A sampling-free resistance square wave control overcurrent protection method

Abstract

The application discloses a kind of no sampling resistance square wave control overcurrent protection method, method includes: real-time acquisition digital sampling sequence of direct current bus voltage, synchronous analysis logic state combination of three-phase PWM driving signal;According to the mapping table of pre-stored commutation state machine, the logic state combination is converted into current effective conduction phase identification, and the signal of corresponding current detection enable window is generated, and enable window is limited to the interval of voltage oscillation attenuation completion after PWM opening to before off before freewheeling protection;During the effective period of enable window, the digital band-pass filtering is carried out to bus voltage digital sampling sequence, the voltage drop component of characteristic frequency band is extracted, and the time-domain integral value and the rate differential value of voltage drop component are calculated.The application is reconstructed by the current of no sampling resistance technology and dynamic protection strategy, while reducing hardware cost, significantly improve system reliability, effectively solve the protection failure problem of traditional scheme under commutation interference, high-speed working condition and transient impact.

Description

Technical Field

[0001] This application relates to the field of motor drive protection technology, and in particular to an overcurrent protection method with square wave control without sampling resistor. Background Technology

[0002] In the field of brushless DC motor drives employing square wave control strategies, overcurrent protection mechanisms to ensure safe system operation are crucial. This control method is simple in structure and highly efficient, but its inherent commutation process generates strong voltage and current transients, posing a significant challenge to real-time and accurate monitoring of phase current. Currently, mainstream technologies generally rely on inserting a physical sampling resistor in series in the motor circuit to obtain the current signal; however, this traditional method suffers from two mutually restrictive and difficult-to-overcome core drawbacks:

[0003] 1. The fundamental conflict between hardware cost and system reliability: The introduction of sampling resistors and their associated signal conditioning circuits directly increases the system's material cost. More importantly, the power dissipation of the resistors during operation causes them to heat up, resulting in resistance drift and current measurement errors. This error directly affects the accuracy of overcurrent protection thresholds, potentially leading to malfunctions or failures of the protection function. Under continuous high current or high temperature conditions, the resistors themselves may even burn out, becoming a potential weak point in system reliability. The pursuit of low cost often comes at the expense of stable and reliable protection capabilities.

[0004] 2. Severe Insufficiency of Protection Performance under Dynamic Operating Conditions: The voltage spikes and current oscillations generated during each commutation of the square wave control are easily misidentified as genuine overcurrent faults, leading to frequent false triggering of the protection system and interfering with normal motor operation. When the motor speed exceeds its base speed, the back electromotive force generated inside the motor increases significantly. At this time, the effective voltage applied to the motor windings (bus voltage minus back electromotive force) decreases relatively. Traditional fixed threshold overcurrent protection cannot detect this change and may fail to detect dangerous operating conditions that are already overcurrent relative to the current effective drive capability in the high-speed range, causing protection failure. For inrush currents with extremely high change rates that occur in a short time (such as short circuits), the response speed of traditional sampling resistor-based solutions is insufficient due to signal filtering and algorithm processing time limitations, making it difficult to prevent damage to power devices in a timely manner. Summary of the Invention

[0005] To address the aforementioned problems, embodiments of the present invention provide a square wave control overcurrent protection method without sampling resistor, the method comprising:

[0006] The digital sampling sequence of DC bus voltage is acquired in real time, and the logic state combination of the three-phase PWM drive signal is analyzed synchronously.

[0007] According to the pre-stored commutation state machine mapping table, the logic state combination is converted into the current effective conducting phase identifier, and a corresponding current detection enable window signal is generated. The enable window is limited to the freewheeling protection range from the completion of voltage oscillation decay after PWM is turned on to before the turn-off.

[0008] During the effective period of the enable window, the digital sampling sequence of the bus voltage is digitally bandpass filtered to extract the voltage drop component of the characteristic frequency band, and the time-domain integral value and the differential value of the rate of change of the voltage drop component are calculated.

[0009] The voltage drop component is input into the discretized state space model of the motor winding. The discretized state space model includes a back EMF observer compensation unit and a power transistor on-state voltage drop adaptive compensation unit. The model parameters are updated in real time through a recursive least squares algorithm, and the reconstructed current value of the winding current is output.

[0010] A dynamic overcurrent protection threshold function is constructed, whose independent variables are the real-time speed and the moving average of the bus voltage calculated based on the Hall event. The output value of the dynamic overcurrent protection threshold function remains constant below the base speed and decays inversely proportionally to the back electromotive force above the base speed.

[0011] When the reconstructed current value exceeds the output value of the dynamic overcurrent protection threshold function within N consecutive electrical cycles, where N > 1, and the excess amount satisfies the preset cumulative energy criterion, the protection state machine is triggered to forcibly switch all PWM outputs to a safe state.

[0012] Furthermore, the processing methods for digital sampling sequences include:

[0013] A multi-stage cascaded digital filter bank is used. The first stage is a moving average filter to suppress switching noise, and the second stage is a phase-compensated FIR filter to retain voltage drop characteristics.

[0014] The logical state combination is obtained by directly reading the binary status word of the microcontroller's timer compare register.

[0015] Furthermore, the method for generating the enabling window includes:

[0016] The window start delay is dynamically calculated based on the product of the motor phase inductance and the current speed;

[0017] The window ends early based on the ratio of the freewheeling time constant to the PWM period.

[0018] Furthermore, the implementation of the back EMF observer compensation unit includes:

[0019] A back EMF tracker triggered by Hall edge events is established, and a continuous electrical angle signal is generated in adjacent Hall intervals using angle interpolation.

[0020] Based on the electrical angle signal and the magnetic flux linkage mapping table, the back electromotive force compensation value is calculated in real time.

[0021] Furthermore, the online calibration method for the magnetic flux linkage map table includes:

[0022] During the no-load acceleration process of the motor, the characteristic quantity of bus voltage fluctuation at different speeds is collected;

[0023] The coefficients of the flux linkage-angle relationship polynomial are updated by curve fitting.

[0024] Furthermore, the dynamic overcurrent protection threshold function is implemented using a two-dimensional interpolation table, and the implementation method includes:

[0025] A non-uniform grid of points is constructed with rotational speed as the x-axis and bus voltage as the y-axis.

[0026] The grid point threshold data was obtained through the software calibration process of the motor temperature rise test.

[0027] Furthermore, the method for calculating the cumulative energy criterion includes:

[0028] The cumulative energy value is obtained by integrating the excess over time.

[0029] Protection is triggered when the accumulated energy value exceeds the preset safety margin.

[0030] Furthermore, the recovery process executed after protection is triggered includes:

[0031] During the period when the PWM output is blocked, the software sample values ​​of the three-phase terminal voltage are continuously monitored;

[0032] Rotor position information is reconstructed using a back EMF zero-crossing detection algorithm;

[0033] The protection state is released after the commutation sequence is initialized based on the reconstructed position.

[0034] Furthermore, the dynamic calculation of the window start delay introduces a parameter self-learning mechanism, which includes:

[0035] Inject a test pulse sequence during the motor startup phase;

[0036] Analyze the voltage response waveform characteristics to automatically optimize the delay parameters.

[0037] A method for overcurrent protection using square wave control without sampling resistor, the method further includes:

[0038] Add a feedforward prediction mechanism to the back EMF observer compensation unit; establish a graded response mechanism for current change rate.

[0039] The technical effects and advantages of the non-sampling resistor square wave controlled overcurrent protection method provided by this invention are as follows:

[0040] This invention significantly improves system reliability while reducing hardware costs through current reconstruction technology without sampling resistors and a dynamic protection strategy, effectively solving the protection failure problems of traditional solutions under commutation interference, high-speed conditions, and transient impacts. This invention suppresses commutation spike interference through voltage characteristic analysis, greatly reducing the protection false trigger rate. Based on adaptive thresholds of real-time speed and voltage, it matches the back EMF changes in the high-speed weak magnetic region to solve the high-speed leakage protection problem. The fast response mechanism effectively captures transient inrush currents, preventing damage to power devices. The current reconstruction technology is independent of motor parameters, adapting to different operating conditions and device aging states. Protection commands and PWM control are linked in real time, automatically adjusting the drive strategy to suppress current during overcurrent. The fault feature marking function helps quickly locate the root cause of abnormalities. Eliminating the sampling resistor and its associated circuitry directly reduces material costs and circuit complexity, avoids measurement deviations caused by resistor temperature drift, eliminates the risk of system failure caused by resistor burnout, reduces heat sources, and improves overall system energy efficiency. Attached Figure Description

[0041] Figure 1 This is a flowchart of a square wave control overcurrent protection method without sampling resistor in Example 1;

[0042] Figure 2 This is a flowchart of a square wave control overcurrent protection method without sampling resistor in Example 2. Detailed Implementation

[0043] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0044] Example 1:

[0045] Please see Figure 1 As shown, an embodiment of the present invention provides a square wave control overcurrent protection method without sampling resistor, the method comprising:

[0046] S101: Real-time acquisition of the digital sampling sequence of DC bus voltage, and synchronous analysis of the logic state combination of three-phase PWM drive signals;

[0047] S102: Based on the pre-stored commutation state machine mapping table, the logic state combination is converted into the current effective conducting phase identifier, and a corresponding current detection enable window signal is generated. The enable window is limited to the freewheeling protection range from the completion of voltage oscillation decay after PWM is turned on to before the turn-off.

[0048] S103: During the effective period of the enable window, digital bandpass filtering is performed on the digital sampling sequence of the bus voltage to extract the voltage drop component of the characteristic frequency band, and the time-domain integral value and the differential value of the rate of change of the voltage drop component are calculated.

[0049] S104: Input the voltage drop component into the discretized state space model of the motor winding. The discretized state space model includes a back EMF observer compensation unit and a power transistor on-state voltage drop adaptive compensation unit. The model parameters are updated in real time through a recursive least squares algorithm, and the reconstructed current value of the winding current is output.

[0050] S105: Construct a dynamic overcurrent protection threshold function, whose independent variables are the real-time speed and the moving average of the bus voltage calculated based on the Hall event. The output value of the dynamic overcurrent protection threshold function remains constant below the base speed and decays inversely proportionally to the back electromotive force above the base speed.

[0051] S106: When the reconstructed current value exceeds the output value of the dynamic overcurrent protection threshold function within N consecutive electrical cycles, N>1, and the exceedance amount satisfies the preset cumulative energy criterion, the protection state machine is triggered to forcibly switch all PWM outputs to a safe state.

[0052] In the digital sampling sequence processing (S101), a multi-stage cascaded digital filter bank is used to achieve signal preprocessing and feature enhancement. The first-stage moving average filter performs window smoothing on the original sampling sequence. By calculating the arithmetic mean of sampling points of a fixed length, it effectively suppresses high-frequency pulse noise introduced by the switching action of the power transistor. For example, in an application scenario with a switching frequency of 100kHz, a sliding window of 20 sampling points (corresponding to a time width of 0.2ms) can be used to attenuate the switching noise by more than 6dB. The second-stage phase compensation FIR filter adopts a linear phase design. Its passband range is set according to the characteristic frequency band of the voltage drop signal. The typical configuration is to pass through the 0.5-5kHz frequency band signal. While eliminating group delay distortion, this filter completely preserves the time-domain waveform characteristics of the voltage drop component, ensuring the timing accuracy of subsequent processing.

[0053] The acquisition of logic state combinations is achieved by directly accessing the microcontroller's timer compare register. In the square wave control architecture, the states of the six PWM signals of the three-phase bridge arm are mapped in real time to binary state words in the register. Taking a three-phase half-bridge drive as an example, the state words are arranged in the format [UH_UL, VH_VL, WH_WL] (U / V / W represent phase sequence, H / L represent high / low side transistors). By periodically reading the state word and masking invalid state combinations (such as interlocking states to prevent shoot-through of upper and lower transistors), valid power transistor conduction combinations can be resolved in real time. This direct register access mechanism avoids additional hardware sampling circuits, and the state update is strictly synchronized with the PWM waveform generation, with a typical delay of no more than one CPU clock cycle.

[0054] The method for generating the enable window in step (S102) includes:

[0055] The window start delay is dynamically calculated based on the product of the motor phase inductance and the current speed;

[0056] The generation of the enable window relies on software modeling of the motor's electromagnetic characteristics and the dynamics of the power switch. The dynamic calculation of the window start delay is based on the following physical principle: the motor phase inductance generates an induced electromotive force when the current changes, and its time constant is positively correlated with the inductance. By converting the current speed into electrical angular velocity and combining it with the pre-stored phase inductance parameter table, the delay time is calculated in real time. For example, under the condition of rated speed of 3000 rpm and phase inductance of 10mH, the start delay can be dynamically adjusted to 15% to 20% of the PWM cycle to ensure that the voltage oscillation caused by the current rise after the power transistor is turned on is fully attenuated, so that the starting point of the detection window is located in the voltage stability plateau period.

[0057] The window ends early based on the ratio of the freewheeling time constant to the PWM period.

[0058] The determination of the window end advance takes into account the energy release characteristics of the freewheeling process. The freewheeling time constant characterizes the speed at which the energy stored in the winding is released through the freewheeling diode. This parameter is related to the equivalent impedance of the motor winding. By establishing a proportional relationship model between the freewheeling time constant and the current PWM cycle, the window end advance coefficient is dynamically generated. In a typical configuration, this coefficient is set to ensure that the window has sufficient margin before the PWM turn-off signal is issued, avoiding the detection time from entering the freewheeling current sudden change region at the moment of turn-off. Taking a 50kHz PWM frequency as an example, the advance can be adaptively adjusted to 5% to 8% of the cycle width, which can effectively avoid voltage spike interference caused by reverse recovery current.

[0059] The implementation method of the back EMF observer compensation unit in step (S104) includes:

[0060] A back EMF tracker triggered by Hall edge events is established, and continuous electrical angle signals are generated in adjacent Hall intervals using angle interpolation. The core of the back EMF observer compensation unit is to construct a real-time tracking model of the motor back EMF. When a Hall sensor edge event is detected, the back EMF tracker uses that moment as a synchronization reference point and generates continuous electrical angle signals in two adjacent Hall event intervals using angle interpolation. In specific implementation, the number of electrical angle partitions is determined based on the number of Hall sensors. For example, with a configuration of three Hall sensors, each 60-degree electrical angle interval uses a linear interpolation algorithm to convert discrete Hall position signals into continuously changing electrical angle values, with a resolution of up to 0.1 degrees, ensuring the phase continuity of the back EMF calculation.

[0061] Based on the electrical angle signal, the back EMF compensation value is calculated in real time by querying a pre-stored flux linkage mapping table. The flux linkage mapping table uses the electrical angle as an index and stores the standard flux linkage value at the corresponding position. This table is generated during the no-load reverse drag test when the motor leaves the factory. During the real-time calculation, a table lookup operation is performed based on the current electrical angle signal, and the instantaneous value of the back EMF e(t) is calculated in combination with the real-time speed: e(t) = flux linkage value × angular velocity, where t is the current time. For example, under the condition of an electrical angle of 30 degrees and a speed of 1500 rpm, if the flux linkage value is found to be 0.05 Wb, the calculated back EMF compensation value is 7.85 V. This compensation value is injected into the discretized state space model in real time to offset the interference of the winding induced EMF on the bus voltage drop component.

[0062] Online calibration methods for flux linkage maps include:

[0063] During the no-load acceleration of the motor, the bus voltage fluctuation characteristics at different speeds are collected. The online calibration of the flux linkage mapping table is automatically triggered when the motor has no torque output. When the system detects the no-load acceleration process, it controls the inverter to output a six-step commutation sequence but does not apply load torque. During this process, the bus voltage fluctuation characteristics are collected at fixed speed intervals. The bus voltage fluctuation characteristics specifically refer to the envelope amplitude and phase offset of the voltage drop component. For example, during the process of increasing the speed from 1000 rpm to 3000 rpm, a set of data points is collected every 200 rpm. Random interference is eliminated by statistically analyzing the distribution of characteristics over multiple electrical cycles.

[0064] Based on the collected feature dataset, the coefficients of the flux linkage-angle relationship polynomial are dynamically updated using a curve fitting algorithm. Specifically, the least squares fitting method is employed to model the flux linkage function as an nth-order polynomial of the electrical angle. ,include: ;

[0065] The polynomial order n is adaptively selected based on the number of pole pairs of the motor. For example, a 5th-order polynomial is used for a 4-pole motor. The fitting process constructs a back electromotive force equation that includes a speed term. ,include: ;

[0066] The measured voltage fluctuation amplitude is used as... The observed values, combined with the corresponding rotational speed and electrical angle Inverse solution The gradient matrix is ​​ultimately used to update the polynomial coefficient vector. The new coefficients are smoothed and filtered before being written into non-volatile memory to achieve online calibration of the flux linkage mapping table.

[0067] The dynamic overcurrent protection threshold function realizes real-time threshold decision through a two-dimensional interpolation table structure. A non-uniform grid coordinate system is constructed with speed as the horizontal axis and bus voltage as the vertical axis. The grid points are divided into non-uniform intervals according to the distribution law of motor thermal characteristics. Dense grid distribution is adopted in the motor base speed transition area (such as 1200-1800rpm) and low voltage operating area (such as bus voltage <24V), and the grid spacing can be reduced to 1 / 3 of the standard area. Sparse distribution is adopted in the flat operating area to balance the storage resources and protection accuracy requirements.

[0068] The threshold data corresponding to the grid points were obtained through the software calibration process of the motor temperature rise test. The calibration process was performed in a constant temperature environment chamber, and the execution method included:

[0069] Operating condition traversal: Control the motor to run at speed-voltage grid points. For example, set 10 speed levels on the speed axis [500rpm, 1000rpm, 1500rpm, ..., 5000rpm] and 4 voltage levels on the voltage axis [12, 24, 36, 48V], forming 40 calibration points.

[0070] Threshold determination: At each operating point, the circuit is continuously run until thermal equilibrium is reached (winding temperature rise ΔT ≤ 2℃ / min), and the load is gradually increased until the insulation critical point is reached. The reconfiguration current value at this point is recorded as the threshold.

[0071] Data encapsulation: The calibration data is packaged into a two-dimensional lookup table and burned into the microcontroller's flash memory. During runtime, the dynamic threshold of any operating point is calculated through bilinear interpolation.

[0072] The calculation of the cumulative energy criterion in step (S106) includes:

[0073] The cumulative energy value is obtained by integrating the excess over time. The cumulative energy criterion quantifies overcurrent energy through real-time integral overcurrent; the overcurrent is defined as the instantaneous deviation of the reconstructed current value from the dynamic overcurrent protection threshold. Over N consecutive electrical cycles, the excess quantity within each control cycle is integrated over time. ;

[0074] in The integral is obtained by discretizing and accumulating the equivalent thermal resistance parameter of the winding. For index scalars, the first The cumulative energy value within one electrical cycle, i.e.: ;

[0075] In the formula, For the control period, for example at a control frequency of 10kHz. =0.1ms, the integration process focuses on accumulating the actual overcurrent energy rather than the instantaneous value, effectively distinguishing between short-time pulse interference and real overload faults.

[0076] Protection is triggered when the accumulated energy value exceeds a preset safety margin; the preset safety margin is set based on the motor's thermal capacity characteristics, and its physical essence is the maximum Joule thermal energy that the winding insulation material can withstand; when the accumulated energy value Exceeding the safety margin threshold When this happens, the protection state machine is triggered. For example, if the winding heat capacity of a certain type of motor is 150J, the setting is... =120J (with a 20% design margin) to ensure that the protection action is completed before the insulation is damaged; this mechanism avoids the "false trigger-reset oscillation" problem of traditional fixed threshold protection and achieves precise management of overcurrent energy.

[0077] The recovery process executed after protection is triggered includes:

[0078] During the PWM output blocking period, the software sampling values ​​of the three-phase terminal voltage are continuously monitored; the recovery process after protection triggering achieves safe restart through sensorless position reconstruction technology; when the system blocks the PWM output, all power transistors of the inverter remain off, and the real-time sampling values ​​of the three-phase terminal voltage are continuously collected through the analog-to-digital conversion channel of the microcontroller; as the motor rotor continues to rotate under inertia, its permanent magnet magnetic field induces a three-phase symmetrical back electromotive force in the stator winding, and the phase change of this electromotive force directly reflects the rotor position information.

[0079] The rotor position information is reconstructed using a back EMF zero-crossing detection algorithm. The core of this algorithm is identifying the correspondence between the voltage waveform and the rotor position, including:

[0080] The collected UVW three-phase voltages are subjected to coordinate transformation to eliminate three-phase coupling interference and obtain independent voltage components;

[0081] When the voltage value at a certain phase terminal crosses the midpoint of the bus voltage (e.g., the 24V threshold in a 48V system), that moment is marked as the zero-crossing point of the back electromotive force.

[0082] Based on the time sequence of the three-phase zero crossings, determine the 60-degree electrical angle sector where the rotor is located (for example, if the V phase crosses zero after the U phase crosses zero, then the rotor is determined to be in sector 1).

[0083] Based on the reconstructed rotor sector position, the initial phase of the six-step commutation sequence is initialized; if the rotor is detected to be in sector 1, the initial conduction mode is set to a combination of the upper bridge arm of phase U and the lower bridge arm of phase V; after capturing three complete zero-crossing event sequences in succession, the system releases the protection state in a limited power mode, including: first outputting the drive signal again with a 50% PWM duty cycle, and then returning to normal operation mode after two electrical cycles.

[0084] The dynamic calculation of the window start delay introduces a parameter self-learning mechanism, including:

[0085] After the rotor positioning is completed in the initial stage of motor startup, the system automatically injects a low-energy test pulse sequence into the three-phase windings. The pulse adopts a short-time square wave mode, the duration is controlled in the order of hundreds of microseconds, and the amplitude is strictly limited to within 20% of the rated operating current to ensure that it will not cause torque fluctuations. The pulse is applied to each phase in the commutation sequence, and the total injection time does not exceed 200 milliseconds to maintain startup efficiency.

[0086] During the pulse action, the target phase-terminal voltage waveform is acquired at high speed using an analog-to-digital converter; three key response characteristics are analyzed in detail:

[0087] The rise-edge overshoot amplitude reflects the instantaneous impact intensity of line parasitic parameters on voltage; the plateau establishment time characterizes the transient response process of winding inductance and resistance; and the decay response rate reflects the energy release characteristics of the winding.

[0088] A delay optimization model is established based on feature analysis. When the establishment time of the measured platform is longer than the reference value, the window delay is increased proportionally. If the voltage overshoot amplitude exceeds the nominal range, a high-frequency oscillation compensation is added. The sampling interval during the turn-off stage is dynamically adjusted according to the change in the decay rate.

[0089] The optimized parameters are updated to the current reconfiguration controller in real time. For example, when the platform setup time is detected to increase from the baseline value of 15 microseconds to 18 microseconds, the system automatically increases the window start delay by 20% in sync.

[0090] The parameter self-learning mechanism runs automatically at each startup, effectively compensating for parameter differences and operating temperature drift of different motor windings, while avoiding the introduction of additional hardware resources; the self-learning process is equipped with a safety fuse mechanism: when the voltage characteristic value exceeds the safety threshold, the learning is automatically terminated and the factory preset parameters are activated to ensure system robustness.

[0091] Example 2:

[0092] like Figure 2 As shown, this embodiment further improves upon the design of Embodiment 1. The difference is that in actual operation of Embodiment 1, it was found that the calculation delay of the back EMF observer compensation unit caused phase lag in the current reconstruction value under conditions of high-speed motor operation or sudden large load changes, resulting in insufficient overcurrent protection response speed and failure to effectively suppress transient overcurrent impacts. Based on this, a square wave control overcurrent protection method without sampling resistors further includes:

[0093] Add a feedforward prediction mechanism to the back EMF observer compensation unit; establish a graded response mechanism for current change rate.

[0094] The feedforward prediction mechanism is executed in the back EMF observer compensation unit, and the execution method includes:

[0095] Based on the timestamps of three consecutive Hall edge events ( Calculate instantaneous angular acceleration ,Right now: ;in, ; The Hall sensor interval electrical angle;

[0096] Constructing a second-order location prediction model ,include: ;

[0097] in, The reference electrical angle for the most recent Hall event. The time offset between the current moment and the nearest Hall event;

[0098] when When (angular acceleration threshold) is reached, with Retrieve the compensation value from the magnet link mapping table ,include: ;

[0099] In the formula, This is the flux linkage value at the predicted angle;

[0100] A graded response is performed before the cumulative energy criterion. The response methods include:

[0101] Real-time calculation of reconstructed current change rate ;

[0102] In the formula, This is the reconfiguration current value for the nth control cycle. To control the cycle;

[0103] Actions are triggered based on the rate of change range, including:

[0104] like If so, the original protection process will remain unchanged;

[0105] like Then tighten the dynamic threshold to , This is the output value of the dynamic overcurrent protection threshold function;

[0106] like This triggers a pre-protection state, limiting the PWM duty cycle to... And enable 2ms fast sampling, This represents the current actual PWM duty cycle. This is the minimum duty cycle allowed by the system.

[0107] In pre-protection mode, if the condition is met within 3ms and If so, the protection will be lifted.

[0108] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

[0109] The above description is merely a preferred embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present application, based on the technical solution and concept of the present application, should be covered within the scope of protection of the present application.

Claims

1. A method for overcurrent protection using square wave control without sampling resistor, characterized in that the method... include: The digital sampling sequence of DC bus voltage is acquired in real time, and the logic state combination of the three-phase PWM drive signal is analyzed synchronously. According to the pre-stored commutation state machine mapping table, the logic state combination is converted into the current effective conducting phase identifier, and a corresponding current detection enable window signal is generated. The enable window is limited to the freewheeling protection range from the completion of voltage oscillation decay after PWM is turned on to before the turn-off. During the effective period of the enable window, the digital sampling sequence of the bus voltage is digitally bandpass filtered to extract the voltage drop component of the characteristic frequency band, and the time-domain integral value and the differential value of the rate of change of the voltage drop component are calculated. The voltage drop component is input into the discretized state space model of the motor winding. The discretized state space model includes a back EMF observer compensation unit and a power transistor on-state voltage drop adaptive compensation unit. The model parameters are updated in real time through a recursive least squares algorithm, and the reconstructed current value of the winding current is output. A dynamic overcurrent protection threshold function is constructed, whose independent variables are the real-time speed and the moving average of the bus voltage calculated based on the Hall event. The output value of the dynamic overcurrent protection threshold function remains constant below the base speed and decays inversely proportionally to the back electromotive force above the base speed. When the reconstructed current value exceeds the output value of the dynamic overcurrent protection threshold function within N consecutive electrical cycles, where N > 1, and the exceedance amount satisfies the preset cumulative energy criterion, the protection state machine is triggered to forcibly switch all PWM outputs to a safe state.

2. The overcurrent protection method with square wave control without sampling resistor according to claim 1, characterized in that, Methods for processing digital sampling sequences include: A multi-stage cascaded digital filter bank is used. The first stage is a moving average filter to suppress switching noise, and the second stage is a phase-compensated FIR filter to retain voltage drop characteristics. The logical state combination is obtained by directly reading the binary status word of the microcontroller's timer compare register.

3. The overcurrent protection method with square wave control without sampling resistor according to claim 1, characterized in that, The method for generating the enabled window includes: The window start delay is dynamically calculated based on the product of the motor phase inductance and the current speed; The window ends early based on the ratio of the freewheeling time constant to the PWM period.

4. The overcurrent protection method with square wave control without sampling resistor according to claim 1, characterized in that, The implementation methods of the back EMF observer compensation unit include: A back EMF tracker triggered by Hall edge events is established, and a continuous electrical angle signal is generated in adjacent Hall intervals using angle interpolation. Based on the electrical angle signal and the magnetic flux linkage mapping table, the back electromotive force compensation value is calculated in real time.

5. The overcurrent protection method with square wave control without sampling resistor according to claim 4, characterized in that, The online calibration method for the magnetic flux linkage map table includes: During the no-load acceleration of the motor, the characteristic quantity of bus voltage fluctuation at different speeds is collected. The coefficients of the flux linkage-angle relationship polynomial are updated by curve fitting.

6. The overcurrent protection method with square wave control without sampling resistor according to claim 1, characterized in that, The dynamic overcurrent protection threshold function is implemented using a two-dimensional interpolation table. The implementation method includes: A non-uniform grid is constructed with rotational speed as the x-axis and bus voltage as the y-axis. The grid point threshold data was obtained through the software calibration process of the motor temperature rise test.

7. The overcurrent protection method with square wave control without sampling resistor according to claim 1, characterized in that, The calculation method for the cumulative energy criterion includes: The cumulative energy value is obtained by integrating the excess over time. Protection is triggered when the accumulated energy value exceeds the preset safety margin.

8. The overcurrent protection method with square wave control without sampling resistor according to claim 1, characterized in that, The recovery process executed after protection is triggered includes: During the period when the PWM output is blocked, the software sample values ​​of the three-phase terminal voltage are continuously monitored; Rotor position information is reconstructed using a back EMF zero-crossing detection algorithm; The protection state is released after the commutation sequence is initialized based on the reconstructed position.

9. The overcurrent protection method with square wave control without sampling resistor according to claim 3, characterized in that, The dynamic calculation of the window start delay introduces a parameter self-learning mechanism, which includes: Inject a test pulse sequence during the motor startup phase; Analyze the voltage response waveform characteristics to automatically optimize the delay parameters.

10. The overcurrent protection method with square wave control without sampling resistor according to claim 1, characterized in that, The method also includes: Add a feedforward prediction mechanism to the back EMF observer compensation unit; establish a graded response mechanism for current change rate.

Images