Explicit Model Predictive Control Systems and Methods

Abstract

This application provides an explicit model predictive control system and method. It acquires the current yaw angle and yaw rate of a three-degree-of-freedom helicopter and generates an error vector for the helicopter. It determines the current yaw motion state of the helicopter and, based on the error vector and the yaw motion state, determines the control state vector for explicit model predictive control. It pre-calculates the state space region of the helicopter offline based on explicit model predictive control and identifies the target state space partition corresponding to the control state vector based on the state space region. It retrieves the control law associated with the identified state space partition and applies the retrieved control law to generate a yaw control signal to control the yaw motion of the three-degree-of-freedom helicopter. The technical solution provided in this application enables timely updates of the control signal during yaw control of a three-degree-of-freedom helicopter.

Description

Technical Field

[0001] This application relates to the field of explicit model predictive control technology, and more specifically, to an explicit model predictive control system and method. Background Technology

[0002] Against the backdrop of increasing demand for complex system control and the development of computer technology, Model Predictive Control (MMDC) has emerged. Leveraging its ability to predict future system dynamics and handle multivariable and constrained problems, MMDC began to gain prominence in industrial processes in the late 1970s and has since been widely applied in high-tech fields such as robotics and aircraft. However, the large computational demands of online optimization problem solving in MMDC limit its real-time applications. To address this, explicit MMDC was proposed. Through parametric programming, explicit MMDC transforms online optimization calculations into offline operations, significantly improving online computation speed and effectively solving the computational bottleneck of MMDC. This paves the way for its application in more scenarios with high real-time requirements.

[0003] In existing explicit model predictive control (EMC), the core is offline computation of the control law. First, based on the system model and constraints, multi-parameter programming is used to solve the optimization problem corresponding to all initial states, yielding the explicit expression of the control law. This expression is stored as state partitions and corresponding linear control laws. During online execution, it is only necessary to determine the partition to which the current state belongs and call the corresponding control law to output the control quantity, thus achieving explicit model predictive control. However, in existing three-degree-of-freedom helicopter yaw control based on EMC, the generation of the control signal depends on solving a constrained optimization problem online at each sampling time. This involves iterative computation (such as quadratic programming or...). Nonlinear programming is used for fast dynamic systems like three-degree-of-freedom helicopters. Their yaw motion is highly nonlinear, strongly coupled, and has a fast response. Online optimization requires a large amount of computational resources, resulting in significant computational delays. These delays stem from the complexity of the optimization problem and the limited computing power of the processor. Especially when the system state dimension is high (such as including yaw angle and yaw rate), the online solution time may exceed the system sampling time, which in turn causes the control signal to be updated untimely during yaw control of a three-degree-of-freedom helicopter. Therefore, how to achieve timely updates of control signals during yaw control of a three-degree-of-freedom helicopter has become a difficult problem for the industry. Summary of the Invention

[0004] This application provides an explicit model predictive control system and method that enables timely updates of control signals during yaw control of a three-degree-of-freedom helicopter.

[0005] In a first aspect, this application provides an explicit model predictive control method, comprising the following steps:

[0006] Obtain the current yaw angle and yaw rate of the three-degree-of-freedom helicopter, and generate the error vector of the three-degree-of-freedom helicopter when yawing based on the current yaw angle and yaw rate;

[0007] The current yaw motion state of the three-degree-of-freedom helicopter is determined, and then the control state vector for explicit model predictive control is determined based on the error vector and the yaw motion state.

[0008] The state space region of the three-degree-of-freedom helicopter is pre-calculated offline based on the explicit model prediction control, and the target state space partition corresponding to the control state vector is identified based on the state space region. The state space region consists of multiple state space partitions of the three-degree-of-freedom helicopter, and each state space partition corresponds to a control law.

[0009] The system retrieves control laws associated with the identified state space partitions from a pre-stored control law database, then applies the retrieved control laws to generate a yaw control signal for the three-degree-of-freedom helicopter, and uses the yaw control signal to control the yaw motion of the three-degree-of-freedom helicopter.

[0010] In some embodiments, generating the error vector for the yaw of the three-degree-of-freedom helicopter based on the current yaw angle and yaw angular velocity specifically includes:

[0011] Obtain the target yaw angle and target yaw rate of the three-degree-of-freedom helicopter;

[0012] Calculate the difference between the current yaw angle and the target yaw angle to obtain the yaw angle error;

[0013] Calculate the difference between the current yaw rate and the target yaw rate to obtain the yaw rate error;

[0014] The yaw angle error and the yaw rate error are combined in a preset order to generate the error vector of the three-degree-of-freedom helicopter.

[0015] In some embodiments, determining the current yaw motion state of the three-degree-of-freedom helicopter specifically includes:

[0016] The yaw angle and yaw rate of the three-degree-of-freedom helicopter are collected over a continuous preset time period to obtain the time-domain feature sequence of yaw motion;

[0017] The yaw motion time-domain feature sequence is subjected to frequency domain transformation and feature extraction to obtain the yaw motion frequency domain feature parameters;

[0018] Based on the matching results between the yaw motion frequency domain feature parameters and the preset modal feature library, the current yaw motion state of the three-degree-of-freedom helicopter is determined.

[0019] In some embodiments, determining the control state vector for explicit model predictive control based on the error vector and the yaw motion state specifically includes:

[0020] The error vector is normalized to obtain a normalized error vector;

[0021] Based on the current yaw motion state, the state transformation matrix corresponding to the normalized error vector is invoked, and the modal adaptation error vector is determined by the normalized error vector and the state transformation matrix.

[0022] Obtain the reference control state vector corresponding to the current yaw motion state;

[0023] A control state vector for explicit model predictive control is constructed based on the modal adaptation error vector and the reference control state vector.

[0024] In some embodiments, the offline calculation of the state space region of the three-degree-of-freedom helicopter based on explicit model prediction control specifically includes:

[0025] Establish a linear time-invariant state-space model of the yaw motion of the three-degree-of-freedom helicopter, and determine the model state variables;

[0026] Define the constraint boundaries for yaw angle, yaw rate, and control input to form a set of constraint conditions;

[0027] Based on the linear time-invariant state-space model and the set of constraints, an offline optimization problem for explicit model predictive control is constructed.

[0028] The offline optimization problem is analyzed to obtain a piecewise linear function of the optimal control law with respect to the state variables;

[0029] The state space region of the three-degree-of-freedom helicopter is generated based on the domain of the piecewise linear function.

[0030] In some embodiments, identifying the target state space partition corresponding to the control state vector based on the state space region specifically includes:

[0031] Extract the components of each dimension of the control state vector to form a group of state components to be mapped;

[0032] Invoke the pre-stored set of boundary inequalities for each state space partition in the state space region;

[0033] The set of state components to be mapped is substituted into the set of boundary inequalities of each state space partition for verification.

[0034] Select the state space partitions that satisfy all boundary inequalities as the target state space partitions mapped by the control state vector.

[0035] In some embodiments, retrieving the control law associated with the identified state space partition from a pre-stored control law database specifically includes:

[0036] Extract the unique identifier information of the identified state space partitions;

[0037] A database retrieval index is constructed based on the unique identifier information;

[0038] The database retrieval index is used to perform a matching query in the pre-stored control law database, and the control law that matches the unique identifier information of the state space partition is extracted as the retrieved control law.

[0039] In some embodiments, the current yaw angle and yaw rate of the three-degree-of-freedom helicopter are obtained by a sensor array mounted on the yaw axis of the three-degree-of-freedom helicopter fuselage.

[0040] In some embodiments, the sensor group includes a magnetic heading sensor and a rate gyroscope.

[0041] Secondly, this application provides an explicit model predictive control system, comprising:

[0042] The acquisition module is used to acquire the current yaw angle and yaw rate of the three-degree-of-freedom helicopter, and generate the error vector of the three-degree-of-freedom helicopter when yawing based on the current yaw angle and yaw rate.

[0043] The processing module is used to determine the current yaw motion state of the three-degree-of-freedom helicopter, and then determine the control state vector for explicit model predictive control based on the error vector and the yaw motion state.

[0044] The processing module is also used to pre-calculate the state space region of the three-degree-of-freedom helicopter based on the explicit model prediction control offline calculation, and to identify the target state space partition corresponding to the control state vector based on the state space region. The state space region is composed of multiple state space partitions of the three-degree-of-freedom helicopter, and each state space partition corresponds to a control law.

[0045] The execution module is used to retrieve control laws associated with the identified state space partitions from a pre-stored control law database, and then apply the retrieved control laws to generate the yaw control signal of the three-degree-of-freedom helicopter, and control the yaw motion of the three-degree-of-freedom helicopter by the yaw control signal.

[0046] The technical solutions provided by the embodiments disclosed in this application have the following beneficial effects:

[0047] The explicit model predictive control system and method provided in this application firstly acquires the current yaw angle and yaw rate of a three-degree-of-freedom helicopter, and generates an error vector for the yaw motion of the three-degree-of-freedom helicopter based on the current yaw angle and yaw rate; secondly, determines the current yaw motion state of the three-degree-of-freedom helicopter, and then determines a control state vector for explicit model predictive control based on the error vector and the yaw motion state; then, pre-calculates the state space region of the three-degree-of-freedom helicopter offline based on explicit model predictive control, and identifies the target state space partition corresponding to the control state vector based on the state space region, wherein the state space region consists of multiple state space partitions of the three-degree-of-freedom helicopter, and each state space partition corresponds to a control law; finally, retrieves the control law associated with the identified state space partition from a pre-stored control law database, and then applies the retrieved control law to generate a yaw control signal for the three-degree-of-freedom helicopter, and controls the yaw motion of the three-degree-of-freedom helicopter by the yaw control signal.

[0048] Therefore, this application can achieve timely updates of control signals during yaw control of a three-degree-of-freedom helicopter. Firstly, by acquiring the current yaw angle and yaw rate of the three-degree-of-freedom helicopter and generating an error vector in conjunction with the target yaw command, the difference between the actual yaw parameters and the target command is quantified to accurately capture the degree to which the three-degree-of-freedom helicopter deviates from the expected state, providing a clear deviation benchmark for subsequent control adjustments and avoiding control inaccuracies caused by ambiguous state descriptions. Secondly, by determining the current yaw motion state and constructing a control state vector, the dynamic motion characteristics and yaw errors of the three-degree-of-freedom helicopter can be integrated into the state representation, making the yaw state description more comprehensively aligned with actual operating conditions. This provides a basis for adapting control strategies to different scenarios, thereby avoiding the calculation errors of explicit model predictive control caused by yaw motion states under different flight conditions. The computational complexity increases; then, the state space region is pre-calculated offline and the target state space partition corresponding to the control state vector is identified. By discretizing the continuous state space into state space partitions and establishing a mapping with the control law, it is possible to quickly locate the applicable control law through state matching without solving complex optimization problems in real time during online control, and to provide a basis for timely updating of the control signal when the three-degree-of-freedom helicopter performs yaw control. Finally, the associated control law is retrieved and the yaw control signal is generated to control the yaw motion. This can directly convert the analytical control strategy obtained by offline optimization into executable mechanism drive instructions, so as to realize timely updating of the control signal when the three-degree-of-freedom helicopter performs yaw control. In summary, the technical solution provided in this application can realize timely updating of the control signal when the three-degree-of-freedom helicopter performs yaw control. Attached Figure Description

[0049] Figure 1This is a schematic diagram of an application scenario architecture for an explicit model predictive control method according to some embodiments of this application;

[0050] Figure 2 This is an exemplary flowchart of an explicit model predictive control method according to some embodiments of this application;

[0051] Figure 3 This is an exemplary flowchart illustrating the determination of a control state vector according to some embodiments of this application;

[0052] Figure 4 This is a schematic diagram of the structure of an explicit model predictive control system according to some embodiments of this application;

[0053] Figure 5 This is a schematic diagram of the structure of a computer device implementing an explicit model predictive control method according to some embodiments of this application. Detailed Implementation

[0054] To better understand the technical solution of this application, the technical solution of this application will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0055] refer to Figure 1 This figure is a schematic diagram of an application scenario architecture for the explicit model predictive control method according to some embodiments of this application. The application scenario architecture includes a data acquisition terminal, a communication network, and a server. The data acquisition terminal and the server are directly or indirectly connected through the communication network. The data acquisition terminal acquires the current yaw angle and yaw rate of a three-degree-of-freedom helicopter and uploads them to the server terminal. The server generates an error vector for the yaw motion of the three-degree-of-freedom helicopter based on the current yaw angle and yaw rate; determines the current yaw motion state of the three-degree-of-freedom helicopter; and then determines the parameters for explicit model prediction based on the error vector and the yaw motion state. The control state vector is controlled; the state space region of the three-degree-of-freedom helicopter is pre-calculated offline based on an explicit model prediction control, and the target state space partition corresponding to the control state vector is identified based on the state space region. The state space region consists of multiple state space partitions of the three-degree-of-freedom helicopter, and each state space partition corresponds to a control law; the control law associated with the identified state space partition is retrieved from a pre-stored control law database, and then the retrieved control law is applied to generate the yaw control signal of the three-degree-of-freedom helicopter, and the yaw motion of the three-degree-of-freedom helicopter is controlled by the yaw control signal.

[0056] refer to Figure 2 The figure is an exemplary flowchart of an explicit model predictive control method according to some embodiments of this application. The explicit model predictive control method mainly includes the following steps:

[0057] In step 101, the current yaw angle and yaw rate of the three-degree-of-freedom helicopter are obtained, and the error vector of the three-degree-of-freedom helicopter when yaw is generated based on the current yaw angle and yaw rate.

[0058] In practice, the current yaw angle and yaw rate of the three-degree-of-freedom helicopter are obtained by a sensor group installed on the yaw axis of the helicopter fuselage. The sensor group includes a magnetic heading sensor and a rate gyroscope. The yaw angle is measured by the magnetic heading sensor in the sensor group, and the yaw rate is measured by the rate gyroscope. The details will not be elaborated here. By obtaining the yaw angle and yaw rate, a state feedback basis can be effectively provided for subsequent control algorithms.

[0059] It should be noted that, in this application, yaw angle refers to an angular parameter describing the rotational attitude of the helicopter body about its perpendicular position to the ground, and yaw angular velocity refers to a parameter describing the speed of the helicopter body's rotation about its yaw axis.

[0060] In some embodiments, generating the error vector for the yaw of the three-degree-of-freedom helicopter based on the current yaw angle and yaw rate is achieved through the following steps:

[0061] Obtain the target yaw angle and target yaw rate of the three-degree-of-freedom helicopter;

[0062] Calculate the difference between the current yaw angle and the target yaw angle to obtain the yaw angle error;

[0063] Calculate the difference between the current yaw rate and the target yaw rate to obtain the yaw rate error;

[0064] The yaw angle error and the yaw rate error are combined in a preset order to generate the error vector of the three-degree-of-freedom helicopter.

[0065] In specific implementation, firstly, conventional command decoding technology can be used to extract the target yaw angle and target yaw rate from the received target yaw command signal. The target yaw angle refers to the yaw direction angle that the three-degree-of-freedom helicopter is expected to achieve, and the target yaw rate refers to the expected rate of change of the yaw angle. Secondly, the target yaw angle is subtracted from the current yaw angle to obtain the yaw angle error, which is the deviation between the current yaw angle and the target yaw angle. Then, the target yaw rate is subtracted from the current yaw rate to obtain the yaw rate error, which is the deviation between the current yaw rate and the target yaw rate. Finally, the yaw angle error is used as the first element of the error vector, and the yaw rate error is used as the second element of the error vector, thereby forming the error vector of the three-degree-of-freedom helicopter.

[0066] It should be noted that the error vector in this application represents a vector used to describe the deviation between the yaw state and the target state of a three-degree-of-freedom helicopter. It includes error information in terms of both yaw angle and yaw rate. The error vector intuitively quantifies the degree of deviation between the current yaw state of the helicopter and the target state, providing an accurate error benchmark for subsequent control decisions. The yaw angle error reflects the deviation at the position level, while the yaw rate error reflects the difference at the dynamic response level. The combination of the two can comprehensively describe the control error state of the yaw motion, providing a basis for the design and execution of the control law.

[0067] In step 102, the current yaw motion state of the three-degree-of-freedom helicopter is determined, and then the control state vector for explicit model predictive control is determined based on the error vector and the yaw motion state.

[0068] In some embodiments, determining the current yaw motion state of the three-degree-of-freedom helicopter is achieved by the following steps:

[0069] The yaw angle and yaw rate of the three-degree-of-freedom helicopter are collected over a continuous preset time period to obtain the time-domain feature sequence of yaw motion;

[0070] The yaw motion time-domain feature sequence is subjected to frequency domain transformation and feature extraction to obtain the yaw motion frequency domain feature parameters;

[0071] Based on the matching results between the yaw motion frequency domain feature parameters and the preset modal feature library, the current yaw motion state of the three-degree-of-freedom helicopter is determined.

[0072] In specific implementation, firstly, the yaw angle and yaw rate are continuously sampled within a preset time period using angular velocity and angle sensors mounted on the three-degree-of-freedom helicopter, thereby forming a yaw motion time-domain feature sequence. This yaw motion time-domain feature sequence is a set of raw data reflecting the changes in yaw angle and yaw rate of the three-degree-of-freedom helicopter over time within the preset time period. Secondly, the time-series signal in the yaw motion time-domain feature sequence is converted into a frequency-domain signal using a fast Fourier transform algorithm, and the dominant frequency and amplitude spectrum peak value in the frequency-domain signal are obtained. The combination of the dominant frequency and the peak value of the amplitude spectrum is used as the yaw motion frequency domain characteristic parameter. The yaw motion frequency domain characteristic parameter is a quantitative index describing the yaw motion in the frequency domain. Then, the similarity between the yaw motion frequency domain characteristic parameter and the frequency domain characteristic parameters of different motion modes in the preset modal feature library is calculated by Euclidean distance. The motion mode with the highest similarity is taken as the current yaw motion state of the three-degree-of-freedom helicopter. The preset modal feature library is a set of frequency domain characteristic parameters corresponding to different motion modes (such as stable cruise, steering maneuver, etc.) that are stored in advance.

[0073] It should be noted that the yaw motion state in this application refers to the motion type of the current yaw motion of a three-degree-of-freedom helicopter. Determining the yaw motion state can clarify the current yaw motion type of the helicopter (such as stable hovering, acceleration turning, etc.), providing a basis for the control system to match and adapt the control strategy. The dynamic characteristics of the helicopter differ under different modes, and different control parameters and response speed requirements are required.

[0074] In some embodiments, reference Figure 3 As shown in the figure, this is an exemplary flowchart of determining the control state vector according to some embodiments of this application. In this embodiment, the determination of the control state vector for explicit model predictive control based on the error vector and the yaw motion state can be achieved by the following steps:

[0075] First, in step 1021, the error vector is normalized to obtain a normalized error vector;

[0076] Secondly, in step 1022, the state transformation matrix corresponding to the normalized error vector is called based on the current yaw motion state, and the modal adaptation error vector is determined by the normalized error vector and the state transformation matrix.

[0077] Then, in step 1023, the reference control state vector corresponding to the current yaw motion state is obtained; finally, in step 1024, the control state vector for explicit model predictive control is constructed based on the modal adaptation error vector and the reference control state vector.

[0078] In specific implementation, firstly, the error vector is normalized using minimum-maximum normalization (MMR). Specifically, each component of the error vector is mapped to the [0,1] interval to obtain a normalized error vector. This normalized error vector eliminates the differences in magnitude between the components. Secondly, based on the current yaw motion state, the corresponding state transformation matrix is ​​retrieved from a preset matrix library. This state transformation matrix is ​​an orthogonal transformation matrix pre-designed according to the dynamic characteristics under a specific mode (details omitted here). It is used to transform the normalized error vector to a coordinate space that better suits the corresponding mode control requirements. Matrix multiplication is performed on the normalized error vector and the state transformation matrix to obtain the modal adaptation error vector, which is the error representation vector for adapting to the current yaw motion state. Then, the reference control state vector corresponding to the current yaw motion state is obtained. The reference control state vector is a control state vector that has been set in advance after simulation control of different yaw states, which will not be elaborated here. Finally, the modal adaptation error vector is horizontally concatenated with the modal adaptation error vector to form the control state vector for explicit model predictive control, which will not be elaborated here.

[0079] It should be noted that, in this application, the control state vector represents the state vector used for explicit model predictive control calculation. The control state vector reflects the control state of the yaw motion of a three-degree-of-freedom helicopter. The determination of the control state vector can effectively take into account the system deviation under specific yaw modes, thereby achieving faster partition mapping and more accurate control law invocation in explicit model predictive control, thus improving the adaptability and anti-interference capability of helicopter yaw control under complex operating conditions.

[0080] In step 103, the state space region of the three-degree-of-freedom helicopter is pre-calculated offline based on the explicit model predictive control, and the target state space partition corresponding to the control state vector is identified based on the state space region. The state space region consists of multiple state space partitions of the three-degree-of-freedom helicopter, and each state space partition corresponds to a control law.

[0081] In some embodiments, the offline calculation of the state space region of the three-degree-of-freedom helicopter based on explicit model predictive control is achieved by the following steps:

[0082] Establish a linear time-invariant state-space model of the yaw motion of the three-degree-of-freedom helicopter, and determine the model state variables;

[0083] Define the constraint boundaries for yaw angle, yaw rate, and control input to form a set of constraint conditions;

[0084] Based on the linear time-invariant state-space model and the set of constraints, an offline optimization problem for explicit model predictive control is constructed.

[0085] The offline optimization problem is analyzed to obtain a piecewise linear function of the optimal control law with respect to the state variables;

[0086] The state space region of the three-degree-of-freedom helicopter is generated based on the domain of the piecewise linear function.

[0087] In specific implementation, firstly, the yaw dynamics equations of a three-degree-of-freedom helicopter are linearized, using yaw angle and yaw rate as model state variables and yaw control signal as input variables. This constructs state equations of the form ẋ=Ax+Bu and output equations of y=Cx. Then, the state equations and output equations are combined to form a linear time-invariant state-space model of the yaw motion of the three-degree-of-freedom helicopter. Here, ẋ represents the first-order derivative vector of the model state variables, reflecting the rate of change of the state variables over time; A is the model matrix, a constant matrix composed of the dynamic parameters of the three-degree-of-freedom helicopter, used to describe the coupling relationship between the internal states of the three-degree-of-freedom helicopter; and x is the model state variable vector, typically containing the yaw angle. Yaw angle and yaw rate are used to fully characterize the dynamic state of a three-degree-of-freedom helicopter. B is the input matrix, reflecting the degree of influence of the control input on the state of the three-degree-of-freedom helicopter. u is the input vector, corresponding to the yaw control signal in yaw control. In the output equation y=Cx, y is the output variable vector, which is a directly measurable system output quantity. C is the output matrix, used to map state variables to output variables. The linear time-invariant state-space model is a mathematical model describing the linear change of the state of a three-degree-of-freedom helicopter with time and whose parameters do not change with time. The model state variables are the smallest set of variables characterizing the dynamic behavior of a three-degree-of-freedom helicopter. Next, the constraint boundaries of yaw angle, yaw rate, and control input are set. The yaw angle and yaw rate are related to the yaw speed. The constraint boundary of the flight angular velocity is determined based on the physical motion range of the three-degree-of-freedom helicopter, and the constraint of the control input is set based on the maximum output capability of the actuator, thus forming a set of constraint conditions including inequality constraints. This set of constraint conditions is the sum of rules restricting the range of states and input values ​​of the three-degree-of-freedom helicopter. Furthermore, based on the linear time-invariant state-space model and the set of constraint conditions, an offline optimization problem for explicit model predictive control is constructed. The optimization objective is to minimize a preset performance index (such as the weighted sum of state error and control quantity), and the constraint conditions in the set of constraint conditions are integrated into the constraint terms of the optimization objective, thus obtaining the offline optimization problem for explicit model predictive control. The offline optimization problem refers to... The mathematical problem aimed at finding the optimal control strategy before the operation of a three-degree-of-freedom helicopter is addressed by: First, an existing multi-parameter programming algorithm is used to solve the offline optimization problem. This algorithm treats state variables as parameters and, under constraints, finds the control quantity that minimizes the performance index, resulting in a piecewise linear function of the optimal control law with respect to the state variables. This piecewise linear function is composed of multiple linear segments describing the relationship between the control quantity and the state variables. Finally, based on the domain of the piecewise linear function, the range of state variable values ​​corresponding to each linear segment is defined as a state space partition. All state space partitions are combined to form the state space region of the three-degree-of-freedom helicopter.

[0088] It should be noted that, in this application, the state space region refers to the region in the state space that satisfies linear constraints. The determination of the state space region provides a structured state mapping framework for explicit model predictive control. Its core function is to divide the continuous system state space into multiple discrete state space partitions, each partition corresponding to a unique piecewise linear optimal control law. This eliminates the need to solve complex optimization problems in real time during online control. Instead, the pre-stored control law can be directly invoked simply by determining the partition to which the current state belongs, significantly improving the response speed of control decisions. At the same time, the partition boundaries strictly follow the set of constraint conditions, ensuring that the control law in each partition can meet the system's physical constraints and performance requirements, thus avoiding situations where the control input exceeds the actuator's capabilities or the system state deviates from the safe range.

[0089] In some embodiments, identifying the target state space partition corresponding to the control state vector based on the state space region is achieved by the following steps:

[0090] Extract the components of each dimension of the control state vector to form a group of state components to be mapped;

[0091] Invoke the pre-stored set of boundary inequalities for each state space partition in the state space region;

[0092] The set of state components to be mapped is substituted into the set of boundary inequalities of each state space partition for verification.

[0093] Select the state space partitions that satisfy all boundary inequalities as the target state space partitions mapped by the control state vector.

[0094] In specific implementation, firstly, the specific values ​​of each dimension are extracted from the control state vector and arranged in the original dimensional order of the control state vector to form a set of state components to be mapped. The set of state components to be mapped is the set of specific values ​​of the control state vector in each dimension. Secondly, the boundary inequality set of all state space partitions in the state space region, which has been pre-calculated offline, is invoked. The boundary inequality set of each state space partition consists of multiple linear inequalities, used to define the boundary range of the state space partition in the state space. The boundary inequality set is a set of linear inequalities describing the boundary of the state space partition. Then, the inequality verification method is used to substitute each value of the set of state components to be mapped into each inequality in the boundary inequality set of each state space partition for calculation, and to determine whether each inequality is true. The inequality verification method is a mathematical verification means to determine whether the inequality is satisfied by substituting numerical values ​​for calculation. Finally, based on the verification results, the partitions that satisfy all the inequalities of the state space partition are determined as the target state space partitions mapped by the control state vector.

[0095] It should be noted that, in this application, the target state space partition refers to the partition in the state space to which the control state vector belongs. The scheme for determining the target state space partition is to directly call the pre-stored set of state space partition boundary inequalities, substitute each dimension component of the control state vector into the verification to filter matching partitions. This breaks through the limitations of existing technologies that rely on complex search algorithms or iterative optimization for partition matching. Its originality lies in realizing partition mapping by "direct substitution of component sets + full verification of inequality sets", rather than the traditional approximate matching based on distance or probability. This method ensures a strict correspondence between the control state vector and the target state space partition, avoids the control law calling deviation that may be caused by approximate matching, and eliminates the need for complex iterative calculations, which significantly improves the real-time performance of partition identification. It is especially suitable for dynamic systems such as three-degree-of-freedom helicopters that have high requirements for control response speed.

[0096] In step 104, control laws associated with the identified state space partitions are retrieved from a pre-stored control law database, and then the retrieved control laws are applied to generate a yaw control signal for the three-degree-of-freedom helicopter, which controls the yaw motion of the three-degree-of-freedom helicopter.

[0097] In some embodiments, retrieving the control law associated with the identified state space partition from a pre-stored control law database is achieved through the following steps:

[0098] Extract the unique identifier information of the identified state space partitions;

[0099] A database retrieval index is constructed based on the unique identifier information;

[0100] The database retrieval index is used to perform a matching query in the pre-stored control law database, and the control law that matches the unique identifier information of the state space partition is extracted as the retrieved control law.

[0101] In specific implementation, firstly, unique identifier information is extracted from the identified state space partitions. This unique identifier information is a unique code pre-assigned to each partition using an existing binary tree encoding method, used to uniquely distinguish different state space partitions. This unique identifier information is specific information that can individually identify a particular state space partition. Secondly, a hash algorithm can be used to convert the unique identifier information into a fixed-length hash value using a hash function, and this hash value is used as a database retrieval index. This database retrieval index is a guiding identifier used to quickly find target data in the database. Then, the constructed database retrieval index is applied to a pre-stored control law database. Existing index matching query technology is used to compare the hash value with the control law in the control law database that corresponds to the database retrieval index, and this retrieved control law is used as the control law. The control law database, which is a pre-stored structured data set used to store the optimal control strategies associated with each state space partition, is built on the results of offline solved explicit model predictive control. Each record in the database contains two core parts: a unique identifier for the state space partition and the corresponding control law. The control law is stored in the form of piecewise linear function parameters, which are obtained by solving offline optimization problems using multi-parameter programming algorithms. These parameters directly represent the linear mapping relationship between control input and system state. The database adopts an indexed storage structure, using the unique identifier of the state space partition as the primary key to build an index table. The index table records the physical storage address of each control law in the database. This structure supports fast retrieval and retrieval of control laws through unique identifier information.

[0102] It should be noted that the control law in this application refers to the control strategy of a three-degree-of-freedom helicopter in the form of an explicit mathematical expression. It reflects the mapping relationship between the state of the three-degree-of-freedom helicopter and the control input, and can be directly invoked without online optimization calculation. The determination of the control law, on the one hand, breaks through the computational burden and real-time limitation brought about by the online solution of optimization problems in traditional model predictive control. By pre-storing the control strategy as an explicit expression, the control system can quickly invoke the corresponding control law according to the current state, which significantly improves the control response speed. On the other hand, its strict correspondence with the state space region ensures the optimality and constraint satisfaction of the control strategy under various operating conditions. It can ensure the yaw motion control accuracy of the three-degree-of-freedom helicopter while avoiding the control input from exceeding the capability range of the actuator, providing a reliable guarantee for the stable operation of the system. At the same time, it simplifies the online control logic and reduces the requirements for the computational performance of the control unit.

[0103] In some embodiments, generating the yaw control signal for the three-degree-of-freedom helicopter using the retrieved control law is achieved through the following steps:

[0104] Obtain the state feedback gain matrix and offset vector corresponding to the control law;

[0105] The control state vector is multiplied by the state feedback gain matrix to obtain the state feedback component.

[0106] The original control quantity is determined based on the state feedback component and the offset vector;

[0107] The original control quantity is subjected to amplitude limiting processing to meet the preset control input constraint range, thereby generating the yaw control signal for the three-degree-of-freedom helicopter.

[0108] In specific implementation, firstly, the state feedback gain matrix and offset vector are extracted from the retrieved control law. The state feedback gain matrix is ​​a parameter matrix describing the degree of influence of the system state on the control input, and the offset vector is a constant vector used to compensate for the static deviation of the system. Secondly, the control state vector is multiplied by the state feedback gain matrix using matrix multiplication to obtain the state feedback component. The state feedback component is the basic control component formed by the feedback gain after the state of the three-degree-of-freedom helicopter is subjected to the feedback gain. Then, the state feedback component is combined with the offset vector using addition to obtain the original control quantity. The original control quantity is the initial control output without constraint processing. Finally, the original control quantity is processed using a limiting algorithm. That is, when the original control quantity exceeds the preset control input constraint range (which is determined by the physical performance of the actuator), the control quantity is truncated to the maximum value within the constraint range, and the truncated control quantity is used to generate the yaw control signal of the three-degree-of-freedom helicopter.

[0109] It should be noted that the yaw control signal in this application refers to the command signal that can directly drive the action of the three-degree-of-freedom helicopter actuator. By determining the yaw control signal, the control law can be transformed into a specific command that can be directly applied to the three-degree-of-freedom helicopter actuator. This command is the direct basis for the action of the actuator, and its function is to ensure the stability of the helicopter's yaw motion. Through precise control signal adjustment, the helicopter can quickly recover to the target yaw state when it is subjected to external disturbances or changes in its own state.

[0110] In some embodiments, controlling the yaw motion of the three-degree-of-freedom helicopter by the yaw control signal is achieved through the following steps:

[0111] The yaw control signal is converted into a drive command for the three-degree-of-freedom helicopter yaw actuator;

[0112] Send the drive command to the yaw actuator to control the yaw actuator to output the corresponding torque;

[0113] The torque output by the yaw actuator acts on the three-degree-of-freedom helicopter fuselage, generating a yaw torque;

[0114] The three-degree-of-freedom helicopter adjusts its yaw attitude under the action of yaw moment.

[0115] In practice, firstly, the digital yaw control signal is converted into an analog voltage signal using a D / A converter. This analog voltage signal is the drive command for the three-degree-of-freedom helicopter yaw actuator (such as a motor), and the drive command is an electrical signal used to control the actuator's movement. Secondly, the drive command is sent to the control interface of the yaw actuator. The drive circuit inside the actuator adjusts the output current according to the drive command to control the motor rotor speed, causing the actuator to output torque corresponding to the drive command. This torque is the rotational force that causes the helicopter to yaw. Then, the torque is transmitted to the yaw axis of the three-degree-of-freedom helicopter fuselage through the drive shaft, forming a yaw torque around the yaw axis. This yaw torque is the dynamic force that causes the fuselage to rotate around the yaw axis. Finally, under the continuous action of the yaw torque, the three-degree-of-freedom helicopter fuselage overcomes air resistance and rotational inertia, generating rotational motion around the vertical axis, thereby adjusting the yaw attitude.

[0116] It should be noted that in this application, yaw attitude refers to the angular position of the three-degree-of-freedom helicopter fuselage around the yaw axis. By adjusting the yaw attitude, the fuselage of the three-degree-of-freedom helicopter can be rotated around the yaw axis to the target angular position. The core function of this process is to ensure the directional control accuracy of the helicopter during flight or hovering, and to ensure that it maintains the correct heading according to the preset trajectory or command.

[0117] Furthermore, in another aspect of this application, in some embodiments, this application provides an explicit model predictive control system, referencing... Figure 4 The figure is a schematic diagram of the structure of an explicit model predictive control system according to some embodiments of this application. The explicit model predictive control system includes: an acquisition module 201, a processing module 202, and an execution module 203, which are described below:

[0118] The acquisition module 201 in this application is mainly used to acquire the current yaw angle and yaw rate of the three-degree-of-freedom helicopter, and generate the error vector of the three-degree-of-freedom helicopter when yawing based on the current yaw angle and yaw rate.

[0119] Processing module 202, in this application, is mainly used to determine the current yaw motion state of the three-degree-of-freedom helicopter, and then determine the control state vector for explicit model predictive control based on the error vector and the yaw motion state.

[0120] The processing module 202 is further configured to pre-calculate the state space region of the three-degree-of-freedom helicopter based on the explicit model prediction control offline calculation, and identify the target state space partition corresponding to the control state vector based on the state space region, wherein the state space region is composed of multiple state space partitions of the three-degree-of-freedom helicopter, and each state space partition corresponds to a control law.

[0121] The execution module 203 in this application is mainly used to retrieve the control law associated with the identified state space partition from the pre-stored control law database, and then apply the retrieved control law to generate the yaw control signal of the three-degree-of-freedom helicopter, and control the yaw motion of the three-degree-of-freedom helicopter by the yaw control signal.

[0122] In addition, this application also provides a computer device, the computer device including a memory and a processor, the memory storing code, and the processor being configured to acquire the code and execute the above-described explicit model predictive control method.

[0123] In some embodiments, reference Figure 5 The figure is a schematic diagram of the structure of a computer device implementing an explicit model predictive control method according to some embodiments of this application. The explicit model predictive control method in the above embodiments can... Figure 5 The computer device shown is used to implement this, and the computer device includes at least one processor 301, a communication bus 302, a memory 303, and at least one communication interface 304.

[0124] The processor 301 may be a general-purpose central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more devices used to control the execution of the explicit model predictive control method in this application.

[0125] The communication bus 302 can be used to transmit information between the aforementioned components.

[0126] The memory 303 may be a read-only memory (ROM) or other type of static storage device capable of storing static information and instructions, random access memory (RAM) or other type of dynamic storage device capable of storing information and instructions, or electrically erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disks or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but not limited thereto. The memory 303 may exist independently and be connected to the processor 301 via the communication bus 302. The memory 303 may also be integrated with the processor 301.

[0127] The memory 303 stores program code for executing the scheme of this application, and its execution is controlled by the processor 301. The processor 301 executes the program code stored in the memory 303. The program code may include one or more software modules. In the above embodiments, the determination of the explicit model predictive control method can be implemented by the processor 301 and one or more software modules in the program code in the memory 303.

[0128] Communication interface 304 uses any transceiver-like device for communicating with other devices or communication networks, such as Ethernet, radio access network (RAN), wireless local area networks (WLAN), etc.

[0129] In a specific implementation, as one example, a computer device may include multiple processors, each of which may be a single-core (single-CPU) processor or a multi-core (multi-CPU) processor. Here, a processor may refer to one or more devices, circuits, and / or processing cores used to process data (e.g., computer program instructions).

[0130] The aforementioned computer device can be a general-purpose computer device or a special-purpose computer device. In specific implementations, the computer device can be a desktop computer, a portable computer, a network server, a handheld digital assistant (PDA), a mobile phone, a tablet computer, a wireless terminal device, a communication device, or an embedded device. This application does not limit the type of computer device.

[0131] In addition, this application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described explicit model predictive control method.

[0132] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.

[0133] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. An explicit model predictive control method, characterized by, The method comprises the following steps: obtaining the current yaw angle and yaw angle velocity of the three-degree-of-freedom helicopter, and generating an error vector of the three-degree-of-freedom helicopter in yaw according to the current yaw angle and yaw angle velocity; determining the current yaw motion state of the three-degree-of-freedom helicopter, and then determining a control state vector for explicit model predictive control according to the error vector combined with the yaw motion state; pre-calculating the state space region of the three-degree-of-freedom helicopter based on explicit model predictive control offline, and identifying the target state space partition corresponding to the control state vector based on the state space region, wherein the state space region is composed of multiple state space partitions of the three-degree-of-freedom helicopter, and each state space partition corresponds to a control law; retrieving the control law associated with the identified state space partition from the pre-stored control law database, and then generating a yaw control signal of the three-degree-of-freedom helicopter by applying the retrieved control law, and controlling the yaw motion of the three-degree-of-freedom helicopter by the yaw control signal; wherein determining the current yaw motion state of the three-degree-of-freedom helicopter specifically comprises: collecting the yaw angle and yaw angle velocity of the three-degree-of-freedom helicopter within a continuous preset time period to obtain a yaw motion time-domain feature sequence; performing frequency domain transformation and feature extraction on the yaw motion time-domain feature sequence to obtain yaw motion frequency domain feature parameters; determining the current yaw motion state of the three-degree-of-freedom helicopter based on the matching result of the yaw motion frequency domain feature parameters and a preset modal feature library; wherein determining a control state vector for explicit model predictive control according to the error vector combined with the yaw motion state specifically comprises: normalizing the error vector to obtain a normalized error vector; calling a state transformation matrix corresponding to the normalized error vector based on the current yaw motion state, and determining a modal adaptation error vector based on the normalized error vector and the state transformation matrix; obtaining a reference control state vector corresponding to the current yaw motion state; constructing a control state vector for explicit model predictive control according to the modal adaptation error vector and the reference control state vector.

2. The method of claim 1, wherein, generating an error vector of the three-degree-of-freedom helicopter in yaw according to the current yaw angle and yaw angle velocity specifically comprises: obtaining a target yaw angle and a target yaw angle velocity of the three-degree-of-freedom helicopter; calculating the difference between the current yaw angle and the target yaw angle to obtain a yaw angle error; calculating the difference between the current yaw angle velocity and the target yaw angle velocity to obtain a yaw angle velocity error; combining the yaw angle error and the yaw angle velocity error in a preset order to generate an error vector of the three-degree-of-freedom helicopter.

3. The method of claim 1, wherein, pre-calculating the state space region of the three-degree-of-freedom helicopter based on explicit model predictive control offline specifically comprises: establishing a linear time-invariant state space model of the yaw motion of the three-degree-of-freedom helicopter, and determining model state variables; setting the constraint boundaries of the yaw angle, the yaw angle velocity and the control input to form a constraint condition set; constructing an offline optimization problem of explicit model predictive control based on the linear time-invariant state space model and the constraint condition set; solving the offline optimization problem to obtain a piecewise linear function of the optimal control law with respect to the state variables; generating a state space region of the three-degree-of-freedom helicopter according to a definition domain of the piecewise linear function.

4. The method of claim 1, wherein, Identifying the target state space partition corresponding to the control state vector based on the state space region specifically includes: extracting each dimension component of the control state vector to form a group of to-be-mapped state components; calling a group of boundary inequalities of each state space partition in the pre-stored state space region; substituting the group of to-be-mapped state components into the group of boundary inequalities of each state space partition for verification; filtering out the state space partition that satisfies all boundary inequalities as the target state space partition to which the control state vector is mapped.

5. The method of claim 1, wherein, Retrieving the control law associated with the identified state space partition from the pre-stored control law database specifically includes: extracting the unique identification information of the identified state space partition; constructing a database retrieval index based on the unique identification information; using the database retrieval index to perform a matching query in the pre-stored control law database, and extracting the control law matching the unique identification information of the state space partition as the retrieved control law.

6. The method of claim 1, wherein, A sensor group installed on the yaw axis of the three-degree-of-freedom helicopter is used to obtain the current yaw angle and yaw angular velocity of the three-degree-of-freedom helicopter.

7. The method of claim 6, wherein, The sensor group includes a magnetic heading sensor and a rate gyroscope.

8. An explicit model predictive control system employing the method of any one of claims 1 to 7 for explicit model predictive control, characterized by The system includes: an acquisition module for acquiring the current yaw angle and yaw angular velocity of the three-degree-of-freedom helicopter, and generating an error vector of the three-degree-of-freedom helicopter during yaw according to the current yaw angle and yaw angular velocity; a processing module for determining the current yaw motion state of the three-degree-of-freedom helicopter, and then determining a control state vector for explicit model predictive control according to the error vector combined with the yaw motion state; the processing module is also used to pre-compute a state space region of the three-degree-of-freedom helicopter based on explicit model predictive control offline, and identify a target state space partition corresponding to the control state vector based on the state space region, wherein the state space region is composed of a plurality of state space partitions of the three-degree-of-freedom helicopter, and each state space partition corresponds to a control law; an execution module for retrieving the control law associated with the identified state space partition from a pre-stored control law database, and then generating a yaw control signal of the three-degree-of-freedom helicopter by applying the retrieved control law, and controlling the yaw motion of the three-degree-of-freedom helicopter by the yaw control signal.

Images