vibration sensor fixed to the outer surface of the thrust ring, and the adaptive actuation module includes actuators symmetrically arranged on opposite sides of the thrust ring and the thrust disk, with axial output force in the same direction as the longitudinal vibration of the bearing; an oil film is formed between the thrust ring and the thrust disk, and thrust pads are evenly distributed circumferentially on one side of the thrust disk. The thermoelectric energy harvesting device is embedded in the thrust bearing. Its hot end is connected to the oil film for thermal conduction, and its cold end is connected to the ship's lubrication circulation system. Its output end supplies power to the sensor, actuator and control module through the energy storage module. The vibration sensor collects the vibration signal of the thrust ring and transmits it to the control module. The control module drives the actuator to generate a compensating displacement that is equal in amplitude but opposite in phase to the vibration.

9. The hybrid adaptive vibration reduction device for oil film thermoelectric energy harvesting of marine thrust bearings according to claim 8, characterized in that... The axial output direction of the actuator is consistent with the longitudinal vibration direction of the thrust bearing.

10. The hybrid adaptive vibration reduction device for oil film thermoelectric energy harvesting of marine thrust bearings according to claim 8, characterized in that... The energy storage module is connected to the vibration sensor, actuator, and control module for power supply respectively; the signal output terminal of the vibration sensor is connected to the control module for signal acquisition of the real-time vibration displacement signal of the thrust ring and transmission to the control module; the control signal output terminal of the control module is connected to the actuator drive, and drives the actuator to generate a compensation displacement with the same amplitude and opposite phase as the vibration displacement by outputting an adaptive control signal.

Images