Coupling optimization method for energy state and attitude state under high dynamic load

Abstract

The application relates to the technical field of unmanned plane control and discloses a coupling optimization method for energy state and attitude state under high dynamic load, which comprises the following steps: acquiring battery operation data, motor operation data and flight attitude data, and establishing a multi-domain joint state model; generating a coupling sensitivity index; constructing a nonlinear disturbance observation decoupling control layer and generating a disturbance estimation result; constructing an energy state and attitude state collaborative optimization target and establishing a model prediction controller; dynamically adjusting a target weight; generating an optimal control sequence and performing rolling horizon optimization; and generating a motor driving control instruction according to a current control amount and updating a system state. According to the application, the multi-domain joint state model is established, and the nonlinear disturbance observation decoupling control layer is constructed, so that the disturbance state caused by the high dynamic load can be compensated, thereby reducing the cross interference between the energy state and the attitude state.

Description

Technical Field

[0001] This invention relates to the field of unmanned aerial vehicle (UAV) control technology, and in particular to a method for coupled optimization of energy state and attitude state under high dynamic loads. Background Technology

[0002] Unmanned aerial vehicle (UAV) control technology belongs to the field of flight control and autonomous control technology, and is widely used in scenarios such as inspection and monitoring, emergency rescue, target reconnaissance, logistics transportation, and autonomous flight in complex environments. As the complexity of UAV missions continues to increase, they often need to perform high-dynamic load flight maneuvers such as rapid acceleration, sharp turns, rapid climbs, rapid dives, and continuous maneuvers during actual operation.

[0003] In existing technologies, UAV flight control typically employs separate attitude controllers and energy management strategies. The attitude controller primarily generates control commands based on flight attitude errors, while the energy management strategy adjusts energy consumption based on battery status, motor status, and power consumption. In practical applications, attitude control and energy control processes usually utilize independent control architectures.

[0004] However, in the process of realizing the technical solution of this application, the inventors of this application discovered that the existing UAV control methods usually lack a unified analysis mechanism for the dynamic correlation between energy state and attitude state, which makes it difficult to effectively compensate for the disturbance state generated under high dynamic load conditions, resulting in a large cross-interference between energy control and attitude control. Summary of the Invention

[0005] To overcome the above shortcomings, this invention provides a coupling optimization method for energy state and attitude state under high dynamic loads, aiming to improve the problem of large cross-interference between energy control and attitude control in the prior art.

[0006] In a first aspect, the present invention provides the following technical solution: a method for coupled optimization of energy state and attitude state under high dynamic load, comprising the following steps:

[0007] S1. Acquire battery operation data, motor operation data, and flight attitude data, and establish a multi-domain joint state model;

[0008] S2. Generate a coupling sensitivity index based on the multi-domain joint state model;

[0009] S3. Construct a nonlinear disturbance observation decoupling control layer based on the coupling sensitivity index, and generate disturbance estimation results;

[0010] S4. Based on the multi-domain joint state model, the coupling sensitivity index, and the disturbance estimation results, construct a collaborative optimization objective for energy state and attitude state, and establish a model prediction controller;

[0011] S5. Dynamically adjust the target weights corresponding to the energy state and attitude state co-optimization target based on voltage deviation, temperature margin, attitude error, and angular velocity error;

[0012] S6. The model predictive controller generates the optimal control sequence based on the adjusted target weights, and performs rolling time-domain optimization based on the system state corresponding to the current control cycle to obtain the current control quantity.

[0013] S7. Generate motor drive control commands based on the current control quantity, and control the UAV to perform flight actions based on the motor drive control commands, while updating the system status corresponding to the next control cycle.

[0014] Preferably, in step S1, the step of establishing the multi-domain joint state model includes:

[0015] Obtain battery terminal voltage, battery output current, and motor temperature to construct an energy state model;

[0016] Obtain the roll angle, pitch angle, yaw angle, and corresponding angular velocity to construct an attitude state model;

[0017] Obtain drive control quantities;

[0018] The multi-domain joint state model is established based on the energy state model, the attitude state model, and the driving control quantity.

[0019] Preferably, in step S2, the step of generating the coupling sensitivity index based on the multi-domain joint state model includes:

[0020] The energy state change rate is generated based on the battery operating data and the motor operating data;

[0021] Generate the attitude state change rate based on the flight attitude data;

[0022] A coupling sensitivity index is generated based on the energy state change rate and the attitude state change rate;

[0023] The coupling level corresponding to the energy state and attitude state is determined based on the coupling sensitivity index.

[0024] Preferably, in step S3, the step of generating the disturbance estimation result includes:

[0025] Construct a disturbance observation model;

[0026] The disturbance observation model is used to obtain the disturbance state corresponding to the high dynamic load;

[0027] A disturbance compensation amount is generated based on the disturbance state;

[0028] The decoupling control input is generated based on the disturbance compensation amount.

[0029] Preferably, in step S4, the step of constructing the co-optimization objective of energy state and attitude state includes:

[0030] Construct an attitude tracking target; construct an energy management target; construct a control input smoothing target;

[0031] The energy state and attitude state co-optimization objective is constructed based on the attitude tracking objective, the energy management objective, and the control input smoothing objective.

[0032] Preferably, in step S5, the step of dynamically adjusting the target weights corresponding to the energy state and attitude state co-optimization target includes:

[0033] Obtain the voltage deviation, the temperature margin, the attitude error, and the angular velocity error;

[0034] A weighting adjustment factor is generated based on the voltage deviation, the temperature margin, the attitude error, and the angular velocity error;

[0035] Adjust the attitude target weight and energy target weight according to the weight adjustment factor.

[0036] Preferably, in step S6, the step of generating the optimal control sequence using the model predictive controller based on the adjusted target weights includes:

[0037] Obtain the energy state prediction results within the future prediction time domain;

[0038] Obtain the attitude state prediction results in the future prediction time domain; construct optimization constraints;

[0039] The optimal control sequence is generated based on the optimization constraints and the adjusted target weights.

[0040] Preferably, in step S6, the step of performing rolling time-domain optimization based on the system state corresponding to the current control cycle includes:

[0041] Obtain the coupling sensitivity index corresponding to the current control cycle;

[0042] Window adjustment parameters are generated based on the coupling sensitivity index;

[0043] Determine the optimal window length based on the window adjustment parameters;

[0044] Update the prediction window based on the optimized window length;

[0045] The optimal control sequence is updated based on the updated prediction window;

[0046] Extract the current control quantity from the updated optimal control sequence.

[0047] Preferably, in step S7, the step of updating the system state corresponding to the next control cycle includes:

[0048] Reacquire battery operation data, motor operation data, and flight attitude data;

[0049] Update the multi-domain joint state model;

[0050] The updated multi-domain joint state model is used as the predicted initial state for the next control cycle.

[0051] Secondly, the present invention provides the following technical solution for a coupled optimization system of energy state and attitude state under high dynamic load, comprising:

[0052] The state modeling module is used to acquire battery operation data, motor operation data, and flight attitude data, and to establish a multi-domain joint state model.

[0053] A coupling analysis module, connected to the state modeling module, is used to generate a coupling sensitivity index based on the multi-domain joint state model.

[0054] The disturbance decoupling module, connected to the coupling analysis module, is used to construct a nonlinear disturbance observation decoupling control layer based on the coupling sensitivity index and generate disturbance estimation results;

[0055] The collaborative optimization module, connected to the state modeling module, the coupling analysis module, and the disturbance decoupling module, is used to construct a collaborative optimization objective for energy state and attitude state based on the multi-domain joint state model, the coupling sensitivity index, and the disturbance estimation results, and to establish a model prediction controller.

[0056] The weight adjustment module, connected to the collaborative optimization module, is used to dynamically adjust the target weights corresponding to the collaborative optimization targets of energy state and attitude state based on voltage deviation, temperature margin, attitude error, and angular velocity error.

[0057] The rolling optimization module, connected to the collaborative optimization module and the weight adjustment module, is used to generate the optimal control sequence according to the adjusted target weight, and update the prediction window and generate the current control quantity according to the system state corresponding to the current control cycle.

[0058] The control output module is connected to the rolling optimization module and the state modeling module respectively. It is used to generate motor drive control commands based on the current control quantity and control the UAV to perform flight actions, while updating the system state corresponding to the next control cycle.

[0059] The present invention has the following beneficial effects:

[0060] 1. This invention establishes a multi-domain joint state model and constructs a nonlinear disturbance observation decoupling control layer based on the coupling sensitivity index, enabling unified analysis of the dynamic correlation between energy state and attitude state, and compensating for disturbance states caused by high dynamic loads, thereby reducing cross-interference between energy control and attitude control.

[0061] 2. This invention constructs a collaborative optimization target for energy state and attitude state, establishes a model predictive controller, and dynamically adjusts the target weights based on voltage deviation, temperature margin, attitude error, and angular velocity error, enabling the energy management target and attitude control target to participate collaboratively in the control process, thereby improving the control stability under high dynamic load conditions.

[0062] 3. This invention dynamically adjusts the prediction window length during the rolling time-domain optimization process based on the coupling sensitivity index, enabling the control strategy to be updated as the current coupling state changes, thereby improving the UAV's adaptability to changes in operating state under conditions such as penetration maneuvers and rapid acceleration. Attached Figure Description

[0063] Figure 1 This is a flowchart of the coupling optimization method for energy state and attitude state under high dynamic load proposed in this invention.

[0064] Figure 2 This is a schematic diagram illustrating the construction of a multi-domain joint state model proposed in this invention;

[0065] Figure 3 This is a schematic diagram of a coupling sensitivity analysis and disturbance decoupling process proposed in this invention;

[0066] Figure 4 This is a schematic diagram of a collaborative optimization structure for energy state and attitude state proposed in this invention;

[0067] Figure 5 This is a schematic diagram of a rolling time-domain optimization process proposed in this invention;

[0068] Figure 6 This is a diagram of a system architecture for the coupled optimization of energy state and attitude state under high dynamic load proposed in this invention. Detailed Implementation

[0069] The technical solutions in 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.

[0070] Reference Figure 1 and Figure 2 This invention provides a coupled optimization method for energy state and attitude state under high dynamic loads, comprising the following steps:

[0071] S1. Acquire battery operation data, motor operation data, and flight attitude data, and establish a multi-domain joint state model;

[0072] Preferably, in step S1, the steps for establishing the multi-domain joint state model include:

[0073] Obtain battery terminal voltage, battery output current, and motor temperature to construct an energy state model;

[0074] Obtain the roll angle, pitch angle, yaw angle, and corresponding angular velocity to construct an attitude state model;

[0075] Obtain drive control quantities;

[0076] A multi-domain joint state model is established based on the energy state model, attitude state model, and driving control variables.

[0077] Specifically, the UAV flight controller is connected to the battery management unit, electronic speed controller (ESC) drive unit, and inertial measurement unit (IMU) via an onboard communication bus. The battery management unit collects battery terminal voltage and battery output current; the ESC drive unit collects drive control quantities and motor temperatures for each motor; and the IMU collects roll angle, pitch angle, yaw angle, and corresponding angular velocity. The flight controller synchronously acquires the above data according to a preset sampling period and performs timestamp alignment processing on data from different sources, ensuring that battery operation data, motor operation data, and flight attitude data within the same control period correspond to the system operating state at the same moment.

[0078] After obtaining the battery terminal voltage, battery output current, and motor temperature, an energy state model is constructed. The energy state model characterizes the operating states of the power system and propulsion system within the current control cycle. The energy state vector is represented as follows:

[0079] ;

[0080] in, Represents the energy state vector; This indicates the battery terminal voltage corresponding to the current control cycle; This indicates the battery output current corresponding to the current control cycle; This indicates the motor temperature corresponding to the current control cycle.

[0081] After obtaining the roll angle, pitch angle, yaw angle, and corresponding angular velocity, an attitude state model is constructed. The attitude state model characterizes the attitude changes of the UAV within the current control cycle. The attitude state vector is represented as follows:

[0082] ;

[0083] in, Represents the attitude state vector; Indicates the roll angle; ψt represents the pitch angle; ψt represents the yaw angle. , , These represent the roll rate, pitch rate, and yaw rate, respectively.

[0084] After acquiring the drive control quantity, it is used as the control input for the propulsion system. The drive control quantity is expressed as: ;in, Represents the set of drive control variables; Indicates the first The drive control quantities corresponding to each motor; Indicates the number of motors.

[0085] Subsequently, the energy state model, attitude state model, and drive control variables are unified and correlated to establish a multi-domain joint state model. This multi-domain joint state model describes the state change relationships between the power system, propulsion system, and flight attitude system, and its state update relationship is expressed as follows:

[0086] ;

[0087] in, This represents the system state vector corresponding to the current control cycle; This represents the system state vector corresponding to the next control cycle; This represents the state update function; This represents the disturbance state corresponding to the current control cycle. The system state vector is composed of both the energy state vector and the attitude state vector. ;

[0088] By employing the above method, battery operating data, motor operating data, flight attitude data, and drive control variables are uniformly mapped to the same state space. This allows subsequent calculations, including the generation of coupling sensitivity indicators, disturbance state estimation, and collaborative optimization, to be performed based on a unified system state, providing fundamental data input for the correlation analysis between energy state and attitude state. Furthermore, by establishing a multi-domain joint state model, the power system state, propulsion system state, and flight attitude state can be uniformly described, providing a unified data foundation for subsequent coupling relationship analysis and control decisions.

[0089] Reference Figure 3 Furthermore, S2 generates a coupling sensitivity index based on a multi-domain joint state model;

[0090] Preferably, in step S2, the step of generating the coupling sensitivity index based on the multi-domain joint state model includes:

[0091] The energy state change rate is generated based on battery and motor operating data.

[0092] Generate attitude state change rate based on flight attitude data;

[0093] A coupling sensitivity index is generated based on the rate of change of energy state and the rate of change of attitude state.

[0094] The coupling level corresponding to the energy state and attitude state is determined based on the coupling sensitivity index.

[0095] Specifically, after obtaining the multi-domain joint state model, the flight controller calculates the rate of change of energy state and the rate of change of attitude state based on system state data over multiple consecutive control cycles. The rate of change of energy state characterizes the degree of state change of the power system and propulsion system between adjacent control cycles. The flight controller first acquires the battery terminal voltage, battery output current, and motor temperature data corresponding to the current control cycle and the previous control cycle, and generates the energy state change. The rate of change of energy state is expressed as:

[0096] ;

[0097] in, Represents the rate of change of energy state; This represents the change in energy state between adjacent control cycles; This indicates the time interval between adjacent control cycles.

[0098] Simultaneously, the flight controller acquires the roll angle, pitch angle, yaw angle, and corresponding angular velocity data corresponding to the current control cycle and the previous control cycle, and generates attitude state change parameters. The attitude state change rate is expressed as:

[0099] ;

[0100] in, Indicates the rate of change of attitude state; Indicates the change in attitude state between adjacent control cycles; This indicates the time interval between adjacent control cycles.

[0101] After obtaining the rate of change of energy state and the rate of change of attitude state, the flight controller generates a coupling sensitivity index based on these two types of rates of change. The coupling sensitivity index characterizes the degree to which the energy state change affects the attitude state change and the degree of feedback from the attitude state change to the energy state change within the current control cycle. The coupling sensitivity index is expressed as:

[0102] ;

[0103] in, This indicates the coupling sensitivity index; Indicates the energy state weighting coefficient; Indicates the attitude state weight coefficient; Represents the rate of change of energy state; This represents the rate of change of attitude state.

[0104] Subsequently, the coupling level corresponding to the energy state and attitude state is determined based on the coupling sensitivity index. Specifically, a coupling level interval table is pre-established, mapping the coupling sensitivity index to low, medium, and high coupling levels. When the coupling sensitivity index is in the low coupling level interval, it indicates a low correlation between the energy state change and the attitude state change within the current control cycle; when the coupling sensitivity index is in the medium coupling level interval, it indicates a significant correlation between the energy state change and the attitude state change within the current control cycle; and when the coupling sensitivity index is in the high coupling level interval, it indicates a strong correlation between the energy state change and the attitude state change within the current control cycle.

[0105] In some implementations, to avoid the impact of fluctuations at a single sampling point on the determination of the coupling level, the flight controller can also generate a coupling sensitivity sequence based on the coupling sensitivity index corresponding to multiple consecutive control cycles, and determine the coupling level corresponding to the current control cycle based on the coupling sensitivity sequence.

[0106] In this way, we can use the energy state information and attitude state information in the multi-domain joint state model to generate a unified coupling sensitivity index, and use the coupling sensitivity index to quantify the degree of correlation between the energy state and the attitude state, so as to provide a basis for the construction of the nonlinear disturbance observation decoupling control layer and the parameter adjustment in the rolling time domain optimization process.

[0107] Furthermore, S3, a nonlinear disturbance observation decoupling control layer is constructed based on the coupling sensitivity index, and disturbance estimation results are generated;

[0108] Preferably, in step S3, the step of generating the disturbance estimation result includes:

[0109] Construct a disturbance observation model;

[0110] The disturbance state corresponding to a highly dynamic load is obtained using a disturbance observation model;

[0111] Generate disturbance compensation amount based on disturbance state;

[0112] The decoupled control input is generated based on the disturbance compensation amount.

[0113] Specifically, after obtaining the coupling sensitivity index, the flight controller constructs a nonlinear disturbance observation decoupling control layer based on the coupling level corresponding to the current control cycle. This nonlinear disturbance observation decoupling control layer is used to estimate disturbances caused by changes in energy state and attitude state under high dynamic load conditions and generate corresponding compensation information. The flight controller first establishes a disturbance observation model based on a multi-domain joint state model, using the system state vector, drive control quantity, and coupling sensitivity index corresponding to the current control cycle as model inputs, and tracks the deviation between the actual system state and the model's predicted state. The disturbance state is represented as:

[0114] ;

[0115] in, Indicates a disturbance state; This represents the torque disturbance component caused by load changes; This represents the disturbance component caused by power fluctuations in the propulsion system. This represents the coupled disturbance component caused by attitude changes.

[0116] The flight controller uses a disturbance observation model to compare the actual state with the predicted state for the current control cycle, and generates a disturbance estimate based on the state deviation. The disturbance estimate is expressed as: ;

[0117] in, Indicates the estimated disturbance value; This indicates the actual system state corresponding to the current control cycle; This indicates the predicted system state corresponding to the current control cycle.

[0118] Subsequently, a disturbance compensation amount is generated based on the disturbance estimate. This compensation amount is used to offset the additional effects caused by energy state fluctuations and attitude state changes under high dynamic load conditions. The disturbance compensation amount is expressed as:

[0119] ;

[0120] in, Indicates the amount of disturbance compensation; Indicates the disturbance compensation coefficient; This represents the estimated disturbance value.

[0121] In some implementations, the disturbance compensation coefficient can be dynamically adjusted based on the coupling sensitivity index. When the coupling sensitivity index increases, it indicates that the correlation between the energy state and the attitude state in the current control cycle has increased, and the disturbance compensation coefficient is increased accordingly. When the coupling sensitivity index decreases, the disturbance compensation coefficient is decreased to avoid excessive compensation causing fluctuations in the control input.

[0122] After obtaining the disturbance compensation amount, it is superimposed on the original control input to generate the decoupled control input. The decoupled control input is expressed as:

[0123] ;

[0124] in, Indicates decoupling of control input; Indicates the raw control input; This represents the disturbance compensation amount. The generated decoupled control input is used as an input parameter in the subsequent energy state and attitude state co-optimization process, enabling the subsequent control process to be optimized based on the state that has been compensated for the disturbance.

[0125] By using the above method, the disturbance state under high dynamic load conditions can be estimated using the coupling sensitivity index, and the corresponding disturbance compensation amount and decoupling control input can be generated based on the disturbance state, thereby reducing the impact of the additional disturbance caused by the mutual influence between energy state changes and attitude state changes on the subsequent control process.

[0126] Reference Figure 4 Furthermore, S4, based on the multi-domain joint state model, coupling sensitivity index and disturbance estimation results, constructs a collaborative optimization objective for energy state and attitude state, and establishes a model predictive controller;

[0127] Preferably, in step S4, the step of constructing the joint optimization objective of energy state and attitude state includes:

[0128] Construct an attitude tracking target; construct an energy management target; construct a control input smoothing target;

[0129] A collaborative optimization objective for energy state and attitude state is constructed based on the attitude tracking objective, energy management objective, and control input smoothing objective.

[0130] Specifically, after obtaining the multi-domain joint state model, coupling sensitivity index, and disturbance estimation results, the flight controller constructs a co-optimization objective for energy state and attitude state, and establishes a model predictive controller. The co-optimization objective is used to simultaneously consider flight attitude control requirements and energy consumption constraints within the same optimization framework, enabling subsequent control processes to be optimized based on a unified objective.

[0131] The flight controller first constructs an attitude tracking target based on the target attitude parameters corresponding to the flight mission. The attitude tracking target characterizes the degree of deviation between the current attitude state and the target attitude state. The attitude tracking target is represented as: ;in, Indicates the attitude tracking target; Indicates the target's attitude state; Indicates the current attitude state.

[0132] Subsequently, energy management targets are constructed based on battery and motor operating data. These targets are used to constrain battery energy consumption and propulsion system power changes during the control process. The energy management targets are expressed as follows: ;in, Indicate energy management goals; Indicates the current energy state; Indicates the reference energy state.

[0133] After obtaining the attitude tracking and energy management objectives, a control input smoothing objective is further constructed. The control input smoothing objective is used to limit excessive changes in control input between adjacent control cycles, reducing abrupt changes in control input. The control input smoothing objective is expressed as: ;in, Indicates the target for smoothing the control input; This indicates the change in control input between adjacent control cycles.

[0134] After obtaining the above three objectives, the flight controller combines the attitude tracking objective, energy management objective, and control input smoothing objective to generate a co-optimization objective for energy state and attitude state. The co-optimization objective is represented as:

[0135] ;

[0136] in, This represents the objective of co-optimizing energy state and attitude state; Indicates the weight of the attitude target; Indicates the weight of the energy target; Indicates the target weights for controlling input smoothing; Indicates the attitude tracking target; Indicate energy management goals; This indicates the target for controlling the smoothness of the input.

[0137] After constructing the collaborative optimization objective, a model predictive controller is established. The flight controller uses a multi-domain joint state model as the predictive model, decoupled control inputs as control variables, and disturbance estimation results as predictive correction parameters. System state constraints and control input constraints are set. The system state constraints constrain the range of attitude state changes and energy state changes, while the control input constraints constrain the range of changes in the corresponding drive control quantities of each motor.

[0138] Subsequently, the model predictive controller predicts the state change trend in the next few control cycles based on the system state corresponding to the current control cycle, and solves the collaborative optimization objective based on the prediction results, providing an optimization basis for subsequent objective weight adjustment and optimal control sequence generation.

[0139] By adopting the above approach, attitude tracking requirements, energy management requirements, and control input constraint requirements can be integrated into the same optimization framework. A model predictive controller can be established using a multi-domain joint state model, so that the subsequent control process can simultaneously consider the correlation between energy state and attitude state, providing an optimization target basis for collaborative control under high dynamic load conditions.

[0140] Furthermore, S5 dynamically adjusts the target weights corresponding to the energy state and attitude state co-optimization target based on voltage deviation, temperature margin, attitude error, and angular velocity error.

[0141] Preferably, in step S5, the step of dynamically adjusting the target weights corresponding to the co-optimization objective of energy state and attitude state includes:

[0142] Obtain voltage deviation, temperature margin, attitude error, and angular velocity error;

[0143] A weighting adjustment factor is generated based on voltage deviation, temperature margin, attitude error, and angular velocity error;

[0144] Adjust the attitude target weight and energy target weight according to the weight adjustment factor.

[0145] Specifically, after establishing a collaborative optimization objective for energy state and attitude state, the flight controller dynamically adjusts the weight parameters of each item in the collaborative optimization objective based on the system operating state corresponding to the current control cycle. This enables the subsequent model predictive control process to adjust the optimization ratio between attitude control requirements and energy management requirements according to changes in the current operating conditions. The flight controller first obtains the battery terminal voltage corresponding to the current control cycle and generates a voltage deviation based on the difference between the current battery terminal voltage and the reference voltage; it then obtains the remaining temperature space between the current motor temperature and the preset temperature threshold to generate a temperature margin; it obtains the deviation between the current attitude state and the target attitude state to generate an attitude error; and it obtains the deviation between the current angular velocity and the target angular velocity to generate an angular velocity error. The attitude error is expressed as:

[0146] ;

[0147] in, Indicates attitude error; Indicates the target's attitude state; Indicates the current attitude state.

[0148] Angular velocity error is expressed as: ;in, Indicates angular velocity error; Indicates the target angular velocity; This indicates the current angular velocity.

[0149] After obtaining the voltage deviation, temperature margin, attitude error, and angular velocity error, the flight controller generates a weighting adjustment factor based on these parameters. The weighting adjustment factor characterizes the system's emphasis on attitude control and energy management requirements within the current control cycle. The weighting adjustment factor is expressed as:

[0150] ;

[0151] in, Indicates the weight adjustment factor; Indicates voltage deviation; Indicates temperature margin; Indicates attitude error; Indicates angular velocity error; This represents the weight adjustment function.

[0152] In some implementations, the flight controller establishes a weight adjustment rule table. When attitude error and angular velocity error increase, the weight of attitude targets is increased; when voltage deviation increases or temperature margin decreases, the weight of energy targets is increased. For situations where both attitude control and energy management requirements exist simultaneously, the weights of the two types of targets are allocated in a coordinated manner according to a weight adjustment factor.

[0153] Subsequently, the attitude target weights and energy target weights are updated according to the weight adjustment factor. The adjusted attitude target weights are expressed as follows: The adjusted energy target weights are expressed as follows:

[0154] ;in, This indicates the adjusted attitude target weights; This indicates the adjusted energy target weights; This indicates the attitude target weights before adjustment; This indicates the energy target weights before adjustment; This indicates the adjustment amount of the attitude target weight; This indicates the amount of weight adjustment for the energy target.

[0155] The updated attitude target weights and energy target weights are written into the model predictive controller and participate in the subsequent optimal control sequence solution process, enabling the model predictive controller to adjust the optimization direction according to the current system state in different control cycles.

[0156] The above method enables the dynamic adjustment of the target weights corresponding to the energy state and attitude state co-optimization objectives based on voltage deviation, temperature margin, attitude error, and angular velocity error. This allows the attitude control requirements and energy management requirements to be dynamically allocated according to changes in the current operating conditions, providing optimized parameters that match the current operating state for the generation of the subsequent optimal control sequence.

[0157] Reference Figure 5 Furthermore, S6, the model predictive controller generates the optimal control sequence based on the adjusted target weights, and performs rolling time-domain optimization based on the system state corresponding to the current control cycle to obtain the current control quantity;

[0158] Preferably, in step S6, the step of generating the optimal control sequence using the model predictive controller based on the adjusted target weights includes:

[0159] Obtain the energy state prediction results within the future prediction time domain;

[0160] Obtain the attitude state prediction results in the future prediction time domain;

[0161] Construct optimization constraints;

[0162] The optimal control sequence is generated based on the optimization constraints and the adjusted target weights.

[0163] Preferably, in step S6, the step of performing rolling time-domain optimization based on the system state corresponding to the current control cycle includes:

[0164] Obtain the coupling sensitivity index corresponding to the current control cycle;

[0165] Window adjustment parameters are generated based on the coupling sensitivity index;

[0166] Determine the optimal window length based on window adjustment parameters;

[0167] Update the prediction window based on the optimized window length;

[0168] Update the optimal control sequence based on the updated prediction window;

[0169] Extract the current control quantity from the updated optimal control sequence.

[0170] Specifically, after adjusting the target weights, the flight controller invokes the model predictive controller to predict and analyze the system's operating state in the future prediction time domain. The model predictive controller uses the multi-domain joint state model corresponding to the current control cycle as the prediction basis, the decoupled control input as the control variable, and the adjusted attitude target weights and energy target weights as optimization parameters to recursively predict the system state over multiple future control cycles, generating energy state prediction results and attitude state prediction results in the future prediction time domain. The predicted state is represented as follows:

[0171] ;

[0172] in, Indicates the first The predicted state corresponding to each prediction time; Indicates the first The predicted state corresponding to each prediction time; Indicates the first The control input corresponding to each predicted time point; This represents the state prediction function.

[0173] During the prediction process, the flight controller simultaneously constructs optimization constraints. These constraints include energy state constraints, attitude state constraints, and control input constraints. The energy state constraints limit the variations in battery terminal voltage, battery output current, and motor temperature; the attitude state constraints limit the variations in roll angle, pitch angle, yaw angle, and corresponding angular velocity; and the control input constraints limit the variations in the corresponding drive control quantities of each motor. Subsequently, the co-optimization objective is solved based on the optimization constraints and adjusted target weights to generate the optimal control sequence. The optimal control sequence is expressed as:

[0174] ;

[0175] in, Represents the optimal control sequence; Indicates the first The control input corresponding to each predictive control cycle; This indicates the length of the prediction time domain.

[0176] After obtaining the optimal control sequence, the flight controller performs a rolling time-domain optimization process. The flight controller first acquires the coupling sensitivity index corresponding to the current control cycle and generates window adjustment parameters based on the coupling sensitivity index. The window adjustment parameters reflect the degree of coupling between the current energy state and attitude state. The window adjustment parameters are expressed as: ;in, Indicates window adjustment parameters; This indicates the coupling sensitivity index; This represents the window adjustment function.

[0177] Subsequently, the optimal window length is determined based on the window adjustment parameters. Increasing the window adjustment parameters indicates a greater correlation between the energy state and attitude state within the current control cycle, thus expanding the optimal window length; conversely, decreasing the window adjustment parameters results in a correspondingly shorter optimal window length. The optimal window length is expressed as:

[0178] ;

[0179] in, This indicates an optimized window length; This represents the window length mapping function.

[0180] After determining the optimization window length, the flight controller updates the prediction window based on the optimization window length and re-executes the prediction optimization calculation based on the updated prediction window to correct the optimal control sequence. Subsequently, the control input corresponding to the current control cycle is extracted from the updated optimal control sequence as the current control quantity. Only the first control input in the optimal control sequence is applied to the current control cycle; the remaining control inputs are retained as prediction reference quantities for subsequent control cycles and recalculated in the next control cycle.

[0181] In some implementations, after each control cycle is completed, the flight controller reacquires the current system state and uses the updated system state as the new initial state for prediction to repeat the above prediction and optimization process, thereby forming a continuously rolling predictive control mechanism.

[0182] In this way, the model predictive controller can generate the optimal control sequence based on the adjusted target weights and dynamically adjust the prediction window length according to the coupling sensitivity index, so that the prediction time domain corresponds with the current system coupling state, thereby enabling the generated current control quantity to simultaneously meet the energy state control requirements and the attitude state control requirements.

[0183] Furthermore, S7 generates motor drive control commands based on the current control input, and controls the UAV to perform flight maneuvers according to the motor drive control commands, while updating the system state for the next control cycle.

[0184] Preferably, in step S7, the step of updating the system state corresponding to the next control cycle includes:

[0185] Reacquire battery operation data, motor operation data, and flight attitude data;

[0186] Update the multi-domain joint state model;

[0187] The updated multi-domain joint state model is used as the predicted initial state for the next control cycle.

[0188] Specifically, after obtaining the current control input, the flight controller generates corresponding motor drive control commands and sends them to each electronic speed controller (ESC) unit via the onboard communication bus. The ESC units adjust the output speed and torque of the corresponding motors based on the received motor drive control commands, causing the propulsion system to generate thrust distribution and attitude control torque corresponding to the current control input. The flight controller then controls the UAV to perform flight maneuvers such as acceleration, deceleration, climb, descent, roll, pitch, and yaw based on the thrust and torque changes of each motor, thereby achieving real-time adjustment of the flight status.

[0189] During the execution of motor drive control commands, the flight controller synchronously initiates the data acquisition process for the next control cycle, reacquiring battery operating data, motor operating data, and flight attitude data. Battery operating data includes battery terminal voltage and battery output current; motor operating data includes drive control quantities and motor temperature; and flight attitude data includes roll angle, pitch angle, yaw angle, and corresponding angular velocity. The acquired data serves as system feedback information after the current control action is executed, participating in subsequent state update processes.

[0190] After receiving the latest system feedback data, the flight controller updates the multi-domain joint state model based on the control input corresponding to the current control cycle and the latest acquired data. The updated system state vector is represented as follows:

[0191] ;

[0192] in, This represents the updated system state vector; This represents the updated energy state vector; This represents the updated attitude state vector.

[0193] The updated energy state vector is represented as follows: The updated attitude state vector is represented as follows: ;in, , , These represent the updated roll angle, pitch angle, and yaw angle, respectively. , , These represent the updated roll rate, pitch rate, and yaw rate, respectively.

[0194] Subsequently, the flight controller uses the updated multi-domain joint state model as the predicted initial state for the next control cycle and inputs the predicted initial state into the model predictive controller. At the start of the next control cycle, the model predictive controller re-executes the processes of generating coupling sensitivity indices, estimating disturbance states, adjusting target weights, and generating optimal control sequences based on the updated predicted initial state, thus forming a continuous closed-loop control process.

[0195] In some implementations, the flight controller can also record the system state update results corresponding to multiple consecutive control cycles and establish a state history sequence for subsequent prediction model parameter updates and control parameter correction processes.

[0196] In this way, motor drive control commands can be generated based on the current control variables and the UAV can be controlled to perform corresponding flight actions. At the same time, the multi-domain joint state model can be updated using the feedback data after execution, so that the prediction process of the next control cycle is based on the latest system state, thereby forming a closed-loop control process that coordinates the optimization of energy state and attitude state.

[0197] Example 2: Refer to Figure 6 In a second embodiment of the present invention, the present invention provides a coupled optimization system for energy state and attitude state under high dynamic loads, comprising:

[0198] The state modeling module is used to acquire battery operation data, motor operation data, and flight attitude data, and to establish a multi-domain joint state model.

[0199] The coupling analysis module, connected to the state modeling module, is used to generate coupling sensitivity indices based on a multi-domain joint state model.

[0200] The disturbance decoupling module, connected to the coupling analysis module, is used to construct a nonlinear disturbance observation decoupling control layer based on the coupling sensitivity index and generate disturbance estimation results.

[0201] The collaborative optimization module, connected to the state modeling module, coupling analysis module, and disturbance decoupling module, is used to construct a collaborative optimization objective for energy state and attitude state based on a multi-domain joint state model, coupling sensitivity index, and disturbance estimation results, and to establish a model predictive controller.

[0202] The weight adjustment module, connected to the collaborative optimization module, is used to dynamically adjust the target weights corresponding to the collaborative optimization objectives of energy state and attitude state based on voltage deviation, temperature margin, attitude error, and angular velocity error.

[0203] The rolling optimization module, connected to the collaborative optimization module and the weight adjustment module, is used to generate the optimal control sequence based on the adjusted target weights, and to update the prediction window and generate the current control quantity based on the system state corresponding to the current control cycle.

[0204] The control output module is connected to the rolling optimization module and the state modeling module respectively. It is used to generate motor drive control commands based on the current control quantity and control the UAV to perform flight actions, while updating the system state corresponding to the next control cycle.

[0205] Specifically, the state modeling module includes a data acquisition unit and a state modeling unit. The data acquisition unit is used to receive battery terminal voltage and battery output current data output by the battery management system, motor temperature and drive control data output by the ESC drive system, and flight attitude data output by the inertial measurement system. The state modeling unit is used to establish a multi-domain joint state model based on the acquired data and send the multi-domain joint state model to the coupling analysis module, the collaborative optimization module, and the control output module.

[0206] The coupling analysis module includes a rate of change analysis unit and a coupling level analysis unit. The rate of change analysis unit is used to generate the energy state change rate based on battery operation data and motor operation data, and to generate the attitude state change rate based on flight attitude data. The coupling level analysis unit is used to generate a coupling sensitivity index based on the energy state change rate and attitude state change rate, and to determine the coupling level corresponding to the current control cycle based on the coupling sensitivity index.

[0207] The disturbance decoupling module includes a disturbance observation unit and a compensation generation unit. The disturbance observation unit is used to construct a nonlinear disturbance observation decoupling control layer based on the coupling sensitivity index and to obtain the disturbance state corresponding to the high dynamic load. The compensation generation unit is used to generate a disturbance compensation amount based on the disturbance state and to generate a decoupling control input based on the disturbance compensation amount.

[0208] The collaborative optimization module includes a target construction unit and a prediction control unit. The target construction unit is used to construct attitude tracking targets, energy management targets, and control input smoothing targets, and to generate collaborative optimization targets for energy state and attitude state. The prediction control unit is used to establish a model predictive controller and to perform predictive control calculations based on the multi-domain joint state model and disturbance estimation results.

[0209] The weight adjustment module includes a state evaluation unit and a weight adjustment unit. The state evaluation unit is used to obtain voltage deviation, temperature margin, attitude error and angular velocity error. The weight adjustment unit is used to generate a weight adjustment factor based on the above parameters, and adjust the attitude target weight and energy target weight based on the weight adjustment factor.

[0210] The rolling optimization module includes a sequence generation unit and a window update unit. The sequence generation unit generates the optimal control sequence based on the adjusted target weights. The window update unit generates window adjustment parameters based on the coupling sensitivity index, determines the optimization window length based on the window adjustment parameters, updates the prediction window based on the optimization window length, and extracts the current control quantity from the updated optimal control sequence.

[0211] The control output module includes a drive control unit and a state update unit. The drive control unit is used to generate motor drive control commands based on the current control quantity and send them to the ESC drive system for execution. The state update unit is used to reacquire battery operation data, motor operation data and flight attitude data, update the multi-domain joint state model, and use the updated multi-domain joint state model as the predicted initial state for the next control cycle.

[0212] In this embodiment, the modules interact with each other via an airborne communication bus. Specifically, the state modeling module is connected to the coupling analysis module, the collaborative optimization module, and the control output module; the coupling analysis module is connected to the disturbance decoupling module; the disturbance decoupling module is connected to the collaborative optimization module; the collaborative optimization module is connected to the weight adjustment module and the rolling optimization module; and the rolling optimization module is connected to the control output module, thus forming a closed-loop control process from state modeling, coupling analysis, disturbance compensation, collaborative optimization, weight adjustment, rolling optimization to control output.

[0213] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A coupled optimization method for energy state and attitude state under high dynamic loads, characterized in that, Includes the following steps: S1. Acquire battery operation data, motor operation data, and flight attitude data, and establish a multi-domain joint state model; S2. Generate a coupling sensitivity index based on the multi-domain joint state model; S3. Construct a nonlinear disturbance observation decoupling control layer based on the coupling sensitivity index, and generate disturbance estimation results; S4. Based on the multi-domain joint state model, the coupling sensitivity index, and the disturbance estimation results, construct a collaborative optimization objective for energy state and attitude state, and establish a model prediction controller; S5. Dynamically adjust the target weights corresponding to the energy state and attitude state co-optimization target based on voltage deviation, temperature margin, attitude error, and angular velocity error; S6. The model predictive controller generates the optimal control sequence based on the adjusted target weights, and performs rolling time-domain optimization based on the system state corresponding to the current control cycle to obtain the current control quantity. S7. Generate motor drive control commands based on the current control quantity, and control the UAV to perform flight actions based on the motor drive control commands, while updating the system status corresponding to the next control cycle.

2. The coupled optimization method for energy state and attitude state under high dynamic loads according to claim 1, characterized in that, In step S1, the step of establishing the multi-domain joint state model includes: Obtain battery terminal voltage, battery output current, and motor temperature to construct an energy state model; Obtain the roll angle, pitch angle, yaw angle, and corresponding angular velocity to construct an attitude state model; Obtain drive control quantities; The multi-domain joint state model is established based on the energy state model, the attitude state model, and the driving control quantity.

3. The coupled optimization method for energy state and attitude state under high dynamic loads according to claim 1, characterized in that, In step S2, the step of generating a coupling sensitivity index based on the multi-domain joint state model includes: The energy state change rate is generated based on the battery operating data and the motor operating data; Generate the attitude state change rate based on the flight attitude data; A coupling sensitivity index is generated based on the energy state change rate and the attitude state change rate; The coupling level corresponding to the energy state and attitude state is determined based on the coupling sensitivity index.

4. The coupled optimization method for energy state and attitude state under high dynamic loads according to claim 1, characterized in that, In step S3, the step of generating the perturbation estimation result includes: Construct a disturbance observation model; The disturbance observation model is used to obtain the disturbance state corresponding to the high dynamic load; A disturbance compensation amount is generated based on the disturbance state; The decoupling control input is generated based on the disturbance compensation amount.

5. The coupled optimization method for energy state and attitude state under high dynamic loads according to claim 1, characterized in that, In step S4, the step of constructing the joint optimization objective of energy state and attitude state includes: Construct an attitude tracking target; construct an energy management target; construct a control input smoothing target; The energy state and attitude state co-optimization objective is constructed based on the attitude tracking objective, the energy management objective, and the control input smoothing objective.

6. The coupled optimization method for energy state and attitude state under high dynamic loads according to claim 1, characterized in that, In step S5, the step of dynamically adjusting the target weights corresponding to the co-optimization objective of the energy state and attitude state includes: Obtain the voltage deviation, the temperature margin, the attitude error, and the angular velocity error; A weighting adjustment factor is generated based on the voltage deviation, the temperature margin, the attitude error, and the angular velocity error; Adjust the attitude target weight and energy target weight according to the weight adjustment factor.

7. The coupled optimization method for energy state and attitude state under high dynamic loads according to claim 1, characterized in that, In step S6, the step of using the model predictive controller to generate the optimal control sequence based on the adjusted target weights includes: Obtain the energy state prediction results within the future prediction time domain; Obtain the attitude state prediction results in the future prediction time domain; construct optimization constraints; The optimal control sequence is generated based on the optimization constraints and the adjusted target weights.

8. The coupled optimization method for energy state and attitude state under high dynamic loads according to claim 1, characterized in that, In step S6, the step of performing rolling time-domain optimization based on the system state corresponding to the current control cycle includes: Obtain the coupling sensitivity index corresponding to the current control cycle; Window adjustment parameters are generated based on the coupling sensitivity index; Determine the optimal window length based on the window adjustment parameters; Update the prediction window based on the optimized window length; The optimal control sequence is updated based on the updated prediction window; Extract the current control quantity from the updated optimal control sequence.

9. The coupled optimization method for energy state and attitude state under high dynamic loads according to claim 1, characterized in that, In step S7, the step of updating the system state corresponding to the next control cycle includes: Reacquire battery operation data, motor operation data, and flight attitude data; Update the multi-domain joint state model; The updated multi-domain joint state model is used as the predicted initial state for the next control cycle.

10. A coupled optimization system for energy state and attitude state under high dynamic loads, characterized in that, The coupled optimization method for energy state and attitude state under high dynamic loads, as described in any one of claims 1-9, includes: The state modeling module is used to acquire battery operation data, motor operation data, and flight attitude data, and to establish a multi-domain joint state model. A coupling analysis module, connected to the state modeling module, is used to generate a coupling sensitivity index based on the multi-domain joint state model. The disturbance decoupling module, connected to the coupling analysis module, is used to construct a nonlinear disturbance observation decoupling control layer based on the coupling sensitivity index and generate disturbance estimation results; The collaborative optimization module, connected to the state modeling module, the coupling analysis module, and the disturbance decoupling module, is used to construct a collaborative optimization objective for energy state and attitude state based on the multi-domain joint state model, the coupling sensitivity index, and the disturbance estimation results, and to establish a model prediction controller. The weight adjustment module, connected to the collaborative optimization module, is used to dynamically adjust the target weights corresponding to the collaborative optimization targets of energy state and attitude state based on voltage deviation, temperature margin, attitude error, and angular velocity error. The rolling optimization module, connected to the collaborative optimization module and the weight adjustment module, is used to generate the optimal control sequence according to the adjusted target weight, and update the prediction window and generate the current control quantity according to the system state corresponding to the current control cycle. The control output module is connected to the rolling optimization module and the state modeling module respectively. It is used to generate motor drive control commands based on the current control quantity and control the UAV to perform flight actions, while updating the system state corresponding to the next control cycle.

Images