A multi-working-condition-oriented intelligent commutation timing dynamic adjustment method for brushless motor

Abstract

The application discloses a multi-working-condition-oriented intelligent commutation timing dynamic adjustment method for a brushless motor, and relates to the technical field of motor control. The application comprises the following steps: establishing a lookup table of the brushless motor, wherein the lookup table comprises a plurality of lookup cells, each of which corresponds to a type of operating condition and a reference commutation angle, wherein the operating condition comprises rotating speed, load torque, bus voltage and temperature. The application can quickly output a disturbance interval by respectively outputting positive and negative disturbance simulation results through two calculation chains, so as to quickly narrow the range of the optimal commutation angle. The application can quickly find the optimal commutation angle by concentrating all resources of the calculation center to search for optimization in the disturbance interval. The application can facilitate uninterrupted disturbance optimization of subsequent real-time operating conditions by the two calculation chains after the optimal commutation angle is found.

Description

Technical Field

[0001] This invention relates to the field of motor control technology, specifically to a method for intelligent dynamic adjustment of commutation timing of brushless motors under multiple operating conditions. Background Technology

[0002] Brushless motors are widely used in electric vehicles, aerospace, industrial robots, and high-end manufacturing equipment due to their high power density, long lifespan, and excellent speed regulation performance. In actual operation, the commutation angle (commutation advance angle or lag angle) of a brushless motor has a decisive impact on torque ripple, system efficiency, and vibration noise. However, the operating conditions of the motor (such as speed, load torque, bus voltage, and winding temperature) in complex application scenarios often exhibit variability, nonlinearity, and strong coupling characteristics, making it difficult for a fixed commutation angle strategy to guarantee that the motor is in its optimal performance state across the entire operating range.

[0003] In existing technologies, two main methods are typically used to achieve dynamic adjustment of the commutation angle: One is a lookup table method based on bench calibration. This involves testing different operating points offline beforehand, recording the optimal commutation angle, and creating a lookup table. While this method allows for fast response times by directly looking up the table based on the current operating condition, its accuracy is limited by the discrete density of the calibrated operating points and it cannot cover uncalibrated intermediate operating conditions or scenarios where environmental parameters change, resulting in performance blind spots. The other method is online optimization, which involves real-time acquisition of signals such as current and speed during motor operation and using strategies such as gradient descent and perturbation observation to search for the optimal commutation angle online. However, existing online optimization methods still have the following shortcomings: First, most online optimization strategies only target a single performance index (such as efficiency, torque ripple, etc.), making it difficult to achieve optimal comprehensive performance across multiple objectives. Second, online disturbance processes typically employ fixed single-step or unidirectional search mechanisms. When the operating conditions vary significantly, the optimization convergence speed is slow, the number of iterations is high, and it is even prone to getting trapped in local extrema, making it difficult to meet real-time control requirements. Third, existing methods lack the ability to dynamically allocate resources across different search intervals when facing multi-dimensional changes in operating conditions, resulting in low utilization of computational resources and potential interference with the motor control task.

[0004] Based on this, the present invention proposes a method for dynamic adjustment of intelligent commutation timing of brushless motors under multiple operating conditions to solve the above problems. Summary of the Invention

[0005] The purpose of this invention is to provide a method for dynamic adjustment of intelligent commutation timing of brushless motors under multiple operating conditions, so as to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a method for intelligent commutation timing dynamic adjustment of a brushless motor under multiple operating conditions, comprising the following steps: A lookup table for the brushless motor is established, which includes multiple lookup cells. Each lookup cell corresponds to a type of operating condition and a reference commutation angle. The operating conditions include speed, load torque, bus voltage, and temperature. The real-time operating conditions of the brushless motor are obtained, and the corresponding lookup grid and reference commutation angle are determined based on the real-time operating conditions. A virtual model of the brushless motor is pre-built. Based on the virtual model, the reference commutation angle corresponding to the real-time operating conditions is perturbed to obtain the optimal commutation angle. In this process, a calculation center is pre-established, which includes multiple calculation grids. Based on the calculation grids, the reference commutation angle is perturbed in parallel until the perturbation result reaches the preset conditions. The commutation timing is obtained based on the optimal commutation angle, and the commutation action corresponding to the optimal commutation angle is executed based on the commutation timing.

[0007] In a preferred embodiment, the step of establishing a lookup table for the brushless motor, the lookup table comprising multiple lookup cells, each lookup cell corresponding to a type of operating condition and a reference commutation angle, includes: Using speed, load torque, bus voltage, and winding temperature as operating condition characteristic dimensions, the range of multiple operating condition characteristic dimensions is divided according to the operating range, and multiple lookup grids are formed by cross-combination. Each lookup grid corresponds to a type of operating condition of the brushless motor. For each lookup cell and its corresponding operating conditions, the reference commutation angle with the best overall performance value is determined through bench testing or electromagnetic simulation, and then mapped and bound to the lookup cells one by one. The mapping relationship between operating conditions and reference commutation angle is stored in a multidimensional array to obtain a lookup table.

[0008] In a preferred embodiment, the steps of acquiring the real-time operating conditions of the brushless motor and determining the corresponding lookup grid and reference commutation angle based on the real-time operating conditions are as follows: The real-time speed, load torque, bus voltage, and temperature of the brushless motor are collected synchronously to obtain the real-time operating conditions. Determine the lookup cell in the lookup table that corresponds to the real-time operating condition, and use the reference commutation angle in the corresponding lookup cell as the reference commutation angle corresponding to the real-time operating condition.

[0009] In a preferred embodiment, the step of pre-constructing a virtual model of the brushless motor, perturbing the reference commutation angle corresponding to the real-time operating condition based on the virtual model to obtain the optimal commutation angle, wherein a calculation center is pre-established, comprising multiple calculation grids, and the reference commutation angle is perturbed in parallel based on the calculation grids until the perturbation result meets a preset condition includes: Using the physical equations of a brushless motor as the main framework, and superimposing unmodeled dynamics for error compensation, a virtual model of the brushless motor is obtained. The virtual model takes the reference commutation angle as input and outputs a comprehensive performance value, which includes the effective value of phase current, torque ripple coefficient, and system efficiency. A computing center is built inside the controller of the brushless motor. The computing center includes multiple computing cells, which are connected to form two computing chains. The reference commutation angle is perturbed based on a virtual model and two computational chains until a preset condition is met. The commutation angle that meets the preset condition is taken as the optimal commutation angle.

[0010] In a preferred embodiment, the step of constructing a computing center within the controller of the brushless motor, the computing center comprising multiple computing cells connected to form two computing chains, includes: Obtain the single-run resources of the virtual model, and divide the total running resources of the computing center based on the single-run resources to obtain multiple computing grids; Divide multiple computation cells into two equal parts to obtain two sets. Connect the multiple computation cells in the two sets in sequence to obtain a computation chain, wherein the number of computation cells in the two sets is the same.

[0011] In a preferred embodiment, the step of perturbing the reference commutation angle based on the virtual model and two computational chains until a preset condition is met, and taking the commutation angle that meets the preset condition as the optimal commutation angle, includes: Two calculation chains are configured with positive and negative offset directions respectively, and each calculation cell in the calculation chain is configured with a corresponding offset angle. The offset angles of multiple sequentially connected calculation cells in the two calculation chains are set at equal arithmetic progressions. The starting point of both calculation chains is the reference commutation angle corresponding to the real-time operating condition. The virtual model is projected into each computing cell in the two computing chains. The virtual model in each computing cell simulates the brushless motor based on the reference commutation angle and preset offset angle corresponding to the computing cell, so as to perform the initial disturbance. Calculate the overall performance value of each computation cell in the two computation chains, draw the corresponding first trend graph based on the overall performance value of each computation cell in the two computation chains, determine the disturbance interval based on the first trend graph, take the computation chain with the overall performance value of the first trend graph as the main computation chain, and take the computation chain with the overall performance value of the first trend graph as the secondary computation chain. The secondary calculation chain is merged with the main calculation chain to obtain a merged chain. The disturbance interval is subjected to a second disturbance based on the merged chain until the preset conditions are met. The commutation angle that meets the preset conditions is taken as the optimal commutation angle. When the optimal commutation angle is reached, the merged chain is restored to two calculation chains.

[0012] In a preferred embodiment, the steps of merging the secondary computation chain with the main computation chain to obtain a merged chain, performing a secondary perturbation on the perturbation interval based on the merged chain until a preset condition is reached, taking the commutation angle that meets the preset condition as the optimal commutation angle, and restoring the merged chain to two computation chains when the optimal commutation angle is reached include: Using the disturbance interval as the upper and lower boundaries, two calculation cells corresponding to the upper and lower boundaries are determined, namely the upper calculation cell and the lower calculation cell. All computation cells in the two computation chains except for the upper and lower computation cells are transferred to the space between the upper and lower computation cells and connected sequentially to form a merge chain; The offset angle is re-adapted within each computation cell in the merged chain. The offset angle corresponding to each computation cell in the merged chain is simulated based on the virtual model projected within each computation cell. The comprehensive performance value of each computation cell in the merged chain is calculated. The corresponding second trend graph is plotted based on multiple comprehensive performance values. Based on the second trend graph, the sum of the reference commutation angle and the offset angle corresponding to the computation cell that meets the preset conditions is determined as the optimal commutation angle. After determining the optimal commutation angle, the merged chain is restored to two computation chains.

[0013] In a preferred embodiment, the step of obtaining the commutation timing based on the optimal commutation angle and executing the commutation action corresponding to the optimal commutation angle based on the commutation timing includes: Obtain the optimal commutation angle and the commutation trigger angle, and calculate the commutation time of each rotor sector based on the commutation trigger angle; The commutation time is written into the controller timer to generate six matching PWM drive signals, which control the power transistor to perform the corresponding commutation action.

[0014] The technical effects and advantages provided by the present invention in the above technical solution are as follows: 1. This invention uses two computational chains, each carrying a set of offset directions, to output positive and negative disturbance simulation results respectively, thereby quickly outputting the disturbance range and rapidly narrowing down the range of the optimal commutation angle. By merging the two computational chains into one merged chain, all resources of the computing center can be concentrated to optimize within the disturbance range, facilitating the rapid finding of the optimal commutation angle. After finding the optimal commutation angle, the merged chain is restored to two computational chains, enabling the two computational chains to continuously optimize disturbances for subsequent real-time operating conditions. 2. By setting up two computing chains, this invention can quickly merge the two computing chains after determining the disturbance range of the reference commutation angle, and restore them to two computing chains after determining the optimal commutation angle. This allows for more accurate utilization of computing resources and maintains extremely low resource switching overhead during the merging and restoration of the two computing chains. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0016] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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, 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.

[0018] Example 1, please refer to Figure 1 As shown in this embodiment, a method for dynamic adjustment of intelligent commutation timing of a brushless motor under multiple operating conditions includes the following steps: A lookup table for the brushless motor is established, which includes multiple lookup cells. Each lookup cell corresponds to a type of operating condition and a reference commutation angle. The operating conditions include speed, load torque, bus voltage, and temperature. The real-time operating conditions of the brushless motor are obtained, and the corresponding lookup grid and reference commutation angle are determined based on the real-time operating conditions. A virtual model of the brushless motor is pre-built. Based on the virtual model, the reference commutation angle corresponding to the real-time operating conditions is perturbed to obtain the optimal commutation angle. In this process, a calculation center is pre-established, which includes multiple calculation grids. Based on the calculation grids, the reference commutation angle is perturbed in parallel until the perturbation result reaches the preset conditions. The commutation timing is obtained based on the optimal commutation angle, and the commutation action corresponding to the optimal commutation angle is executed based on the commutation timing.

[0019] In this embodiment, the present invention uses two computational chains, each carrying a set of offset directions. The two chains output positive and negative disturbance simulation results respectively, enabling rapid output of the disturbance range and quickly narrowing down the range of the optimal commutation angle. By merging the two computational chains into a single merged chain, all resources of the computing center can be concentrated on optimization within the disturbance range, facilitating the rapid finding of the optimal commutation angle. After finding the optimal commutation angle, the merged chain is restored to two computational chains, allowing for uninterrupted disturbance optimization for subsequent real-time operating conditions. This invention, through the setup of two computational chains, can quickly merge the two chains after determining the disturbance range of the reference commutation angle, and restore them to two chains again after determining the optimal commutation angle. This allows for more precise utilization of computing resources while maintaining extremely low resource switching overhead during the merging and restoration of the two computational chains.

[0020] In one embodiment, the step of establishing a lookup table for the brushless motor, the lookup table including multiple lookup cells, each lookup cell corresponding to a type of operating condition and a reference commutation angle, includes: Using speed, load torque, bus voltage, and winding temperature as operating condition characteristic dimensions, the range of multiple operating condition characteristic dimensions is divided according to the operating range, and multiple lookup grids are formed by cross-combination. Each lookup grid corresponds to a type of operating condition of the brushless motor. For each lookup cell and its corresponding operating conditions, the reference commutation angle with the best overall performance value is determined through bench testing or electromagnetic simulation, and then mapped and bound to the lookup cells one by one. The mapping relationship between operating conditions and reference commutation angle is stored in a multidimensional array to obtain a lookup table.

[0021] It should be noted that four core parameters—speed, load torque, bus voltage, and winding temperature—are used as operating condition characteristic dimensions. Each dimension is further divided into graded intervals according to the actual operating range of the project. For example, the speed dimension is divided into four levels: low speed, medium speed, high speed, and low field speed; the load torque is divided into four levels: light load, medium load, rated load, and overload; the bus voltage is divided into three levels: undervoltage, rated voltage, and overvoltage; and the winding temperature is divided into three levels: low temperature, normal temperature, and high temperature. The various dimensions are combined to form a multi-dimensional operating condition grid. Each grid cell is a lookup cell, corresponding to a type of operating condition interval, rather than a single parameter point. This ensures that even with small fluctuations in parameters during actual operation, the corresponding grid can still be stably matched. For each lookup cell corresponding to an operating condition range, the reference commutation angle is calibrated through motor bench testing or finite element electromagnetic simulation. Under typical parameter points of that operating condition, the optimization objective is to maximize the comprehensive performance value calculated from the effective value of phase current, torque ripple coefficient, and system efficiency. The motor operating performance under different commutation advance angles is tested traversally, and the commutation angle with the best comprehensive performance is selected as the reference commutation angle for that lookup cell and written into the lookup table. The lookup table is stored in the controller's memory unit in the form of a multidimensional array, supporting fast addressing via the gear code of the operating condition parameters. Simultaneously, a boundary condition fallback mechanism is reserved; when real-time parameters exceed the calibration range, the nearest boundary lookup cell is automatically matched to avoid lookup failure. Furthermore, the lookup table setting provides a benchmark for subsequent disturbances to the optimal commutation angle, shortening the optimization time.

[0022] In one embodiment, the steps of obtaining the real-time operating conditions of the brushless motor and determining the corresponding lookup grid and reference commutation angle based on the real-time operating conditions are as follows: The real-time speed, load torque, bus voltage, and temperature of the brushless motor are collected synchronously to obtain the real-time operating conditions. Determine the lookup cell in the lookup table that corresponds to the real-time operating condition, and use the reference commutation angle in the corresponding lookup cell as the reference commutation angle corresponding to the real-time operating condition.

[0023] It should be noted that the rotational speed is obtained by differential calculation after acquiring the rotor position through Hall sensors or encoders; the load torque is estimated by combining the phase current amplitude with the flux linkage observer, without the need for an additional torque sensor; the bus voltage is obtained by sampling through the voltage divider circuit at the front end of the controller; the winding temperature is acquired by the NTC thermistor built into the brushless motor, and the sampling frequency is adapted to the temperature change rate; the four types of operating condition parameters acquired in real time are compared with the gear ranges of each dimension to determine the gear combination to which the current operating condition belongs, and the corresponding lookup cell and reference commutation angle are located in the lookup table through the addressing logic; by comparing the real-time operating condition with the lookup table to obtain the corresponding lookup cell and reference commutation angle, it is easier to narrow down the disturbance range corresponding to the subsequent acquisition of the optimal commutation angle.

[0024] In one embodiment, the step of pre-constructing a virtual model of the brushless motor, perturbing the reference commutation angle corresponding to the real-time operating condition based on the virtual model to obtain the optimal commutation angle, wherein a calculation center is pre-established, comprising multiple calculation grids, and the reference commutation angle is perturbed in parallel based on the calculation grids until the perturbation result meets a preset condition includes: Using the physical equations of a brushless motor as the main framework, and superimposing unmodeled dynamics for error compensation, a virtual model of the brushless motor is obtained. The virtual model takes the reference commutation angle as input and outputs a comprehensive performance value, which includes the effective value of phase current, torque ripple coefficient, and system efficiency. A computing center is built inside the controller of the brushless motor. The computing center includes multiple computing cells, which are connected to form two computing chains. The reference commutation angle is perturbed based on a virtual model and two computational chains until a preset condition is met. The commutation angle that meets the preset condition is taken as the optimal commutation angle.

[0025] In one embodiment, the step of constructing a computing center within the controller of the brushless motor, the computing center comprising multiple computing cells connected to form two computing chains, includes: Obtain the single-run resources of the virtual model, and divide the total running resources of the computing center based on the single-run resources to obtain multiple computing grids; Divide multiple computation cells into two equal parts to obtain two sets. Connect the multiple computation cells in the two sets in sequence to obtain a computation chain, wherein the number of computation cells in the two sets is the same.

[0026] It should be noted that the physical equations of the brushless motor are mainly based on the voltage balance equation, electromagnetic torque equation, and mechanical motion equation in a two-phase rotating coordinate system. A virtual model can be directly built using known parameters such as stator resistance, phase inductance, and permanent magnet flux linkage. Unmodeled dynamic compensation terms are used to correct nonlinear characteristics that the physical equations cannot accurately cover. These include inverter dead-zone effects, stator cogging torque, winding eddy current losses, and bearing friction losses. The compensation parameters are obtained by fitting error data from bench tests, which can control the simulation error within the engineering allowable range and meet the accuracy requirements of online optimization. The specific method for calculating the comprehensive performance value is as follows: the reciprocals of system efficiency, torque ripple, and effective phase current are mapped to the [0,1] interval, and then weighted and summed according to preset weights to obtain the comprehensive performance value. A higher value indicates better overall motor performance; for example, the system efficiency weight is set to 0.5, the torque ripple weight to 0.3, and the effective phase current weight to 0.2.

[0027] The single-run resource of the virtual model refers to the controller computing resources required to complete one commutation angle performance evaluation, including fixed RAM usage, CPU clock cycles, and computing stack space. The total computing center's operating resources are dedicated resources pre-allocated by the controller to the commutation angle optimization task, and do not occupy the computing power of core control tasks such as the motor current loop and speed loop. Virtual model projection essentially involves copying and loading the same reduced-order model algorithm into the computing resources of each computing cell. All computing cells share the current real-time operating parameters (speed, load torque, bus voltage, winding temperature), differing only in the input commutation angle offset angle. Each computing cell independently completes a full simulation calculation, outputting the corresponding effective value of phase current, torque ripple coefficient, and system efficiency—three raw indicators. The calculation processes are independent and data is isolated, with no resource contention or computational conflicts. Each computing cell is a logically independent computing unit, capable of independently handling a complete virtual model calculation. Resource isolation and non-interference between units avoid data conflicts during parallel computing. Initially, the two computation chains are independent and topologically symmetrical, each possessing half of the computational resources. Computational cells are sequentially linked within the chain according to their offset angles, forming a traversable computational link. For example, with a total of 10 computational cells, after being bisected, each computational chain contains 5 computational cells, and the two chains can start computation simultaneously.

[0028] In one embodiment, the step of perturbing the reference commutation angle based on a virtual model and two computational chains until a preset condition is met, and then taking the commutation angle that meets the preset condition as the optimal commutation angle, includes: Two calculation chains are configured with positive and negative offset directions respectively, and each calculation cell in the calculation chain is configured with a corresponding offset angle. The offset angles of multiple sequentially connected calculation cells in the two calculation chains are set at equal arithmetic progressions. The starting point of both calculation chains is the reference commutation angle corresponding to the real-time operating condition. The virtual model is projected into each computing cell in the two computing chains. The virtual model in each computing cell simulates the brushless motor based on the reference commutation angle and preset offset angle corresponding to the computing cell, so as to perform the initial disturbance. Calculate the overall performance value of each computation cell in the two computation chains, draw the corresponding first trend graph based on the overall performance value of each computation cell in the two computation chains, determine the disturbance interval based on the first trend graph, take the computation chain with the overall performance value of the first trend graph as the main computation chain, and take the computation chain with the overall performance value of the first trend graph as the secondary computation chain. The secondary calculation chain is merged with the main calculation chain to obtain a merged chain. The disturbance interval is subjected to a second disturbance based on the merged chain until the preset conditions are met. The commutation angle that meets the preset conditions is taken as the optimal commutation angle. When the optimal commutation angle is reached, the merged chain is restored to two calculation chains.

[0029] In one embodiment, the steps of merging the secondary computation chain with the main computation chain to obtain a merged chain, performing a secondary perturbation on the perturbation interval based on the merged chain until a preset condition is reached, taking the commutation angle that meets the preset condition as the optimal commutation angle, and restoring the merged chain to two computation chains when the optimal commutation angle is reached include: Using the disturbance interval as the upper and lower boundaries, two calculation cells corresponding to the upper and lower boundaries are determined, namely the upper calculation cell and the lower calculation cell. All computation cells in the two computation chains except for the upper and lower computation cells are transferred to the space between the upper and lower computation cells and connected sequentially to form a merge chain; The offset angle is re-adapted within each computation cell in the merged chain. The offset angle corresponding to each computation cell in the merged chain is simulated based on the virtual model projected within each computation cell. The comprehensive performance value of each computation cell in the merged chain is calculated. The corresponding second trend graph is plotted based on multiple comprehensive performance values. Based on the second trend graph, the sum of the reference commutation angle and the offset angle corresponding to the computation cell that meets the preset conditions is determined as the optimal commutation angle. After determining the optimal commutation angle, the merged chain is restored to two computation chains.

[0030] It should be noted that the positive offset direction corresponds to the commutation angle advance adjustment direction, that is, the commutation time is advanced after superimposing the preset offset angle; the negative offset direction corresponds to the commutation angle lag adjustment direction, that is, the commutation time is delayed after superimposing the preset offset angle. The offset angles are all in electrical degrees. For example, with the reference commutation angle as the zero point, the preset offset angles of the 5 calculation cells of the positive calculation chain are set to 1°, 2°, 3°, 4°, and 5° respectively, with a step size of 1°; the preset offset angles of the 5 calculation cells of the negative calculation chain are set to -1°, -2°, -3°, -4°, and -5° respectively, with a step size of 1° as well. In the initial stage, a large arithmetic step size is used to quickly cover the search intervals in both positive and negative directions and to initially determine the distribution direction of the optimal commutation angle. During the initial perturbation, the two computation chains use the reference commutation angle as a common zero point and superimpose a preset angle along their own offset direction: the positive computation chain calculates the comprehensive performance value corresponding to "reference commutation angle + 1°", "reference commutation angle + 2°"... "reference commutation angle + 5°" in sequence, while the negative computation chain calculates the comprehensive performance value corresponding to "reference commutation angle - 1°", "reference commutation angle - 2°"... "reference commutation angle - 5°" simultaneously. Regarding the acquisition of the first trend graph and the disturbance range: Using the reference commutation angle as the origin of the horizontal axis, the preset offset angle of each calculation cell as the horizontal axis coordinate, and the corresponding comprehensive performance value as the vertical axis coordinate, an angle-performance curve is plotted to obtain the first trend graph. The calculation chain in the positive offset direction corresponds to the performance change trend of increasing offset angle, while the calculation chain in the negative offset direction corresponds to the performance change trend of decreasing offset angle. If the comprehensive performance value of the calculation chain in the positive offset direction increases with the offset angle (an upward trend), then this chain is marked as the main calculation chain, indicating that the optimal commutation angle is in the direction of increasing offset angle; conversely, if the calculation chain in the negative offset direction shows an upward trend, it is marked as the main calculation chain, indicating that the optimal commutation angle is in the direction of decreasing offset angle. The disturbance interval is specifically defined as follows: based on the first trend map obtained from the initial disturbance, the rate of change of the comprehensive performance value between two adjacent calculation cells is calculated, and the two adjacent calculation cells with the largest absolute value of the rate of change (i.e. at the peak value) are selected. The angle range after superimposing the reference commutation angles corresponding to the two calculation cells with their respective offset angles is determined as the disturbance interval. The upper and lower boundaries of the perturbation interval are the superposition angles corresponding to the two computation grids, with the larger one being the upper boundary and the smaller one being the lower boundary. For example, if the perturbation interval is from the reference angle to the reference angle +5°, then the lower computation grid is the computation grid corresponding to the offset of 0°, and the upper computation grid is the computation grid corresponding to the offset of +5°. These two computation grids are retained as boundary nodes of the merged chain. Subsequently, all the computation grids in the two computation chains except for the upper and lower computation grids are transferred to the space between the upper and lower computation grids and connected in series in ascending order of offset angle to form a single merged chain. The total number of computation grids in the merged chain is consistent with the total number of the initial two computation chains. Only through resource reorganization (merging) are all the computation grids originally distributed in the invalid interval transferred to the perturbation interval, thereby improving the search density. For example, initially there are a total of 10 calculation cells. Two calculation cells are reserved at the top and bottom boundaries, and the remaining 8 calculation cells are filled into the perturbation range of 0° to 5°, so that the number of calculation cells in the perturbation range is increased from the original 2 to 10, doubling the search density, which makes it easier to determine the optimal commutation angle. The merged chain is perturbed again (secondary perturbation, i.e., the virtual model is projected onto each calculation cell in the merged chain with a preset offset angle), and the comprehensive performance value corresponding to each calculation cell in the merged chain is obtained. The corresponding angle-performance curve is plotted (in the same way as the curve plotted in the initial perturbation), resulting in a second trend graph. The rate of change between two adjacent calculation cells is calculated, and the superposition angle (reference commutation angle + preset offset angle) corresponding to the calculation cell with the comprehensive performance value at its peak and the rate of change decreasing is taken as the optimal commutation angle (find the commutation angle that meets the preset conditions, and take its superposition value with the reference commutation angle as the optimal commutation angle for the current working condition). After determining the optimal commutation angle, the merged chain is split back into two calculation chains and reset to the initial standby state. Among them, the offset angle of the merged chain is optimized by using a smaller arithmetic step size to achieve high-precision fine search; for example, 10 calculation cells are distributed in the perturbation range of 0°~5°, corresponding to a step size of about 0.56°, which significantly improves the optimization accuracy compared to the initial coarse step size of 1°. The preset conditions are as follows: In the trend graph, when traversing the calculation grid along the direction of increasing or decreasing offset angle, if the comprehensive performance value first appears in any of the following situations, it is determined that the preset conditions are met: (1) The comprehensive performance value stops increasing with the same direction of offset angle, that is, the increment of the comprehensive performance value of adjacent calculation grids changes from positive to negative or tends to zero; (2) The rate of change of comprehensive performance value continues to decrease and is lower than the preset threshold, indicating that the performance improvement tends to saturate; When the preset conditions are met, the commutation angle corresponding to the peak point is the optimal commutation angle under the current working condition. After optimization is completed (i.e., the optimal commutation angle is obtained), it is split and restored into two calculation chains, so that bidirectional parallel coarse search can be directly started for the next real-time running condition without rebuilding the link topology, ensuring the response speed of the next optimization.

[0031] Furthermore, by having two computational chains each carry a set of offset directions, and outputting positive and negative disturbance simulation results respectively, the disturbance range can be quickly output, thereby rapidly narrowing down the range of the optimal commutation angle. By merging the two computational chains into a single merged chain, all resources of the computational center can be concentrated on optimization within the disturbance range to quickly find the optimal commutation angle. After finding the optimal commutation angle, the merged chain is restored to two computational chains, allowing for continuous disturbance optimization of subsequent real-time operating conditions. By continuously optimizing the real-time operating conditions, the optimal commutation angle can be directly prepared for application in the next commutation operation, facilitating timely commutation of the brushless motor using the optimal commutation angle.

[0032] Specifically, the preset offset templates are divided into two categories based on the disturbance stage (initial disturbance and secondary disturbance): coarse search templates and fine search templates. The coarse search template uses a larger step size (e.g., 1°) to quickly determine the disturbance range and the direction of the optimal commutation angle during the initial disturbance. The fine search template uses a smaller step size (e.g., 0.56°) and is called back and overwritten with the original offset angle after the merged chain is formed, performing a high-precision secondary search on the disturbance range. The two adaptations share the same set of preset offset template distribution-bit allocation-independent execution operation flow, with only the parameters of the preset offset templates being called being different. The essence of the computing grid is a virtual machine or instance. Connection and merging are not physical lines or network connections, but rather logical pointer rearrangement and task queue reorganization. The specific mechanism is as follows: Initial connection (forming a double chain): The controller maintains a global index array. The IDs (e.g., ID0~ID9) of the 10 VM instances (computing grids) are stored in the array in order of offset angle, forming a linear logical chain that is traversed sequentially. For example, the array of forward computation chains is [VM3, VM5, VM1...], and the sorting within the chain is based on its assigned offset angle. Merging into a single chain (between boundaries): This is a cut-and-paste process of logical pointers, not a physical movement of the VMs. Locking boundaries: Determine the upper computation cell (e.g., VM_A) and the lower computation cell (e.g., VM_B) as the fixed head and tail nodes of the merged chain. Reorganizing the index: Extract all other VM instance IDs from the two original computation chains except for VM_A and VM_B, and clear their original offset angle assignments. Re-chaining: Sort these extracted VM instance IDs according to the preset offset angle from smallest to largest, and move them back into the logical interval between VM_A and VM_B, forming a new single linear linked list (i.e., the merged chain). The entire process only modifies the task scheduling table in the controller; the VM entities themselves remain unchanged in memory. The adaptation process is a parameter injection process for the offset angle driven by the main control logic. Specifically: Preset offset template issuance: The main controller reads the preset offset template (e.g., coarse search template: step size 1°, 5 grids) and calculates the total step size range to be covered in this optimization (e.g., +1°~+5°). Bit-by-bit allocation (angle mapping): Based on the sequence number of each VM instance in the current calculation chain (1st, 2nd, ...), the calculated offset angles (e.g., +1°, +2°, ...) are written to the dedicated input registers of the corresponding VM instances. This is a one-to-one mapping; the main controller directly assigns values ​​to each VM instance through address offset. Independent execution: After receiving the offset angle from the main controller, each VM instance, combined with the shared reference commutation angle, independently starts virtual model calculations, outputs performance values, and stores them back in its dedicated output register, thus completing the adaptation of the offset angle and the calculation grid.

[0033] In one embodiment, the step of obtaining the commutation timing based on the optimal commutation angle and executing the commutation action corresponding to the optimal commutation angle based on the commutation timing includes: Obtain the optimal commutation angle and the commutation trigger angle, and calculate the commutation time of each rotor sector based on the commutation trigger angle; The commutation time is written into the controller timer to generate six matching PWM drive signals, which control the power transistor to perform the corresponding commutation action.

[0034] It should be noted that the six-step commutation of the brushless motor corresponds to six rotor position sectors and six commutation points. The optimal commutation angle is essentially the commutation trigger angle value of each commutation point. Based on the current rotor position and the optimal commutation angle, the commutation time for each step is calculated: if it is an early commutation, the commutation is triggered at the corresponding angle ahead of the original commutation position; if it is a delayed commutation, the trigger is delayed by the corresponding angle, completing the conversion from angle value to timing node. The calculated commutation time is written into the controller's timer comparison register, and the switching state of the three-phase bridge arm is updated in real time within each PWM cycle, generating six PWM drive signals matching the optimal commutation angle to control the on and off of the power transistors, ultimately executing the corresponding commutation action. The update of the commutation timing is synchronized with the motor's main control interrupt to ensure the real-time performance of angle adjustment. Subsequently, an independent optimal value buffer is established to bind and store the optimal commutation angle obtained from each optimization with the corresponding operating condition (lookup grid). When the brushless motor re-enters the cached operating range, it directly reads the historically optimal commutation angle from the cache, skipping the complete disturbance optimization process, further reducing computational latency and improving dynamic response speed. If an optimization failure occurs (e.g., abnormal virtual model calculation, iteration exceeding the limit, non-convergence), to ensure the continuous and stable operation of the brushless motor, commutation can be performed at the time corresponding to a pre-calibrated reference commutation angle, thus avoiding operational failures caused by optimization failures.

[0035] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for intelligent commutation timing dynamic adjustment of a brushless motor under multiple operating conditions, characterized in that, Includes the following steps: A lookup table for the brushless motor is established, which includes multiple lookup cells. Each lookup cell corresponds to a type of operating condition and a reference commutation angle. The operating conditions include speed, load torque, bus voltage, and temperature. The real-time operating conditions of the brushless motor are obtained, and the corresponding lookup grid and reference commutation angle are determined based on the real-time operating conditions. A virtual model of the brushless motor is pre-built. Based on the virtual model, the reference commutation angle corresponding to the real-time operating conditions is perturbed to obtain the optimal commutation angle. In this process, a calculation center is pre-established, which includes multiple calculation grids. Based on the calculation grids, the reference commutation angle is perturbed in parallel until the perturbation result reaches the preset conditions. The commutation timing is obtained based on the optimal commutation angle, and the commutation action corresponding to the optimal commutation angle is executed based on the commutation timing.

2. The method for intelligent commutation timing dynamic adjustment of a brushless motor under multiple operating conditions according to claim 1, characterized in that, The step of establishing a lookup table for the brushless motor, wherein the lookup table includes multiple lookup cells and each lookup cell corresponds to a type of operating condition and a reference commutation angle, includes: Using speed, load torque, bus voltage, and winding temperature as operating condition characteristic dimensions, the range of multiple operating condition characteristic dimensions is divided according to the operating range, and multiple lookup grids are formed by cross-combination. Each lookup grid corresponds to a type of operating condition of the brushless motor. For each lookup cell and its corresponding operating conditions, the reference commutation angle with the best overall performance value is determined through bench testing or electromagnetic simulation, and then mapped and bound to the lookup cells one by one. The mapping relationship between operating conditions and reference commutation angle is stored in a multidimensional array to obtain a lookup table.

3. The method for intelligent commutation timing dynamic adjustment of a brushless motor under multiple operating conditions according to claim 1, characterized in that, The steps of obtaining the real-time operating conditions of the brushless motor and determining the corresponding lookup grid and reference commutation angle based on the real-time operating conditions are as follows: The real-time speed, load torque, bus voltage, and temperature of the brushless motor are collected synchronously to obtain the real-time operating conditions. Determine the lookup cell in the lookup table that corresponds to the real-time operating condition, and use the reference commutation angle in the corresponding lookup cell as the reference commutation angle corresponding to the real-time operating condition.

4. The method for intelligent commutation timing dynamic adjustment of a brushless motor under multiple operating conditions according to claim 1, characterized in that, The process of pre-constructing a virtual model of the brushless motor, perturbing the reference commutation angle corresponding to the real-time operating conditions based on the virtual model to obtain the optimal commutation angle, includes the following steps: A pre-established computational center comprising multiple computational cells is used to perturb the reference commutation angle in parallel based on the computational cells until the perturbation result meets preset conditions. Using the physical equations of a brushless motor as the main framework, and superimposing unmodeled dynamics for error compensation, a virtual model of the brushless motor is obtained. The virtual model takes the reference commutation angle as input and outputs a comprehensive performance value, which includes the effective value of phase current, torque ripple coefficient, and system efficiency. A computing center is built inside the controller of the brushless motor. The computing center includes multiple computing cells, which are connected to form two computing chains. The reference commutation angle is perturbed based on a virtual model and two computational chains until a preset condition is met. The commutation angle that meets the preset condition is taken as the optimal commutation angle.

5. The method for intelligent commutation timing dynamic adjustment of a brushless motor under multiple operating conditions according to claim 4, characterized in that, The step of constructing a computing center within the controller of the brushless motor, the computing center comprising multiple computing cells connected to form two computing chains, includes: Obtain the single-run resources of the virtual model, and divide the total running resources of the computing center based on the single-run resources to obtain multiple computing grids; Divide multiple computation cells into two equal parts to obtain two sets. Connect the multiple computation cells in the two sets in sequence to obtain a computation chain, wherein the number of computation cells in the two sets is the same.

6. The method for intelligent commutation timing dynamic adjustment of a brushless motor under multiple operating conditions according to claim 5, characterized in that, The step of perturbing the reference commutation angle based on a virtual model and two computational chains until a preset condition is met, and taking the commutation angle that meets the preset condition as the optimal commutation angle, includes: Two calculation chains are configured with positive and negative offset directions respectively, and each calculation cell in the calculation chain is configured with a corresponding offset angle. The offset angles of multiple sequentially connected calculation cells in the two calculation chains are set at equal arithmetic progressions. The starting point of both calculation chains is the reference commutation angle corresponding to the real-time operating condition. The virtual model is projected into each computing cell in the two computing chains. The virtual model in each computing cell simulates the brushless motor based on the reference commutation angle and preset offset angle corresponding to the computing cell, so as to perform the initial disturbance. Calculate the overall performance value of each computation cell in the two computation chains, draw the corresponding first trend graph based on the overall performance value of each computation cell in the two computation chains, determine the disturbance interval based on the first trend graph, take the computation chain with the overall performance value of the first trend graph as the main computation chain, and take the computation chain with the overall performance value of the first trend graph as the secondary computation chain. The secondary calculation chain is merged with the main calculation chain to obtain a merged chain. The disturbance interval is subjected to a second disturbance based on the merged chain until the preset conditions are met. The commutation angle that meets the preset conditions is taken as the optimal commutation angle. When the optimal commutation angle is reached, the merged chain is restored to two calculation chains.

7. The method for intelligent commutation timing dynamic adjustment of a brushless motor under multiple operating conditions according to claim 6, characterized in that, The steps of merging the secondary computation chain with the main computation chain to obtain a merged chain, performing a secondary perturbation on the perturbation interval based on the merged chain until a preset condition is reached, taking the commutation angle that meets the preset condition as the optimal commutation angle, and restoring the merged chain to two computation chains when the optimal commutation angle is reached include: Using the disturbance interval as the upper and lower boundaries, two calculation cells corresponding to the upper and lower boundaries are determined, namely the upper calculation cell and the lower calculation cell. All computation cells in the two computation chains except for the upper and lower computation cells are transferred to the space between the upper and lower computation cells and connected sequentially to form a merge chain; The offset angle is re-adapted within each computation cell in the merged chain. The offset angle corresponding to each computation cell in the merged chain is simulated based on the virtual model projected within each computation cell. The comprehensive performance value of each computation cell in the merged chain is calculated. The corresponding second trend graph is plotted based on multiple comprehensive performance values. Based on the second trend graph, the sum of the reference commutation angle and the offset angle corresponding to the computation cell that meets the preset conditions is determined as the optimal commutation angle. After determining the optimal commutation angle, the merged chain is restored to two computation chains.

8. The method for intelligent commutation timing dynamic adjustment of a brushless motor under multiple operating conditions according to claim 7, characterized in that, The steps of obtaining the commutation timing based on the optimal commutation angle and executing the commutation action corresponding to the optimal commutation angle based on the commutation timing include: Obtain the optimal commutation angle and the commutation trigger angle, and calculate the commutation time of each rotor sector based on the commutation trigger angle; The commutation time is written into the controller timer to generate six matching PWM drive signals, which control the power transistor to perform the corresponding commutation action.

Images