Throttle adjustment method, device and equipment of unmanned aerial vehicle engine and storage medium

Abstract

The application discloses a throttle adjusting method, device and equipment of a UAV engine and a storage medium, relates to the technical field of UAVs, and realizes high-precision, fast-response and strong-anti-interference adjustment of the throttle of the UAV engine through the cooperative matching of a fractional order proportional-integral-derivative control algorithm, a radial basis function neural network feedforward compensation and a sliding mode control algorithm. In the method, S1 realizes accurate conversion of a flight control signal to a target opening degree through instruction analysis; S2 realizes real-time monitoring of the opening degree and the current through double feedback collection; S3 realizes adaptive adjustment of a position deviation and generation of a target current instruction through the fractional order proportional-integral-derivative control algorithm; S4 realizes prediction and compensation of nonlinear friction characteristics through the radial basis function neural network feedforward compensation; S5 realizes fast tracking and robust control of a current deviation through the sliding mode control algorithm; and S6 realizes accurate adjustment of the throttle opening degree through pulse width modulation driving.

Description

Technical Field

[0001] This application relates to the field of unmanned aerial vehicle (UAV) technology, and in particular to a method, apparatus, device, and storage medium for adjusting the throttle valve of an UAV engine. Background Technology

[0002] Piston-type aero engines are widely used in UAV power systems due to their high energy density and long endurance. Engine throttle control is a core component in regulating engine output power. Traditional solutions typically use a DC motor to drive the throttle actuator, adjusting the throttle opening by changing the motor's voltage polarity and duty cycle.

[0003] Currently, the following technical problems still exist in the throttle control of UAV engines: the traditional DC motor solution has a slow response speed, with the full stroke response time of the throttle usually exceeding 800 milliseconds, which is difficult to meet the requirements of rapid maneuvering flight; the anti-interference capability is weak, and the engine is prone to position drift under airflow disturbances such as gusts and eddies, resulting in unstable engine speed; the control precision is low, and the feedback accuracy of ordinary potentiometers is limited, making it difficult to achieve precise adjustment of small openings, resulting in steady-state errors. Summary of the Invention

[0004] The main objective of this application is to provide a method, device, equipment, and storage medium for adjusting the throttle valve of an unmanned aerial vehicle (UAV) engine, in order to solve the problem of the difficulty in accurately controlling traditional DC motor solutions in the prior art.

[0005] To achieve the above objectives, this application provides the following technical solution:

[0006] A method for adjusting the throttle valve of an unmanned aerial vehicle (UAV) engine, the method comprising:

[0007] Step S1: Receive the target opening command signal sent by the UAV flight control system, and parse the target opening command signal to obtain the target throttle opening value;

[0008] Step S2: The actual opening feedback signal of the drone's throttle is collected in real time by the angle sensor, and the actual working current signal of the throttle motor is collected in real time by the current sampling circuit.

[0009] Step S3: The target throttle opening value is compared with the actual opening feedback signal to obtain the position deviation signal, and the position deviation signal is adaptively adjusted by the fractional proportional-integral-derivative control algorithm to obtain the target current command.

[0010] Step S4: Construct a radial basis function neural network feedforward compensation model, calculate the feedforward compensation current using the target throttle opening value and the actual opening feedback signal as input, and superimpose the feedforward compensation current onto the target current command to obtain the corrected target current command;

[0011] Step S5: Compare the corrected target current command with the actual working current signal to obtain the current deviation signal, and calculate the pulse width modulation control signal according to the sign of the pre-constructed sliding mode switching function through the sliding mode control algorithm.

[0012] Step S6: Control the throttle motor through the pulse width modulation control signal to adjust the throttle opening.

[0013] Beneficial effects of steps S1 to S6:

[0014] This method establishes a multi-loop closed-loop control architecture consisting of a position loop, a feedforward compensation loop, and a current loop through the coordinated use of fractional proportional-integral-derivative control algorithm, radial basis function neural network feedforward compensation, and sliding mode control algorithm. This enables high-precision, fast-response, and strong disturbance rejection regulation of the UAV engine throttle.

[0015] Specifically, step S1 achieves accurate conversion of flight control signals to target throttle opening through command parsing; step S2 achieves real-time monitoring of throttle opening and current through dual feedback acquisition; step S3 achieves adaptive adjustment of position deviation and generation of target current command through fractional proportional-integral-derivative control algorithm; step S4 achieves prediction and compensation of nonlinear friction characteristics through radial basis function neural network feedforward compensation; step S5 achieves rapid tracking and robust control of current deviation through sliding mode control algorithm; and step S6 achieves precise adjustment of throttle opening through pulse width modulation drive.

[0016] As a further improvement to this application, step S1, receiving a target throttle opening command signal sent by the UAV flight control system, and parsing the target throttle opening command signal to obtain the target throttle opening value, includes:

[0017] Step S1.1: Receive the target opening command signal sent by the UAV flight control system and extract the pulse width parameter of the target opening command signal;

[0018] Step S1.2: Determine whether the pulse width parameter is within a preset valid range. If it is within the preset valid range, proceed to step S1.3; otherwise, keep the throttle opening unchanged.

[0019] Step S1.3: Based on the pulse width parameter and the preset opening degree-pulse width correspondence, map the pulse width parameter to the target throttle opening value.

[0020] Beneficial effects of steps S1.1 to S1.3:

[0021] This series of steps establishes a reliable conversion link from flight control commands to target throttle opening values ​​through a combination of pulse width parameter extraction, validity judgment, and linear mapping, thereby achieving accurate acquisition of target throttle opening values ​​and protection against abnormal signals.

[0022] Specifically, step S1.1 extracts the pulse width parameter to provide raw data for subsequent analysis; step S1.2 eliminates abnormal instructions through validity judgment to ensure the safe operation of the system; and step S1.3 achieves accurate conversion from pulse width to opening value through linear mapping.

[0023] As a further improvement to this application, step S2 involves real-time acquisition of the actual throttle opening feedback signal of the UAV via an angle sensor, and real-time acquisition of the actual operating current signal of the throttle motor via a current sampling circuit, including:

[0024] Step S2.1: Obtain the real-time angle value of the throttle valve through an angle sensor installed on the throttle valve shaft, and convert the real-time angle value into an actual opening feedback signal;

[0025] Step S2.2: Obtain the instantaneous voltage signal across the motor terminals through a current sampling circuit connected in series in the throttle motor drive circuit, and calculate the actual operating current signal based on the instantaneous voltage signal and the preset sampling resistor value;

[0026] Step S2.3: Filter the actual opening feedback signal to obtain a smoothed actual opening feedback signal;

[0027] Step S2.4: Perform a moving average filtering process on the actual operating current signal to obtain a smoothed actual operating current signal.

[0028] Beneficial effects of steps S2.1 to S2.4:

[0029] This series of steps establishes a high-precision real-time monitoring mechanism for throttle opening and motor current through dual feedback acquisition using an angle sensor and a current sampling circuit, combined with filtering technology, achieving low noise and high reliability of the feedback signal.

[0030] Specifically, step S2.1 uses an angle sensor to accurately acquire the actual throttle opening; step S2.2 uses a current sampling circuit to acquire the motor operating current in real time; step S2.3 uses filtering to eliminate high-frequency noise interference in the opening feedback signal; and step S2.4 uses moving average filtering to smooth the current signal.

[0031] As a further improvement of this application, step S3 involves comparing the target throttle opening value with the actual opening feedback signal to obtain a position deviation signal, and then adaptively adjusting the position deviation signal using a fractional-order proportional-integral-derivative control algorithm to obtain a target current command, including:

[0032] Step S3.1: Calculate the difference between the target throttle opening value and the actual opening feedback signal to obtain the position deviation signal;

[0033] Step S3.2: Perform fractional integration on the position deviation signal, wherein the order λ of the fractional integration operation ranges from 0.5 to 1.0.

[0034] Step S3.3: Perform fractional derivative operation on the position deviation signal, wherein the order μ of the fractional derivative operation ranges from 0.1 to 0.5;

[0035] Step S3.4: Adjust the order λ of the fractional integral operation and the order μ of the fractional derivative operation according to the rate of change of the position deviation signal. When the rate of change is greater than a preset rate of change threshold, increase μ and decrease λ. When the rate of change is less than the preset rate of change threshold, decrease μ and increase λ.

[0036] Step S3.5: The target current command is obtained by weighted summation of the proportional coefficient of the position deviation signal, the result of fractional integral operation, and the result of fractional derivative operation.

[0037] Beneficial effects of steps S3.1 to S3.5:

[0038] This series of steps establishes an adaptive adjustment mechanism for the position deviation signal through a fractional-order proportional-integral-derivative control algorithm, thereby achieving accurate generation of the target current command and optimization of dynamic response performance.

[0039] Specifically, step S3.1 obtains the position deviation signal through difference calculation; step S3.2 expands the control degrees of freedom through fractional integral operation; step S3.3 improves the dynamic response capability through fractional derivative operation; step S3.4 adaptively adjusts the integral and derivative orders according to the deviation change rate to achieve parameter self-tuning under different working conditions; and step S3.5 outputs the target current command through weighted summation.

[0040] As a further improvement to this application, step S4 involves constructing a radial basis function neural network feedforward compensation model. The feedforward compensation current is calculated using the target throttle opening value and the actual opening feedback signal as inputs. This feedforward compensation current is then superimposed on the target current command to obtain the corrected target current command, including:

[0041] Step S4.1: Construct a radial basis function neural network model, which includes an input layer, a hidden layer, and an output layer. The input variables of the input layer include the target throttle opening value and the actual opening feedback signal.

[0042] Step S4.2: Define the center vector and width parameter of the radial basis function of the hidden layer, wherein the radial basis function is in Gaussian form;

[0043] Step S4.3: Obtain the Euclidean distance between the input variable and the center vector of the radial basis function, and calculate the output value of the radial basis function of each node in the hidden layer;

[0044] Step S4.4: The output values ​​of each node in the hidden layer are weighted and summed to obtain the feedforward compensation current;

[0045] Step S4.5: Add the feedforward compensation current to the target current command to obtain the corrected target current command.

[0046] Beneficial effects of steps S4.1 to S4.5:

[0047] This series of steps establishes a prediction and compensation mechanism for the nonlinear friction characteristics of the throttle valve through a radial basis function neural network feedforward compensation model, realizing accurate calculation of the feedforward compensation current and correction and enhancement of the target current command.

[0048] Specifically, step S4.1 constructs a radial basis function neural network model to provide a computational framework for feedforward compensation; step S4.2 defines radial basis function parameters to determine network mapping characteristics; step S4.3 calculates the activation output of each node in the hidden layer using Euclidean distance; step S4.4 obtains the feedforward compensation current through weighted summation; and step S4.5 superimposes the feedforward compensation current onto the target current command to eliminate the influence of nonlinear friction on control accuracy.

[0049] As a further improvement of this application, step S5 involves comparing the corrected target current command with the actual operating current signal to obtain a current deviation signal, and calculating a pulse width modulation control signal based on the sign of the pre-constructed sliding mode switching function using a sliding mode control algorithm, including:

[0050] Step S5.1: Construct a sliding mode switching function, wherein the sliding mode switching function is a linear combination of the current deviation signal and the derivative of the current deviation signal;

[0051] Step S5.2: Determine the sign of the sliding mode switching function. When the sliding mode switching function is positive, output the first control quantity; when the sliding mode switching function is negative, output the second control quantity.

[0052] Step S5.3: Calculate the sliding mode control output value based on the first control quantity or the second control quantity using the exponential reaching law;

[0053] Step S5.4: Convert the sliding mode control output value into a pulse width modulation control signal, wherein the duty cycle of the pulse width modulation control signal is proportional to the sliding mode control output value.

[0054] Beneficial effects of steps S5.1 to S5.4:

[0055] This series of steps establishes a robust control mechanism for the current deviation signal through a sliding mode control algorithm, enabling rapid calculation of the pulse width modulation control signal and high-precision output of current tracking.

[0056] Specifically, step S5.1 constructs a sliding mode switching function to provide state variables for control law design; step S5.2 outputs the corresponding control quantity by determining the sign to achieve rapid switching of control direction; step S5.3 calculates the sliding mode control output value by using an exponential approach law to suppress system chattering and improve response speed; and step S5.4 converts the sliding mode control output value into a pulse width modulation control signal to achieve precise driving of the current loop.

[0057] As a further improvement to this application, step S6, controlling the throttle motor through the pulse width modulation control signal to adjust the throttle opening, includes:

[0058] Step S6.1: Determine the conduction time of the H-bridge drive circuit of the throttle motor according to the duty cycle of the pulse width modulation control signal;

[0059] Step S6.2: Determine the rotation direction of the throttle motor according to the polarity of the corrected target current command. When the polarity is positive, control the motor to rotate in the forward direction; when the polarity is negative, control the motor to rotate in the reverse direction.

[0060] Step S6.3: The drive current is output to the throttle motor through the H-bridge drive circuit to drive the throttle valve to perform the mechanism action.

[0061] Beneficial effects of steps S6.1 to S6.3:

[0062] This series of steps establishes a precise drive mechanism for the throttle motor through the coordination of pulse width modulation control signals and H-bridge drive circuits, realizing bidirectional adjustment of throttle opening and reliable operation of the actuator.

[0063] Specifically, step S6.1 determines the conduction time based on the duty cycle to achieve precise control of the motor drive power; step S6.2 determines the rotation direction based on the polarity of the target current command to achieve adjustment of the throttle opening; and step S6.3 outputs the drive current through the H-bridge drive circuit to drive the throttle actuator to complete the opening adjustment action.

[0064] To achieve the above objectives, this application also provides the following technical solutions:

[0065] A throttle adjustment device for a drone engine, the throttle adjustment device being applied to the throttle adjustment method described above, the throttle adjustment device comprising:

[0066] The instruction parsing module is used to receive the target opening instruction signal sent by the UAV flight control system and parse the target opening instruction signal to obtain the target throttle opening value.

[0067] The data acquisition module is used to acquire the actual opening feedback signal of the drone's throttle valve in real time through the angle sensor, and to acquire the actual operating current signal of the throttle valve motor in real time through the current sampling circuit.

[0068] The position control module is used to compare the target throttle opening value with the actual opening feedback signal to obtain a position deviation signal, and to adaptively adjust the position deviation signal through a fractional proportional-integral-derivative control algorithm to obtain a target current command.

[0069] The feedforward compensation module is used to construct a radial basis function neural network feedforward compensation model. It calculates the feedforward compensation current using the target throttle opening value and the actual opening feedback signal as inputs, and superimposes the feedforward compensation current onto the target current command to obtain the corrected target current command.

[0070] The current control module is used to compare the corrected target current command with the actual working current signal to obtain a current deviation signal, and to calculate the pulse width modulation control signal according to the sign of the pre-constructed sliding mode switching function through the sliding mode control algorithm.

[0071] The drive output module is used to control the throttle motor through the pulse width modulation control signal to adjust the throttle opening.

[0072] To achieve the above objectives, this application also provides the following technical solutions:

[0073] An electronic device includes a processor and a memory coupled to the processor, the memory storing program instructions executable by the processor; when the processor executes the program instructions stored in the memory, it implements the throttle adjustment method for a drone engine as described above.

[0074] To achieve the above objectives, this application also provides the following technical solutions:

[0075] A computer-readable storage medium storing program instructions that, when executed by a processor, can implement the throttle adjustment method for a drone engine as described above. Attached Figure Description

[0076] Figure 1 This is a schematic flowchart illustrating the steps of an embodiment of a throttle adjustment method for an unmanned aerial vehicle engine according to this application.

[0077] Figure 2 This is a schematic diagram of the functional modules of one embodiment of a throttle adjustment device for an unmanned aerial vehicle engine according to this application;

[0078] Figure 3 This is a schematic diagram of the structure of an embodiment of the electronic device of this application;

[0079] Figure 4 This is a schematic diagram of the structure of one embodiment of the storage medium of this application;

[0080] Explanation of reference numerals in the attached figures: 1, instruction parsing module; 2, data acquisition module; 3, position control module; 4, feedforward compensation module; 5, current control module; 6, drive output module; 7, electronic equipment; 71, processor; 72, memory; 8, storage medium; 81, program instructions. Detailed Implementation

[0081] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0082] The terms "first," "second," and "third" in this application are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined as "first," "second," or "third" may explicitly or implicitly include at least one of that feature. In the description of this application, "multiple" means at least two, such as two, three, etc., unless otherwise explicitly specified. All directional indications (such as up, down, left, right, front, back, etc.) in the embodiments of this application are only used to explain the relative positional relationships and movements between components in a specific orientation (e.g., as shown in the figures). If the specific orientation changes, the directional indications also change accordingly. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or devices.

[0083] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a mutually exclusive, independent, or alternative embodiment. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0084] It should be noted that, due to the limited types and number of symbols or letters that can represent specific meanings, for embodiments with many formulas or codes, there may be situations where symbols or letters cannot meet the usage requirements. Therefore, the interpretation of formula symbols in the steps or sub-steps of the embodiments is only valid for the current step or sub-step.

[0085] If the same symbol has different interpretations in different steps or sub-steps, the interpretation in the current step or sub-step shall prevail; if the same symbol appears in different steps or sub-steps, but no interpretation is given in subsequent steps or sub-steps after its first appearance, the interpretation in the first step or sub-step shall be used.

[0086] like Figure 1 As shown, a throttle adjustment method for a drone engine includes:

[0087] Step S1: Receive the target throttle opening command signal sent by the UAV flight control system, and parse the target throttle opening command signal to obtain the target throttle opening value.

[0088] Furthermore, step S1 specifically includes the following steps:

[0089] Step S1.1: Receive the target opening command signal sent by the UAV flight control system and extract the pulse width parameter of the target opening command signal.

[0090] Preferably, the core of this step is to receive the modulation signal from the UAV flight control system and extract the pulse width parameter, identify the rising and falling edges by edge detection, and calculate the time interval to obtain the pulse width parameter.

[0091] Regarding the signal format, the UAV flight control system transmits the target throttle opening command signal in the form of a pulse width modulation (PWM) signal. The pulse width parameter of the PWM signal has a linear correspondence with the target throttle opening value. After receiving the target throttle opening command signal, the rising and falling edges of the PWM signal are identified using an edge detection method, and the time interval between the rising and falling edges is calculated to obtain the pulse width parameter.

[0092] For the edge detection method, the rising and falling edges of the signal are detected using external interrupts of the digital I / O port or the timer input capture function. When a rising edge is detected, the current timer count value is recorded; when a falling edge is detected, the count value is recorded again. The difference between the two counts is the pulse width parameter. The timer clock frequency is set to 1 MHz to ensure that the pulse width detection accuracy is better than 0.1 milliseconds.

[0093] Specifically, for invalid signal processing, when the signal pulse width is less than 0.5 milliseconds or greater than 2.5 milliseconds, it is judged as an abnormal signal, and no analysis is performed, keeping the current throttle opening unchanged.

[0094] Preferably, the standard period of the pulse width modulation signal is 20 milliseconds, and the effective range of the pulse width parameter is 1 millisecond to 2 milliseconds, which correspond to 0% to 100% of the throttle opening value, respectively.

[0095] For example, when the UAV flight control system sends a target opening of 50%, the corresponding pulse width is 1.5 milliseconds. The controller detects the rising edge at T1=1000 microseconds and the falling edge at T2=2500 microseconds, and calculates the pulse width parameter T=1500 microseconds=1.5 milliseconds.

[0096] Step S1.2: Determine whether the pulse width parameter is within the preset valid range. If it is within the preset valid range, proceed to step S1.3; otherwise, keep the throttle opening unchanged.

[0097] Preferably, the core of this step is to determine the validity of the extracted pulse width parameters and eliminate abnormal instructions to ensure system security.

[0098] Specifically, the preset valid range is 1 millisecond to 2 milliseconds. When the pulse width parameter is within the preset valid range, the target opening command signal is determined to be valid, and the process proceeds to the next step for analysis. When the pulse width parameter exceeds the preset valid range, the target opening command signal is determined to be abnormal, the current throttle opening remains unchanged, and no further analysis is performed.

[0099] Regarding the anomaly handling strategy, when the pulse width parameter is detected to be outside the valid range, an anomaly flag is set and the anomaly type (too small or too large) is recorded, but a fault alarm is not triggered immediately because short-term signal interference may cause intermittent anomalies. The command signal is only considered to have truly failed after three consecutive anomaly detections.

[0100] Preferably, the threshold parameter is set as follows: lower limit of the effective pulse width interval T min =1 millisecond, upper limit of the effective pulse width interval T max =2 milliseconds, number of consecutive anomaly checks N abnormal =3.

[0101] For example, when a pulse width parameter T = 0.8 milliseconds is detected, which is less than the lower limit of 1 milliseconds, it is determined to be an invalid command, and the current throttle opening remains unchanged. When T = 2.3 milliseconds is detected three times consecutively, it is determined to be a continuous abnormality, and an alarm signal is output.

[0102] Step S1.3: Based on the pulse width parameter and the preset opening degree-pulse width correspondence, map the pulse width parameter to the target throttle opening value.

[0103] Preferably, the core of this step is to linearly map the pulse width parameter to the target throttle opening value, establishing a conversion relationship between the signal and the opening command.

[0104] Specifically, regarding the throttle opening-pulse width mapping relationship, the pulse width parameter is linearly mapped to the target throttle opening value based on the pulse width parameter and the preset correspondence between throttle opening and pulse width. The formula for the throttle opening-pulse width correspondence is as follows:

[0105]

[0106] in, The target throttle opening value (percentage, ranging from 0% to 100%). This represents the minimum throttle opening (0%). This is the maximum throttle opening (100%). This is the pulse width parameter (milliseconds). The minimum effective pulse width is 1 millisecond. The maximum effective pulse width is 2 milliseconds.

[0107] For the reverse mapping formula, the target throttle opening value is calculated from the pulse width parameter:

[0108]

[0109] The meanings of each symbol are the same as in the above formula.

[0110] Preferably, the parameters are set as follows: millisecond, millisecond, , .

[0111] For example, when the pulse width parameter In milliseconds, substitute the values ​​into the formula to calculate: The target opening value is 50%.

[0112] Beneficial effects of steps S1.1 to S1.3:

[0113] This series of steps establishes a reliable conversion link from flight control commands to target throttle opening values ​​through a combination of pulse width parameter extraction, validity judgment, and linear mapping, thereby achieving accurate acquisition of target throttle opening values ​​and protection against abnormal signals.

[0114] Specifically, step S1.1 extracts the pulse width parameter to provide raw data for subsequent analysis; step S1.2 eliminates abnormal instructions through validity judgment to ensure the safe operation of the system; and step S1.3 achieves accurate conversion from pulse width to opening value through linear mapping.

[0115] Step S2: The actual opening feedback signal of the drone's throttle is collected in real time by the angle sensor, and the actual operating current signal of the throttle motor is collected in real time by the current sampling circuit.

[0116] Furthermore, step S2 specifically includes the following steps:

[0117] Step S2.1: Obtain the real-time angle value of the throttle valve through an angle sensor installed on the throttle valve shaft, and convert the real-time angle value into an actual opening feedback signal.

[0118] Preferably, the core of this step is to complete the acquisition of angle sensor data and conversion of opening signal, and to map the mechanical rotation angle into a percentage opening value through the angle-opening ratio relationship.

[0119] For the selection of the angle sensor, a high-resolution magnetic encoder or photoelectric encoder is used, with a resolution of no less than 12 bits, meaning an angle detection accuracy of 0.088 degrees. The sensor is mounted on the throttle body shaft and outputs an angle signal as the shaft rotates. The sensor output interface is digital SPI or I2C, with a sampling frequency of no less than 100 Hz.

[0120] Specifically, for the angle-to-opening conversion relationship, the real-time angle value is converted into an actual opening feedback signal, and the conversion formula is as follows:

[0121]

[0122] in, This is the actual opening feedback signal (percentage, ranging from 0% to 100%). This is the real-time angle value (in degrees) output by the angle sensor, where 90° is the angle value corresponding to the maximum throttle opening.

[0123] Specifically, for sensor fault detection, a sensor fault is determined when the angle sensor feedback value exceeds the range of [0°, 90°]. clamp it to 0°; if It clamps the sensor to 90°; at the same time, it switches to the opening estimation mode based on motor current, uses the empirical correspondence between current and opening to make real-time estimation, and triggers the sensor fault flag.

[0124] Preferably, the parameters are set as follows: Sensor resolution ≥ 12 bits, sensor sampling frequency ≥ 100Hz.

[0125] For example, when the angle sensor provides feedback When, substitute into the formula to calculate: The actual opening feedback signal obtained was 50%.

[0126] Step S2.2: The instantaneous voltage signal at both ends of the motor is obtained by the current sampling circuit connected in series in the throttle motor drive circuit, and the actual working current signal is calculated based on the instantaneous voltage signal and the preset sampling resistor value.

[0127] Preferably, the core of this step is to complete the real-time acquisition of motor current and convert the sampled voltage into a working current signal using Ohm's law.

[0128] The current sampling circuit employs a high-precision sampling resistor with a resistance of 0.01 ohms and a power rating of at least 2 watts, connected in series in the throttle motor drive circuit. The output of the current sampling circuit is amplified and then fed into the controller's ADC channel. The sampling frequency of the current sampling circuit is at least 10 kHz to ensure the ability to capture rapid changes in motor current.

[0129] The current calculation formula is as follows: The actual operating current signal is calculated based on the instantaneous voltage signal and the preset sampling resistor value.

[0130]

[0131] in, This is the actual operating current signal (amperes). This is the instantaneous voltage signal (volts) across the sampling resistor. The value of the sampling resistor is 0.01 ohms.

[0132] Among them, for abnormal current protection, when When the current continuously exceeds 150% of the rated current of 2 amps (i.e., 3 amps), it is determined that the motor is stalled or the drive circuit is short-circuited, triggering overcurrent protection and shutting down the H-bridge output. If the current remains below 10 mA, it is determined to be an open circuit in the motor or a fault in the sampling circuit, triggering a circuit breaker alarm.

[0133] Preferably, the parameters are set as follows: Ohms, sampling frequency ≥10kHz, rated current = 2A, overcurrent protection threshold = 3A, circuit breaker protection threshold = 10mA.

[0134] For example, when the voltage across the sampling resistor volt, When the value is in ohms, substitute it into the formula to calculate: Ampere. The actual operating current of the motor is 12 amperes × 0.01 ohms = 0.12 volts, which conforms to Ohm's law.

[0135] Step S2.3: Filter the actual opening feedback signal to obtain a smoothed actual opening feedback signal.

[0136] Preferably, the core of this step is to perform low-pass filtering on the opening feedback signal and eliminate high-frequency noise interference through a first-order inertial filtering algorithm.

[0137] Regarding the selection of the filtering algorithm, a first-order inertial low-pass filter is applied to the actual opening feedback signal. This algorithm has low computational complexity, good real-time performance, and is suitable for embedded controllers. The core idea of ​​the first-order inertial filter is to use the weighted average of the current sampled value and the historical filtered value as the new filter output, and to attenuate high-frequency components through the inertial effect to achieve smoothing.

[0138] For the first-order inertial filter formula:

[0139]

[0140] in, This represents the filtered actual opening feedback signal (percentage) at time k. This represents the actual opening feedback signal (percentage) at time k. This represents the actual opening feedback signal (percentage) after filtering at time k-1. is the filter coefficient (value range 0 to 1), and k is the index of the discrete time step.

[0141] Among them, the impact on the filter coefficients, The closer the value is to 1, the closer the filter output is to the current sampled value, resulting in a faster response but a worse smoothing effect. The closer the filter output is to 0, the more it relies on historical values, resulting in better smoothing but a slower response. In practical applications, a balance must be struck between response speed and noise suppression.

[0142] Preferably, the parameters are set as follows: Filtering time constant milliseconds, control cycle millisecond, It satisfies the relationship with the time constant .

[0143] For example, when hour:

[0144] If the filter value at the previous moment... Current sample value Then the current filter value High-frequency fluctuations are effectively suppressed.

[0145] Step S2.4: Perform moving average filtering on the actual operating current signal to obtain a smoothed actual operating current signal.

[0146] Preferably, the core of this step is to perform sliding window smoothing of the current signal and reduce current sampling noise by using the arithmetic mean method.

[0147] The moving average filtering algorithm applies a moving average filter to the actual operating current signal. The moving window length is set to N sampling points, and the arithmetic mean of the most recent N current samples is used as the filtered output for the current moment. Moving average filtering effectively suppresses random white noise and improves the signal-to-noise ratio of the current signal.

[0148] For the moving average filter formula:

[0149]

[0150] in, This is the smoothed actual operating current signal (amperes) at time k. Let be the actual operating current signal (amperes) at time ki, N be the sliding window length, and k be the discrete time index.

[0151] Regarding the choice of window length, a larger N results in better filtering but also greater computational latency and slower dynamic response; a smaller N results in faster response but limited smoothing effect. The value of N needs to strike a balance between filtering effect and response speed.

[0152] Preferably, the parameters are set as follows: N=10 (i.e., the arithmetic mean of the most recent 10 current sampling values), current sampling frequency=10kHz, and time window corresponding to 10 sampling points=1 millisecond.

[0153] For example, when the sliding window length N=10, if the most recent 10 current sampling values ​​are [0.82A, 0.78A, 0.85A, 0.81A, 0.79A, 0.83A, 0.80A, 0.84A, 0.77A, 0.81A], then Ampere, to obtain the smoothed actual operating current signal.

[0154] Beneficial effects of steps S2.1 to S2.4:

[0155] This series of steps establishes a high-precision real-time monitoring mechanism for throttle opening and motor current through dual feedback acquisition using an angle sensor and a current sampling circuit, combined with filtering technology, achieving low noise and high reliability of the feedback signal.

[0156] Specifically, step S2.1 uses an angle sensor to accurately acquire the actual throttle opening; step S2.2 uses a current sampling circuit to acquire the motor operating current in real time; step S2.3 uses filtering to eliminate high-frequency noise interference in the opening feedback signal; and step S2.4 uses moving average filtering to smooth the current signal.

[0157] Step S3: The target throttle opening value is compared with the actual opening feedback signal to obtain the position deviation signal. The position deviation signal is then adaptively adjusted using a fractional proportional-integral-derivative control algorithm to obtain the target current command.

[0158] Furthermore, step S3 specifically includes the following steps:

[0159] Step S3.1: Calculate the difference between the target throttle opening value and the actual opening feedback signal to obtain the position deviation signal.

[0160] Preferably, the core of this step is to calculate the deviation between the target value and the feedback value, providing error input for fractional-order PID control.

[0161] In the deviation calculation method, the target throttle opening value is compared with the actual opening feedback signal, and the difference between the two is calculated to obtain the position deviation signal. This deviation signal is the sole input to the subsequent fractional-order PID control algorithm, determining the direction and intensity of the controller's adjustment.

[0162] The formula for calculating the position deviation signal is as follows:

[0163]

[0164] Where e(t) is the position deviation signal (percentage) at time t. The target throttle opening value (percentage) at time t. This represents the actual opening feedback signal (percentage) at time t.

[0165] Among them, the physical meaning of the deviation direction is that when When the target opening is greater than the actual opening, the throttle needs to be opened wider; when When the target opening is less than the actual opening, the throttle needs to be closed slightly; when e(t) = 0, it means that the opening tracking is complete.

[0166] Preferably, the parameters are set as follows: the deviation signal is expressed as a percentage, with a value range of -100% to 100%.

[0167] For example, when the target opening Actual opening hour, The throttle needs to be opened by 45 percentage points.

[0168] Step S3.2: Perform fractional integration on the position deviation signal. The order λ of the fractional integration operation ranges from 0.5 to 1.0.

[0169] Preferably, the core of this step is to perform fractional integral calculation of the position deviation signal. The numerical calculation of the fractional integral is achieved through the GL discretization method, thereby expanding the integral degrees of freedom of the control system.

[0170] For the selection of fractional integrals, the Grünwald-Letnikov (GL) definition is adopted. This definition is based on generalized binomial coefficients and is suitable for discrete real-time computation. Compared with integer integrals, fractional integrals have memory characteristics, which can integrate the entire process information of historical deviation signals and improve the steady-state accuracy of the control system.

[0171] Among them, the discretization formula for the fractional integral GL is:

[0172]

[0173] in, The result of the fractional integral at time k (percentage × milliseconds^λ). The sampling period is 1 millisecond. It represents the fractional order of the integral (ranging from 0.5 to 1.0). Let be the position deviation signal at time kj, where j is the index of the generalized binomial summation, and k is the index of the current discrete time.

[0174] For the generalized binomial coefficients:

[0175]

[0176] Where j! is the factorial of j, when When it is a positive integer, it satisfies , , This is a gamma function.

[0177] Regarding the physical meaning of the order λ, when λ=1, the above equation degenerates into a standard integer-order integral; when λ=0.5, the integral has weak memory characteristics, and the weight of the recent deviation signal is relatively large; when λ→0, the integral effect gradually disappears. The larger λ is, the stronger the integral effect, and the faster the steady-state error is eliminated, but the greater the risk of overshoot.

[0178] Preferably, the parameters are set as follows: The default value is 0.7, and the range is limited to [0.5, 1.0]. If the value is outside the range, it will be clamped to the nearest boundary value. Milliseconds, at the initial time k=0 .

[0179] For example, when , The millisecond and five consecutive time-time deviation signals are all When, the generalized binomial coefficients are respectively , , , , Substituting into the formula and summing them up will give the result. .

[0180] Step S3.3: Perform fractional derivative operation on the position deviation signal. The order μ of the fractional derivative operation ranges from 0.1 to 0.5.

[0181] Preferably, the core of this step is to perform fractional derivative operations on the position deviation signal. Fractional derivatives are achieved through the GL discretization method, which enables advanced prediction of the trend of the deviation signal.

[0182] For the selection of fractional derivatives, the Grünwald-Letnikov (GL) definition is also adopted. Compared with integer derivatives, fractional derivatives have smoothing properties, can suppress high-frequency noise while predicting changing trends, and provide a more detailed description of the changing process of the deviation signal.

[0183] Among them, the discretization formula for the fractional differential GL is:

[0184]

[0185] in, This is the fractional derivative result at time k (percentage in milliseconds^{-μ}). The sampling period is 1 millisecond. The fractional derivative order (ranging from 0.1 to 0.5) is used, with the other symbols remaining the same.

[0186] Among them, the coefficients of the generalized binomial are the same as those of the fractional integral. .

[0187] Regarding the physical meaning of the order μ, when μ=1, the above equation degenerates into a standard integer-order differential (difference); when μ=0.1, the differential action is weak, but the ability to suppress high-frequency noise is strong; when μ=0.5, the differential action is moderate, balancing prediction ability and noise suppression. The larger μ is, the stronger the differential action and the faster the dynamic response, but the more obvious the noise amplification effect.

[0188] Preferably, the parameters are set as follows: The default value is 0.3, and the range is limited to [0.1, 0.5]. If the value is outside the range, it will be clamped to the nearest boundary value. milliseconds, initial moment .

[0189] For example, when , In milliseconds, Generalized binomial coefficients , , , Substitute into the formula to calculate. .

[0190] Step S3.4: Adjust the order λ of the fractional integral operation and the order μ of the fractional derivative operation according to the rate of change of the position deviation signal. When the rate of change is greater than the preset rate of change threshold, increase μ and decrease λ. When the rate of change is less than the preset rate of change threshold, decrease μ and increase λ.

[0191] Preferably, the core of this step is to adaptively adjust the fractional integral and derivative orders, dynamically allocate the weights of the integral and derivative actions according to the rate of change of deviation, and achieve parameter self-tuning under different working conditions.

[0192] The formula for calculating the rate of change is as follows:

[0193]

[0194] in, The rate of change of the position deviation signal (percentage / millisecond). This is the position deviation signal at time k. This is the position deviation signal at time k-1. The sampling period is 1 millisecond.

[0195] Specifically, for the adaptive adjustment rule, when the rate of change of the position deviation signal is greater than the preset rate of change threshold, the system is determined to be in a fast response condition (such as a step command), and the fractional derivative order μ is increased to improve the dynamic response speed, while the fractional integral order λ is decreased to reduce the risk of overshoot; when the rate of change of the position deviation signal is less than the preset rate of change threshold, the system is determined to be in a steady-state tracking condition, and the fractional derivative order μ is decreased to suppress noise sensitivity, while the fractional integral order λ is increased to eliminate steady-state error.

[0196] Specifically, the adjustment strategy employs tiered control rather than continuous adjustment, dividing the rate of change into three zones: a rapid response zone, a transition zone, and a steady-state zone. This is a rapid response zone. Reduce to 0.5. Increase it to 0.5; This is a transitional zone. and Use the default values ​​of 0.7 and 0.3; The steady-state region is defined as ≤1% / ms. Increase to 1.0. Reduce the value to 0.1. Each adjustment step is 0.05, and the adjustment frequency is limited to once every 10 milliseconds to prevent frequent parameter jumps.

[0197] Preferably, the parameters are set as follows: preset change rate threshold = 5% per millisecond, λ adjustment range is 0.5 to 1.0, μ adjustment range is 0.1 to 0.5, adjustment step size = 0.05, and adjustment period = 10 milliseconds.

[0198] For example, when the target opening suddenly jumps from 30% to 75% at time k, e(k) jumps from 0% to 45%. =(45%-0%) / 1ms = 45% / ms > 5% / ms, indicating a fast response condition. The order adjustment is triggered: λ decreases from 0.7 to 0.65 (a decrease of 0.05), and μ increases from 0.3 to 0.35 (an increase of 0.05). After the deviation gradually converges... When reduced to the steady-state region, λ and μ automatically revert to their default values.

[0199] Step S3.5: The target current command is obtained by weighted summation of the proportional coefficient of the position deviation signal, the result of fractional integral operation, and the result of fractional derivative operation.

[0200] Preferably, the core of this step is to complete the output calculation of the fractional proportional-integral-derivative controller, and to sum the weighted proportional term, fractional integral term and fractional derivative term of the deviation signal to generate the target current command for the drive current loop.

[0201] For the fractional-order PID control output formula:

[0202]

[0203] in, The target current command (amperes). This is a proportionality coefficient (amperes per percentage). The integral coefficient is (amperes per percentage milliseconds^{-λ}). The coefficient is the differential coefficient (ampere per percentage millisecond^{μ}), and the other symbols are the same as before.

[0204] The physical meaning of each gain coefficient is as follows: This determines the strength of the system's response to the current deviation. The larger the value, the faster the response, but the greater the overshoot. The strength of the cumulative compensation for historical deviations is determined. The larger the value, the faster the steady-state error is eliminated, but the higher the risk of integral saturation. The strength of the advance adjustment to the rate of change of deviation is determined. The larger the value, the stronger the ability to suppress overshoot, but the more obvious the noise amplification effect.

[0205] Among them, for integral saturation protection, when When the current exceeds the rated current range of the motor ([-2A,+2A]), it is forcibly clamped to the boundary value, and the historical cumulative value of the integral term is appropriately attenuated to prevent integral saturation from causing the system response to be slow.

[0206] Preferably, the parameters are set as follows: Ampere per percentage, Amperes per percentage milliseconds^{-λ} (λ takes the default value of 0.7). Ampere per percentage millisecond^{μ} (μ takes the default value of 0.3), the target current command limit range is [-2A,+2A].

[0207] Beneficial effects of steps S3.1 to S3.5:

[0208] This series of steps establishes an adaptive adjustment mechanism for the position deviation signal through a fractional-order proportional-integral-derivative control algorithm, thereby achieving accurate generation of the target current command and optimization of dynamic response performance.

[0209] Specifically, step S3.1 obtains the position deviation signal through difference calculation; step S3.2 expands the control degrees of freedom through fractional integral operation; step S3.3 improves the dynamic response capability through fractional derivative operation; step S3.4 adaptively adjusts the integral and derivative orders according to the deviation change rate to achieve parameter self-tuning under different working conditions; and step S3.5 outputs the target current command through weighted summation.

[0210] Step S4: Construct a radial basis function neural network feedforward compensation model. Calculate the feedforward compensation current using the target throttle opening value and the actual opening feedback signal as inputs. Then, superimpose the feedforward compensation current onto the target current command to obtain the corrected target current command.

[0211] Furthermore, step S4 specifically includes the following steps:

[0212] Furthermore, step S4 specifically includes the following steps:

[0213] Step S4.1: Construct a radial basis function neural network model. The radial basis function neural network model includes an input layer, a hidden layer, and an output layer. The input variables of the input layer include the target throttle opening value and the actual opening feedback signal.

[0214] Preferably, the core of this step is to complete the structural definition of the radial basis function neural network and establish a nonlinear mapping framework from the opening signal to the feedforward compensation current.

[0215] The radial basis function neural network model comprises three layers: an input layer, a hidden layer, and an output layer. The input layer's input variables include the target throttle opening value and the actual opening feedback signal, forming a two-dimensional input vector. The hidden layer contains M radial basis function neurons, each corresponding to a Gaussian radial basis function. The output layer outputs a feedforward compensation current, forming a single-output structure. The network input vector is defined as:

[0216]

[0217] in, The network input vector (two-dimensional column vector). The target throttle opening value (percentage). This is the actual opening feedback signal (percentage).

[0218] In terms of the physical meaning of the network, the throttle actuator exhibits nonlinear static and dynamic friction, the magnitude of which is closely related to the current opening position and direction of movement. and As input, it can simultaneously reflect the target position and current position information, enabling the network to predict friction compensation requirements under different opening ranges and motion directions.

[0219] Preferably, the parameters are set as follows: input layer dimension = 2, hidden layer node number M is 10 to 20, in this embodiment M = 15, and output layer dimension = 1.

[0220] For example, when , At that time, the network input vector is (Percentage), sent to each node of the hidden layer for radial basis function calculation.

[0221] Step S4.2: Define the center vector and width parameter of the radial basis function of the hidden layer. The radial basis function is in Gaussian form.

[0222] Preferably, the core of this step is to define the Gaussian radial basis function parameters of each node in the hidden layer and determine the nonlinear mapping characteristics of the network.

[0223] Specifically, for the Gaussian radial basis function expression, for the j-th hidden layer node, the radial basis function adopts the Gaussian function form:

[0224]

[0225] in, This is the output value of the radial basis function of the Jth hidden layer node (ranging from 0 to 1). For network input vectors, Let J be the center vector of the J-th hidden layer node (a two-dimensional column vector, and...) (same dimensions) is the width parameter (positive real number) of the J-th hidden layer node. Describing the Euclidean norm, .

[0226] Among them, for the initialization of the center vector, the center vector The components are evenly distributed within the range of [0%, 100%] of the target throttle opening value and the actual opening feedback signal. When M=15, [0%, 100%] is evenly divided into 15 intervals, and the two components of each central vector take the midpoint value of the corresponding interval, forming a uniform distribution covering the entire opening range.

[0227] Among them, the width parameter setting... Controlling the width of the Gaussian function's response curve, The smaller the value, the sharper the basis function and the stronger the local response; The larger the value, the flatter the basis functions and the stronger the global response. The width parameter is taken as half the minimum distance between adjacent center vectors to ensure a moderate overlap between adjacent basis functions.

[0228] Preferably, the parameters are set as follows: M=15, and the components of the center vector are evenly distributed in [0%, 100%]. A uniform 15% is used (the spacing between adjacent center vectors is approximately 7%, and the width parameter is twice the spacing to ensure sufficient coverage).

[0229] For example, when M=15, the center vector of the first node The center vector of the 8th node The center vector of the 15th node Width parameter .

[0230] Step S4.3: Obtain the Euclidean distance between the input variable and the center vector of the radial basis function, and calculate the output value of the radial basis function of each node in the hidden layer.

[0231] Preferably, the core of this step is to complete the activation calculation from the input vector to each hidden layer node. The similarity between the input and each center vector is measured by Euclidean distance, and the closer the distance, the larger the activation value.

[0232] Among them, for the Euclidean distance calculation formula, for the input vector The Euclidean distance to the Jth hidden layer node is calculated using the following formula:

[0233]

[0234] in, For input vectors The center vector corresponding to the Jth hidden layer node The Euclidean distance between them The first component of the input vector is... (Target opening, percentage) The second component of the input vector is... (Actual opening, percentage) Let be the first component of the center vector corresponding to the Jth hidden layer node. It is the second component of the center vector corresponding to the hidden layer node.

[0235] Specifically, the calculation of the radial basis function output value involves using the Euclidean distance. Substituting the Gaussian radial basis function defined in step S4.2, we obtain the output value of the Jth hidden layer node:

[0236]

[0237] The meanings of the symbols are the same as before. When hour, This indicates that the input vector completely coincides with the center of the node, resulting in the maximum activation value; when Much larger hour, This indicates that the node is almost unresponsive to the current input.

[0238] Preferably, the Euclidean distance and radial basis function output values ​​are calculated in parallel for all M nodes to obtain the hidden layer output vector. .

[0239] For example, when the input vector The center vector of the 8th node , hour, , The low activation value of this node indicates that the current input is far from the center.

[0240] Step S4.4: The output values ​​of each node in the hidden layer are weighted and summed to obtain the feedforward compensation current.

[0241] Preferably, the core of this step is to complete the linear mapping from the hidden layer output to the feedforward compensation current, and obtain the final compensation amount by weighted summation of the output values ​​of each node.

[0242] Among them, the formula for calculating the feedforward compensation current is:

[0243]

[0244] in, This is the feedforward compensation current (amperes). Let J be the connection weight (Amperes) from the J-th hidden layer node to the output layer. is the radial basis function output value of the J-th hidden layer node, and M is the number of hidden layer nodes (15).

[0245] Among them, the connection weights are obtained by... The data was obtained through offline training. Offline training employed the gradient descent algorithm, using experimental data from tests on the friction characteristics of the throttle actuator as the training set, and the prediction error of the feedforward compensation current as the loss function, iteratively updating the data. Until convergence. After training is complete, It is permanently stored in the controller's non-volatile memory and can only be read and not written during runtime.

[0246] For the training parameter settings, the learning rate is set to 0.01, the number of training iterations is no less than 1000, and the training data covers the forward and reverse motion conditions within the full throttle opening range (0% to 100%) to ensure that the network has a good fit to the friction characteristics of different motion directions.

[0247] Preferably, the parameters are set as follows: M=15, learning rate=0.01, and training iterations ≥1000. The amplitude limit range is [-0.5A, +0.5A] to prevent over-compensation of feedforward.

[0248] For example, when the radial basis function output value of M=15 nodes The calculations have been completed, and the connection weights are now set. The feedforward compensation current under the current operating condition can be obtained by substituting the values ​​into the formula and summing the results, which has been determined through offline training. .like Ampere indicates that the frictional force in the current opening range requires an additional 0.3 amperes of drive current to compensate.

[0249] Step S4.5: Add the feedforward compensation current to the target current command to obtain the corrected target current command.

[0250] Preferably, the core of this step is to superimpose the feedforward compensation current and the target current command, thereby overcoming the nonlinear frictional resistance of the throttle actuator in advance through feedforward compensation and improving the tracking accuracy of the current loop.

[0251] The corrected target current command calculation formula is as follows:

[0252]

[0253] in, The corrected target current command (Amperes). The target current command (amperes) output in step S3. The feedforward compensation current (Amperes) is calculated in step S4.4.

[0254] In terms of the physical meaning of feedforward compensation, the throttle actuator experiences varying degrees of static and dynamic friction at different opening positions. Relying solely on feedback control (the fractional-order PID in step S3) requires waiting until a deviation occurs before responding, resulting in an inherent response delay. Feedforward compensation, by pre-adding a compensation current, provides the additional driving force needed to overcome friction before a deviation occurs, enabling the current loop control output to overcome static friction more quickly and improving the throttle opening response speed and positioning accuracy.

[0255] Among them, the limiting of the modified target current command, When the current exceeds the rated current range of the motor ([-2A,+2A]), it will be clamped to the boundary value to prevent excessive current from damaging the motor drive circuit.

[0256] Preferably, the parameters are set as follows: The limiting range is [-2A, +2A], and the feedforward compensation current is... The amplitude limit range is [-0.5A, +0.5A].

[0257] For example, when ampere, Ampere, The current output is within the limit range and is directly output to the current loop control in step S5. If... ampere, Ampère, then Ampere, exceeding the +2 ampere limit, clamped to +2 amperes before output.

[0258] Beneficial effects of steps S4.1 to S4.5:

[0259] This series of steps establishes a prediction and compensation mechanism for the nonlinear friction characteristics of the throttle valve through a radial basis function neural network feedforward compensation model, realizing accurate calculation of the feedforward compensation current and correction and enhancement of the target current command.

[0260] Specifically, step S4.1 constructs a radial basis function neural network model to provide a computational framework for feedforward compensation; step S4.2 defines radial basis function parameters to determine network mapping characteristics; step S4.3 calculates the activation output of each node in the hidden layer using Euclidean distance; step S4.4 obtains the feedforward compensation current through weighted summation; and step S4.5 superimposes the feedforward compensation current onto the target current command to eliminate the influence of nonlinear friction on control accuracy.

[0261] Step S5: Compare the corrected target current command with the actual working current signal to obtain the current deviation signal, and calculate the pulse width modulation control signal according to the sign of the pre-constructed sliding mode switching function through the sliding mode control algorithm.

[0262] Furthermore, step S5 specifically includes the following steps:

[0263] Step S5.1: Construct the sliding mode switching function, which is a linear combination of the current deviation signal and the derivative of the current deviation signal.

[0264] Preferably, the core of this step is to calculate the current deviation signal and construct the sliding mode switching function, providing state variables for the subsequent design of the sliding mode control law.

[0265] The formula for calculating the current deviation signal is as follows:

[0266]

[0267] in, This is the current deviation signal (amperes). The corrected target current command (Amperes) is output from step S4.5. The smoothed actual operating current signal (amperes) output from step S2.4.

[0268] Specifically, the sliding mode switching function is constructed by combining the current deviation signal with its derivative, and its expression is:

[0269]

[0270] Where s is the sliding mode switching function, These are the design parameters (positive real numbers) for the sliding surface. This is the current deviation signal (amperes). This is the derivative of the current deviation signal (amperes / milliseconds), which is the rate of change of the current deviation.

[0271] Among them, the discretization calculation of the derivative of the current deviation is as follows:

[0272]

[0273] in, Let be the derivative of the current deviation signal at time k (amperes / milliseconds). Let be the current deviation signal at time k. This represents the current deviation signal at time k-1. The sampling period is 1 millisecond.

[0274] In terms of the physical meaning of the sliding mode switching function, s=0 defines the sliding surface in the state space. When the system state is on the sliding surface, the current deviation converges to zero at an exponential rate. Determine the slope of the sliding surface. The larger the value, the steeper the sliding surface, the faster the state converges, but the more severe the chattering. The smaller the value, the weaker the chattering, but the slower the convergence speed.

[0275] Preferably, the parameters are set as follows: The value is between 100 and 500, in this embodiment. =200, this value can ensure a balance between the convergence speed of the sliding surface and the robustness of the system.

[0276] For example, when ampere, Ampere, Ampere; if the previous moment Ampère, then Amperes / milliseconds; Substitute into the sliding mode switching function: , Proceed to step S5.2 to determine the control direction.

[0277] Step S5.2: Determine the sign of the sliding mode switching function. When the sliding mode switching function is positive, output the first control quantity; when the sliding mode switching function is negative, output the second control quantity.

[0278] Preferably, the core of this step is to determine the sign of the sliding mode switching function, determine the control direction based on the sign, and output the corresponding first control quantity or second control quantity.

[0279] The definitions of the first control variable and the second control variable are as follows:

[0280]

[0281]

[0282] in, As the first control variable, This is the second control variable. To control the output value, This is the minimum control output value.

[0283] With the introduction of symbolic functions, the above piecewise definitions can be uniformly represented by symbolic functions:

[0284]

[0285] in, For a sign function, when The output +1 corresponds to the first control quantity. ,when The output of -1 corresponds to the second control quantity. When s=0, the output 0 indicates that the system has reached the sliding surface.

[0286] Among them, the physical meaning of the upper and lower limits of the control quantity is as follows: The maximum positive drive corresponding to the H-bridge drive circuit, The value range corresponding to the maximum reverse drive is normalized to [-1.0, +1.0], which facilitates the conversion to duty cycle in the subsequent step S5.4.

[0287] Preferably, the parameters are set as follows: , The sign function takes the value 0 when s=0, corresponding to a duty cycle of 50% (motor stops).

[0288] For example, continuing from the previous example s=119.8>0, the first control quantity is output. This corresponds to the maximum positive drive, which drives the throttle valve to move in the direction of wider opening.

[0289] Step S5.3: Calculate the sliding mode control output value based on the first control quantity or the second control quantity using the exponential approach law.

[0290] Preferably, the core of this step is to calculate the sliding mode control output value, and suppress system chattering by using the exponential approach law to ensure rapid approach to the sliding surface.

[0291] Among them, the expression for the exponential reaching law is:

[0292]

[0293] in, The derivative of the sliding mode switching function. To approximate the velocity coefficient (a positive real number). It is the exponential convergence coefficient (a positive real number). is the sign function, and s is the sliding mode switching function.

[0294] Among them, the physical meaning of the exponential reaching law, The term is a constant-rate approaching term, ensuring that the system moves at a constant rate. Approaching the sliding surface, The larger the value, the faster it approaches the target, but the stronger the chattering becomes. The term is an exponential approaching term, which enables the system to approach rapidly when far from the sliding surface and decelerate when approaching it, effectively suppressing chattering. The two terms work synergistically to balance approach speed and chattering suppression.

[0295] The derivation of the sliding mode control output value is achieved by combining the exponential reaching law with the system dynamic equations. The resulting sliding mode control output value is:

[0296]

[0297] in, is the sliding mode control output value (normalized to [-1.0, +1.0]), and b is the system control gain parameter (Amperes / Volts, determined by motor parameters). To approximate the velocity coefficient, Here, is the exponential reaching coefficient, and s is the sliding mode switching function. It is a symbolic function.

[0298] Among them, for The amplitude limiting processing, calculated When the output exceeds the range of [-1.0, +1.0], it is forcibly clamped to the boundary value to prevent excessive control output from damaging the drive circuit.

[0299] Preferably, the parameters are set as follows: The value is between 50 and 200, in this embodiment. ; The value is between 100 and 500, in this embodiment. b = 0.5 amperes / volt (determined based on the motor winding resistance of 6 ohms and the bus voltage of 12 volts); The amplitude limit range is [-1.0, +1.0].

[0300] Step S5.4: Convert the sliding mode control output value into a pulse width modulation control signal. The duty cycle of the pulse width modulation control signal is proportional to the sliding mode control output value.

[0301] Preferably, the core of this step is to complete the linear conversion from the sliding mode control output value to the duty cycle, thereby generating the pulse width modulation control signal to drive the H-bridge circuit.

[0302] Among them, the duty cycle conversion formula is as follows:

[0303] %

[0304] Where D is the duty cycle (percentage, ranging from 0% to 100%) of the pulse width modulation control signal. The output value for sliding mode control (normalized to [-1.0, +1.0]). To control the lower limit of the output (-1.0). To control the upper limit of output (+1.0).

[0305] Regarding the physical meaning of duty cycle, D=50% corresponds to... That is, the motor stops; D=100% corresponds to That is, maximum positive drive; D=0% corresponds to This refers to maximum reverse drive. The duty cycle controls the on-time of the power switching transistors via signals from the H-bridge drive circuit, thereby controlling the average voltage across the motor.

[0306] Regarding the selection of carrier frequency, a higher carrier frequency results in lower motor current ripple and smoother drive, but also greater switching losses; conversely, a lower carrier frequency results in lower switching losses, but also greater current ripple. Considering both motor response bandwidth and drive efficiency, a carrier frequency of 10 kHz to 20 kHz is chosen.

[0307] Preferably, the parameters are set as follows: , The carrier frequency is 15 kHz (corresponding to a carrier period of approximately 66.7 microseconds), and the duty cycle resolution is not less than 0.1%.

[0308] For example, when (When clamped, maximum positive drive) With the H-bridge drive circuit fully activated, the motor rotates forward with maximum driving force, rapidly opening the throttle. When When D=50%, the motor stops.

[0309] Beneficial effects of steps S5.1 to S5.4:

[0310] This series of steps establishes a robust control mechanism for the current deviation signal through a sliding mode control algorithm, enabling rapid calculation of the pulse width modulation control signal and high-precision output of current tracking.

[0311] Specifically, step S5.1 constructs a sliding mode switching function to provide state variables for control law design; step S5.2 outputs the corresponding control quantity by determining the sign to achieve rapid switching of control direction; step S5.3 calculates the sliding mode control output value by using an exponential approach law to suppress system chattering and improve response speed; and step S5.4 converts the sliding mode control output value into a pulse width modulation control signal to achieve precise driving of the current loop.

[0312] Step S6: The throttle motor is controlled by the pulse width modulation control signal to adjust the throttle opening.

[0313] Furthermore, step S6 specifically includes the following steps:

[0314] Step S6.1: Determine the conduction time of the H-bridge drive circuit of the throttle motor based on the duty cycle of the pulse width modulation control signal.

[0315] Preferably, the core of this step is to convert the duty cycle into the conduction time of the H-bridge drive circuit and determine the conduction duration of the power switch in each carrier cycle.

[0316] The formula for calculating the conduction time is as follows:

[0317]

[0318] in, D is the conduction time (microseconds) of the H-bridge drive circuit, and D is the duty cycle (percentage, ranging from 0% to 100%) of the pulse width modulation control signal output in step S5.4. The carrier period (microseconds) is the carrier cycle. , For carrier frequency.

[0319] The physical meaning of the conduction time is as follows: in each carrier cycle... Within the H-bridge drive circuit, the power switch conduction time is... The shutdown time is The longer the conduction time, the higher the average voltage across the motor, the greater the drive current, and the faster the throttle valve moves; the shorter the conduction time, the lower the average voltage, the smaller the drive current, and the slower the throttle valve moves.

[0320] Regarding the dead time setting, to prevent a power short circuit caused by simultaneous conduction of the upper and lower bridge arms of the H-bridge, a dead time needs to be inserted between the turn-off of the upper bridge arm and the turn-on of the lower bridge arm. The dead time is determined based on the turn-off delay time of the power switching transistors, typically ranging from 100 nanoseconds to 500 nanoseconds. During the dead time, both bridge arms are turned off, and the motor current freewheels through the freewheeling diode.

[0321] Preferably, the parameters are set as follows: carrier frequency kilohertz, corresponding to the carrier period Microseconds; dead time is 200 nanoseconds; duty cycle resolution is not less than 0.1%, corresponding to a conduction time resolution of approximately 0.067 microseconds.

[0322] For example, when step S5.4 outputs a duty cycle D of 75%, The H-bridge drive circuit is on for 50.0 microseconds and off for 16.7 microseconds in each carrier cycle, and the average voltage across the motor is... volt.

[0323] Step S6.2: Determine the rotation direction of the throttle motor according to the polarity of the corrected target current command. When the polarity is positive, control the motor to rotate in the forward direction. When the polarity is negative, control the motor to rotate in the reverse direction.

[0324] Preferably, the core of this step is to determine the direction of motor rotation and configure the state of the H-bridge switch, thereby achieving bidirectional adjustment of the throttle valve through different switch combinations.

[0325] Among them, the rotation direction determination rule is based on the modified target current command. The polarity determines the direction of rotation of the throttle motor: when When the polarity is positive, it controls the motor to rotate in the forward direction to increase the throttle opening; when When the polarity is negative, the motor is controlled to rotate in the opposite direction to reduce the throttle opening; when When the current is less than 5% of the rated current (i.e., 0.1 amperes), it is determined to be a zero current command, triggering the braking stop state.

[0326] Specifically, for the H-bridge switching state configuration, the H-bridge drive circuit includes four power switches Q1, Q2, Q3, and Q4, and the switching states corresponding to different rotation directions are as follows:

[0327] Forward rotation (opening degree increases): Q1 and Q4 are on, Q2 and Q3 are off, and the current flows from the positive terminal of the motor to the negative terminal; Reverse rotation (opening degree decreases): Q2 and Q3 are on, Q1 and Q4 are off, and the current flows from the negative terminal of the motor to the positive terminal; Braking and stopping: Q1 and Q3 are on, Q2 and Q4 are off, and the two ends of the motor are short-circuited to achieve rapid braking.

[0328] Specifically, regarding the direction switching timing, when the rotation direction changes, all currently active switches must be turned off first, and after waiting for the dead time (200 nanoseconds), the switch combination corresponding to the target direction must be turned on to prevent the upper and lower bridge arms from being directly connected and causing a power short circuit.

[0329] Preferably, the parameters are set as follows: zero current threshold = 0.1 amperes (5% of the rated current of 2 amperes), direction switching dead time = 200 nanoseconds, the switching transistor is an N-channel enhancement-mode metal-oxide-semiconductor field-effect transistor, and the on-resistance is not greater than 10 milliohms.

[0330] For example, when When the ampere is applied, the polarity is positive, Q1 and Q4 are turned on, and Q2 and Q3 are turned off, the motor rotates in the forward direction, and the throttle valve moves in the direction of widening; when When the ampere is applied, the polarity is negative, and Q2 and Q3 are turned on while Q1 and Q4 are turned off. The motor rotates in the opposite direction, and the throttle moves in the direction of closing.

[0331] Step S6.3: The drive current is output to the throttle motor through the H-bridge drive circuit to drive the throttle valve to perform the mechanism action.

[0332] Preferably, the core of this step is to calculate the current output of the H-bridge drive circuit, convert the duty cycle and bus voltage into actual drive current, and drive the throttle actuator to complete the opening adjustment action.

[0333] The formula for calculating the motor drive current is as follows:

[0334]

[0335] in, This represents the motor drive current (amperes). is the H-bridge bus voltage (volts), and D is the duty cycle of the pulse width modulation control signal (percentage, in decimal form, ranging from 0 to 1). The resistance of the motor windings is in ohms.

[0336] Regarding the physical meaning of the drive current, when the H-bridge drive circuit is in the on state, the average voltage across the motor is: According to Ohm's law, the steady-state current flowing through the motor windings is... The actual drive current has a rise time due to the inductance of the motor windings; the rise time constant is... ,in This is the inductance of the motor windings.

[0337] Among them, for the mechanical action of the motor, the drive current Electromagnetic torque is generated by flowing through the motor windings. This electromagnetic torque overcomes the friction and spring return force of the throttle actuator, driving the throttle shaft to rotate and changing the throttle opening. The change in throttle opening is fed back in real time by the angle sensor in step S2.1, forming a complete closed-loop control circuit.

[0338] Among them, for overcurrent protection, when When the current exceeds 150% of the rated current of 2 amps (i.e., 3 amps), all H-bridge switches are immediately turned off, triggering overcurrent protection. The system restarts after waiting for 10 milliseconds to prevent the motor windings from being damaged by overheating due to overcurrent.

[0339] Preferably, the parameters are set as follows: volt, ohm, millihenry, current rise time constant Milliseconds, rated current = 2 amps, overcurrent protection threshold = 3 amps. The throttle motor uses a coreless servo motor with a rated voltage of 12 volts and a rated speed of 15,000 revolutions per minute. Under rated voltage conditions, the throttle response time for the entire stroke does not exceed 200 milliseconds.

[0340] For example, when Volts, D=75%=0.75 Ohm time, Amperes, not exceeding the rated current of 2 Ampers, the motor operates normally; when D=100%, Amperes, just reaching the upper limit of the rated current.

[0341] Beneficial effects of steps S6.1 to S6.3:

[0342] This series of steps establishes a precise drive mechanism for the throttle motor through the coordination of pulse width modulation control signals and H-bridge drive circuits, realizing bidirectional adjustment of throttle opening and reliable operation of the actuator.

[0343] Specifically, step S6.1 determines the conduction time based on the duty cycle to achieve precise control of the motor drive power; step S6.2 determines the rotation direction based on the polarity of the target current command to achieve adjustment of the throttle opening; and step S6.3 outputs the drive current through the H-bridge drive circuit to drive the throttle actuator to complete the opening adjustment action.

[0344] This method uses the target opening command from the UAV flight control system as input and the actual throttle motor opening as output, forming a multi-loop coordinated closed-loop control system. The entire control process executes steps S1 to S6 sequentially within each control cycle, with each step tightly connected via signal flow to form a complete adjustment link. The typical duration of the control cycle is 1 to 5 milliseconds, depending on the processor's computing power and control bandwidth requirements.

[0345] In summary, the overall beneficial effects of steps S1 to S6 in this embodiment are as follows:

[0346] This method establishes a multi-loop closed-loop control architecture consisting of a position loop, a feedforward compensation loop, and a current loop through the coordinated use of fractional proportional-integral-derivative control algorithm, radial basis function neural network feedforward compensation, and sliding mode control algorithm. This enables high-precision, fast-response, and strong disturbance rejection regulation of the UAV engine throttle.

[0347] Specifically, step S1 achieves accurate conversion of flight control signals to target throttle opening through command parsing; step S2 achieves real-time monitoring of throttle opening and current through dual feedback acquisition; step S3 achieves adaptive adjustment of position deviation and generation of target current command through fractional proportional-integral-derivative control algorithm; step S4 achieves prediction and compensation of nonlinear friction characteristics through radial basis function neural network feedforward compensation; step S5 achieves rapid tracking and robust control of current deviation through sliding mode control algorithm; and step S6 achieves precise adjustment of throttle opening through pulse width modulation drive.

[0348] like Figure 2 As shown, this embodiment provides an example of a throttle adjustment device for a drone engine. In this embodiment, the throttle adjustment device is applied to the throttle adjustment method as described in the above embodiment.

[0349] Specifically, the throttle adjustment device includes, in sequence, an instruction parsing module 1, a data acquisition module 2, a position control module 3, a feedforward compensation module 4, a current control module 5, and a drive output module 6, which are electrically or communicatively connected.

[0350] The system comprises the following modules: Module 1 receives the target throttle opening command signal from the UAV flight control system and parses it to obtain the target throttle opening value; Module 2 acquires the actual throttle opening feedback signal from the UAV using an angle sensor and the actual operating current signal from the throttle motor using a current sampling circuit; Module 3 compares the target throttle opening value with the actual opening feedback signal to obtain a position deviation signal, and adaptively adjusts the position deviation signal using a fractional proportional-integral-derivative control algorithm to obtain the target current command; Module 4 constructs a radial basis function neural network feedforward compensation model, calculates the feedforward compensation current using the target throttle opening value and the actual opening feedback signal as input, and superimposes the feedforward compensation current onto the target current command to obtain the corrected target current command; Module 5 compares the corrected target current command with the actual operating current signal to obtain a current deviation signal, and calculates the pulse width modulation control signal using a sliding mode control algorithm based on the sign of a pre-constructed sliding mode switching function; and Module 6 controls the throttle motor using the pulse width modulation control signal to adjust the throttle opening.

[0351] like Figure 3 As shown, the electronic device 7 includes a processor 71 and a memory 72 coupled to the processor 71.

[0352] The memory 72 stores program instructions for implementing the throttle adjustment method of the UAV engine in any of the above embodiments.

[0353] The processor 71 is used to execute program instructions stored in the memory 72 to implement throttle adjustment of the UAV engine.

[0354] The processor 71 can also be referred to as a CPU (Central Processing Unit). The processor 71 may be an integrated circuit chip with signal processing capabilities. The processor 71 can also be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. A general-purpose processor can be a microprocessor or any conventional processor.

[0355] Furthermore, Figure 4 This is a schematic diagram of the structure of a storage medium according to an embodiment of this application. See also: Figure 4The storage medium 8 in this embodiment stores program instructions 81 capable of implementing all the above methods. These program instructions 81 can be stored in the storage medium as a software product, including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute all or part of the steps of the methods in each embodiment of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks, or terminal devices such as computers, servers, mobile phones, and tablets.

[0356] In the several embodiments provided in this application, it should be understood that the disclosed systems, methods, and approaches can be implemented in other ways. For example, the system embodiments described above are merely illustrative. For instance, the throttle adjustment of a unit is only a logical function of throttle adjustment; in actual implementation, there may be other throttle adjustment methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or omitted. Furthermore, the mutual coupling or direct coupling or communication connection shown or discussed may be through some interface; the indirect coupling or communication connection of the system or unit may be electrical, mechanical, signal, or other forms.

[0357] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated units described above can be implemented in hardware or as software functional units. The above are merely embodiments of this application and do not limit the patent scope of this application. Any equivalent structural or procedural transformations made based on the description and drawings of this application, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of this application.

Claims

1. A method for adjusting the throttle valve of an unmanned aerial vehicle (UAV) engine, characterized in that, The throttle adjustment method includes: Step S1: Receive the target opening command signal sent by the UAV flight control system, and parse the target opening command signal to obtain the target throttle opening value; Step S2: The actual opening feedback signal of the drone's throttle is collected in real time by the angle sensor, and the actual working current signal of the throttle motor is collected in real time by the current sampling circuit. Step S3: The target throttle opening value is compared with the actual opening feedback signal to obtain the position deviation signal, and the position deviation signal is adaptively adjusted by the fractional proportional-integral-derivative control algorithm to obtain the target current command. Step S4: Construct a radial basis function neural network feedforward compensation model, calculate the feedforward compensation current using the target throttle opening value and the actual opening feedback signal as input, and superimpose the feedforward compensation current onto the target current command to obtain the corrected target current command; Step S5: Compare the corrected target current command with the actual working current signal to obtain the current deviation signal, and calculate the pulse width modulation control signal according to the sign of the pre-constructed sliding mode switching function through the sliding mode control algorithm. Step S6: Control the throttle motor through the pulse width modulation control signal to adjust the throttle opening.

2. The throttle adjustment method according to claim 1, characterized in that, Step S1: Receive the target throttle opening command signal sent by the UAV flight control system, and parse the target throttle opening command signal to obtain the target throttle opening value, including: Step S1.1: Receive the target opening command signal sent by the UAV flight control system and extract the pulse width parameter of the target opening command signal; Step S1.2: Determine whether the pulse width parameter is within a preset valid range. If it is within the preset valid range, proceed to step S1.3; otherwise, keep the throttle opening unchanged. Step S1.3: Based on the pulse width parameter and the preset opening degree-pulse width correspondence, map the pulse width parameter to the target throttle opening value.

3. The throttle adjustment method according to claim 1, characterized in that, Step S2 involves acquiring the actual throttle opening feedback signal of the UAV in real time using an angle sensor, and acquiring the actual operating current signal of the throttle motor in real time using a current sampling circuit, including: Step S2.1: Obtain the real-time angle value of the throttle valve through an angle sensor installed on the throttle valve shaft, and convert the real-time angle value into an actual opening feedback signal; Step S2.2: Obtain the instantaneous voltage signal across the motor terminals through a current sampling circuit connected in series in the throttle motor drive circuit, and calculate the actual operating current signal based on the instantaneous voltage signal and the preset sampling resistor value; Step S2.3: Filter the actual opening feedback signal to obtain a smoothed actual opening feedback signal; Step S2.4: Perform a moving average filtering process on the actual operating current signal to obtain a smoothed actual operating current signal.

4. The throttle adjustment method according to claim 1, characterized in that, Step S3: Compare the target throttle opening value with the actual opening feedback signal to obtain a position deviation signal, and adaptively adjust the position deviation signal using a fractional-order proportional-integral-derivative control algorithm to obtain a target current command, including: Step S3.1: Calculate the difference between the target throttle opening value and the actual opening feedback signal to obtain the position deviation signal; Step S3.2: Perform fractional integration on the position deviation signal, wherein the order λ of the fractional integration operation ranges from 0.5 to 1.

0. Step S3.3: Perform fractional derivative operation on the position deviation signal, wherein the order μ of the fractional derivative operation ranges from 0.1 to 0.5; Step S3.4: Adjust the order λ of the fractional integral operation and the order μ of the fractional derivative operation according to the rate of change of the position deviation signal. When the rate of change is greater than a preset rate of change threshold, increase μ and decrease λ. When the rate of change is less than the preset rate of change threshold, decrease μ and increase λ. Step S3.5: The target current command is obtained by weighted summation of the proportional coefficient of the position deviation signal, the result of fractional integral operation, and the result of fractional derivative operation.

5. The throttle adjustment method according to claim 1, characterized in that, Step S4: Construct a radial basis function neural network feedforward compensation model. Using the target throttle opening value and the actual opening feedback signal as inputs, calculate the feedforward compensation current. Then, superimpose the feedforward compensation current onto the target current command to obtain the corrected target current command, including: Step S4.1: Construct a radial basis function neural network model, which includes an input layer, a hidden layer, and an output layer. The input variables of the input layer include the target throttle opening value and the actual opening feedback signal. Step S4.2: Define the center vector and width parameter of the radial basis function of the hidden layer, wherein the radial basis function is in Gaussian form; Step S4.3: Obtain the Euclidean distance between the input variable and the center vector of the radial basis function, and calculate the output value of the radial basis function of each node in the hidden layer; Step S4.4: The output values ​​of each node in the hidden layer are weighted and summed to obtain the feedforward compensation current; Step S4.5: Add the feedforward compensation current to the target current command to obtain the corrected target current command.

6. The throttle adjustment method according to claim 1, characterized in that, Step S5 involves comparing the corrected target current command with the actual operating current signal to obtain a current deviation signal, and calculating a pulse width modulation control signal based on the sign of the pre-constructed sliding mode switching function using a sliding mode control algorithm, including: Step S5.1: Construct a sliding mode switching function, wherein the sliding mode switching function is a linear combination of the current deviation signal and the derivative of the current deviation signal; Step S5.2: Determine the sign of the sliding mode switching function. When the sliding mode switching function is positive, output the first control quantity; when the sliding mode switching function is negative, output the second control quantity. Step S5.3: Calculate the sliding mode control output value based on the first control quantity or the second control quantity using the exponential reaching law; Step S5.4: Convert the sliding mode control output value into a pulse width modulation control signal, wherein the duty cycle of the pulse width modulation control signal is proportional to the sliding mode control output value.

7. The throttle adjustment method according to claim 1, characterized in that, Step S6, controlling the throttle motor through the pulse width modulation control signal to adjust the throttle opening, includes: Step S6.1: Determine the conduction time of the H-bridge drive circuit of the throttle motor according to the duty cycle of the pulse width modulation control signal; Step S6.2: Determine the rotation direction of the throttle motor according to the polarity of the corrected target current command. When the polarity is positive, control the motor to rotate in the forward direction; when the polarity is negative, control the motor to rotate in the reverse direction. Step S6.3: The drive current is output to the throttle motor through the H-bridge drive circuit to drive the throttle valve to perform the mechanism action.

8. A throttle adjustment device for an unmanned aerial vehicle (UAV) engine, wherein the throttle adjustment device is applied to the throttle adjustment method as described in any one of claims 1 to 7, characterized in that, The throttle adjustment device includes: The instruction parsing module is used to receive the target opening instruction signal sent by the UAV flight control system and parse the target opening instruction signal to obtain the target throttle opening value. The data acquisition module is used to acquire the actual opening feedback signal of the drone's throttle valve in real time through the angle sensor, and to acquire the actual operating current signal of the throttle valve motor in real time through the current sampling circuit. The position control module is used to compare the target throttle opening value with the actual opening feedback signal to obtain a position deviation signal, and to adaptively adjust the position deviation signal through a fractional proportional-integral-derivative control algorithm to obtain a target current command. The feedforward compensation module is used to construct a radial basis function neural network feedforward compensation model. It calculates the feedforward compensation current using the target throttle opening value and the actual opening feedback signal as inputs, and superimposes the feedforward compensation current onto the target current command to obtain the corrected target current command. The current control module is used to compare the corrected target current command with the actual working current signal to obtain a current deviation signal, and to calculate the pulse width modulation control signal according to the sign of the pre-constructed sliding mode switching function through the sliding mode control algorithm. The drive output module is used to control the throttle motor through the pulse width modulation control signal to adjust the throttle opening.

9. An electronic device, characterized in that, The method includes a processor and a memory coupled to the processor, the memory storing program instructions executable by the processor; when the processor executes the program instructions stored in the memory, it implements the throttle adjustment method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores program instructions that, when executed by a processor, enable the throttle adjustment method as described in any one of claims 1 to 7.

Images