A gyro stabilizer rotation speed adaptive control method

Abstract

This invention discloses an adaptive control method for the rotational speed of a gyro-induced roll stabilizer, relating to the field of control technology. It includes steps of roll amplitude detection, control strategy design, and feedback control. By detecting the roll amplitude of the vessel, establishing a mapping model between the roll amplitude and rotational speed, adaptively adjusting the rotational speed of the gyro-induced roll stabilizer, and implementing feedback control, adaptive control of the gyro-induced roll stabilizer's rotational speed is achieved. Because the rotational speed of the gyro-induced roll stabilizer is adaptively controlled according to the vessel's roll amplitude, the gyro-induced roll stabilizer does not need to maintain high-speed rotation continuously, avoiding energy waste caused by continuous high-speed rotation and achieving energy-saving effects. Furthermore, since the gyro-induced roll stabilizer does not need to maintain high-speed rotation continuously, wear caused by continuous high-speed rotation is also avoided, extending its service life.

Description

Technical Field

[0001] This invention relates to the field of control technology, and in particular to an adaptive control method for the rotational speed of a gyroscope anti-roll device. Background Technology

[0002] During navigation, especially when boats are traveling on water, they are inevitably affected by external factors such as waves and winds, resulting in rolling. This rolling not only affects the comfort of the crew but can also threaten the navigational safety of the boat, reducing its handling performance and stability. To reduce the rolling of boats, gyro-based roll dampers have been developed.

[0003] A gyro-based roll stabilizer is a device that utilizes the fixed-axis and precession properties of a gyroscope to provide roll-damping torque. Traditional gyro-based roll stabilizers typically maintain a constant high-speed rotation during operation. This design can effectively reduce the rolling of a vessel and improve its stability to some extent. However, this method of maintaining constant high-speed rotation presents several problems.

[0004] From an energy utilization perspective, maintaining high-speed rotation continuously consumes a significant amount of energy. A vessel's energy supply is typically limited, especially for smaller boats or those relying on batteries or other limited power sources, where energy waste is even more pronounced. This not only increases operating costs but can also limit the vessel's range.

[0005] Considering the lifespan and maintenance costs of the equipment, prolonged high-speed rotation will subject the various components of the gyro stabilizer to significant mechanical stress and wear, thus shortening the equipment's lifespan. Furthermore, frequent repairs and component replacements will increase the boat's maintenance costs.

[0006] Currently, there are some technologies on the market for controlling gyro-based roll stabilizers, but most of them simply involve on / off control or setting a few fixed speed levels, failing to provide real-time and precise adjustments based on the actual rolling motion of the vessel. These control methods still suffer from low energy efficiency and unsatisfactory roll reduction effects. Therefore, developing a method that can adaptively control the speed of the gyro-based roll stabilizer according to the vessel's rolling amplitude is of significant practical importance. Summary of the Invention

[0007] This invention provides a method for adaptive control of the rotational speed of a gyroscope roll stabilizer, which solves the technical problems of how to make real-time adjustments based on the actual rolling conditions of the vessel and low energy utilization efficiency.

[0008] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows:

[0009] A method for adaptive control of the rotational speed of a gyroscope damper includes the following steps:

[0010] Step 1: Obtain the lateral roll amplitude of the boat and longitudinal sway amplitude , to increase the horizontal sway amplitude and longitudinal sway amplitude Synthesize to obtain swing amplitude ;

[0011] Step 2: Obtain the preprocessed experimental data, establish a mapping model between the rolling amplitude and rotational speed, and train and optimize the model using gradient descent to obtain a well-trained mapping model. Then, use the real-time obtained boat rolling amplitude... The input is fed into the trained mapping model to calculate the target rotational speed of the gyro stabilizer under the current swing amplitude. And it is used to adaptively adjust the rotation speed of the gyroscope damper;

[0012] Step 3: Record the real-time boat roll amplitude With safety threshold The speed deviation was calculated by comparison when the threshold was exceeded. The control output is obtained based on the discretized PID control algorithm. Closed-loop feedback continues until the gyroscope stabilizer reaches its optimal operating state.

[0013] A further technical solution is that, in step 1, the swing amplitude is obtained. The steps include the following steps:

[0014] The lateral acceleration of the boat is obtained by monitoring an acceleration sensor installed near the center of gravity of the boat. and longitudinal acceleration ;

[0015] Based on the trapezoidal integral method, according to the lateral acceleration of the boat... Calculate the lateral roll angle of the boat According to the longitudinal acceleration of the boat Calculate the longitudinal roll angle of the boat ; the angle of the boat's lateral roll The filtered lateral sway angle is obtained after processing by a filtering algorithm. The longitudinal roll angle of the boat The longitudinal sway angle is obtained after processing by a filtering algorithm. Based on the filtered lateral sway angle Calculate the lateral roll amplitude of the boat Based on the filtered longitudinal sway angle Calculate the longitudinal roll amplitude of the boat .

[0016] A further technical solution is that, in step 1, the filtering algorithm is the Kalman filtering algorithm.

[0017] A further technical solution is as follows: In step 2, orthogonal experiments are used to collect experimental data, and the experimental data is preprocessed to obtain preprocessed experimental data; the experimental data includes the rotational speed n of the gyroscope stabilizer and the corresponding stabilizer effect. The ship's roll angle, pitch angle, and yaw acceleration. Lateral roll amplitude, longitudinal roll amplitude, and real-time ship roll amplitude The preprocessing steps include cleaning to remove outliers and normalization.

[0018] A further technical solution is as follows: In step 2, a mapping model between sway amplitude and rotational speed is established based on a multilayer perceptron (MLP) neural network; the step of training and optimizing by gradient descent includes using MSE as the error function, updating the weights and biases of the mapping model between sway amplitude and rotational speed using gradient descent, and optimizing the mapping model between sway amplitude and rotational speed by combining cross-validation.

[0019] A further technical solution is that: in step 2, the trained swing amplitude and rotation speed mapping model is given by equation (21).

[0020] (twenty one)

[0021] In equation (21), The target rotational speed of the gyroscope damper under the current swing amplitude. The coefficients of the multinomial regression model are... Let be the order of the polynomial. This represents the real-time sway amplitude of the boat; for The square of, for h times squared.

[0022] A further technical solution is that step 2 also includes obtaining the target rotational speed. The subsequent correction steps include introducing a dynamic adjustment factor α to obtain the final target rotational speed n. final Using the target rotational speed n final As the corrected target speed .

[0023] A further technical solution is as follows: In step 2, the dynamic adjustment factor α is calculated according to equation (28), and the final target rotational speed n is obtained by correcting according to equation (29). final ,

[0024] (28)

[0025] In equation (28), As a dynamic adjustment factor, The adjustment coefficient is dynamically adjusted based on the specific characteristics of the vessel. The angular acceleration of the boat's roll;

[0026] (29)

[0027] In equation (29), For the final target speed, The target rotational speed.

[0028] A further technical solution is that, in step 3, the rotational speed deviation is calculated according to equation (31). ,

[0029] (31)

[0030] In equation (31), Current speed With target speed The deviation between them This is the actual rotational speed of the motor driving the current gyroscope stabilizer.

[0031] A further technical solution is that, in step 3, the discretized PID control algorithm is Equation (33).

[0032] (33)

[0033] In equation (33), It is the current number Control output at the next sampling time; It is a proportionality coefficient used to respond to deviations proportionally, thereby speeding up the system's response. It is the current number Deviation during the second sampling; It is the first Deviation during the second sampling; These are the integral coefficients used to eliminate the steady-state error of the system; The sampling period; It is the first step in the control process Deviation during the second sampling; These are differential coefficients used to predict the trend of deviation changes and improve the dynamic performance of the system.

[0034] The beneficial effects of adopting the above technical solution are as follows:

[0035] By detecting the boat's roll amplitude, establishing a mapping model between roll amplitude and rotational speed, and implementing adaptive adjustment and feedback control of the gyro-stabilized roll stabilizer's rotational speed, adaptive control of the gyro-stabilized roll stabilizer's rotational speed is achieved. Because the gyro-stabilized roll stabilizer's rotational speed is adaptively controlled based on the boat's roll amplitude, it does not need to maintain high-speed rotation continuously, avoiding energy waste caused by continuous high-speed rotation and achieving energy-saving effects. Furthermore, the elimination of the need for continuous high-speed rotation also avoids wear and tear caused by continuous high-speed rotation, extending its service life.

[0036] By installing an acceleration sensor near the center of gravity of the boat, the sway amplitude can be accurately detected. Based on a scientifically established mapping model between sway amplitude and rotational speed, the rotational speed can be accurately adjusted adaptively, further ensuring stable and reliable feedback control.

[0037] Experimental data were obtained using orthogonal experimental data acquisition. Based on the mapping model between gyroscope amplitude and rotational speed, the rotational speed n of the gyroscope stabilizer and the corresponding gyroscope stabilization effect were correlated. By implementing technical linkages, the anti-slip effect has been further improved. Attached Figure Description

[0038] Figure 1 This is a flowchart of the present invention. Detailed Implementation

[0039] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. The following description of at least one exemplary embodiment is merely illustrative and is in no way intended to limit this application or its application or use. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0040] Many specific details are set forth in the following description in order to provide a full understanding of this application. However, this application may also be implemented in other ways different from those described herein. Those skilled in the art can make similar extensions without departing from the spirit of this application. Therefore, this application is not limited to the specific embodiments disclosed below.

[0041] like Figure 1 As shown, this invention discloses an adaptive control method for the rotational speed of a gyroscope roll stabilizer, comprising the following steps:

[0042] Step 1: Swing amplitude detection.

[0043] Input data: The lateral acceleration of the boat is obtained by monitoring an acceleration sensor installed near the center of gravity of the boat. and longitudinal acceleration .

[0044] Processing algorithm: Based on the lateral acceleration of the boat The lateral roll angle of the boat is calculated using formula (1). According to the longitudinal acceleration of the boat The longitudinal roll angle of the boat is calculated using equation (2). The lateral roll angle of the boat The lateral sway angle is obtained after processing with the Kalman filter algorithm. The longitudinal roll angle of the boat The longitudinal sway angle is obtained after processing by the Kalman filter algorithm. Based on the filtered lateral sway angle The lateral roll amplitude of the vessel is calculated using equation (11). Based on the filtered longitudinal sway angle The longitudinal roll amplitude of the vessel is calculated using equation (12). .

[0045] Output data: lateral roll amplitude of the boat Vertical sway amplitude lateral sway amplitude and longitudinal sway amplitude Synthetic swing amplitude .

[0046] Step 1 is detailed below.

[0047] Step 1.1: Selection and installation of the accelerometer.

[0048] For detecting the roll amplitude on boats, a high-precision and suitable accelerometer is crucial. The appropriate accelerometer should be selected based on the specific type, size, and application of the boat, ensuring its range, accuracy, and sensitivity are suitable. Generally, for small boats, a sensor with a range of ±2g to ±5g and an accuracy of approximately ±0.01g can be used; for large boats, the range may need to be extended to ±5g to ±10g.

[0049] The installation location should be chosen near the boat's center of gravity and in a place that is as free from vibration interference from other equipment to ensure accurate detection of the boat's rolling acceleration. Assuming the boat rolls in a two-dimensional plane, the acceleration sensor can measure acceleration components in two directions, denoted as... and The two directions are typically horizontal and vertical.

[0050] Step 1.2: Acceleration data acquisition.

[0051] Connect the accelerometer to a high-performance data acquisition module. The data acquisition module needs to have a high sampling frequency and high-precision analog-to-digital conversion capability. Let the sampling frequency be... Unit: Hz, which is the number of samples collected per second. There are several acceleration data points. The collected acceleration data will form a discrete time series, for lateral acceleration... In the The data at each sampling time is denoted as Longitudinal acceleration In the The data at each sampling time is denoted as .

[0052] Step 1.3: Calculate the swing angle.

[0053] The collected acceleration data represents the acceleration information of the boat, which needs to be converted into a rolling angle through integration. The trapezoidal integral method is used here, which divides the integration interval into multiple smaller trapezoids and approximates the integral value by calculating the sum of the areas of these trapezoids.

[0054] For lateral acceleration From the initial moment To the Each sampling time Integral calculation of the swing angle The formula is:

[0055] (1)

[0056] In equation (1), The lateral roll angle of the boat. This represents the lateral roll angle of the boat at the previous moment. The sampling period is , Sampling frequency, For the first The lateral acceleration of the boat at each sampling time. For the first The lateral acceleration of the boat at each sampling moment.

[0057] Similarly, for longitudinal acceleration Longitudinal sway angle The calculation formula is:

[0058] (2)

[0059] In equation (2), The longitudinal roll angle of the boat. This represents the longitudinal roll angle of the vessel at the previous moment. Let x be the longitudinal acceleration of the boat at the k-th sampling time. For the first The longitudinal acceleration of the vessel at each sampling moment.

[0060] Step 1.4: Data filtering processing.

[0061] Since the actual collected acceleration data may contain noise interference, which can affect the accuracy of the swing angle calculation, filtering is required. Here, the Kalman filter algorithm is used; it is an optimal recursive filter that can optimally estimate the system state based on the system's state equation and observation equation.

[0062] Assume the system's state vector is:

[0063] (3)

[0064] In equation (3), Let be the lateral state vector of the vessel. The lateral angular velocity, The transpose symbol indicates that the row vector is transposed. Convert to column vectors.

[0065] Equation (4) establishes the evolution relationship of the state from time k to time k+1 based on the dynamic characteristics of the ship's lateral rolling. The state equation of the system is:

[0066] (4)

[0067] In equation (4), Let k+1 be the state equation of the boat at sampling time k+1, where k+1 is the sampling time progression of the Kalman filter, representing the prediction of the state at time k+1 based on the state at time k. , , The noise is a transverse process noise, with a mean of zero and a covariance of... The Gaussian distribution.

[0068] The representative (4) model does not include all the disturbance factors. It is a mathematical model of random disturbances that the model does not capture, rather than physical quantities obtained through measurement.

[0069] Equation (4) is the state prediction: based on , and transverse process noise predict , To model system uncertainty.

[0070] Equation (5) establishes the relationship between the observed value and the actual state. The observation equation is:

[0071] (5)

[0072] In equation (5), The observed value is the lateral sway angle. , The lateral observation noise follows a pattern with a mean of zero and a covariance of... The Gaussian distribution.

[0073] It is Gaussian noise following the N(0,R) pattern, representing the observed value. The deviation from the observable portion of the true state is obtained through calculation. .

[0074] Equation (5) is the observation model: observations are obtained through sensors. Its deviation from the true state is due to describe, That is, modeling and measuring uncertainty.

[0075] Kalman filtering predicts the current state based on the optimal estimate from the previous time step. The steps of Kalman filtering include prediction and updating:

[0076] Prediction steps:

[0077] (6)

[0078] In equation (6), This is the posterior optimal estimate at time k−1. This is the prior state estimate at time k.

[0079] (7)

[0080] Equation (7), Let be the posterior covariance at time k-1. Let be the prior covariance at time k. Process noise covariance.

[0081] Update steps:

[0082] (8)

[0083] In equation (8), , express The inverse matrix.

[0084] (9)

[0085] In equation (9), .

[0086] (10)

[0087] In equation (10), time Covariance estimation It is an identity matrix.

[0088] After Kalman filtering, an accurate posterior estimate of the transverse state vector is obtained. .

[0089] Similarly, the filtered longitudinal sway angle The calculation is completely consistent with the horizontal calculation. It is necessary to replace the subscript x in equations (3) to (10) with y. Finally, the longitudinal state vector is estimated from the posterior. Extracted from .

[0090] Step 1.5: Obtain the swing amplitude.

[0091] According to the preset time interval The filtered sway angle is sampled. Let the sampling time be... Then in the first Lateral swing amplitude at each sampling time and longitudinal sway amplitude These represent the difference between the maximum and minimum swing angles within that time interval:

[0092] (11)

[0093] In equation (11), This refers to the horizontal sway amplitude. Let i be the k-th sampling time, and i is the discrete-time index of the angle sampling in the Kalman filter.

[0094] (12)

[0095] In equation (12), This represents the longitudinal sway amplitude.

[0096] The final result and This refers to the data on the lateral and longitudinal roll amplitudes of the vessel, specifically the lateral roll amplitude. and longitudinal sway amplitude Synthesize to obtain swing amplitude .

[0097] Step 2: Control strategy design.

[0098] Input data: Input the manually set swing amplitude through the parameter setting module of the boat swing test platform. Initial speed and sailing speed sway amplitude Set as Initial speed Set as sailing speed Set as Furthermore, the actual rotational speed n of the gyro stabilizer and the corresponding anti-roll effect were obtained through actual measurements on a boat swaying test platform for each orthogonal experiment. .

[0099] Processing Algorithm: An orthogonal experimental design is used to arrange experiments and record data. The collected experimental data, including actual rotational speed, actual acceleration, actual roll angle and pitch angle, etc., mentioned in step 1, are first cleaned to remove outliers, and then normalized using equation (13). An MLP neural network is selected to establish a rotational speed mapping model. The normalized roll amplitude setting value A and the sailing speed setting value v are used as inputs. The number of neurons in the input layer is the number of input variables p. The hidden layer output z is calculated using equation (14). j Equation (15) calculates the neural network fitting speed of the output layer. Using the MSE of equation (16) as the error function, the model weights and biases are updated by the gradient descent method of equations (17) to (20), and the model is optimized by cross-validation; the real-time swing amplitude is... = or Input the trained model (21) to obtain the target rotational speed n target By introducing the dynamic adjustment factor α of equation (28), n is obtained based on the correction of equation (29). final The target PWM duty cycle D is calculated according to equations (22) and (23). target The speed error Δn is calculated according to equation (24), and the adjustment amount u(t) is calculated by the integral separation PID algorithm of equations (25) and (26). The final PWM duty cycle D is obtained by correction by equations (27) to (30). final .

[0100] Output model: The trained rotational speed mapping model (21), the target rotational speed n of the gyro stabilizer. target The final target rotational speed n final The target PWM duty cycle D of the drive motor target The final PWM duty cycle D final and the actual speed n of the drive motor under real-time monitoring real Rotational speed error Δn.

[0101] Step 2 is detailed below.

[0102] Step 2.1: Establish the rotational speed mapping model.

[0103] Step 2.1.1: Experimental data acquisition and experimental design.

[0104] To establish an accurate mapping model between the sway amplitude and the rotational speed of the gyro damper, numerous ship sway experiments are required. Considering both efficiency and accuracy, orthogonal experimental design is employed to optimize the experimental scheme. Orthogonal experimental design is an efficient and rapid experimental design method that can obtain comprehensive experimental information with a relatively small number of experiments.

[0105] Assume that the factors affecting the roll reduction effect of the gyroscope include the ship's roll amplitude. The initial speed of the gyroscope damper The speed of the boat Each factor can be set to several levels. For example, the amplitude of the swing. Set as m represents the horizontal amplitude of the oscillation; initial rotational speed Set as b represents the initial rotational speed; sailing speed Set as , l represents the level of sailing speed.

[0106] According to orthogonal array Arrange experiments, among which To record the number of experiments, the rotational speed of the gyroscope stabilizer was recorded under different combinations of factor levels in each experiment. and the corresponding anti-roll effect The anti-roll effect can be expressed as the rate of reduction in the ship's roll amplitude.

[0107] Step 2.1.2: Data preprocessing.

[0108] The collected experimental data may contain noise and outliers, requiring preprocessing. First, the experimental data is cleaned to remove obvious outliers. Then, the data is normalized, mapping the data for different variables to... Intervals are used to eliminate the influence of different variable units.

[0109] Let any original data that needs to be normalized be... , ; , Let be the number of data points, and i be the i-th data point. Let be the total number of variables, and j be the j-th variable. The normalization formula is:

[0110] (13)

[0111] In equation (13), The normalized value. This is the original data. and The first The minimum and maximum values ​​of each variable.

[0112] Step 2.1.3: Model selection and establishment.

[0113] Based on the characteristics of the data, a suitable model is selected to describe the relationship between the sway amplitude and the rotational speed. Considering the complexity of the ship's sway and the operation of the gyroscopic anti-roll device, a neural network model is adopted. Neural networks have powerful nonlinear mapping capabilities and can handle complex nonlinear relationships well.

[0114] Here, we choose a multilayer perceptron neural network, abbreviated as MLP. An MLP consists of an input layer, hidden layers, and an output layer. The number of neurons in the input layer is equal to the number of input variables; in this problem, the input variable is the given boat's sway amplitude. Given the speed of the boat Given the initial rotational speed of the gyroscope, let the number of input variables be... Then the number of neurons in the input layer is The output layer has only one neuron, meaning the output is the rotational speed of the gyroscope stabilizer fitted by the neural network. The number of neurons in the hidden layer can be determined experimentally and empirically. Let the number of neurons in the hidden layer be... .

[0115] The input-output relationship of an MLP neural network can be represented as:

[0116] , (14)

[0117] In equation (14), For the hidden layer The output value of each neuron For the input layer The input value of each neuron, For the input layer The first neuron is connected to the hidden layer. The weights of each neuron, For the hidden layer Bias of each neuron.

[0118] (15)

[0119] In equation (15), For the hidden layer The weights from each neuron to the output layer neurons. This is the bias for the output layer neurons. For activation functions, commonly used activation functions include the sigmoid function and the ReLU function.

[0120] Step 2.1.4: Model training and optimization.

[0121] The MLP neural network was trained using preprocessed experimental data. The goal of the training was to minimize the predicted output. Compared with actual output The error between them. The commonly used error function is Mean Squared Error, abbreviated as MSE:

[0122] (16)

[0123] In equation (16), MSE is the mean square error. The number of training data. For the i-th predicted output, This is the i-th actual output.

[0124] Gradient descent is used to update the weights and biases of the neural network to minimize the mean squared error (MSE). The update formula for gradient descent is:

[0125] (17)

[0126] In equation (17), For the updated input layer The first neuron is connected to the hidden layer. The weights of each neuron, For the input layer before the update The first neuron is connected to the hidden layer. The weights of each neuron, The learning rate controls the step size for updating weights and biases. For MSE The partial derivatives of .

[0127] (18)

[0128] In equation (18), For the updated hidden layer Bias of each neuron For the hidden layer before the update Bias of each neuron For MSE The partial derivatives of .

[0129] (19)

[0130] In equation (19), For the updated hidden layer The weights from each neuron to the output layer neurons. For the hidden layer before the update The weights from each neuron to the output layer neurons. For MSE The partial derivatives of .

[0131] (20)

[0132] In equation (20), To update the bias of the output layer neurons, To update the bias of the output layer neurons before the update. Let g be the partial derivative of MSE with respect to the bias g of the output layer neurons.

[0133] To avoid overfitting during training, cross-validation can be used. This involves dividing the training data into training and validation sets, training the model on the training set, and evaluating its performance on the validation set. By continuously adjusting the neural network structure and training parameters, the model's performance is optimized until it achieves satisfactory accuracy and generalization ability.

[0134] Step 2.2: Target speed calculation.

[0135] Real-time data on the ship's roll amplitude The input is fed into the trained rotational speed mapping model to calculate the target rotational speed of the gyro stabilizer under the current swing amplitude. Assuming the rotational speed mapping model is a nonlinear model based on multinomial regression, its expression is:

[0136] (twenty one)

[0137] In equation (21), The target rotational speed of the gyroscope damper under the current swing amplitude. These are the coefficients of the multinomial regression model, which were obtained through training and optimization using experimental data during the model building phase of the speed mapping model. The order of the polynomial can be adjusted based on the fitting effect of the experimental data. This represents the real-time sway amplitude of the boat; for The square of, for h times squared.

[0138] Step 2.3: Drive motor control strategy.

[0139] The controller calculates the target speed. Adjust the input voltage or current of the drive motor. Pulse Width Modulation (PWM) technology is used here to precisely control the speed of the drive motor. The duty cycle of the PWM signal... There is a certain relationship between it and the speed of the drive motor.

[0140] Let the speed of the drive motor be With PWM duty cycle Satisfies a linear relationship:

[0141] (twenty two)

[0142] In equation (22), K is the proportionality coefficient. It is a constant. The speed of the drive motor can be determined by experimentally measuring the speed of the drive motor under different duty cycles. and The value of .

[0143] In order to make the drive motor reach the target speed The corresponding PWM duty cycle needs to be calculated. From the above linear relationship, we can obtain:

[0144] (twenty three)

[0145] In equation (23), To make the drive motor reach the target speed The corresponding PWM duty cycle, This is the proportionality coefficient. It is a constant.

[0146] Step 2.4: Real-time speed monitoring and error compensation.

[0147] Step 2.4.1: Real-time speed monitoring and error compensation.

[0148] During the adjustment process, it is necessary to monitor the actual speed of the drive motor in real time. This ensures that the target speed is accurately achieved. The actual speed can be obtained through an encoder mounted on the drive motor shaft.

[0149] actual speed With target speed Error between for:

[0150] (twenty four)

[0151] In equation (24), Actual rotational speed With target speed The error between them.

[0152] To reduce this error, an integral-separated PID control algorithm is used to adjust the PWM duty cycle. The expression for the integral-separated PID control algorithm is:

[0153] when hour, (25)

[0154] In equation (25), The integral separation threshold, This is the output of the PID controller, used to adjust the PWM duty cycle; This is the proportionality coefficient. Let be the error at time t.

[0155] when hour,

[0156] (26)

[0157] In equation (26), The integral coefficient is... These are the differential coefficients; The error at time t-1, Let be the error at time j.

[0158] Adjusted PWM duty cycle for:

[0159] (27)

[0160] In equation (27), This is the adjusted PWM duty cycle. The target PWM duty cycle.

[0161] Step 2.4.2: Adjust the speed considering the dynamic characteristics of the boat.

[0162] Because vessels exhibit dynamic characteristics during navigation, their rolling motion changes with variations in their navigation conditions. Therefore, the dynamic characteristics of the vessel must be considered during adaptive speed adjustment.

[0163] Taking the second derivative of the roll and pitch angles ϕ and θ of the boat, we obtain the roll acceleration as follows: and This can be obtained by differentiating the oscillation acceleration. A dynamic adjustment factor is then introduced. Its expression is:

[0164] (28)

[0165] In equation (28), As a dynamic adjustment factor, The adjustment coefficient is dynamic and can be adjusted according to the specific characteristics of the vessel. This refers to the angular acceleration of the boat's roll.

[0166] The final target speed for:

[0167] (29)

[0168] In equation (29), For the final target speed, The target rotational speed.

[0169] Accordingly, the final PWM duty cycle for:

[0170] (30)

[0171] In equation (30), This is the final PWM duty cycle. This is the proportionality coefficient. It is a constant.

[0172] Through step 2 above, the rotational speed of the gyro damper can be adjusted in real time and accurately according to the boat's rolling amplitude. At the same time, the dynamic characteristics of the boat are taken into account, which improves the accuracy and effectiveness of the adaptive speed adjustment.

[0173] Step 3: Feedback control.

[0174] Input data: The lateral roll amplitude of the boat obtained in step 1. Longitudinal sway amplitude Preset safety threshold for boat roll amplitude The current speed of the gyro-driven motor is collected in real time by a speed sensor. ; Parameters of the rotational speed mapping model established based on the correlation between the ship's rolling amplitude and the gyroscope's rotational speed; Preset sampling period T of the digital control system; Maximum input voltage U of the gyroscope roll stabilizer drive motor. ax The PID controller has preset proportional coefficient K, integral coefficient Kᵢ, and derivative coefficient K. d .

[0175] Processing algorithm: The real-time swing amplitude A is calculated from the swing acceleration using an integral algorithm. real , will A real With A safe Compare; if A real >A safeThe deviation is calculated according to equation (31), and Δy is used as the deviation signal e(t). The control output u(k) is obtained through the PID control algorithm of equation (32) and the discretization processing of equation (33). Then, the PWM duty cycle is calculated according to equation (34) to adjust the motor speed. The above process is repeated until A real ≤A safe .

[0176] Output Model: Output of a PID Controller Discretized output of PID controller The horizontal PWM duty cycle D(k) of the kth sample during PID control.

[0177] The core formula of the feedback controller is the continuous form of the PID control algorithm formula (32). The integral separation PID controller and dynamic adjustment mechanism in step 2 are closely related to the feedback controller in step 3. Both are based on PID control and form a closed-loop control link of preliminary adjustment and fine correction. The controller in step 2 outputs the target speed based on the speed mapping model. It achieves the initial speed adaptation through integral separation PID formulas (25) and (26) and dynamic adjustment factor formulas (28) and (29), providing a basic target and initial adjustment for the anti-sway control. The feedback controller in step 3, on the basis of this, continuously monitors the deviation between the swing amplitude and the safety threshold, inputs the real-time speed error into the PID algorithm formulas (32) and (33), and further corrects the control quantity, i.e. the PWM duty cycle, to solve the speed deviation problem that may be caused by model error and external interference in step 2. In terms of technical function, the controller in step 2 is responsible for establishing the mapping relationship between the swing amplitude and the target speed and realizing the initial dynamic adjustment to ensure that the control direction meets the requirements of the gyro roll reduction; the feedback controller continuously optimizes the speed through the cyclic feedback and convergence mechanism, and finally stabilizes the swing amplitude within the safe threshold. The two work together to improve the accuracy, anti-interference and stability of the gyro roll reducer control.

[0178] For real-time adjustments based on the actual rolling motion of the vessel, the formula uses the comparison between the real-time monitored rolling amplitude and the safety threshold as the trigger condition. When the rolling amplitude exceeds the limit, the current rotational speed is obtained through the speed sensor, and the deviation Δn is calculated by combining it with the target rotational speed obtained from the speed mapping model. The proportional term of the PID controller is then used to adjust the deviation. It can quickly respond to the current deviation and accelerate the adjustment speed; integral term Accumulated deviations are used to eliminate steady-state errors and ensure long-term stability; differential terms The system predicts the trend of deviation changes and makes adjustments in advance to adapt to the dynamic changes of the swing. The discretization formula is adapted to the real-time sampling characteristics of the digital control system, ensuring that each adjustment is based on the latest swing state and achieving real-time response.

[0179] To address the issue of low energy efficiency, the controller precisely calculates the control quantity u(t) and adjusts the PWM duty cycle using equation (34) to ensure that the drive motor speed strictly matches the target speed required for the current swing: when the swing amplitude is within the safe threshold, the current speed is maintained without additional adjustment; when the swing exceeds the limit, only the necessary control quantity is output according to the magnitude of the deviation to avoid energy waste caused by excessive acceleration; at the same time, the derivative term reduces frequent speed fluctuations, and the integral term avoids ineffective energy consumption under continuous deviation. Through precise control that adjusts as needed, the energy efficiency is significantly improved.

[0180] Step 3 is detailed below.

[0181] Step 3.1: Monitoring the swing amplitude and comparing it with the threshold.

[0182] After adjusting the rotation speed of the gyro damper, the accelerometer continuously monitors the boat's rolling acceleration, and the real-time rolling amplitude is calculated according to the integral algorithm used in the rolling amplitude detection phase. Integration algorithms, such as the trapezoidal rule or Simpson's rule. A preset safety threshold is used. Real-time swing amplitude With safety threshold Compare. If This indicates that the boat's rolling amplitude is within a safe range, and the current rotation speed of the gyro damper can be maintained; if If so, the rotation speed of the gyroscope damper needs further adjustment.

[0183] Step 3.2: Deviation calculation.

[0184] when At that time, calculate the current speed. With target speed Deviation between The calculation formula is as follows:

[0185] (31)

[0186] In equation (31), Current speed With target speed The deviation between them It is based on the rotational speed mapping model, determined by the real-time oscillation amplitude. The calculated optimal rotational speed of the gyroscope damper; It is the actual rotational speed of the motor driving the current gyroscope anti-roll device, which can be obtained in real time through a speed sensor.

[0187] Step 3.3 PID: Control Algorithm Principle.

[0188] The rotational speed of the gyro stabilizer is further adjusted using a proportional-integral-derivative (PID) control algorithm. The output of the PID controller... It is a control variable used to adjust the input voltage or current of the drive motor, and its calculation formula is:

[0189] (32)

[0190] In equation (32), This is the output of the PID controller; This is a deviation signal. ; It is a proportionality coefficient used to respond to deviations proportionally, thereby speeding up the system's response. These are the integral coefficients used to eliminate the steady-state error of the system; These are the differential coefficients, used to predict the trend of deviation changes and improve the dynamic performance of the system. The integral term for the deviation; This represents the rate of change of the deviation.

[0191] Step 3.4: Discretization.

[0192] In practical applications, since the control system is based on computer-based digital control, the continuous PID control algorithm needs to be discretized. Let the sampling period be... , No. The deviation at the second sampling time is The discretized PID control algorithm formula is:

[0193] (33)

[0194] In equation (33), It is the current number Control output at the next sampling time; It is the current number The deviation at the time of sampling, i.e. ; It is the first Deviation during the second sampling It is the first step in the control process Deviation during the second sampling.

[0195] Step 3.5: Control output and speed adjustment.

[0196] The control output is calculated based on the discretized PID control algorithm. The input voltage or current of the drive motor is adjusted using pulse width modulation (PWM) technology. The duty cycle of the PWM signal... With control output There is a certain relationship, which can be represented as:

[0197] (34)

[0198] In equation (34), This refers to the duty cycle of the PWM signal. It is the maximum value of the input voltage to the drive motor.

[0199] By adjusting the duty cycle of the PWM signal It can precisely control the speed of the drive motor, gradually bringing it closer to the target speed. .

[0200] Step 3.6: Loop Feedback and Convergence.

[0201] After each speed adjustment, steps 3.1 to 3.5 are repeated: monitoring the sway amplitude, calculating the deviation, applying a PID control algorithm to adjust the control input, and adjusting the drive motor speed. As the number of cycles increases, the boat's sway amplitude gradually decreases, eventually reducing it to a safe threshold. The following steps achieve convergence of closed-loop feedback control, ensuring that the gyro damper is always in optimal working condition based on the actual rolling conditions of the boat.

[0202] The beneficial technical effects are explained below.

[0203] First, the energy-saving effect is significant. By adaptively adjusting the rotation speed of the gyro damper based on the real-time rolling amplitude of the vessel, the energy waste caused by the gyro constantly rotating at high speed in traditional solutions is avoided, thereby significantly reducing the overall energy consumption of the vessel and improving energy utilization efficiency. Details are as follows.

[0204] Threshold trigger control: In step 3, equation (31) triggers speed adjustment based on the deviation between the target speed and the actual speed corresponding to the real-time swing amplitude; when the real-time monitored swing amplitude is lower than the set safety threshold, the system maintains the gyroscope anti-swing device at low speed to avoid ineffective high-speed operation when no high power output is required, thereby reducing energy waste from the source.

[0205] Precise speed matching: In step 2, a nonlinear positive correlation mapping relationship between the target speed and the swing amplitude is established through equation (21). Combined with the introduction of the dynamic adjustment factor in equation (29), the target speed can adapt to the current swing intensity in real time. The smaller the swing amplitude, the lower the calculated target speed, and the smaller the corresponding motor input power, thus avoiding excessive energy consumption due to over-powered motors. Further explanation is as follows.

[0206] In step 2, the speed mapping model is established and the speed of the gyroscope damper is controlled. Equation (21) calculates the basic target speed based on the real-time swing amplitude, realizing the nonlinear mapping of swing amplitude to target speed. Equations (28) to (29) correct the target speed, realizing the precise adaptation of dynamic swing amplitude to adaptive speed. In steps 2.5 to 2.8, precise speed matching is achieved. Equation (21) realizes the positive correlation between target speed and swing amplitude, and Equation (29) realizes the dynamic adjustment factor to adapt to the swing intensity, so that the speed strictly matches the current swing requirement, that is, the smaller the swing amplitude, the lower the target speed, and the smaller the motor input power.

[0207] Closed-loop energy consumption optimization: In step 3, the PWM duty cycle is dynamically adjusted using equation (34) to accurately convert the speed error caused by the sway amplitude into motor control commands, thereby achieving closed-loop speed control. This strategy effectively prevents redundant energy consumption caused by over-response, hysteresis, or nonlinear errors, further improving energy utilization efficiency. Therefore, the adaptive speed regulation strategy based on sway sensing and closed-loop control proposed in this application can significantly reduce unnecessary motor energy consumption while ensuring the stability of the vessel, demonstrating the high-efficiency energy-saving advantages of this application under long-duration, multi-condition, and complex sea conditions. Further explanation follows.

[0208] In step 3, the gyro damper speed is controlled in a closed-loop feedback manner. The deviation related to the swing amplitude in equations (32) to (34) is converted into a motor speed adjustment command, thus completing the adaptive control closed loop. In step 3.1, threshold trigger control is used. When the real-time swing amplitude is less than the safety threshold, the gyro damper is kept at a low speed, which means that high power output is not required, thus avoiding unnecessary high-speed operation. In step 3.5, closed-loop energy consumption optimization is performed. The PWM duty cycle is precisely controlled by equation (34) to avoid motor overshoot or redundant output, thereby further reducing ineffective energy consumption. Since the speed of the gyro damper is adaptively adjusted according to the swing amplitude of the boat, the energy waste caused by maintaining high-speed rotation is avoided. Therefore, the energy consumption of the boat is greatly reduced, and the energy utilization efficiency is improved.

[0209] Secondly, it extends the service life of the equipment. By reducing the operating time of the gyroscope stabilizer in unnecessary high-speed rotation, it reduces mechanical stress and wear on the equipment, thereby extending its service life and reducing maintenance costs.

[0210] Third, it improves the roll reduction effect. This application achieves precise adjustment of the gyro-based roll stabilizer speed through an adaptive control strategy based on the ship's roll amplitude, ensuring its efficient operation under different sea conditions and significantly improving the roll reduction effect and ship stability. Details are as follows.

[0211] Target speed adaptive matching: In step 2, a nonlinear mapping relationship between the swing amplitude and the target speed is established according to equation (21), and a dynamic adjustment factor is introduced in combination with equation (29) so that the target speed is automatically adjusted with the change of swing intensity to ensure that the magnitude of the counter torque matches the disturbance.

[0212] Closed-loop control dynamic response: In step 3, a PID closed-loop control mechanism based on real-time deviation is constructed through equations (31) to (34). The motor speed is controlled by the PWM duty cycle to achieve rapid closed-loop adjustment from swing to target to actual, thereby enhancing the system's response capability to sudden disturbances.

[0213] It can adjust the rotation speed in real time according to the actual rolling amplitude of the boat, as detailed below.

[0214] In step 3, the real-time monitoring and adjustment logic of the closed-loop feedback control of the gyroscope speed damper is explained as follows.

[0215] In step 3.1, the real-time sway amplitude is obtained by calculating the sway acceleration using an integral algorithm, which serves as the input basis for real-time adjustment.

[0216] In step 3.2, the real-time deviation is calculated: Equation (31) associates the target speed corresponding to the real-time swing amplitude with the actual speed to generate a deviation signal for real-time adjustment;

[0217] In steps 3.3 to 3.4, the real-time control output is: Equation (32) is continuous PID control, which responds to the deviation in real time; Equation (33) is discretized and adapted to the real-time sampling of the digital system to realize the real-time conversion of deviation to control quantity.

[0218] In step 3.5, real-time speed adjustment: Equation (34) converts the real-time control quantity into PWM duty cycle, directly adjusts the motor speed, and completes the closed loop of real-time swing amplitude → real-time speed.

[0219] This ensures that the gyro damper is always in optimal working condition, improving the damping effect and enhancing the stability and comfort of the boat, as detailed below.

[0220] The core is a closed-loop control system that achieves precise speed matching, efficient counter-torque output, and real-time suppression of swaying.

[0221] Speed-swing precision matching: In steps 2.5 to 2.9, equation (21) is a nonlinear mapping based on experimental data, and equation (29) is a dynamic adjustment factor to ensure that the counter torque output by the gyroscope anti-swing device matches the magnitude and phase of the ship's swing torque, thus avoiding insufficient speed or excessive speed. Insufficient speed means weak anti-swing, while excessive speed will cause new disturbances.

[0222] Real-time deviation compensation: In steps 3.2 to 3.6, the PID closed-loop control of equations (31) to (34) quickly corrects the speed deviation. For example, if the sway amplitude increases suddenly due to the sudden change of the waves, the counter torque is ensured to track the sway change in real time and suppress the increase of sway.

[0223] Swing convergence control: In step 3.6, through a cycle of monitoring → adjustment → re-monitoring, the swing amplitude is quickly converged to below the safe threshold, reducing the swing frequency and amplitude, and improving stability and comfort.

[0224] Fourth, rapid roll convergence. Through continuous monitoring and dynamic adjustment, the control system can converge the roll amplitude to below the safe threshold in a short time, reducing the roll frequency and maximum amplitude, thereby improving the ship's navigation stability and passenger comfort.

[0225] Fifth, accurate swing amplitude detection. In step 1, this application uses a data acquisition module with high sampling frequency and high-precision analog-to-digital conversion capability to achieve high-quality acquisition of the analog signal output by the accelerometer, and completes analog-to-digital conversion through a high-performance ADC to ensure the accuracy of subsequent digital calculations. Furthermore, the swing angle is obtained by integrating the acceleration using the trapezoidal integral method (1) and (2). The high sampling frequency can effectively reduce the integration error and improve the angle estimation accuracy; combined with the Kalman filter algorithm, continuous state updates are achieved, significantly enhancing the anti-noise performance. The above-described analog signal → digital signal → swing angle link realizes high-precision, low-noise swing amplitude detection, providing a reliable basis for subsequent speed control. Details are as follows.

[0226] A high-sampling-frequency and high-precision analog-to-digital conversion data acquisition module is used to ensure the accuracy of data input and provide reliable raw data for subsequent mathematical calculations, corresponding to step 1.2 acceleration data acquisition in step 1 swing amplitude detection.

[0227] In step 1.2, the accelerometer is connected to a high-performance data acquisition module. The data acquisition module must have a high sampling frequency and high-precision analog-to-digital conversion capability. Let the sampling frequency be , in Hz, meaning it collects acceleration data points per second. This step directly defines the core performance requirements of the data acquisition module; high sampling frequency and high-precision analog-to-digital conversion are the foundation for subsequent angle calculations and filtering.

[0228] Analog-to-digital conversion is the core hardware function of step 1.2, and the specific logic is as follows.

[0229] Signal type differences: Accelerometers output analog signals, such as voltage signals, typically ranging from 0 to 5V, which are proportional to the acceleration value. For example, 0.1V corresponds to 0.1g acceleration. However, the trapezoidal integration method in step 1.3 and the Kalman filtering in step 1.4 need to be implemented through mathematical calculations in the digital control system, and cannot directly process analog signals. Digital control systems, such as microcontrollers and PLCs, are used for this purpose.

[0230] The role of analog-to-digital conversion: The data acquisition module uses a high-precision ADC chip to convert the analog acceleration signal output by the sensor into a digital signal. For example, a digital value of 1024 corresponds to an analog value of 2.5V, which in turn corresponds to an acceleration of 0.25g. This completes the key conversion from analog physical signal to digital calculation data, which is the prerequisite for all subsequent accurate calculations.

[0231] Step-relatedness: Without analog-to-digital conversion, the analog signal from the accelerometer cannot be recognized by the digital system, and the angle calculation, filtering, and amplitude acquisition in steps 1.3 to 1.5 cannot be carried out. Therefore, the analog-to-digital conversion in step 1.2 is a necessary step in the entire process of swing amplitude detection.

[0232] The sampling frequency has a core impact on the accuracy of swing amplitude detection, and its specific technical contributions are reflected in three dimensions, as explained below.

[0233] The first dimension is to avoid frequency aliasing and eliminate false signals: According to the Nyquist sampling theorem, the sampling frequency must be greater than twice the highest frequency of the acquired signal in order to avoid high-frequency noise / interference from aliasing into the low-frequency band, which would cause distortion in the calculation of the swing angle.

[0234] The second dimension is to reduce integration error and improve the accuracy of angle calculation: the trapezoidal integration method in step 1.3 calculates the angle through discrete acceleration data. The trapezoidal integration method includes equation (1) and equation (2). The smaller the positive correlation between integration error and sampling period, the finer the integration interval, and the closer the trapezoidal area is to the true integration value.

[0235] The third dimension ensures the effectiveness of Kalman filtering and reduces the impact of noise: Kalman filtering in step 1.4 relies on continuous state updates. A high sampling frequency can make the evolution of the state vector closer to the actual rolling dynamics of the boat, reduce the cumulative error of process noise, and improve the stability of the angle after filtering.

[0236] Sixth, the rotational speed mapping model is scientific. In step 2, this application optimizes the sample collection scheme through orthogonal experimental design. In step 2.2, the obtained swing amplitude and rotational speed samples are normalized and outlier removal is performed. In step 2.3, equation (21) constructs a nonlinear swing amplitude-base target rotational speed mapping model based on a multilayer perceptron neural network (MLP), and uses cross-validation and backpropagation algorithms to improve fitting accuracy and generalization ability. Experimental results show that the model can accurately reflect the target rotational speed requirements under different swing conditions, improving the scientific nature and adaptability of the modeling.

[0237] Seventh, the speed adaptive adjustment is precise. In steps 2.5 to 2.8, based on equation (21) and the dynamic factor correction model equation (29), the adaptive target speed is calculated in real time, taking into account the swaying dynamic characteristics caused by the change in the hull force, and realizing the dynamic mapping of swaying amplitude to target speed. In steps 3.1 to 3.4, the integral separation PID algorithm is used to compensate for the error between the actual speed and the target value. The integral separation PID algorithm includes equations (32) and (33), and the drive motor is controlled by PWM signal equation (34) to ensure that the speed adjustment has good real-time performance and response accuracy.

[0238] Eighth, the feedback control is stable and reliable. In step 3, this application introduces a closed-loop mechanism of target-actual-deviation-adjustment. In step 3.1, the sway amplitude is acquired in real time and compared with a preset safety threshold. If the amplitude exceeds the range in equation (31), the controller is triggered to calculate the deviation signal. In equation (34), the PID control module is called to dynamically correct the PWM duty cycle, thereby driving the motor to adjust the speed and realize continuous iterative speed update. This closed-loop control mechanism ensures that the gyroscope anti-roll device always responds quickly to sea state disturbances, so that the sway amplitude converges within the threshold range, thereby improving the stability and reliability of the system.

[0239] To facilitate a concise and systematic understanding of the technical solution of this application, a comprehensive description is provided below.

[0240] The purpose of this application is to solve the problem of energy waste caused by the continuous high-speed rotation of existing gyro dampers, and to achieve energy-saving effect by adaptively controlling the rotation speed of the gyro damper according to the sway amplitude of the boat.

[0241] First, accelerometers are installed on the boat to accurately acquire data on the boat's roll amplitude through data acquisition, integration, and filtering. Then, orthogonal experimental design is used to collect experimental data. After preprocessing, a multilayer perceptron neural network is used to establish a mapping model between roll amplitude and rotational speed, which is then trained and optimized. Next, real-time roll amplitude data is input into the model to calculate the target rotational speed. Pulse width modulation (PWM) technology is used to control the drive motor speed, while the actual rotational speed is monitored in real time. An integral-separated PID algorithm is used to compensate for errors, and the target rotational speed is adjusted considering the boat's dynamic characteristics. Furthermore, after adjusting the rotational speed, the roll amplitude is continuously monitored. When the speed exceeds a safety threshold, the rotational speed deviation is calculated, and a discretized PID control algorithm is used to adjust the PWM duty cycle to achieve closed-loop feedback control.

[0242] This application has the advantages of significant energy saving, extended equipment service life, improved anti-sway effect, and strong innovation. It can accurately detect the sway amplitude, scientifically establish a speed mapping model, and accurately perform speed adaptive adjustment to ensure stable and reliable feedback control.

[0243] The innovative points are as follows.

[0244] Multi-step adaptive control: Adaptive control of the gyro damper's speed is achieved through four key steps: swing amplitude detection, speed mapping model establishment, adaptive speed adjustment, and feedback control. Swing amplitude detection utilizes a high-precision accelerometer, and accurate swing amplitude data is obtained through data acquisition, integration, and filtering. and .

[0245] Precise roll amplitude detection: A suitable accelerometer is installed at a suitable position near the vessel's center of gravity, and acceleration data is collected using a high sampling frequency data acquisition module. , The swing angle was calculated using the trapezoidal integral method. , The data is filtered using a Kalman filter algorithm, and the swing amplitude is obtained at preset time intervals.

[0246] Complex rotational speed mapping model: Experimental data were collected using an orthogonal experimental design. After preprocessing, a multilayer perceptron (MLP) neural network was selected to establish a mapping model between the sway amplitude and rotational speed. , And it is trained and optimized using gradient descent.

[0247] Comprehensive adaptive speed adjustment: Target speed calculated based on a multinomial regression model. The drive motor speed is controlled using PWM technology. The actual speed is monitored in real time by an encoder, and the error is compensated using an integral-separated PID algorithm. At the same time, a dynamic adjustment factor is introduced to consider the dynamic characteristics of the boat. Adjust target speed .

[0248] Closed-loop feedback control: Continuously monitor the ship's roll amplitude With safety threshold Comparison, calculate speed deviation when exceeding threshold Discrete PID control algorithm is adopted Adjusting the PWM duty cycle This achieves closed-loop feedback, enabling the gyroscope stabilizer to reach its optimal operating state.

Claims

1. A method for adaptive control of the rotational speed of a gyroscope damper, characterized in that: Includes the following steps, Step 1: Obtain the lateral roll amplitude of the boat and longitudinal sway amplitude , to increase the horizontal sway amplitude and longitudinal sway amplitude Synthesize to obtain swing amplitude ; Step 2: Obtain the preprocessed experimental data, establish a mapping model between the rolling amplitude and rotational speed, and train and optimize the model using gradient descent to obtain a well-trained mapping model. Then, use the real-time obtained boat rolling amplitude... The input is fed into the trained mapping model to calculate the target rotational speed of the gyro stabilizer under the current swing amplitude. And it is used to adaptively adjust the rotation speed of the gyroscope damper; Step 3: Record the real-time boat roll amplitude With safety threshold The speed deviation was calculated by comparison when the threshold was exceeded. The control output is obtained based on the discretized PID control algorithm. Closed-loop feedback continues until the gyroscope stabilizer reaches its optimal operating state.

2. The adaptive control method for the rotational speed of a gyroscope damper according to claim 1, characterized in that: In step 1, the swing amplitude is obtained. The steps include the following steps: The lateral acceleration of the boat is obtained by monitoring an acceleration sensor installed near the center of gravity of the boat. and longitudinal acceleration ; Based on the trapezoidal integral method, according to the lateral acceleration of the boat... Calculate the lateral roll angle of the boat According to the longitudinal acceleration of the boat Calculate the longitudinal roll angle of the boat ; the angle of the boat's lateral roll The filtered lateral sway angle is obtained after processing by a filtering algorithm. The longitudinal roll angle of the boat The longitudinal sway angle is obtained after processing by a filtering algorithm. Based on the filtered lateral sway angle Calculate the lateral roll amplitude of the boat Based on the filtered longitudinal sway angle Calculate the longitudinal roll amplitude of the boat .

3. The adaptive control method for the rotational speed of a gyroscope damper according to claim 2, characterized in that: In step 1, the filtering algorithm is the Kalman filter algorithm.

4. The adaptive control method for the rotational speed of a gyroscope damper according to claim 1, characterized in that: In step 2, orthogonal experiments are used to collect experimental data, which is then preprocessed to obtain preprocessed experimental data. The experimental data includes the rotational speed n of the gyroscope stabilizer and the corresponding stabilizer effect. The ship's roll angle, pitch angle, and yaw acceleration. Lateral roll amplitude, longitudinal roll amplitude, and real-time ship roll amplitude The preprocessing steps include cleaning to remove outliers and normalization.

5. The adaptive control method for the rotational speed of a gyroscope damper according to claim 1, characterized in that: In step 2, a mapping model between sway amplitude and rotational speed is established based on a multilayer perceptron (MLP) neural network. The step of training and optimizing the model using gradient descent includes using MSE as the error function, updating the weights and biases of the mapping model between sway amplitude and rotational speed using gradient descent, and optimizing the mapping model between sway amplitude and rotational speed using cross-validation.

6. The adaptive control method for the rotational speed of a gyroscope damper according to claim 1, characterized in that: In step 2, the trained swing amplitude and rotation speed mapping model is given by equation (21). （21） In equation (21), The target rotational speed of the gyroscope damper under the current swing amplitude. The coefficients of the multinomial regression model are... Let be the order of the polynomial. This represents the real-time sway amplitude of the boat; for The square of, for h times squared.

7. The adaptive control method for the rotational speed of a gyroscope damper according to claim 1, characterized in that: Step 2 also includes obtaining the target rotational speed. The subsequent correction steps include introducing a dynamic adjustment factor α to obtain the final target rotational speed n. final Using the target rotational speed n final As the corrected target speed .

8. The adaptive speed control method for a gyroscope damper according to claim 7, characterized in that: In step 2, the dynamic adjustment factor α is calculated according to equation (28), and the final target rotational speed n is obtained by correcting it according to equation (29). final , （28） In equation (28), As a dynamic adjustment factor, The adjustment coefficient is dynamically adjusted based on the specific characteristics of the vessel. The angular acceleration of the boat's roll; （29） In equation (29), For the final target speed, The target rotational speed.

9. The adaptive control method for the rotational speed of a gyroscope damper according to claim 1, characterized in that: In step 3, the rotational speed deviation is calculated according to equation (31). , （31） In equation (31), Current speed With target speed The deviation between them This is the actual rotational speed of the motor driving the current gyroscope stabilizer.

10. The adaptive control method for the rotational speed of a gyroscope damper according to claim 1, characterized in that: In step 3, the discretized PID control algorithm is Equation (33). （33） In equation (33), It is the current number Control output at the next sampling time; It is a proportionality coefficient used to respond to deviations proportionally, thereby speeding up the system's response. It is the current number Deviation during the second sampling; It is the first Deviation during the second sampling; These are the integral coefficients used to eliminate the steady-state error of the system; The sampling period; It is the first step in the control process Deviation during the second sampling; These are differential coefficients used to predict the trend of deviation changes and improve the dynamic performance of the system.

Images