A method and apparatus for self-configuration and self-tuning of autopilot parameters based on simplified ship descriptions

By using a self-configuration and self-tuning method for autopilot parameters based on a simplified description, the problem of mismatch between autopilot control gain and ship bow turning dynamics is solved, enabling adaptive control gain configuration without professional knowledge and improving the stability and continuity of heading control.

CN123104903APending Publication Date: 2026-11-17XIAMEN XINNUO TECH
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611285223.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-08-24
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Existing automatic steering control systems require readjustment of control parameters after a ship changes its loading status, speed, or is replaced. Furthermore, non-professionals often lack the expertise to provide professional dynamic parameters, leading to a mismatch between control gain and the ship's bow turning dynamics.

Method used

By obtaining simplified descriptions of ship type, length, tonnage, loading status, and steering preferences, and utilizing dimensionless hydrodynamic prior parameters and model parameter correction factors, the system automatically configures initial values ​​for rudder effect gain and time constant. Combined with online identification and adaptive boundaries, it gradually approximates the target control gain, thereby achieving automatic rudder parameter self-tuning.

Benefits of technology

Matching of autopilot control gain can be achieved without professional knowledge, lowering the threshold for deployment and use, improving the stability and continuity of heading control, and avoiding parameter drift and gain step update disturbances.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123104903A_ABST
    Figure CN123104903A_ABST
Patent Text Reader

Abstract

This invention provides a method and apparatus for self-configuration and self-tuning of automatic steering parameters based on a simplified ship description. The method includes a simplified ship description of the vessel type, length, tonnage, loading status, configured speed, and steering preferences. Based on this description, dimensionless hydrodynamic prior parameters and correction factors are determined. Initial values ​​of rudder effect gain, time constant, and adaptive boundary are calculated. The initial expected closed-loop bandwidth is determined based on steering preferences and speed, and the initial control gain is tuned. During navigation, when excitation conditions are met, the dynamic model of the turning head is identified online, and the results are constrained within the boundary. The target bandwidth and target control gain are then re-determined based on the online parameters, allowing the current gain to gradually approach the target value with single-step limiting. Rudder angle commands are generated based on the heading deviation for steering. This invention obtains initial gain values ​​matching the ship without requiring input of specialized parameters, helps improve the convergence speed of online adjustments, and suppresses identification drift and gain jumps.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of ship automatic steering control technology, and in particular to a method and apparatus for self-configuration and self-tuning of automatic steering parameters based on a popular ship description. Background Technology

[0002] Automatic steering systems maintain or change a ship's course by adjusting the rudder angle based on the deviation between the target course and the actual course. The effectiveness of this course control is closely related to the degree of matching between the control gain and the ship's turning dynamics. Because different ships vary in hull type, size, tonnage, and loading status, and the turning dynamics of the same ship can change under different speeds and loading conditions, a control gain suitable for one type of ship or navigation state may no longer be compatible with the current turning dynamics after the ship or navigation state changes.

[0003] Currently, automatic steering systems typically use manual tuning or preset parameter settings for configuration. Manual tuning requires professionals to repeatedly test the steering at sea and gradually adjust the gain based on experience, which is time-consuming and requires retuning after changes in the vessel's loading status, speed, or even after a change of vessel. While preset parameter settings are convenient, the given parameters are fixed and difficult to adjust according to the vessel's loading status, speed, and individual differences. Other parameter determination methods require users to pre-input specialized characteristic quantities such as the vessel's turning index and time constant, or directly input the control gain. However, ordinary crew members usually lack knowledge of these specialized characteristic quantities and cannot provide them correctly, making these methods difficult to implement on actual vessels.

[0004] In addition, there are schemes that adjust control parameters based on navigation data, but the initial parameters are usually derived from fixed default values ​​or preset parameters. When there is a deviation between the initial parameters and the actual ship turning dynamics, parameter adjustment requires a certain convergence process; while when the ship is in a state of relatively small dynamic changes, such as straight navigation, the collected data is difficult to fully reflect the ship's turning characteristics, which can easily cause the control parameters to continuously deviate from the appropriate operating range during actual navigation.

[0005] In view of the above, this application is hereby submitted. Summary of the Invention

[0006] This invention discloses a method for self-configuration and self-tuning of autopilot parameters based on popular ship descriptions. It aims to solve the problem that when non-professional users cannot provide professional dynamic parameters, the autopilot control gain is difficult to match with the actual bow turning dynamics of the ship during initial commissioning and changes in navigation status.

[0007] The first embodiment of the present invention provides a method for self-configuration and self-tuning of autopilot parameters based on a simplified description of ships, including: Obtain a simplified ship description including ship type, length, tonnage, loading status, configured speed and steering preference; determine dimensionless hydrodynamic prior parameters based on the ship type; and determine model parameter correction factors based on the ship length, tonnage and loading status. The initial values ​​of rudder effect gain, time constant, and adaptive boundary are determined based on the dimensionless hydrodynamic prior parameters, model parameter correction factors, ship length, and configured speed. The initial expected closed-loop bandwidth is determined based on the steering preference, configured speed, and initial value of time constant. The initial value of control gain is determined based on the initial expected closed-loop bandwidth, initial value of rudder effect gain, and initial value of time constant. During navigation, the current heading, current speed, rudder angle, and turning rate are acquired. When the preset excitation conditions are met, the turning dynamic model is identified online. The identified rudder effect gain and time constant are constrained within the adaptive boundary to obtain the online rudder effect gain and online time constant. The target desired closed-loop bandwidth is determined based on the steering preference, current speed, and online time constant. The target control gain is determined based on the target desired closed-loop bandwidth, online rudder effect gain, and online time constant. The initial value of the control gain is used as the initial value of the current control gain, and the current control gain is gradually made to approach the target control gain in a single-step limiting manner. Based on the heading deviation between the target heading and the current heading, and the current control gain, a rudder angle command is generated to control the servo motor.

[0008] The second embodiment of the present invention provides an autopilot parameter self-configuration and self-tuning device based on a popular ship description, comprising: The description acquisition unit is used to acquire a simplified ship description including ship type, length, tonnage, loading status, configured speed and steering preference, determine dimensionless hydrodynamic prior parameters based on the ship type, and determine model parameter correction factors based on the ship length, tonnage and loading status. The initial value configuration unit is used to determine the initial value of the rudder effect gain, the initial value of the time constant, and the adaptive boundary based on the dimensionless hydrodynamic prior parameters, model parameter correction factors, ship length, and configured speed; to determine the initial expected closed-loop bandwidth based on the steering preference, configured speed, and initial value of the time constant; and to determine the initial value of the control gain based on the initial expected closed-loop bandwidth, the initial value of the rudder effect gain, and the initial value of the time constant. The online identification unit is used to acquire the current heading, current speed, rudder angle and turning rate during navigation. When the preset excitation conditions are met, it performs online identification of the turning dynamic model and constrains the identified rudder effect gain and time constant within the adaptive boundary to obtain the online rudder effect gain and online time constant. The gain tuning unit is used to determine the target desired closed-loop bandwidth based on the steering preference, current speed and online time constant, and to determine the target control gain based on the target desired closed-loop bandwidth, online rudder effect gain and online time constant; the initial value of the control gain is used as the initial value of the current control gain, so that the current control gain gradually approaches the target control gain in a single-step limiting manner; The steering control unit is used to generate rudder angle commands for controlling the servo motors based on the heading deviation between the target heading and the current heading, as well as the current control gain.

[0009] Beneficial effects: This invention uses simplified ship descriptions such as ship type, length, tonnage, loading status, configured speed, and steering preferences—which can be provided by users without requiring specialized knowledge—as configuration inputs. Initial values ​​for the ship's rudder gain and time constant are obtained through conversion using dimensionless hydrodynamic prior parameters and model parameter correction factors. These are then combined with the initial expected closed-loop bandwidth determined by steering preferences and configured speed to tune the initial control gain. This allows the autopilot to achieve control gain matching the ship's dimensions, tonnage, loading status, and speed immediately upon power-up, without requiring sea trials or user input of specialized quantities such as turning index, time constant, or control gain. Compared to solutions requiring manual tuning or input of specialized parameters, this significantly lowers the deployment and usage threshold. Furthermore, compared to solutions with fixed tiers based on ship type, it allows initial parameter values ​​to be converted and corrected according to the ship's actual dimensions and loading conditions.

[0010] This invention starts with the aforementioned initial control gain and uses adaptive boundary constraints determined by the initial model value for online identification. The identification results are updated only when preset excitation conditions are met, and the current control gain gradually approaches the target control gain in a single-step limiting manner. Thus, during navigation, the target desired closed-loop bandwidth can be re-determined according to the current speed and online time constant, so that the control gain continuously follows the dynamic changes of the ship's turning. It also helps to suppress problems such as slow convergence of online tuning without prior knowledge, drift of identification results under conditions of lack of excitation such as straight navigation, and disturbance of the servo motor by gain step update, which is beneficial to improving the stability and continuity of the automatic steering heading control process. Attached Figure Description

[0011] Figure 1 This is a flowchart illustrating a method for self-configuration and self-tuning of autopilot parameters based on a simplified ship description, provided in the first embodiment of the present invention. Figure 2 This is a schematic block diagram of the system hardware composition provided by the present invention; Figure 3 This is a schematic diagram of online self-tuning convergence provided by the present invention; Figure 4This is a schematic diagram of a module for an autopilot parameter self-configuration and self-tuning device based on a simplified ship description, provided in the second embodiment of the present invention. Detailed Implementation

[0012] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0013] To better understand the technical solution of the present invention, the embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0014] This invention discloses a method for self-configuration and self-tuning of autopilot parameters based on popular ship descriptions. It aims to solve the problem that when non-professional users cannot provide professional dynamic parameters, the autopilot control gain is difficult to match with the actual bow turning dynamics of the ship during initial commissioning and changes in navigation status.

[0015] Please see Figure 1 and Figure 2 The first embodiment of the present invention provides a method for self-configuration and self-tuning of autopilot parameters based on a popular ship description. The method for self-configuration and self-tuning of autopilot parameters based on a popular ship description is executed by a program deployed in a microprocessor. The microprocessor is connected to a human-machine interaction configuration unit, a heading and turning rate sensor, a ship speed sensor, a rudder angle feedback sensor, and a steering gear.

[0016] S101, Obtain a simplified ship description including ship type, length, tonnage, loading status, configured speed and steering preference, determine dimensionless hydrodynamic prior parameters based on the ship type, and determine model parameter correction factors based on the ship length, tonnage and loading status; In this embodiment, after the system is powered on, it first enters the configuration phase. The user inputs a simplified description of the ship through the human-machine interface configuration unit, and the microprocessor determines the dimensionless hydrodynamic prior parameters and model parameter correction factors of the ship accordingly. The human-machine interface configuration unit can be a touch screen menu on the autopilot control panel, a rotary menu, a physical DIP switch, or a host computer software interface that communicates with the autopilot; its specific form does not constitute a limitation of the present invention.

[0017] The simplified vessel description includes six items: vessel type, length, tonnage, loading status, configured speed, and steering preference. These can be selected or filled in directly. Vessel type is categorized into yachts / speedboats, fishing boats, passenger / ro-ro ships, container ships, tankers / bulk carriers, and tugboats. Length can be selected from small, medium, large, and super-large categories, or the user can directly enter a specific length value. The tonnage can be entered directly or selected according to light, medium, and heavy categories; the loading status can be selected from empty, half-loaded, and fully loaded; the configured speed can be selected according to low, medium, and high speed categories, or the user can directly enter a specific speed value. The values ​​are uniformly converted to meters per second within the microprocessor; steering preferences can be selected from three levels: fuel-efficient, standard, and precise tracking. The above six items are all common-sense information that crew members can directly know in daily operations. They do not include professional hydrodynamic parameters such as turning index and time constant, nor do they include control parameters such as proportional gain, integral gain, and derivative gain. Therefore, users are not required to have professional knowledge of ship maneuvering or automatic control. This is precisely why this invention refers to the above description as a simplified ship description.

[0018] To illustrate how the microprocessor derives the ship's dynamic characteristics from the above simplified description, this embodiment uses a first-order Nomoto model to approximate the ship's turning dynamics, that is, to characterize the dynamic relationship between rudder angle changes and the ship's turning rate response. The expression is as follows:

[0019] In the formula, For the course, For the rate of change of heading, For turning rate, The rate of change of the turning rate, As the rudder angle, For rudder effect gain, It is a time constant. These are nonlinear coefficients. Rudder effect gain. The time constant characterizes the steady-state turning rate produced by a unit rudder angle; the larger the value, the more sensitive the ship's response to the rudder. The time scale required for the turning rate response to reach a steady state is characterized; a larger value indicates stronger inertia and a slower response during the ship's turning. The nonlinear coefficient... This factor is used to characterize the damping enhancement effect under high turn rate conditions. However, its influence is minimal under low turn rate heading-keeping conditions. Therefore, in the subsequent online identification step S103, the linear portion of the model is used as the identification object. It is understandable that... and It is not fixed, but varies with the ship type, ship size, speed and loading status, which is exactly what this invention needs to automatically configure and identify online during navigation.

[0020] because and All of these are dimensional quantities, and their values ​​are simultaneously determined by the captain. With speed The impact of this cannot be directly addressed by providing typical values ​​based on ship type. Therefore, this embodiment uses ship length as the basis for the calculation. and configuration speed Normalizing them yields the dimensionless rudder effect gain prior parameters. and dimensionless time constant prior parameters Both of these are related to the dimensional rudder effect gain. Time constant The conversion relationship between them is as follows:

[0021] After the above normalization process, and It only reflects hydrodynamic characteristics related to the ship type, such as hull form, rudder area ratio, and block coefficient, and is largely independent of the ship's specific dimensions and speed. Therefore, it can be approximated as a constant for the same type of ship. This property allows the microprocessor to retrieve the ship's parameters from a pre-stored table of dimensionless hydrodynamic prior parameters based solely on the user-selected ship type. and The aforementioned prior parameter table can be pre-written after statistical summarization of ship model maneuverability test data, actual ship turning test data, historical navigation data, or maneuverability data from publicly available literature. The general rule is that for ships with full hulls, large block coefficients, and small rudder area ratios, such as tankers and bulk carriers, the... Smaller Larger rudders result in less responsive steering and greater turning inertia; conversely, slender vessels like yachts and speedboats with larger rudder-to-surface-area ratios exhibit the opposite characteristics. Larger The value is relatively small, resulting in rudder sensitivity and rapid bow turning; the value for fishing boats, passenger ships (ro-ro), cargo ships (container ships), tugboats, and other ship types falls between these two values. Preferably, the prior parameter table can also store dimensionless nonlinear prior parameters corresponding to each ship type. This is provided for implementations that require consideration of Norrbin nonlinear terms.

[0022] Results obtained by looking up the table by ship type only and This reflects the common characteristics of this type of ship and cannot reflect the individual differences of this ship in terms of specific length, tonnage, and loading status. Therefore, in this embodiment, the microprocessor also retrieves the rudder effect gain correction factor from a pre-stored correction factor table based on the length, tonnage, and loading status input by the user. and time constant correction factor The correction factor is used to individually correct the common priors obtained by looking up tables by ship type. Its value follows this rule: for the same ship type, the larger the tonnage and the heavier the load, the greater the displacement and moment of inertia. This results in a relatively lower steady-state turning rate per unit rudder angle and a relatively stronger turning inertia. The value is relatively reduced. The value increases relatively; conversely, it decreases under no-load or light-load conditions. The value increases relatively. The value is relatively smaller. When the user provides the ship's length and tonnage in a tiered manner, the microprocessor directly retrieves the corresponding correction factor according to the selected tier; when the user directly enters the ship's length... When dealing with tonnage values, the microprocessor can determine the correction factor using a linear interpolation method between adjacent gears to achieve a more continuous correction effect.

[0023] It should be noted that the method for obtaining dimensionless hydrodynamic prior parameters is not limited to looking up tables by ship type. As an equivalent alternative implementation, they can also be directly calculated using analytical fitting formulas with ship length and tonnage as independent variables. and Alternatively, it can be obtained by interpolation among several sets of known ship type samples; similarly, the correction factor... and The values ​​can be fixed according to the above-described grading method, or further subdivided according to the loading status, actual draft, or trim to improve the correction accuracy. The results obtained by the above methods can all be used as inputs for the subsequent step S102 without affecting the implementation of the present invention.

[0024] S102, determine the initial value of rudder effect gain, the initial value of time constant, and the adaptive boundary based on the dimensionless hydrodynamic prior parameters, model parameter correction factors, ship length, and configured speed; determine the initial expected closed-loop bandwidth based on the steering preference, configured speed, and initial value of time constant, and determine the initial value of control gain based on the initial expected closed-loop bandwidth, initial value of rudder effect gain, and initial value of time constant; After completing the mapping from the simplified ship description to the ship's hydrodynamic prior information, the microprocessor enters the second half of the configuration phase. First, it calculates the initial value of the ship's dynamic model and its adaptive boundary from the dimensionless prior, and then tunes the initial value of the control gain from the initial value of the model.

[0025] Specifically, the microprocessor will use the dimensionless rudder effect gain prior parameters and dimensionless time constant prior parameters According to the captain entered by the user and configuration speed Scale conversion is performed, and the rudder effect gain correction factor is used. and time constant correction factor Individualized adjustments were made to obtain the initial value of the ship's rudder effect gain. and initial value of time constant Its expression is:

[0026] The above two equations are the inverse forms of the normalization relation: , It carries the hydrodynamic characteristics common to similar ship types, factors / and / To restore it to the ship's actual dimensions and speed, and , Then, individual adjustments are made based on the ship's length, tonnage, and loading status. Multiplying these three values ​​yields the initial values ​​of the model parameters that match the ship's current state. This formula shows that the initial value of the time constant for the same ship changes depending on whether the load is increased or the speed is reduced. As the initial value of the rudder effect gain increases, so does the rudder effect gain. This decreases accordingly, which is consistent with the actual law that the ship's bow inertia increases and its rudder response slows down, thus enabling the initial model value given in the configuration stage to change with the operating conditions.

[0027] After obtaining the initial values ​​of the model, the microprocessor uses... and An adaptive boundary is set for the baseline, that is, the rudder effect gain obtained from subsequent online identification is defined. and time constant Values ​​can only be taken from the following range:

[0028] The adaptive boundaries serve to ensure that online identification only makes bounded adjustments near initial values ​​that are already close to the actual characteristics of the vessel, rather than searching freely throughout the entire parameter space. Preferably, the upper and lower bounds are taken as follows:

[0029]

[0030] The microprocessor then determines the initial desired closed-loop bandwidth. The desired closed-loop bandwidth... The response speed of the autopilot's heading control loop is determined by both steering preference and configured speed, and its value is expressed as:

[0031] The formula uses the reciprocal of the initial value of the time constant as the bandwidth reference, and its physical meaning is: the greater the ship's own turning inertia, The larger the bandwidth, the lower the closed-loop bandwidth it can withstand. If the bandwidth is forcibly increased, the rate of change of the rudder angle required by the controller will exceed the actual response capability of the ship and the steering gear, which will cause overrudder and oscillation. Therefore, using the reciprocal of the initial value of the time constant as a reference can make the bandwidth automatically match the inherent time scale of the ship. The steering preference coefficient is obtained by mapping the steering preference gear selected by the user. Preferably, it is 0.6 for fuel saving and steering saving, 0.9 for standard, and 1.3 for precise tracking. The larger the value, the higher the closed-loop bandwidth and the tighter the heading tracking, but the more frequent the servo movements. Based on this, the user can make a trade-off between tracking accuracy and servo load without having to understand the concept of bandwidth itself. For speed coefficient, The pre-set service speed for this vessel; when configured speed Not exceeding the service speed hour Set to 1, bandwidth remains unchanged; when Higher than hour The ratio is less than 1, causing the bandwidth to decrease as speed increases. This is because higher speeds result in a greater turning rate for the same rudder angle, and maintaining the original bandwidth would easily lead to oversteer. Finally, a bandwidth limiting function restricts the result to a preset lower bandwidth limit. With bandwidth limit Between, among The optimal value should not exceed the response bandwidth of the servo motor itself, in order to reduce the risk that the rudder angle command issued by the controller exceeds the servo motor's following capability.

[0032] After determining the initial desired closed-loop bandwidth, the microprocessor uses the ship's dynamic model as the controlled object and employs the pole placement method to tune the initial value of the proportional gain. Initial value of integral gain and initial value of differential gain Its expression is:

[0033]

[0034]

[0035] The above three equations are obtained from the pole placement method, and their derivation process is as follows: Combining the linear part of the first-order Nomoto model with the relationship that the rate of change of heading equals the rate of turn, it can be seen that the controlled object from the rudder angle to the heading is equivalent to an integral element and a time constant. The inertial components are connected in series; the automatic rudder heading controller adopts a proportional-integral-derivative form, with heading deviation as input and rudder angle command as output, thus forming a third-order closed-loop system. The desired poles of this closed-loop system are configured as a pair with a damping ratio... Natural frequency The conjugate dominant pole, and a pole located at that natural frequency. Find the negative real poles at the doubled value, and then make the coefficients of the same-degree terms in the closed-loop characteristic polynomial equal to those in the characteristic polynomial corresponding to the desired poles. This allows us to solve for the expressions for the initial values ​​of the proportional gain, integral gain, and differential gain, which are the three equations mentioned above. Where... For the desired damping, a value of 0.9 is preferred. This value is between critical damping and underdamping, which can achieve a compromise between suppressing heading overshoot and ensuring turning speed. The desired integral pole coefficient is preferably set to 0.6, ensuring the integral pole is located to the left of the dominant pole but not too far away. This eliminates the steady heading error caused by constant crosswinds and flow pressure without significantly reducing the system's response speed. Furthermore, the initial value of the differential gain calculated according to the above formula... If it is less than zero, set it to zero.

[0036] It is worth noting that, substituting the initial value expression of the model and the expression for the expected closed-loop bandwidth into the expression for the initial value of the proportional gain, under the condition that the bandwidth is not limited by upper or lower limits and the configured flight speed is not higher than the service flight speed, we can obtain:

[0037] This formula shows that the initial value of the proportional gain is related to the ship length. and speed The relevant factors have completely canceled each other out, and their values ​​are determined solely by the dimensionless ship type constant. , Correction factor , and steering preference coefficient This decision is precisely the theoretical basis for the present invention's ability to directly provide a reasonable initial gain value based solely on common descriptions such as ship type and steering preference. As for the initial values ​​of integral gain and differential gain, since they have the dimensions of time reciprocal and time, respectively, and retain the time scale factor of the ratio of ship length to speed after normalization, they cannot be determined completely detached from ship dimensions and speed, unlike the initial value of proportional gain.

[0038] As an equivalent alternative implementation, the tuning of the control gain is not limited to the above-mentioned pole placement method. Other model-based tuning methods such as Lambda tuning can also be used. In addition to the first-order Nomoto model, the controlled object model can also be a model containing Norrbin nonlinear terms or a second-order manipulator model. The corresponding gain expression varies with the selected model and tuning method, but all of them belong to the implementation method of directly tuning the initial value of the control gain from the initial value of the model obtained in step S102, which does not affect the implementation of the present invention.

[0039] S103: During navigation, the current heading, current speed, rudder angle and turning rate are acquired. When the preset excitation conditions are met, the turning dynamic model is identified online. The identified rudder effect gain and time constant are constrained within the adaptive boundary to obtain the online rudder effect gain and online time constant. In this embodiment, the microprocessor enters the online self-tuning phase. During navigation, the microprocessor operates according to a preset sampling period. Navigation data is collected periodically: the current heading and turning rate are obtained from a heading and turning rate sensor, which can be a satellite compass, magnetic compass, or a single-antenna combined navigation heading device; the current speed is obtained from a ship speed sensor, which can be a log or a GNSS receiver; and the actual rudder angle is obtained from a rudder angle feedback sensor. The turning rate and rudder angle are used for online identification in this step, the current speed is used in step S104 to schedule the target desired closed-loop bandwidth, and the current heading is used in step S105 to calculate the heading deviation. Preferably, the sampling period is... Take the same value as the control cycle of the autopilot heading controller so that the existing sampling timing of the controller can be directly reused.

[0040] Since the turning dynamic model established in step S102 is a continuous-time model, while the navigation data is collected discretely according to the sampling period, the microprocessor first selects the data according to the sampling period. The model is discretized. Considering the small turning rate under heading hold and small-angle turn-off conditions, and the minimal difference between the Norrbin nonlinear terms and linear terms, the linear part of the first-order Nomoto model is used for discretization and identification. Assuming the rudder angle remains constant within one sampling period (i.e., using the zero-order hold method), the first-order differential equation corresponding to the linear part is solved between two adjacent sampling times, yielding the discrete turning model:

[0041]

[0042] In the formula and They represent the first The turning rate and rudder angle at each sampling time. and These are the coefficients of the discrete model. Therefore, the rudder effect gain in the continuous model... With time constant The identification then transforms into the identification of the coefficients of the two discrete models. and The latter is linear with respect to the observed data and can be solved recursively using the linear regression method.

[0043] Specifically, the microprocessor uses the steering rate and rudder angle from the previous sampling time to construct a regression vector, and the estimated values ​​of the coefficients of the two discrete models to construct a parameter estimation vector, i.e.:

[0044]

[0045] The discrete turning model can then be written in linear regression form, and a recursive least squares algorithm with a forgetting factor is used to recursively estimate it. The recursive process at each sampling time is as follows:

[0046]

[0047]

[0048]

[0049] In the formula, To identify the residual, which is the deviation obtained by subtracting the measured value from the parameter estimate of the previous time from the predicted turning rate at the current time; This is the gain vector, which determines the extent to which the residuals are used to correct the parameter estimates. It is the covariance matrix; As a forgetting factor, its function is to cause earlier historical data to decay exponentially in the recursion, so that the identification results can drift with the dynamic characteristics caused by the ship's loading and speed. The smaller the value, the faster the forgetting and the stronger the tracking ability, but the more sensitive it is to noise. The optimal value is 0.995, at which point the effective memory length of the algorithm is about two hundred sampling points.

[0050] To enable the recursion to begin and to embody the concept of starting with prior initial values, the microprocessor takes a set of discrete coefficients derived from the ship's prior parameters as the initial value of the parameter estimation vector, that is, the initial value of the rudder effect gain obtained in step S102. with initial value of time constant Substituting the above conversion relationship between discrete model coefficients and continuous model parameters, the two resulting discrete coefficients are used as the initial values ​​of the parameter estimation vector, rather than being set to zero or other default values ​​unrelated to this ship.

[0051] The above recursive process is not executed unconditionally at every sampling time. When the ship is sailing in a straight line and its course and rudder angle remain relatively constant, the turning rate is close to zero and the rudder angle is essentially constant. In this case, the components of the regression vector are approximately constant, and the resulting information matrix is ​​nearly singular. The data does not contain sufficient information to distinguish between rudder effect gain and time constant. If parameter estimates are continuously updated under these conditions, the recursive least squares approach will amplify sensor noise as valid information, causing the parameter estimates to slowly deviate from the true values—a phenomenon known as identification drift. Therefore, at each sampling time, the microprocessor first checks whether the preset excitation conditions are met, i.e., whether the current navigation data contains data changes that can be used to update the discrete model coefficients: obtaining the turning rate at the current sampling time and the change in rudder angle between the current and previous sampling times. The above recursive update is only executed when at least one of the following two conditions is met:

[0052]

[0053]

[0054] If neither of the two conditions is met, the discrete model coefficients and covariance matrix remain unchanged, and the update is skipped. In the formula... The turning rate excitation threshold, The excitation threshold for rudder angle change is preferably set as follows: Take 0.3 degrees per second, A 0.5 metric level is used; the two threshold values ​​should be higher than the noise level of the corresponding sensor, but lower than the typical amplitude during normal steering and bow turning processes. This ensures that noise under static conditions is filtered out without missing the actual steering excitation. The "OR" logic is used because either the bow turning rate or the change in rudder angle has a sufficient amplitude, meaning that there has been a valid change in the ship's input and output data, sufficient to support a parameter update.

[0055] After each parameter update, the microprocessor converts the identified discrete model coefficients back into the physical parameters of the continuous-time model, obtaining the identified rudder effect gain before boundary constraints. and identification time constant :

[0056] The converted and Before it can be used directly, the microprocessor projects it onto the adaptive boundary set in step S102 to obtain the online rudder effect gain. and online time constant :

[0057] Please see Figure 3 The projection is a key step in this invention to adjust the constraints online near the prior initial values: since each upper and lower bound is given by the initial model value in step S102 at a fixed multiple, and the initial model value originates from the user-provided simplified ship description, therefore, after projection... and The identification is constrained within the adaptive boundary. This helps the identification converge in a neighborhood close to the actual characteristics of the ship and reduces the risk that a large identification error will cause the control gain to jump to an unreasonable value.

[0058] As an equivalent alternative implementation, online identification is not limited to the recursive least squares algorithm with a forgetting factor. Other recursive identification methods such as Kalman filtering and subspace identification can also be used, as long as the identification results are also constrained within the adaptive boundary. In addition, for ships that are put into operation for the first time, a set of model parameters can be obtained near the dock or in open waters using relay self-tuning as a starting point, and then the online identification in this step can be fine-tuned.

[0059] S104, determine the target expected closed-loop bandwidth based on the steering preference, current speed and online time constant, and determine the target control gain based on the target expected closed-loop bandwidth, online rudder effect gain and online time constant; use the initial value of the control gain as the initial value of the current control gain, and make the current control gain gradually approach the target control gain in a single-step limiting manner; In this embodiment, the microprocessor first determines the target desired closed-loop bandwidth. The formula used in step S102 to determine the initial desired closed-loop bandwidth is formally identical, the only difference being that two input quantities have been replaced by measured and identified values ​​during the navigation process, instead of prior values ​​from the configuration phase: one is based on the first... Current speed at each sampling time Replaces the configured speed entered by the user Based on this, the current speed coefficient is determined. Second, based on the online time constant Replace the initial value of the time constant This serves as the bandwidth benchmark. The target desired closed-loop bandwidth is then obtained. :

[0060] The steering preference coefficient in the formula Preset service speed and bandwidth minimum With bandwidth limit All values ​​are taken from step S102, where The reason for keeping it unchanged is that steering preference reflects the user's trade-off between tracking accuracy and steering gear load; it is a user setting, not a ship state, and should not change automatically with navigation conditions. After the above replacement, the target expected closed-loop bandwidth is simultaneously subject to real-time scheduling from two aspects: firstly, when the ship's turning inertia increases due to changes in navigation conditions such as loading status, the online time constant... As the speed increases, the bandwidth automatically decreases, helping to match the controller's response requirements with the ship's turning dynamics; on the other hand, when the ship accelerates to speeds above its service speed, Consequently, the bandwidth is also reduced, which helps to decrease the risk of oversteer during high-speed navigation. It should be noted that an online time constant must be used here. Non-time constant initial value If it is still used as a bandwidth benchmark If this is the case, the expected closed-loop bandwidth will be fixed at the level calculated from the common description at the time of power-on, and cannot be adjusted according to the changes in the actual dynamic characteristics of the ship. Therefore, the online self-tuning function of this invention will be difficult to fully realize.

[0061] After determining the target desired closed-loop bandwidth, the microprocessor adopts the exact same pole configuration as when tuning the initial value of the control gain in step S102, that is, it uses the same set of desired damping. With expected integral pole coefficients Replace the desired closed-loop bandwidth, rudder effect gain, and time constant with the target desired closed-loop bandwidth. Online rudder effect gain With online time constant Thus, the proportional target gain can be calculated. Integral target gain and differential target gain :

[0062]

[0063]

[0064] When the calculated differential target gain When the value is less than zero, it is also set to zero.

[0065] However, the target control gain is a quantity that is constantly updated with each online identification result and each speed change. If the actual control gain used by the autopilot is directly set to the target value at each sampling moment, the control gain will frequently jump with the fluctuations of the identification results. The jump in control gain will cause a step in the rudder angle command, which will in turn cause a step disturbance to the rudder, resulting in frequent actions of the steering mechanism and additional disturbances in the heading. This problem is particularly prominent when the ship finishes a long period of straight-line navigation and meets the excitation conditions for the first time, causing the parameter estimation to be significantly corrected at once. Therefore, this invention does not directly replace the control gain, but rather makes it gradually approach the target value in a single-step limiting manner. Specifically, when the self-tuning phase starts, the microprocessor uses the initial value of the proportional gain obtained in step S102. Initial value of integral gain and initial value of differential gain Each is used as the current proportional gain Current integral gain and current differential gain The initial value is then used; subsequently, at each sampling time, the current proportional gain is updated using the following formula:

[0066] The current integral gain and the current derivative gain are respectively applied to the target integral gain using the same single-step limiting method. and the differential target gain Adjust gradually. In the formula... This sets a preset upper limit for single-step gain change. The meaning of this formula is: when the absolute value of the difference between the target gain and the current gain does not exceed... When the current gain is equal to the target gain, the update immediately sets the target gain to the target gain in one step; when the absolute value of the difference exceeds... In this case, the update only proceeds along the direction of the difference. The remaining portion is reserved for subsequent sampling times. Therefore, by gradually approaching the target value with a limiting step size, gain jumps can be suppressed, and the possibility of overshoot during the transition can be reduced. Preferably, Take a gain on the order of two percent of the target gain. This is understandable. It can be determined according to the target gain in the above proportion, or fixed values ​​can be preset for the proportional, integral, and derivative gains respectively.

[0067] The aforementioned single-step limiting mechanism, together with the adaptive boundary set in step S102, constitutes a dual protection for the control gain: the adaptive boundary ensures the online rudder effect gain in terms of its value range. With online time constant It will not deviate from the prior initial value by a reasonable multiple, thus the target control gain calculated from it is itself bounded; single-step amplitude limiting ensures that the transition process of the current control gain to the bounded target value is gradual in terms of the rate of change. The combination of the two ensures that even if a certain online identification is subject to strong interference and gives a result with a large deviation, its impact on the actual steering behavior of the autopilot is limited to the amplitude and rate.

[0068] As an equivalent alternative implementation, the smooth transition of the control gain from the current value to the target value is not limited to the step-limiting method described above. It can also be achieved by using a first-order low-pass filter or other methods, that is, making the current gain exponentially approach the target gain by a certain time constant. The effect is the same: to avoid a step change in the gain. This is an equivalent substitution of the present invention.

[0069] S105, Generate a rudder angle command for controlling the servo motor based on the heading deviation between the target heading and the current heading and the current control gain.

[0070] In this embodiment, the microprocessor first obtains the target heading and the current heading. The current heading is provided by the heading and turn rate sensor data collected in step S103. Subtracting the current heading from the target heading yields the heading deviation.

[0071] Subsequently, the heading controller takes the heading deviation as input and the rudder angle command as output, and generates the rudder angle command using a proportional-integral-derivative form. That is, the rudder angle command is composed of the sum of the proportional term of the heading deviation, the integral term of the heading deviation with respect to time, and the derivative term of the heading deviation with respect to time, and the proportional gain multiplied by each of the three terms. Integral gain and differential gain This refers to the current control gain output in step S104 at this sampling moment, which is refreshed with each sampling period. This is precisely the working principle of the parameter self-configuration and self-tuning results described in this invention. In the discrete implementation, the integral term consists of the heading deviation and the sampling period. The product is obtained by accumulating it periodically.

[0072] The rudder angle command is sent from the microprocessor to the servo motor, which then executes steering actions to change the actual rudder angle, thereby changing the turning torque on the ship and causing the heading to converge toward the target heading.

[0073] Please see Figure 4 The second embodiment of the present invention provides an autopilot parameter self-configuration and self-tuning device based on a simplified description of ships, comprising: The description acquisition unit 201 is used to acquire a simplified ship description including ship type, ship length, tonnage, loading status, configured speed and steering preference, determine dimensionless hydrodynamic prior parameters based on the ship type, and determine model parameter correction factors based on the ship length, tonnage and loading status. The initial value configuration unit 202 is used to determine the initial value of the rudder effect gain, the initial value of the time constant, and the adaptive boundary based on the dimensionless hydrodynamic prior parameters, model parameter correction factors, ship length, and configured speed; to determine the initial expected closed-loop bandwidth based on the steering preference, configured speed, and initial value of the time constant; and to determine the initial value of the control gain based on the initial expected closed-loop bandwidth, the initial value of the rudder effect gain, and the initial value of the time constant. The online identification unit 203 is used to acquire the current heading, current speed, rudder angle and turning rate during navigation. When the preset excitation conditions are met, it performs online identification of the turning dynamic model and constrains the identified rudder effect gain and time constant within the adaptive boundary to obtain the online rudder effect gain and online time constant. The gain tuning unit 204 is used to determine the target desired closed-loop bandwidth based on the steering preference, current speed and online time constant, and to determine the target control gain based on the target desired closed-loop bandwidth, online rudder effect gain and online time constant; using the initial value of the control gain as the initial value of the current control gain, the current control gain gradually approaches the target control gain in a single-step limiting manner; The steering control unit 205 is used to generate a rudder angle command for controlling the servo motor based on the heading deviation between the target heading and the current heading and the current control gain.

[0074] The above description is merely a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for self-configuration and self-tuning of autopilot parameters based on simplified ship descriptions, characterized in that, include: Obtain a simplified ship description including ship type, length, tonnage, loading status, configured speed and steering preference; determine dimensionless hydrodynamic prior parameters based on the ship type; and determine model parameter correction factors based on the ship length, tonnage and loading status. The initial values ​​of rudder effect gain, time constant, and adaptive boundary are determined based on the dimensionless hydrodynamic prior parameters, model parameter correction factors, ship length, and configured speed. The initial expected closed-loop bandwidth is determined based on the steering preference, configured speed, and initial value of time constant. The initial value of control gain is determined based on the initial expected closed-loop bandwidth, initial value of rudder effect gain, and initial value of time constant. During navigation, the current heading, current speed, rudder angle, and turning rate are acquired. When the preset excitation conditions are met, the turning dynamic model is identified online. The identified rudder effect gain and time constant are constrained within the adaptive boundary to obtain the online rudder effect gain and online time constant. The target desired closed-loop bandwidth is determined based on the steering preference, current speed, and online time constant. The target control gain is determined based on the target desired closed-loop bandwidth, online rudder effect gain, and online time constant. The initial value of the control gain is used as the initial value of the current control gain, and the current control gain is gradually made to approach the target control gain in a single-step limiting manner. Based on the heading deviation between the target heading and the current heading, and the current control gain, a rudder angle command is generated to control the servo motor.

2. The method for self-configuration and self-tuning of autopilot parameters based on simplified ship description as described in claim 1, characterized in that, The turning dynamic model is a first-order Nomoto model, used to characterize the dynamic relationship between rudder angle changes and the ship's turning rate response. Its expression is: in, Indicates the course. Indicates the rate of change of heading. Indicates the turning rate, Indicates the rate of change of the turning rate. represents the rudder angle, Indicates the rudder effect gain. Represents the time constant. This represents the nonlinear coefficient.

3. The method for self-configuration and self-tuning of autopilot parameters based on simplified ship description as described in claim 2, characterized in that, The determination of dimensionless hydrodynamic prior parameters based on the ship type, and the determination of model parameter correction factors based on the ship length, tonnage, and loading status, specifically involves: Based on the ship type, obtain the dimensionless rudder effect gain prior parameters. and dimensionless time constant prior parameters The dimensionless rudder effect gain prior parameter and dimensionless time constant prior parameters The quantities, normalized according to the ship length and configured speed, are respectively, and their conversion relationships with the rudder effect gain and the time constant are as follows: The rudder effectiveness gain correction factor is then retrieved based on the ship's length, tonnage, and loading status. and time constant correction factor ;in, This refers to the captain in the simplified description of the ship. This indicates the configured speed in the simplified ship description.

4. The method for self-configuration and self-tuning of autopilot parameters based on simplified ship description as described in claim 3, characterized in that, The determination of the initial values ​​of the rudder effect gain, the initial values ​​of the time constant, and the adaptive boundary based on the dimensionless hydrodynamic prior parameters, model parameter correction factors, ship length, and configured speed is specifically as follows: The dimensionless hydrodynamic prior parameters are scaled according to the ship length and configured speed, and then corrected using the model parameter correction factor to obtain the initial value of the rudder effect gain. and initial value of time constant Its expression is: With the initial value of the rudder effect gain and initial value of time constant The adaptive boundary is set for the reference, and its expression is: in, and These represent the adaptive lower bound and adaptive upper bound of the rudder effect gain, respectively. and These represent the adaptive lower bound and the adaptive upper bound of the time constant, respectively.

5. The method for self-configuration and self-tuning of autopilot parameters based on simplified ship description as described in claim 4, characterized in that, The process of determining the initial expected closed-loop bandwidth based on the steering preference, configured speed, and initial time constant, and determining the initial control gain based on the initial expected closed-loop bandwidth, initial rudder effect gain, and initial time constant, specifically involves: Determine the steering preference coefficient based on the steering preference. According to the configured speed With preset service speed The relationship between the speed coefficients determines the speed coefficients. And in conjunction with the initial value of the time constant Determine the initial expected closed-loop bandwidth Its expression is: Based on the initial expected closed-loop bandwidth Initial value of rudder effect gain and initial value of time constant The initial value of the proportional gain is determined using the pole placement method. Initial value of integral gain and initial value of differential gain Its expression is: When the initial value of the differential gain is calculated according to the above formula When it is less than zero, the initial value of the differential gain is... Set to zero. in, This represents the steering preference coefficient. Indicates the speed coefficient. Indicates the preset service speed. and These represent the lower limit and upper limit of the closed-loop bandwidth, respectively. Indicates desired damping, Represents the expected integral pole coefficients; This means restricting the input value to a given upper or lower limit.

6. The method for self-configuration and self-tuning of autopilot parameters based on simplified ship description according to claim 5, characterized in that, The online identification of the turning dynamic model when the preset excitation conditions are met specifically involves: according to a preset sampling period. The linear part of the first-order Nomoto model is discretized to obtain discrete model coefficients. and The discrete turning model, as represented, is expressed as follows: The regression vector is constructed using the turning rate and rudder angle, and the coefficients of the discrete model are calculated using a recursive least squares algorithm with a forgetting factor. and A recursive estimation is performed, and the estimated results of the discrete model coefficients are used as the online identification results of the turning dynamic model. The recursive process satisfies: in, Indicates the sampling time sequence number. Indicates the sampling period. Represents the regression vector. Indicates the first The parameter estimation vector at each sampling time point. and They represent the coefficients of the discrete model, respectively. and In the The estimated value at each sampling time point, Indicates the identification of residuals, Represents the gain vector. Represents the covariance matrix. This represents the forgetting factor.

7. The method for self-configuration and self-tuning of autopilot parameters based on simplified ship description as described in claim 6, characterized in that, The preset excitation condition is used to determine whether the current navigation data has a data change amount that can be used to update the discrete model coefficients; the turning rate at the current sampling time and the change in rudder angle between the current sampling time and the previous sampling time are obtained. When at least one of the following conditions is met, the discrete model coefficients are updated; otherwise, the discrete model coefficients are kept unchanged: or in, This indicates the turning rate incentive threshold. Indicates the excitation threshold for rudder angle change. and They represent the first The sampling time and the first sampling time The rudder angle at each sampling moment.

8. The method for self-configuration and self-tuning of autopilot parameters based on simplified ship description according to claim 7, characterized in that, The step of constraining the identified rudder effect gain and time constant within the adaptive boundary to obtain the online rudder effect gain and online time constant specifically involves: based on the discrete model coefficients obtained online... and The identification rudder effect gain before boundary constraints is calculated. and identification time constant Its expression is: The identified rudder effect gain and identification time constant Projecting these values ​​onto the corresponding adaptive boundaries yields the online rudder effect gain. and online time constant Its expression is: in, Indicates the rudder effect gain before boundary constraints. This represents the identification time constant before boundary constraints. Indicates online rudder effect gain. This represents the online time constant.

9. The method for self-configuration and self-tuning of autopilot parameters based on simplified ship description as described in claim 8, characterized in that, The step of determining the target desired closed-loop bandwidth based on the steering preference, current speed, and online time constant, and determining the target control gain based on the target desired closed-loop bandwidth, online rudder effect gain, and online time constant, specifically involves: determining the target control gain based on the current speed... With the preset service speed Determine the current speed coefficient And in combination with the steering preference coefficient and online time constant Determine the target expected closed-loop bandwidth Its expression is: Using the same pole placement relationship as the initial value of the control gain, based on the target desired closed-loop bandwidth... Online rudder effect gain and online time constant Determine the proportional target gain Integral target gain and differential target gain ; According to the preset single-step gain change upper limit For the current proportional gain To perform an update, the expression is: The current integral gain and the current derivative gain are adjusted stepwise towards the target integral gain and the target derivative gain, respectively, using the same single-step limiting method. in, Indicates the first stage of the voyage The current speed at each sampling time. This represents the current speed coefficient corresponding to the current speed. This represents the target expected closed-loop bandwidth. , and These represent the proportional target gain, integral target gain, and derivative target gain, respectively. Indicates the current proportional gain. This indicates the preset upper limit of single-step gain change.

10. A device for self-configuration and self-tuning of autopilot parameters based on a simplified description of ships, characterized in that, include: The description acquisition unit is used to acquire a simplified ship description including ship type, length, tonnage, loading status, configured speed and steering preference, determine dimensionless hydrodynamic prior parameters based on the ship type, and determine model parameter correction factors based on the ship length, tonnage and loading status. The initial value configuration unit is used to determine the initial value of the rudder effect gain, the initial value of the time constant, and the adaptive boundary based on the dimensionless hydrodynamic prior parameters, model parameter correction factors, ship length, and configured speed; to determine the initial expected closed-loop bandwidth based on the steering preference, configured speed, and initial value of the time constant; and to determine the initial value of the control gain based on the initial expected closed-loop bandwidth, the initial value of the rudder effect gain, and the initial value of the time constant. The online identification unit is used to acquire the current heading, current speed, rudder angle and turning rate during navigation. When the preset excitation conditions are met, it performs online identification of the turning dynamic model and constrains the identified rudder effect gain and time constant within the adaptive boundary to obtain the online rudder effect gain and online time constant. The gain tuning unit is used to determine the target desired closed-loop bandwidth based on the steering preference, current speed and online time constant, and to determine the target control gain based on the target desired closed-loop bandwidth, online rudder effect gain and online time constant; the initial value of the control gain is used as the initial value of the current control gain, so that the current control gain gradually approaches the target control gain in a single-step limiting manner; The steering control unit is used to generate rudder angle commands for controlling the servo motors based on the heading deviation between the target heading and the current heading, as well as the current control gain.