A cross-medium driving blade spmsm position sensorless control method

Abstract

The application discloses a kind of SPMSM position sensorless control methods of cross-media driven paddle, comprising the following steps: step S1: obtaining the stator current information and voltage information of SPMSM;Step S2: using the position observer that is superimposed with virtual injection signal is carried out rotor position estimation, obtains estimated rotor position;Step S3: through speed observer is carried out speed estimation, obtains estimated speed;Step S4: through load torque observer is carried out load torque estimation, obtains estimated load torque;Step S5: identify current medium state;Step S6: torque instruction is carried out feedforward compensation, and according to current medium state adjustment control parameter.The application can realize the cooperative control of rotor position and speed estimation, load torque observation, medium state identification and torque feedforward compensation under the condition of no encoder, and improve position observation robustness under low-speed heavy load and medium mutation working condition, to effectively inhibit the speed fluctuation and current impact in cross-media transition process.

Description

Technical Field

[0001] This invention relates to the field of motor drive control technology for cross-medium unmanned platforms, and specifically to a sensorless control method for cross-medium driven propellers using SPMSM. Background Technology

[0002] A water-air cross-medium unmanned platform is a novel type of unmanned system capable of autonomously alternating between air and water. It typically requires a single propulsion system that can handle both high-speed, light-load operation in air and low-speed, heavy-load operation in water. When using a single motor to directly drive the propeller blades, the fluid resistance, load torque, and thrust characteristics experienced by the blades differ significantly between air and water. The load torque rises sharply upon entering the water and drops rapidly upon exiting, easily leading to motor speed drops, overshoot, current surges, and speed loop integral saturation. This severely impacts the operational stability and safety during the cross-medium transition process.

[0003] Currently, most existing motor control schemes use sensor encoders to obtain stable speeds and then estimate blade loads through load torque observers, or use fixed-parameter speed PI controllers or sliding mode controllers for disturbance rejection control. However, the aforementioned existing technologies still have the following shortcomings: 1. The addition of encoders, torque sensors, or medium sensors increases the system weight, volume, cost, and complexity of waterproof design, which is not conducive to the integration and deployment of miniaturized water-air cross-medium unmanned platforms; 2. Ordinary sensorless control has weak back electromotive force under low-speed, zero-speed, and heavy-load conditions in water, resulting in insufficient position observability and problems such as angle drift, low-speed step loss, and difficulty in starting, making it difficult to meet the requirements of cross-medium full-condition operation; 3. Existing load observers usually assume reliable speed feedback. If the speed is directly estimated without sensors, the observation noise and angle error will be transmitted to the load torque estimation result, reducing the accuracy of load observation; 4. Traditional load torque feedforward methods are mostly used only to compensate for load disturbances and are not designed in conjunction with the observability enhancement of sensorless position observers, lacking overall optimization at the system level; 5. Existing medium judgments mostly rely on external sensors or offline calibration parameters, making it difficult to identify the transition states of air, water, and water-air in real time without increasing hardware costs, thus limiting the improvement of cross-medium adaptive control capabilities.

[0004] Therefore, there is an urgent need to propose a control method that can simultaneously achieve rotor position and speed estimation, cross-medium load torque observation, medium state identification, and torque feedforward compensation without encoders, and improve the robustness of position observation under low-speed heavy load and medium change conditions, so as to effectively solve the many problems existing in the above-mentioned prior art. Summary of the Invention

[0005] To overcome the shortcomings of the prior art, the present invention aims to provide a sensorless SPMSM control method for cross-medium driven blades, which can achieve coordinated control of rotor position and speed estimation, load torque observation, medium state identification and torque feedforward compensation under encoder-free conditions, and improve the robustness of position observation under low-speed heavy load and medium change conditions, so as to effectively suppress speed fluctuations and current surges during cross-medium transition.

[0006] To achieve the objective of this invention, the following solution is adopted: A sensorless control method for SPMSM (Split-Media Driven Blade) propellers across media includes the following steps: Step S1: Obtain stator current and voltage information of SPMSM; Step S2: Based on the stator current information and voltage information, the rotor position is estimated using a position observer superimposed with a virtual injection signal to obtain the estimated rotor position; wherein, the virtual injection signal only acts inside the position observer and is not applied to the actual terminal voltage of the motor. Step S3: Based on the estimated rotor position and the voltage information, estimate the speed using a speed observer to obtain the estimated speed; Step S4: Based on the estimated speed and the stator current information, the load torque is estimated using a load torque observer to obtain the estimated load torque; Step S5: Identify the current medium state based on the estimated load torque and the estimated speed; Step S6: Perform feedforward compensation on the torque command based on the estimated load torque, and adjust the control parameters according to the current medium state.

[0007] Furthermore, the position observer is an active flux linkage estimator, and the virtual injection signal is superimposed on the voltage signal input to the active flux linkage estimator; The virtual injection signal includes a bias component and an alternating component. The bias component is used to provide continuous excitation under low-speed conditions, and the alternating component is used to generate flux disturbances to enhance position observability.

[0008] Furthermore, the intensity of the virtual injection signal is adaptively adjusted according to the motor operating state, which includes at least one of the following: motor speed, load change, flux linkage estimation error, and flux linkage estimation reliability. When the motor is at low speed, under heavy load, experiencing sudden load changes, or when the flux linkage reliability decreases, the strength of the virtual injection signal is increased; when the motor is running stably, the strength of the virtual injection signal is decreased.

[0009] Furthermore, the speed observer is an observer constructed based on the mechanical motion equation and electromagnetic torque equation of the motor. Its inputs include the estimated rotor position, the voltage information, and the estimated load torque, and its output is the estimated speed. The estimated velocity is then low-pass filtered to output the filtered velocity.

[0010] Furthermore, the load torque observer is a sliding mode observer, and its speed observation error is composed of the difference between the estimated speed and the speed estimate inside the load torque observer. Its gain adapts to the magnitude of the speed observation error. When sudden load changes cause an increase in speed observation error, the gain is increased to improve load torque tracking speed; when the system tends to stabilize, the gain is decreased to suppress chattering.

[0011] Furthermore, in step S5, the current medium state includes an air medium state, a water medium state, and a transitional state between air and water; The method for identifying the current medium state is as follows: A discriminant is constructed that is associated with the ratio of the estimated load torque to the square of the speed, and the current medium state is determined by comparing the discriminant with a preset threshold; the preset threshold includes at least an upper threshold and a lower threshold, and stable discrimination of the medium state is achieved through hysteresis comparison.

[0012] Furthermore, a continuously varying medium factor is constructed based on the discriminant, which varies continuously between air and water media; The medium factor, after filtering, is used to smoothly schedule the speed loop control parameters, which include at least one of proportional gain and integral gain.

[0013] Further, in step S6, the feedforward compensation is to convert the estimated load torque into a q-axis current feedforward quantity, which is then superimposed on the q-axis current command output by the speed loop controller. The q-axis current feedforward increases as the estimated load torque increases and decreases as the estimated load torque decreases.

[0014] Furthermore, the position observer estimates the rotor flux linkage by combining a voltage model with amplitude correction, wherein the amplitude correction is used to constrain the estimated flux linkage amplitude to near the known permanent magnet flux linkage amplitude.

[0015] Furthermore, the method forms a closed-loop cooperative control: The estimated load torque is also used for identifying the current medium state, for feedforward compensation of the torque command, and for intensity adjustment of the virtual injection signal; When the blades transition between different media, the three components work in concert: When the estimated load torque increases, the virtual injection signal is enhanced, the feedforward compensation is increased, and the speed loop control parameters are switched to an underwater adaptation mode; when the estimated load torque decreases, the virtual injection signal is weakened, the feedforward compensation is reduced, and the speed loop control parameters are switched to an air adaptation mode.

[0016] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. This invention can achieve rotor position estimation, speed estimation, load torque observation and medium state identification without encoders, torque sensors or medium sensors, effectively reducing system weight, volume and cost, reducing the complexity of waterproof design, and is particularly suitable for miniaturized water and air cross-medium unmanned platforms.

[0017] 2. This invention, by superimposing a virtual injection signal inside the position observer, and without applying the signal to the actual terminal voltage of the motor, can generate controllable flux tangential disturbances under low-speed, zero-speed, and underwater heavy-load conditions without causing additional motor vibration, noise, and current ripple. This enhances the observability of flux position when the back electromotive force is weak, and avoids problems such as angle drift, low-speed step loss, and starting difficulties.

[0018] 3. This invention uses a natural velocity observer as the main velocity observer and outputs the filtered estimated velocity as the velocity input of the load torque observer. This avoids the velocity noise amplification problem caused by directly differentiating the estimated electrical angle, effectively suppresses the transmission of observation noise and angle error to the load torque estimation result, and improves the accuracy of load torque observation under encoder-less conditions.

[0019] 4. This invention constructs a medium discrimination quantity based on the ratio of blade load torque to the square of speed. It can identify air medium, water medium, and water-air transition state in real time without adding any external sensors, and adaptively adjust the speed loop control parameters according to the medium state to match the control parameters with the medium environment, thus avoiding the problem of insufficient adaptability of fixed parameter controllers under cross-medium working conditions.

[0020] 5. This invention converts the estimated load torque into a q-axis current feedforward quantity, which is superimposed on the speed loop output. This allows for an increase in torque current to compensate for the load when the load suddenly increases at the moment of water ingress, and a timely reduction in torque current to prevent overspeed when the load suddenly decreases at the moment of water outlet. This effectively suppresses speed drop, speed overshoot, current surge and speed loop integral saturation during the cross-medium transition process.

[0021] 6. The steps of this invention cooperate and work together: the load torque observation results provide a basis for medium state identification and torque feedforward compensation, the medium identification results are used for smooth scheduling of speed loop control parameters, and the load change information is fed back to the intensity adjustment of the virtual injection signal, so that the system can maintain good speed stability and observation robustness in air, water and transition conditions, and realize stable operation across medium and all working conditions. Attached Figure Description

[0022] Figure 1 This is a flowchart of the SPMSM sensorless control method for cross-medium driven blades in an embodiment of the present invention; Figure 2 This is a schematic diagram of the SPMSM sensorless control method for cross-medium driven blades in an embodiment of the present invention. Detailed Implementation

[0023] The present invention will now be further described in conjunction with the accompanying drawings and specific embodiments. It should be noted that, without conflict, the various embodiments or technical features described below can be arbitrarily combined to form new embodiments.

[0024] like Figure 1-2 As shown, this embodiment of the invention provides a sensorless control method for SPMSM (Split Medium Module) propellers driven across media, comprising the following steps: Step S1: Obtain the stator current and voltage information of the SPMSM.

[0025] Step S2: Based on the stator current information and voltage information, the rotor position is estimated using a position observer superimposed with a virtual injection signal to obtain the estimated rotor position; wherein, the virtual injection signal only acts inside the position observer and is not applied to the actual terminal voltage of the motor.

[0026] Step S3: Based on the estimated rotor position and the voltage information, the speed is estimated using a speed observer to obtain the estimated speed.

[0027] Step S4: Based on the estimated speed and the stator current information, the load torque is estimated by using a load torque observer to obtain the estimated load torque.

[0028] Step S5: Identify the current medium state based on the estimated load torque and the estimated speed.

[0029] Step S6: Perform feedforward compensation on the torque command based on the estimated load torque, and adjust the control parameters according to the current medium state.

[0030] The SPMSM sensorless control method for cross-medium driven propellers according to embodiments of the present invention will be further described in detail below.

[0031] In this embodiment, the SPMSM sensorless control method for cross-medium driven propellers of the present invention is applied to the SPMSM propeller drive system of a water-air cross-medium unmanned platform. A sensorless disturbance rejection control structure is constructed, consisting of "an active flux linkage observer based on virtual hybrid voltage low-speed enhancement + a natural velocity observer + a sensorless ALTSMO + medium judgment + q-axis feedforward". The system samples the three-phase current and bus voltage, and obtains the stationary coordinate system current through Clarke transformation. The voltage command output by the current loop is compensated for dead time, switching transistor voltage drop, and bus voltage correction to obtain the actual compensated voltage.

[0032] In the formula, The stator three-phase current vector, for Phase current, for Phase current, for Phase current, This is the matrix transpose symbol. For the actual compensation voltage vector, This is the voltage vector output by the current loop. For the compensation voltage vector; where, the actual compensation voltage is... It is divided into two paths: one path enters the natural velocity observer and the actual current control link; the other path enters the active flux estimator after being superimposed with a virtual hybrid voltage.

[0033] In this embodiment, the specific details of the active flux linkage estimator design are as follows: For SPMSM, the d-axis and q-axis inductances are approximately equal. The stator flux linkage consists of inductive flux linkage and active flux linkage from the permanent magnets. The direction of the active flux linkage is the same as the rotor flux linkage direction, and therefore can be used to estimate electrical angles.

[0034] In the formula, This represents the d-axis inductance of the stator winding. This represents the q-axis inductance of the stator winding. This is the equivalent inductance value of the stator winding. The stator flux linkage vector is... The stator current vector in the stationary coordinate system. For active flux linkage vectors, It is a permanent magnet flux linkage. The rotor electrical angle.

[0035] The active flux linkage estimator employs a combination of voltage model and amplitude correction: In the formula, To estimate the stator flux linkage vector, To estimate the time derivative of the stator flux linkage, To estimate the active flux linkage vector, The observer voltage is input to the active flux linkage estimator. For stator resistance, The stator current vector in the stationary coordinate system. To estimate the electrical angle, This is the arctangent function in the four quadrants, used to calculate angles from the α-β components of magnetic flux linkage. To estimate the active flux linkage component along the α-axis, To estimate the active flux linkage component on the β-axis, This is the active flux linkage amplitude correction gain; where the amplitude correction term is used to pull the estimated active flux linkage amplitude back to near the permanent magnet flux linkage amplitude, suppressing low-speed voltage integral drift.

[0036] In this embodiment, the specific details of the virtual hybrid voltage low-speed enhancement are as follows: Traditional active flux linkage estimators are prone to flux linkage direction drift under low speed, zero speed, and heavy load conditions in water due to weak back electromotive force and a high proportion of voltage error. This invention injects a virtual hybrid voltage only into the voltage model input of the active flux linkage estimator, without it entering the actual SVPWM output.

[0037] In the formula, For virtual mixed voltage, Injecting intensity coefficients for virtual hybrid voltages. This is the virtual bias voltage coefficient. The virtual sinusoidal voltage coefficient, It is an orthogonal rotation matrix. Inject voltage amplitude into a virtual sine wave. Inject angular frequency into the virtual sine wave. For time; among which, This is a virtual bias voltage used to provide continuous model excitation during low-speed heavy loads; A virtual sinusoidal voltage orthogonal to the estimated active flux direction is used to generate tangential flux disturbance.

[0038] Since the rotor position angle is determined by the direction of the active flux linkage, linearization for small disturbances yields: During a short injection period, the perturbation of the active flux derivative is approximately determined by the virtual voltage: When only the orthogonal sinusoidal injection term is considered, we can obtain: In the formula, To estimate small perturbations in the electrical angle, To estimate the small perturbation vector of the active flux linkage, To estimate the perturbation of the active flux linkage derivative, To estimate the perturbation of the electrical angle derivative, To estimate the maximum value of the electrical angle disturbance; it can be seen that the orthogonal virtual sinusoidal voltage can generate controllable angle dynamic excitation inside the active flux estimator, so that there is still a flux tangential response that can be used for position determination at low speed or zero speed, thereby enhancing the observability of flux position.

[0039] In this embodiment, the specific details of the adaptive adjustment of virtual injection strength are as follows: The intensity of virtual injection is determined by the low speed, sudden load change, flux linkage amplitude error, and flux linkage reliability. In the formula, Calculate the rotational speed weighting coefficient for virtual injection intensity. Calculate the load torque weighting factor for the virtual injection strength. Virtual injection intensity calculation of flux linkage weight coefficient, Virtual injection strength calculation of flux linkage confidence weight coefficient, The threshold for low-speed judgment is set. To estimate the sudden change in load torque, This is the normalization threshold for load mutation. To estimate the confidence level of the magnetic linkage, The flux linkage confidence level is a saturation function; where the flux linkage confidence level can be constructed from the flux linkage amplitude error, the rate of change of angle, and the current residual. In the formula, It is an exponential function. This is the confidence weighting coefficient for the flux linkage amplitude. This is the confidence weighting coefficient for the magnetic flux linkage angle. The virtual injection intensity increases when the motor is at low speed, under heavy load in water, or when there is a sudden change in the inlet / outlet load or when the magnetic flux reliability decreases. When the system is stable, the virtual injection intensity decreases, causing the virtual mixed voltage to gradually withdraw.

[0040] In this embodiment, the specific details of the natural velocity observer design are as follows: The natural velocity observer, acting as the primary velocity observer, takes as input the electrical angle output from the active flux linkage estimator and the real compensation voltage (excluding dummy voltages). The dq-axis quantities are obtained through the Park transform. In the formula, For d-axis current, For q-axis current, This represents the α-axis stator current component. This represents the β-axis stator current component. This is the q-axis voltage. To accurately compensate for the voltage component on the α-axis, This is to provide the actual compensation voltage component on the β axis.

[0041] The natural velocity observer is designed as follows: In the formula, The q-axis current estimated by the natural velocity observer. To estimate the q-axis current derivative, To estimate the electric angular velocity, To estimate the mechanical angular velocity, To estimate the derivative of the mechanical angular velocity, This is the equivalent rotational inertia of the motor and blades. This represents the number of pole pairs of the motor. For SPMSM torque coefficient, To estimate the load torque; where, Load torque estimation is performed by the subsequent SALTSMO stage. To avoid speed noise affecting the load observer, the output speed needs to be low-pass filtered. In the formula, For the first The filtering mechanical speed per sampling period, The filtered mechanical speed of the previous sampling period. For the first The natural velocity observer outputs mechanical velocity per sampling period. For velocity filtering coefficients, This represents the discrete sampling period number.

[0042] In this embodiment, the specific details of the SALTSMO load torque observer are as follows: Traditional water-air trans-medium ALTSMOs typically use encoder speed to construct speed errors. This invention replaces this with a sensorless speed input. The internal speed state and speed observation error of the ALTSMO are defined as follows: In the formula, This is due to the error in the Saltsmo velocity observation. This involves estimating the mechanical angular velocity internally for SALTSMO. Specifically, it involves internally estimating the mechanical angular velocity. Stabilizes internal velocity estimation for SALTSMO.

[0043] SALTSMO is designed as follows: In the formula, Estimate the mechanical angular velocity inside SALTSMO. For symbolic functions, To estimate the derivative of the load torque; where, , Using the K∞ adaptive function: In the formula, Let be the adjustment coefficient for the i-th gain. Let be the error scaling coefficient for the i-th gain. The maximum limit of the i-th gain is defined as follows: when the load changes abruptly upon entering the water, causing an increase in speed error, the adaptive gain automatically increases to improve the speed estimation of the load torque; when the operation is stable, the adaptive gain decreases to reduce chattering.

[0044] In this embodiment, the specific content of the medium determination module is as follows: The blade load torque can be approximately expressed as a square relationship with the rotational speed: In the formula, This represents the actual blade load torque. This is the blade load factor, which is related to the medium density, blade shape, etc. This is the actual mechanical angular velocity.

[0045] The loading factor is smaller in air and larger in water. Therefore, a medium discrimination factor is defined as follows: In the formula, For medium discrimination, To prevent small positive numbers from being divided by zero at low speed, set a hysteresis threshold: In the formula, The upper threshold is used to determine the underwater pattern. The threshold value is used to determine the air mode.

[0046] Simultaneously, a continuous medium factor can be defined and low-pass filtered: In the formula, The continuity factor represents the degree of continuous change in the medium from air to water. The dielectric factor after low-pass filtering is used for smooth scheduling of velocity loop parameters. The medium factor of the previous sampling period. For the first The medium determination output value for each sampling period. The dielectric factor is the low-pass filter coefficient. The dielectric factor is used for speed loop gain scheduling, and the speed PI is adjusted according to the dielectric factor. In the formula, For the velocity loop proportional gain, For velocity loop integral gain, For the velocity loop proportional gain in air mode, The velocity loop integral gain in air mode. For the velocity loop proportional gain in underwater mode, The velocity loop integral gain is for underwater mode. In air, the load is light and the parameters are gentle; in water, the load is heavy and the parameters are enhanced; the transition region changes smoothly.

[0047] In this embodiment, the specific details of the q-axis torque feedforward compensation are as follows: SPMSM electromagnetic torque under id=0 control Approximately: Therefore, the load torque observation can be converted into a q-axis current feedforward. : The speed loop output and the final q-axis current command are: In the formula, This is the q-axis current command output by the speed loop. For speed loop PI controller, Given a mechanical angular velocity, Final q-axis current command; set amplitude and rate of change limits for the final q-axis current command. When entering the water, the load torque increases, and the q-axis feedforward increases rapidly to compensate for the load in advance; when exiting the water, the load torque decreases, and the q-axis feedforward decreases rapidly to suppress overspeed.

[0048] In this embodiment, the specific content of the control flow is as follows: This cross-medium SPMSM sensorless control method is based on an active flux linkage estimator, a natural velocity observer, a SALTSMO load observer, a medium judgment module, and a q-axis feedforward compensation module. The system first samples the three-phase current and bus voltage, and then combines this with the current loop output voltage to compensate for dead zone, tube voltage drop, and bus voltage, obtaining the true compensation voltage. The true compensation voltage and the virtual hybrid voltage are input together to the active flux linkage estimator to estimate the active flux linkage and rotor electrical angle. The virtual hybrid voltage does not actually act on the motor terminals; it only functions within the observer. When the motor is under low-speed heavy load, sudden water ingress, or a decrease in flux linkage reliability, the system enhances the virtual injection strength, improves the observability of the flux linkage position, and avoids instability in low-speed sensorless control.

[0049] The estimated electrical angle is used for Park transformation, converting the stationary coordinate system current and compensation voltage into dq-axis quantities, which are then input to the natural velocity observer to obtain the estimated electrical angular velocity and filtered mechanical velocity. The filtered mechanical velocity and q-axis current are further input to SALTSMO to estimate the blade load torque in real time. The system constructs a medium discrimination quantity based on the load torque and the square of the velocity to identify air, water, and water-air transition states. When the blade enters the water, the load torque increases rapidly, and the system determines that it has entered a transition or water state, and simultaneously increases the q-axis current feedforward, enhances the velocity loop's disturbance rejection capability, and increases the virtual injection intensity to suppress sudden speed drops. During stable operation in water, the controller maintains a strong load compensation capability and appropriately reduces virtual injection based on the flux linkage estimation state to reduce additional disturbances.

[0050] When the propeller blades emerge from the water, the load torque decreases rapidly, and the system promptly reduces the q-axis feedforward current to prevent the motor from overspeeding due to excessive torque. Simultaneously, the speed loop parameters smoothly transition from underwater mode to air mode, and virtual injection gradually exits. Ultimately, this method achieves smooth switching between air, underwater, and transition phases through coordinated control of medium determination, load observation, torque feedforward, and observation enhancement, improving the speed stability, disturbance rejection capability, and sensorless observation robustness of the SPMSM under cross-medium conditions.

[0051] The SPMSM sensorless control method for cross-medium driven propellers in this invention does not employ an active flux estimator, natural velocity observer, load torque observer, or q-axis feedforward individually. Instead, it collaboratively designs these modules according to the cross-medium operating conditions of water and air. In this SPMSM sensorless control method for cross-medium driven propellers, the virtual hybrid voltage is adaptively adjusted by load mutation, flux residual, and reliability, and is injected only into the master position observer; the ALTSMO is changed from encoder speed input to sensorless speed input; load torque estimation is used not only for q-axis feedforward but also for medium judgment and virtual injection to enhance decision-making. Through this closed-loop collaborative mechanism, position observability, load estimation accuracy, and speed disturbance immunity can be improved simultaneously under low-speed heavy load and medium mutation conditions.

[0052] This invention designs a sensorless composite observation and control structure for water-air cross-medium propeller drive. Unlike existing load torque observation methods that rely on encoder speed, this invention uses an active flux linkage estimator as the primary position observer and a natural velocity observer as the primary velocity observer to obtain sensorless electrical angles and electrical angular velocities, providing a sensorless input basis for cross-medium load observation and torque compensation. This invention also designs a load-sensing virtual hybrid voltage low-speed enhancement method. Unlike traditional fixed virtual voltage injection or injection methods only for zero-speed observation, this invention adaptively generates virtual injection intensity based on the low-speed level, load torque mutation, flux linkage amplitude error, and observation reliability. The virtual hybrid voltage is injected only into the voltage model input of the active flux linkage estimator to enhance the observability of flux linkage position under low-speed heavy load and medium mutation conditions. Finally, this invention designs a sensorless input cross-medium sliding mode load torque observer. Unlike existing water-air load torque observation methods based on encoder speed, this invention uses filtered sensorless speed output from a natural velocity observer to construct the observation error and combines it with adaptive sliding mode gain to estimate propeller load torque, achieving water-air load mutation observation without mechanical speed sensors. This invention designs a water-air medium identification method based on the load torque coefficient. This invention utilizes the observed load torque value and the square of the imperceptible velocity to construct a medium discrimination quantity: The invention identifies air, water, and transition states using hysteresis thresholds or continuous medium factors. It employs a load torque feedforward and sensorless observation enhancement collaborative control method. This method converts the observed load torque into a q-axis current feedforward to suppress inflow deceleration and outflow overspeed. Simultaneously, it feeds back load mutation information to a virtual hybrid voltage injection module, achieving closed-loop collaborative control of load identification, observation enhancement, and torque compensation, rather than relying solely on load feedforward or sensorless observation methods.

[0053] The SPMSM sensorless control method for cross-medium driven propellers according to embodiments of the present invention has the following advantages: 1. This invention can achieve rotor position estimation, speed estimation, load torque observation and medium state identification without relying on encoders, torque sensors and water immersion sensors. It can effectively reduce the weight, volume, cost and waterproof design complexity of water-air cross-medium unmanned platforms, and is conducive to the integration and deployment of miniaturized platforms.

[0054] 2. This invention injects the virtual hybrid voltage only into the position master observer and does not actually apply it to the motor end. This can enhance the position observability under low-speed and zero-speed conditions, and avoid the additional vibration, noise and current ripple caused by real high-frequency injection. At the same time, the virtual injection intensity is determined by the load change, flux linkage amplitude error and observation reliability, so as to achieve on-demand enhancement and avoid angle deviation and control disturbance caused by long-term fixed injection.

[0055] 3. This invention uses a natural velocity observer as the main velocity observer. Compared with directly obtaining the velocity by differentiating the estimated electrical angle, it can obtain a smoother velocity signal, which is beneficial to the stable operation of the subsequent load torque observer and avoids the transmission of velocity noise and angle error to the load torque estimation result.

[0056] 4. The SALTSMO of the present invention can quickly track the change of blade load torque when the cross-medium load changes abruptly. Its K∞ adaptive gain automatically improves the response speed under large disturbances and automatically decreases during stable operation to reduce chattering, thus taking into account both dynamic response speed and steady-state accuracy.

[0057] 5. This invention utilizes... It can determine the medium status without the need for an additional medium sensor, and can identify air, water and transition zone in real time based on the motor's own operating status; at the same time, the load torque feedforward can increase the q-axis current in advance when entering water and decrease the q-axis current in time when exiting water, thereby effectively suppressing speed drop, speed overshoot, current surge and speed loop integral saturation, and achieving smooth operation in the cross-medium transition process.

[0058] The above is a detailed description of the preferred embodiments of the present invention. However, the present invention is not limited to the embodiments described. Those skilled in the art can make various equivalent modifications or substitutions without departing from the spirit of the present invention. All such equivalent modifications or substitutions are included within the scope defined by the claims of this application.

Claims

1. A sensorless control method for SPMSM (Split-Media Driven Blade) propellers across media, characterized in that, Includes the following steps: Step S1: Obtain stator current and voltage information of SPMSM; Step S2: Based on the stator current information and voltage information, the rotor position is estimated using a position observer superimposed with a virtual injection signal to obtain the estimated rotor position; wherein, the virtual injection signal only acts inside the position observer and is not applied to the actual terminal voltage of the motor. Step S3: Based on the estimated rotor position and the voltage information, estimate the speed using a speed observer to obtain the estimated speed; Step S4: Based on the estimated speed and the stator current information, the load torque is estimated using a load torque observer to obtain the estimated load torque; Step S5: Identify the current medium state based on the estimated load torque and the estimated speed; Step S6: Perform feedforward compensation on the torque command based on the estimated load torque, and adjust the control parameters according to the current medium state.

2. The SPMSM sensorless control method for cross-medium driven blades according to claim 1, characterized in that, The position observer is an active flux linkage estimator, and the virtual injection signal is superimposed on the voltage signal input to the active flux linkage estimator; The virtual injection signal includes a bias component and an alternating component. The bias component is used to provide continuous excitation under low-speed conditions, and the alternating component is used to generate flux disturbances to enhance position observability.

3. The SPMSM sensorless control method for cross-medium driven blades according to claim 2, characterized in that, The intensity of the virtual injection signal is adaptively adjusted according to the motor operating status, which includes at least one of the following: motor speed, load change, flux linkage estimation error, and flux linkage estimation reliability. When the motor is at low speed, under heavy load, experiencing sudden load changes, or when the flux linkage reliability decreases, the strength of the virtual injection signal is increased; when the motor is running stably, the strength of the virtual injection signal is decreased.

4. The SPMSM sensorless control method for cross-medium driven propellers according to claim 1, characterized in that, The speed observer is an observer constructed based on the mechanical motion equation and electromagnetic torque equation of the motor. Its inputs include the estimated rotor position, the voltage information and the estimated load torque, and its output is the estimated speed. The estimated velocity is then low-pass filtered to output the filtered velocity.

5. The SPMSM sensorless control method for cross-medium driven propellers according to claim 1, characterized in that, The load torque observer is a sliding mode observer, and its speed observation error is composed of the difference between the estimated speed and the speed estimate inside the load torque observer. Its gain adapts to the magnitude of the speed observation error. When sudden load changes cause an increase in speed observation error, the gain is increased to improve load torque tracking speed; when the system tends to stabilize, the gain is decreased to suppress chattering.

6. The SPMSM sensorless control method for cross-medium driven propellers according to claim 1, characterized in that, In step S5, the current medium state includes the air medium state, the water medium state, and the transition state between air and water; The method for identifying the current medium state is as follows: A discriminant is constructed that is associated with the ratio of the estimated load torque to the square of the speed, and the current medium state is determined by comparing the discriminant with a preset threshold; the preset threshold includes at least an upper threshold and a lower threshold, and stable discrimination of the medium state is achieved through hysteresis comparison.

7. The SPMSM sensorless control method for cross-medium driven propellers according to claim 6, characterized in that, A continuously varying medium factor is constructed based on the discriminant, which varies continuously between air and water media. The medium factor, after filtering, is used to smoothly schedule the speed loop control parameters, which include at least one of proportional gain and integral gain.

8. The SPMSM sensorless control method for cross-medium driven propellers according to claim 1, characterized in that, In step S6, the feedforward compensation is to convert the estimated load torque into a q-axis current feedforward quantity, which is then superimposed on the q-axis current command output by the speed loop controller. The q-axis current feedforward increases as the estimated load torque increases and decreases as the estimated load torque decreases.

9. The SPMSM sensorless control method for cross-medium driven propellers according to claim 1, characterized in that, The position observer estimates the rotor flux linkage by combining a voltage model with amplitude correction. The amplitude correction is used to constrain the estimated flux linkage amplitude to the vicinity of the known permanent magnet flux linkage amplitude.

10. The SPMSM sensorless control method for cross-medium driven propellers according to claim 1, characterized in that, The method forms a closed-loop collaborative control: The estimated load torque is also used for identifying the current medium state, for feedforward compensation of the torque command, and for intensity adjustment of the virtual injection signal; When the blades transition between different media, the three components work in concert: When the estimated load torque increases, the virtual injection signal is enhanced, the feedforward compensation is increased, and the speed loop control parameters are switched to an underwater adaptation mode; when the estimated load torque decreases, the virtual injection signal is weakened, the feedforward compensation is reduced, and the speed loop control parameters are switched to an air adaptation mode.

Images