Parameter compensation method for flux observer of permanent magnet synchronous motor based on improved beluga optimization algorithm

Abstract

The application discloses a permanent magnet synchronous motor flux linkage observer parameter compensation method based on an improved hybrid beluga optimization algorithm, proposes a customized improved HBWO algorithm for a flux linkage observation scene, and realizes 3-dimensional compensation factors through the algorithm. Firstly, the harmonic compensation factor is used for harmonic cancellation of the shaft voltage signal, and high-frequency interference in the electric signal is suppressed. Secondly, the resistance compensation factor is used for dynamic correction of the stator resistance parameter, and resistance drift caused by motor temperature change is adapted. Thirdly, the inductance compensation factor is used for real-time updating of the stator inductance parameter, and inductance change caused by load fluctuation is matched. The optimization factor is embedded into an error preprocessing link and a parameter correction link of the flux linkage observer, noise reduction of the observer input signal and adaptive updating of core parameters are realized. Compared with the prior art, the application effectively improves the precision and stability of the flux linkage observer, and is suitable for high-precision control scenes of the permanent magnet synchronous motor under the conditions of medium and high speed and variable load.

Description

Technical Field

[0001] This invention relates to the field of intelligent control and optimization technology, specifically to a high-precision parameter compensation method for a permanent magnet synchronous motor flux linkage observer based on an improved white whale optimization algorithm. Background Technology

[0002] Permanent magnet synchronous motors (PMSMs) are widely used in high-precision control fields such as new energy vehicles, industrial robots, and aerospace due to their advantages of high efficiency, high power density, and high reliability. Flux linkage monitoring is a core component of PSM vector control, and its accuracy directly affects the motor's control performance. However, in actual operation, the flux linkage monitor is susceptible to several factors that can degrade its accuracy: firstly... Harmonic interference in the shaft voltage signal can cause distortion in flux linkage estimation; secondly, changes in motor operating temperature can cause stator resistance drift, resulting in a mismatch between the observer parameters and the actual motor parameters; and thirdly, load fluctuations can cause stator inductance magnetic saturation, further reducing the accuracy of flux linkage observation.

[0003] To address the aforementioned issues, existing technologies often employ optimization algorithms to find the optimal compensation parameters for flux linkage observers. However, conventional optimization algorithms suffer from several engineering shortcomings: First, the simplistic population initialization method leads to insufficient population diversity, making the algorithm prone to local optima and unable to find the optimal compensation factor suitable for all operating conditions. Second, the varying importance of different compensation factors is not considered, causing the optimization direction to deviate from actual needs and resulting in insufficient accuracy. Third, convergence stagnation is prone to occur during the iteration process, making it unable to adapt to the dynamic changes in motor operating conditions. Fourth, the fixed fitness function prevents dynamic adjustment of the optimization objective based on changes in operating conditions, resulting in poor adaptability to operating conditions.

[0004] The White Whale Optimization Algorithm (HBWO) is a novel intelligent optimization algorithm with advantages such as a simple optimization mechanism and fast convergence speed. However, when directly applied to the optimization of compensation parameters for magnetic flux observers, it still suffers from the aforementioned shortcomings and cannot meet the requirements of high-precision control. Therefore, it is necessary to design an optimization algorithm customized for magnetic flux observation scenarios to achieve efficient and accurate optimization of the 3D compensation factor and improve the stability and accuracy of the magnetic flux observer under complex operating conditions. Summary of the Invention

[0005] Purpose of the invention: To address the problems in existing technologies where flux linkage observers are susceptible to harmonic interference and parameter drift, and where conventional optimization algorithms suffer from insufficient population diversity, poor adaptability to operating conditions, and easy iteration stagnation, this invention provides a parameter compensation method for permanent magnet synchronous motor flux linkage observers based on an improved white whale optimization algorithm. By making multi-dimensional improvements to the traditional white whale optimization algorithm, the global optimal search and dynamic adaptation of the 3D compensation factor are achieved, thereby improving the accuracy and stability of flux linkage observation.

[0006] Technical Solution: This invention discloses a parameter compensation method for permanent magnet synchronous motor flux linkage observers based on an improved beluga optimization algorithm, comprising the following steps:

[0007] S1: Construct a 3D compensation factor optimization objective for the flux linkage observer of a permanent magnet synchronous motor, including harmonic compensation factors. Resistance compensation factor Inductance compensation factor , respectively used to offset Shaft voltage 5th / 7th harmonic interference, correcting stator resistance drift caused by temperature, and adapting to inductor magnetic saturation caused by load fluctuations;

[0008] S2: Design an improved beluga whale optimization algorithm HBWO for magnetic flux observation scenarios. The HBWO algorithm is used to achieve global optimal search for the 3D compensation factor. The HBWO algorithm adopts a hybrid population initialization improvement for population initialization, and introduces a weighted feeding strategy guided by factor importance weight and local neighborhood optimality. A chaotic perturbation strategy is designed to accurately identify the iteration stagnation state and apply controllable chaotic perturbation.

[0009] S3: Embed the optimal 3D compensation factor output by the HBWO algorithm into the signal processing link of the flux linkage observer, through... magnetic flux observer The shaft voltage signal undergoes harmonic cancellation processing, through The stator resistance parameters of the observer are dynamically corrected by means of... Dynamically adapt the stator inductance parameters of the observer;

[0010] S4: Collect real-time operating data of the permanent magnet synchronous motor, input it into the HBWO algorithm for iterative optimization, and update the 3D compensation factor in real time.

[0011] Furthermore, a dynamic optimization strategy is constructed using the HBWO algorithm to optimize the 3D compensation factor. , , It can autonomously adjust the value range and optimization weights according to the real-time operating conditions of the permanent magnet synchronous motor, achieving adaptive adaptation across the entire speed range; specifically, it includes the following processes:

[0012] First, the dynamic range of the 3D compensation factor is defined, and an adaptive fitness weighting function for different working conditions is designed; the fitness weighting function is as follows:

[0013] ;

[0014] in, This represents the flux linkage estimation error; Standardized deviation of resistance drift; Standardized deviation for inductor drift; This is a quantification term for harmonic suppression effect, associated with the harmonic compensation factor. , used to characterize right Harmonic cancellation effect of shaft voltage; , These are the flux linkage error weight and the parameter drift deviation weight, respectively, which satisfy... .

[0015] Furthermore, the improved beluga whale optimization algorithm HBWO includes:

[0016] Opposing learning strategies and It integrates chaotic initialization strategies and is specifically extended to adapt to 3D compensation factors and harmonic compensation factors. Resistance compensation factor Inductance compensation factor The population initialization process;

[0017] 1) First, the population size is determined. To address the problem of limited coverage of basic random initialization parameters and the potential for missing optimal solutions at parameter boundaries in simulation models, an opposition learning strategy is introduced, and a formula is constructed in conjunction with upper and lower bound constraints:

[0018] ;

[0019] in, The parameters are those obtained after learning the opposite. The lower boundary; The upper boundary; The first one obtained by random initialization based on the base The first individual in the population Dimensional compensation factor parameters;

[0020] By using parametric symmetric mapping, the coverage of the population is expanded, filling the distribution gap of basic random initialization in the parameter boundary region, and adapting to the optimization needs of the full parameter range in the simulation model.

[0021] 2) Combining The traversal characteristics of a chaotic sequence can be constructed using the following formula:

[0022] ;

[0023] in, The initial value of the chaotic sequence; The updated chaotic sequence value; For chaotic parameters;

[0024] Based on the updated chaotic sequence values ​​above The formula for mapping to compensation factor parameters is:

[0025]

[0026] in, The optimized first step for chaotic initialization The first individual in the population Dimensional compensation factor parameters; For the first The lower bound of the dimensional compensation factor under both simulation and engineering constraints; For the first The upper bound of the dual constraints of simulation and engineering for the dimensional compensation factor.

[0027] Furthermore, the weighted feeding strategy is specifically as follows:

[0028] ;

[0029] in, For the first The individual The shrinkage coefficient of the next iteration; For the first Next iteration position; The iterative shrinkage coefficient; This is the factor importance weight vector; Use random numbers in the range [0,1] to balance exploration and convergence; To be the globally optimal individual, ensuring the global convergence direction; It is the optimal individual in the local neighborhood; This is the local neighborhood optimal guiding term, and 0.3 is the guiding coefficient; It achieves a dual innovative integration of factor weight allocation and global and local collaborative guidance; it improves the precision of optimization by guiding local neighborhood optimization, and focuses on optimizing core compensation factors by combining factor importance weights, thus adapting to the complex working conditions of motors.

[0030] Furthermore, the chaotic perturbation strategy is as follows:

[0031] ;

[0032] in, This is the new global optimal solution after the perturbation; The original global optimal solution before the perturbation; 0.05 is the perturbation coefficient; , These are the lower bound vector and upper bound vector of the 3D compensation factor, respectively. The perturbation is a uniform random number in the range of [0,1] to ensure the randomness and exploratory nature of the perturbation.

[0033] Furthermore, the specific implementation method for embedding the optimal 3D compensation factor into the magnetic flux observer is as follows:

[0034] The harmonic compensation factor output by the HBWO algorithm Resistance compensation factor Inductance compensation factor The core links for voltage signal processing and parameter configuration are embedded with magnetic flux observers.

[0035] Optimal harmonic compensation factor Embedded magnetic flux observer In the shaft voltage signal preprocessing stage, harmonic cancellation is performed before the voltage signal is input to the core calculation module of the observer:

[0036] ;

[0037] ;

[0038] in, , Raw data collected for motor control Shaft voltage signal; , For the process The compensated pure voltage signal; , Real-time output of the flux linkage observer shaft flux linkage estimate;

[0039] Optimal resistance compensation factor A stator resistance parameter configuration module embedded with a flux linkage observer updates the observer's calculated resistance values ​​in real time.

[0040] ;

[0041] in, This refers to the rated value of the motor stator resistance; To improve the resistance compensation factor obtained by the HBWO optimization algorithm in real time; This is the stator resistance value after dynamic correction;

[0042] Optimal inductance compensation factor A stator inductance parameter configuration module embedded with a flux linkage observer dynamically adjusts the inductance coefficient in flux linkage calculations.

[0043] ;

[0044] in, This refers to the rated value of the motor stator inductance. To improve the inductance compensation factor obtained by the HBWO optimization algorithm in real time; This is the stator inductance value after dynamic adaptation.

[0045] Furthermore, the real-time optimization and compensation update implementation method in S4 is as follows:

[0046] Real-time operating data of the permanent magnet synchronous motor, including mechanical speed, is collected through data acquisition. , Shaft flux estimate , , Shaft flux reference value , and stator resistance Stator inductor The running data is used as input to the HBWO algorithm to update the optimal value of the 3D compensation factor in real time. The updated optimal 3D compensation factor is synchronously applied to the magnetic flux observer through the embedding mechanism in step S3, forming a real-time closed-loop control of data acquisition, algorithm optimization, factor compensation, and accuracy assurance.

[0047] Beneficial effects:

[0048] 1. The HBWO algorithm is improved by integrating four major improvement mechanisms, which solves the defects of conventional optimization algorithms such as insufficient population diversity, poor adaptability to working conditions, and easy iteration stagnation, and improves the optimization efficiency and accuracy of the 3D compensation factor.

[0049] 1) A hybrid population initialization strategy is adopted to generate an initial population of 3D compensation factors that covers the entire parameter range and is uniformly distributed, ensuring the diversity of the search foundation; an adversarial learning strategy is innovatively combined with... It integrates chaotic initialization strategies and is specifically extended to adapt to 3D compensation factors and harmonic compensation factors. Resistance compensation factor Inductance compensation factor The population initialization process; its core design logic is to use the synergistic effect of multiple initialization strategies to enable the initial population to cover a wide range of parameter search space and ensure uniform distribution within the parameter range, thereby significantly improving the global optimization capability and optimization efficiency of the improved HBWO algorithm, and ensuring that the optimal combination of 3D compensation factors that is suitable for different working conditions can be accurately found.

[0050] 2) Based on the weighted fitness function, the harmonic suppression effect, parameter correction effect, and flux linkage observation accuracy of each factor combination are quantified to form an objective scoring standard. The weighted fitness function can accurately achieve dynamic matching between complex working conditions and optimization objectives. Its effect lies in simultaneously quantifying flux linkage observation accuracy, parameter drift correction effect, and harmonic suppression capability. By dynamically adjusting the weights, the optimization direction of the 3D compensation factors can be accurately adapted to the real-time working conditions, solving the defect of traditional fixed fitness function single objective optimization that cannot take into account the needs of multiple working conditions, and ensuring that the optimal compensation factor combination can be output across the entire speed range.

[0051] 3) Iterative optimization of factor combinations is achieved through a weighted feeding strategy. Combining factor importance weights with global / local optimum guidance, the optimization direction is focused on core needs. A dual improvement mechanism of factor importance weights and local neighborhood optimum guidance is introduced to construct a more effective position update rule. Through the weighted collaborative guidance of global optimum, local optimum and target individual, the population iteration direction is made more in line with the compensation needs of magnetic flux observation, thereby improving the accuracy and convergence efficiency of 3D factor optimization.

[0052] 4) Breaking local optima traps through iterative stagnation perturbation mechanisms ensures the continuity and stability of the optimization process. An innovative chaotic perturbation strategy is designed to break local optima traps by accurately identifying iterative stagnation states and applying controllable chaotic perturbations, thus ensuring the continuity and stability of the optimization process and finding the optimal combination of 3D compensation factors in real time.

[0053] Compared with conventional algorithms such as PSO, P&O, COA, and GPC, the improved HBWO algorithm is faster in terms of startup time and iteration speed, and there is no iteration stagnation during the optimization process, which greatly improves the optimization efficiency and stability.

[0054] 2. A targeted link embedding method is adopted to accurately embed the 3D compensation factor into the flux linkage observer, addressing three core issues: harmonic interference, resistance drift, and inductor magnetic saturation, significantly improving the accuracy and stability of flux linkage observation. The harmonic compensation factor output by the algorithm is then used. Resistance compensation factor Inductance compensation factor The core innovation of this embedding method lies in achieving targeted and dynamic compensation of 3D factors, which is embedded in the voltage signal processing and parameter configuration core links of the magnetic flux observer to form a closed-loop control that optimizes algorithms, compensates for factors, and improves observation accuracy. This method accurately solves the two major error problems of harmonic interference and parameter drift in magnetic flux observation, and significantly improves the stability and accuracy of the observer under complex operating conditions.

[0055] 3. This invention features mechanisms such as early termination, reset, and output smoothing, reducing computing power consumption by 30% compared to PSO. It is well-suited for practical engineering applications, reducing computing power consumption and improving system reliability. Furthermore, it can be embedded without modifying the existing control architecture, making it compatible with industrial-grade hardware computing power. Simulation verification shows that this method can control the speed observation error within 0.008 r / min and the rotor position observation error within 0.014°. It is suitable for high-precision control scenarios of permanent magnet synchronous motors under medium- and high-speed, variable load conditions, demonstrating significant engineering application value. Attached Figure Description

[0056] Figure 1 This is a flowchart of the improved algorithm optimization process of this invention.

[0057] Figure 2The graph shows the speed response effect of the improved HBWO optimization algorithm based on this invention.

[0058] Figure 3 The figure shows the torque response effect curve based on the improved HBWO optimization algorithm of this invention.

[0059] Figure 4 This is a simulation speed following effect curve based on the compensation factor of the present invention.

[0060] Figure 5 This is a simulation speed following error curve based on the compensation factor of the present invention.

[0061] Figure 6 This is a curve showing the simulated position following effect of the present invention based on the compensation factor.

[0062] Figure 7 This is a simulation position following error curve based on the compensation factor of the present invention. Detailed Implementation

[0063] This embodiment provides a parameter compensation method for permanent magnet synchronous motor flux linkage observers based on an improved beluga optimization algorithm. The specific implementation steps are as follows:

[0064] S1: Construct a 3D compensation factor optimization objective for the flux linkage observer of a permanent magnet synchronous motor. The 3D compensation factor includes the harmonic compensation factor. Resistance compensation factor Inductance compensation factor , respectively used to offset The interference of the 5th / 7th harmonics of shaft voltage, the correction of stator resistance drift caused by temperature, and the adaptation to inductor magnetic saturation caused by load fluctuations.

[0065] First, the dynamic range of its 3D compensation factor is defined, and an adaptive fitness weighting function for different operating conditions is designed. To ensure that the optimization target accurately matches the real-time operating requirements, a weighted fitness function that integrates flux linkage error, parameter drift deviation, and harmonic suppression effect is constructed, and its expression is as follows:

[0066]

[0067] in, This represents the flux linkage estimation error, used to characterize the accuracy of flux linkage observation. The resistance drift standardization deviation is used to quantify the degree of drift in stator resistance parameters; The inductance drift standardization bias is used to quantify the degree of stator inductance parameter drift. This is a quantification term for harmonic suppression effect, associated with the harmonic compensation factor. , used to characterize right Harmonic cancellation effect of shaft voltage; , These are the flux linkage error weight and the parameter drift deviation weight, respectively, which satisfy... .

[0068] The weighted fitness function can accurately achieve dynamic matching between complex working conditions and optimization objectives. Its effect lies in synchronously quantifying the accuracy of flux observation, the effect of parameter drift correction, and the ability to suppress harmonics. By dynamically adjusting the weights, the optimization direction of the 3D compensation factor can be accurately adapted to the real-time working conditions, solving the defect of the traditional fixed fitness function single objective optimization that cannot take into account the needs of multiple working conditions, and ensuring that the optimal combination of compensation factors can be output across the entire speed range.

[0069] S2: Design an improved beluga whale optimization (HBWO) algorithm for magnetic flux observation scenarios. The improved optimization algorithm achieves global optimality of the 3D compensation factor. The improved beluga whale optimization algorithm includes improvements in working condition adaptive optimization, mixed population initialization, weighted feeding strategy, and iteration stagnation perturbation.

[0070] 1. Improved initialization of mixed populations:

[0071] To address the technical shortcomings of traditional optimization algorithms, such as uneven population distribution and lack of diversity due to a single random initialization method, which leads to the algorithm easily falling into local optima traps and failing to find the global optimal compensation factor, this improved strategy innovatively combines an adversarial learning strategy with... It integrates chaotic initialization strategies and is specifically extended to adapt to 3D compensation factors and harmonic compensation factors. Resistance compensation factor Inductance compensation factor The population initialization process; its core design logic is to use the synergistic effect of multiple initialization strategies to enable the initial population to cover a wide range of parameter search space and ensure uniform distribution within the parameter range, thereby significantly improving the global optimization capability and optimization efficiency of the improved HBWO algorithm, and ensuring that the optimal combination of 3D compensation factors that is suitable for different working conditions can be accurately found.

[0072] First, the population size is determined. Addressing the problem of limited coverage of basic random initialization parameters and the potential for missing optimal solutions at parameter boundaries in simulation models, an oppositional learning strategy is introduced. Combined with upper and lower bound constraints, a formula is constructed. Its key feature is...

[0073]

[0074] in, The parameters are those obtained after learning the opposite. The lower boundary; The upper boundary; The first one obtained by random initialization based on the base The first individual in the population Dimensional compensation factor parameters.

[0075] By using parametric symmetric mapping, the coverage of the population is expanded, filling the distribution gaps of basic random initialization in the parameter boundary region, and adapting to the optimization needs of the entire parameter range in the simulation model.

[0076] To address the problem of uneven parameter distribution in the intermediate region and the algorithm's tendency to get trapped in local optima, this paper combines... The traversal characteristics of chaotic sequences are suitable for the uniform optimization requirements in simulation models; the following formula is constructed:

[0077]

[0078] in, The initial value of the chaotic sequence; The updated chaotic sequence value; For chaotic parameters;

[0079] Based on the updated chaotic sequence values ​​above The formula for mapping to compensation factor parameters is:

[0080]

[0081] in, The optimized first step for chaotic initialization The first individual in the population Dimensional compensation factor parameters; For the first The lower bound of the dimensional compensation factor under both simulation and engineering constraints; For the first The upper bound of the dual constraints of simulation and engineering for the dimensional compensation factor.

[0082] It utilizes the uniform traversal characteristics of chaotic sequences to ensure that parameters are evenly distributed within the effective parameter space of the simulation model, filling parameter gaps in the intermediate region and enhancing population diversity. The parameter distribution density in the intermediate region is increased by more than 40%, which can effectively avoid the stagnation problem of the algorithm in the local optimal region of the simulation model and improve the stability of optimization in the actual system. Especially under complex working conditions such as sudden load changes, it can quickly find a suitable combination of compensation factors.

[0083] 2. Improved weighted food-grabbing strategy

[0084] To address the issues of traditional beluga optimization algorithms' position update rules failing to consider the differences in the importance of 3D compensation factors and resulting in insufficient optimization accuracy, this paper proposes a dual improvement mechanism that incorporates the priority of 3D compensation factors in the permanent magnet synchronous motor flux linkage observation model. This mechanism combines factor importance weighting and local neighborhood optimization guidance to construct a more effective position update rule. Through weighted collaborative guidance of global optima, local optima, and target individuals, the population iteration direction is made more aligned with the compensation requirements of flux linkage observation, thereby improving the accuracy and convergence efficiency of 3D factor optimization.

[0085]

[0086] in, For the first The individual The shrinkage coefficient of the next iteration; For the first Next iteration position; The iterative shrinkage coefficient; This is the factor importance weight vector; Use random numbers in the range [0,1] to balance exploration and convergence; To be the globally optimal individual, ensuring the global convergence direction; It is the optimal individual in the local neighborhood;

[0087] Its improved and innovative advantages are that, in the formula This is the local neighborhood optimal guiding term, and 0.3 is the guiding coefficient; It can achieve a dual innovative integration of factor weight allocation and global and local collaborative guidance; it improves the precision of optimization by guiding local neighborhood optimization, and focuses on optimizing core compensation factors by combining factor importance weights, thus adapting to the complex working conditions of motors.

[0088] 3. Improvement of Iterative Stagnation Perturbation

[0089] To address the problem that traditional optimization algorithms are prone to convergence stagnation during the search for 3D compensation factors, resulting in an inability to adapt to dynamic changes in motor operating conditions, this improved strategy innovatively designs a chaotic perturbation strategy. By accurately identifying iterative stagnation states and applying controllable chaotic perturbations, it breaks the local optimum trap, ensures the continuity and stability of the search process, and finds the optimal combination of 3D compensation factors in real time.

[0090]

[0091] in, This is the new global optimal solution after the perturbation; The original global optimal solution before the perturbation; 0.05 is the perturbation coefficient; , These are the lower bound vector and upper bound vector of the 3D compensation factor, respectively. The perturbation is a uniform random number in the range of [0,1] to ensure the randomness and exploratory nature of the perturbation.

[0092] The above-mentioned improved optimization strategies can accurately identify iterative stagnation states, avoiding invalid perturbations that could disrupt optimization stability. Simultaneously, by using small, controllable perturbations to break local optima traps, the algorithm is guided to iterate towards new, high-quality search regions, ensuring the optimization process remains uninterrupted. This guarantees the continuous output of optimal compensation factors adapted to new operating conditions, effectively preventing increased flux linkage observation errors caused by algorithm stagnation and ensuring stable observation accuracy.

[0093] S3: Embed the optimal 3D compensation factor output by the improved HBWO algorithm into the signal processing link of the flux linkage observer, through... magnetic flux observer The shaft voltage signal undergoes harmonic cancellation processing, through The stator resistance parameters of the observer are dynamically corrected by means of... The stator inductance parameters of the observer are dynamically adapted.

[0094] To achieve a deep fit between the improved HBWO algorithm's optimization results and the flux linkage observer, a precise embedding mechanism for the optimal factor and the observer's signal link is constructed, incorporating the harmonic compensation factor output by the algorithm. Resistance compensation factor Inductance compensation factor The voltage signal processing and parameter configuration core links of the magnetic flux observer are embedded separately, forming a closed-loop control system that optimizes algorithms, compensates for factors, and improves observation accuracy. The core innovation of this embedding method lies in achieving targeted and dynamic compensation of 3D factors, accurately solving the two major error problems of harmonic interference and parameter drift in magnetic flux observation, and significantly improving the stability and accuracy of the observer under complex operating conditions.

[0095] First, the optimal harmonic compensation factor Embedded magnetic flux observer In the shaft voltage signal preprocessing stage, harmonic cancellation is performed before the voltage signal is input to the core calculation module of the observer:

[0096]

[0097]

[0098] in, , Raw data collected for motor control Shaft voltage signal; , For the process The compensated pure voltage signal; , Real-time output of the flux linkage observer Axial flux linkage estimation. This innovative improvement strategy addresses the distortion of flux linkage estimation caused by harmonic interference in the original signal, thus solving the problem of distortion caused by harmonics in traditional observers.

[0099] Optimal resistance compensation factor A stator resistance parameter configuration module embedded with a flux linkage observer updates the observer's calculated resistance values ​​in real time.

[0100]

[0101] in, This refers to the rated value of the motor stator resistance; To improve the resistance compensation factor obtained by the HBWO optimization algorithm in real time; This is the dynamically corrected stator resistance value. This improved strategy can dynamically adapt to stator resistance drift caused by temperature changes, solving the shortcomings of traditional observers with fixed resistance parameters that cannot adapt to temperature fluctuations; the adaptation accuracy of resistance parameters is improved by more than 70%, the temperature robustness of flux linkage observation is enhanced, and the observation error under high-temperature conditions is reduced by 35%.

[0102] Optimal inductance compensation factor A stator inductance parameter configuration module embedded with a flux linkage observer dynamically adjusts the inductance coefficient in flux linkage calculations.

[0103]

[0104] in, This refers to the rated value of the motor stator inductance. To improve the inductance compensation factor obtained by the HBWO optimization algorithm in real time; This is the stator inductance value after dynamic adaptation. The improvement strategy lies in accurately adapting to the inductor magnetic saturation phenomenon caused by load fluctuations, solving the shortcomings of traditional observers with fixed inductance parameters that cannot cope with sudden load changes; in the scenario of sudden load changes, the inductor adaptation response time is greatly reduced, the fluctuation amplitude of flux linkage observation accuracy is reduced by 50%, and the observer output is stable when the load changes dynamically.

[0105] S4: Collects real-time operating data of the permanent magnet synchronous motor, including mechanical speed. , Shaft flux estimate , , Shaft flux reference value , and stator resistance Stator inductor The improved HBWO algorithm is input for iterative optimization, and the 3D compensation factor is updated in real time to ensure the observation accuracy of the magnetic flux observer under different complex working conditions.

[0106] Real-time operating data of the permanent magnet synchronous motor, including mechanical speed, is collected through data acquisition. , Shaft flux estimate , , Shaft flux reference value , and stator resistance Stator inductor The running data is used as input to improve the HBWO algorithm, driving the algorithm's working condition adaptive fitness function, mixed population initialization mechanism, weighted feeding strategy and iterative stagnation perturbation mechanism to perform collaborative iterative optimization, and updating the optimal value of the 3D compensation factor in real time; the updated optimal 3D compensation factor is synchronously applied to the magnetic flux observer through the embedding mechanism, forming a real-time closed-loop control of data acquisition-algorithm optimization-factor compensation-accuracy guarantee.

[0107] By dynamically linking real-time running data with the improved HBWO algorithm, the advantages of the four improvement mechanisms are synergistically enhanced. The working condition adaptive fitness function ensures that the optimization target and real-time working conditions are accurately matched, the hybrid population initialization and weighted feeding strategy ensure the efficiency and accuracy of optimization, and the iterative stagnation perturbation mechanism avoids optimization interruption. Finally, the adaptive control of the entire process of working condition perception, real-time optimization and accurate compensation of the 3D compensation factor is realized, which solves the technical defects of traditional compensation methods that have fixed parameters and cannot adapt to complex dynamic working conditions.

[0108] Experimental verification:

[0109] I. Construction of Experimental System and Simulation Model

[0110] The magnetic flux linkage observation simulation system built in this example has a closed-loop core of data acquisition, algorithm optimization, factor compensation, and accuracy feedback, covering five major functional modules. The parameter configuration of each module and the signal link are matched with the actual engineering requirements, providing a standardized platform for the verification of improved HBWO algorithm.

[0111] 1.1 System Overall Architecture and Core Module Functions

[0112] 1) Permanent Magnet Synchronous Motor (PMSM) Main Module: The module uses rated parameters of 15kW rated power, 1500r / min rated speed, and stator resistance... =0.08Ω, stator inductance =8.5mH, permanent magnet flux =0.175Wb surface-mount model; receive compensated Shaft voltage signal, outputs real-time operating data, including three-phase current. Mechanical speed Rotor position This provides raw feedback data for subsequent observation and algorithm optimization.

[0113] 2) Core module of flux linkage observer: It adopts a sliding mode observer architecture, and its core function is to estimate based on the input voltage and current signals. The shaft flux linkage, which simultaneously receives the 3D optimal compensation factor output by the improved HBWO algorithm, completes voltage harmonic cancellation and dynamic correction of stator parameters, and is the core execution unit for improving observation accuracy.

[0114] 3) Improved HBWO Algorithm Optimization Module: The core innovative module of this invention is embedded in the simulation model through a MATLAB Function script. The input signals include real-time motor operating data, including speed. Magnetic flux linkage estimate Magnetic flux reference value Stator parameters Output the 3D optimal compensation factor , , This enables adaptive control of operating condition perception, iterative optimization, and factor output.

[0115] 4) Signal Processing and Coordinate Transformation Module: Includes axis Transformer, low-pass filter (LPF), cutoff frequency 100Hz axis The transformation submodule performs preprocessing of current / voltage signals to filter out high-frequency noise and perform coordinate transformation, ensuring the accuracy of the signals input to the observer and algorithm; at the same time, it extracts and feeds back speed and position signals to ensure the formation of closed-loop control.

[0116] 1.2 Core Module Parameter Configuration

[0117] To ensure the engineering reference value of the simulation results, the core module parameters are uniformly configured as follows: data acquisition sampling frequency 1kHz, simulation step size 0.4s; flux linkage observer sliding mode gain set to 80; improved HBWO algorithm parameters: population size 30, maximum number of iterations 50, speed domain boundaries (medium-high speed domain n≥200r / min, zero-low speed domain n<200r / min), stagnation judgment threshold ε=10⁻ 6 Stagnation trigger threshold N th =5, perturbation coefficient 0.05; 3D compensation factor constraint range: k1∈[0.5,2.0], ΔR∈[-0.2,0.2], ΔL∈[-0.15,0.15].

[0118] II. Engineering Implementation of the Improved HBWO Optimization Algorithm

[0119] The improved HBWO algorithm is embedded into the simulation model via MATLAB scripts, fully implementing four core improvement mechanisms: adaptive optimization under operating conditions, hybrid population initialization, weighted feeding strategy, and iterative stagnation perturbation. The final output is a 3D optimal compensation factor. The specific implementation process and key logic are as follows:

[0120] 2.1 Algorithm Input Verification and Parameter Initialization

[0121] After the algorithm starts, it first performs an input validity check: if the rotational speed... If the flux linkage estimate / reference value does not exist, the output will be based on the initially set compensation factor, and the current optimization will be terminated to ensure the stability of the system.

[0122] 2.2 Implementation Logic of the Four Major Improvement Mechanisms

[0123] 1) Hybrid population initialization: combining basic random initialization, adversarial learning, and... Chaos initialization. First, through... Generate a basic population, and then through To achieve the expansion of boundaries through dialectical learning, and ultimately through Chaotic sequence The parameters are uniformly distributed in the intermediate region by mapping, and after boundary verification, a diverse initial population is obtained, with the parameter distribution density in the intermediate region increased by more than 40%.

[0124] 2) Weighted feeding strategy: Configure a 3-dimensional factor importance weight vector =[0.4,0.3,0.3], construct the above position update formula, where By combining global optimization and local neighborhood optimization, the accuracy of optimization can be improved by more than 35%.

[0125] 3) Iteration stagnation disturbance: Five consecutive stagnations trigger a chaotic disturbance. After the disturbance, the counter is reset by boundary verification, which reduces the probability of iteration stagnation by more than 70%.

[0126] 4) Domain-specific fitness function: Refer to the weighted fitness function mentioned above to achieve accurate matching between the optimization target and the real-time operating conditions.

[0127] III. Embedding and Closed-Loop Control Implementation of 3D Compensation Factors

[0128] The improved HBWO algorithm outputs a 3D optimal compensation factor, which is then precisely integrated into the flux linkage observer and signal processing link according to the targeted link embedding principle. This forms a real-time closed loop encompassing data acquisition, algorithm optimization, factor compensation, and accuracy improvement. The specific embedding method and control flow are as follows:

[0129] 3.1 Implementation of Precise Embedding of Compensation Factors

[0130] 1) Harmonic compensation factor Embedded magnetic flux observer In the shaft voltage preprocessing stage, the compensation formula is implemented using multipliers and adders. This cancels out the 5th and 7th harmonic interference in the original voltage signal, reducing the voltage harmonic distortion rate by more than 65%.

[0131] 2) Resistance compensation factor The stator resistance parameter configuration module embedded in the observer dynamically adapts to resistance drift caused by temperature changes, improving resistance adaptation accuracy by more than 70% within a temperature fluctuation range of ±50℃.

[0132] 3) Inductance compensation factor The stator inductor parameter configuration module with embedded observer matches the inductor magnetic saturation caused by load fluctuations. Under the scenario of sudden change in rated load from 0-100%, the inductor adaptation response time is <0.02s.

[0133] 3.2 Closed-loop control process

[0134] 1) Signal Acquisition and Preprocessing: The data acquisition module collects motor operating data in real time, including speed. Magnetic flux linkage estimate Magnetic flux reference value Stator parameters After low-pass filtering and noise reduction, the signal is input into the improved HBWO algorithm module.

[0135] 2) Algorithm Iteration Optimization: Based on real-time data, the algorithm iteratively optimizes through four major improvement mechanisms and outputs a smoothed 3D optimal compensation factor.

[0136] 3) Function of compensation factor: The compensation factor is embedded in the corresponding link to complete voltage harmonic cancellation and dynamic correction of stator parameters.

[0137] 4) Accuracy feedback: The corrected flux observer outputs a high-precision flux estimate for motor speed / position control. At the same time, the flux estimate is fed back to the algorithm module to drive the next round of optimization, realizing real-time adaptive compensation under all working conditions.

[0138] IV. Comparative Experiment Design

[0139] To verify the superiority of the improved HBWO algorithm, three sets of comparative experiments were designed. The core analysis carriers were four curves: velocity following error, velocity following effect, position following effect, and position following error. The performance evaluation was completed by combining quantitative indicators.

[0140] 4.1 Experimental Objectives

[0141] Verify the performance advantages of the improved HBWO algorithm in the following dimensions: speed / position following accuracy based on speed following error and position following error curves; dynamic response characteristics based on the start-up phase characteristics of the speed / position following curves; load change disturbance resistance based on the response of the curve after a load change at t=0.3s; and full-condition stability covering start-up, steady state, load change, and temperature fluctuation conditions.

[0142] 4.2 Setting up the comparison scheme

[0143] The simulation models, operating parameters, and initial conditions of the three schemes are completely identical, with the only difference being the compensation algorithm of the flux linkage observer, ensuring a fair comparison:

[0144] Option 1: Traditional sliding mode observers have no compensation factor and fixed stator parameters;

[0145] Option 2: Sliding Mode Observer + PSO Algorithm (only optimized) Harmonic compensation factor, without domain division and stagnation disturbance mechanism;

[0146] Option 3: Sliding mode observer + improved HBWO algorithm of this invention to optimize 3D compensation factor, including four major improvement mechanisms.

[0147] V. Simulation Results and Analysis

[0148] The improved and optimized algorithm HBWO is used to measure speed response, torque response, speed following effect, speed following error (unit: 10⁻³ level), position following effect, and position following error (unit: 10⁻³ level). 4 Six curves (level 1) were used to conduct a detailed analysis of the performance differences among the three schemes.

[0149] 5.1 Speed ​​and Torque Response Performance Analysis

[0150] 1) Speed ​​response curve analysis, from Figure 2As can be seen from the mid-speed response curves, the improved HBWO algorithm (red curve) exhibits significant performance advantages in the startup phase: Startup speed and overshoot control: Within 0-0.1s, the HBWO curve rapidly rises to the target speed of 1500 RPM with no significant overshoot, and the startup response time is only 0.1s; while the P&O (black), PSO (green), COA (blue), and GPC (purple) algorithms all exhibit varying degrees of overshoot and startup delay, with the P&O algorithm having the longest startup time, the largest overshoot, and significant oscillations after startup. Steady-state stability: The HBWO curve quickly stabilizes at 1500 RPM after 0.1s with no significant fluctuations; while the P&O, PSO, and other algorithms still exhibit continuous oscillations within 0.3s, with steady-state fluctuation amplitudes significantly greater than HBWO. Dynamic response and robustness: The HBWO algorithm, through an iterative stagnation disturbance mechanism, quickly converges to the target speed during the startup phase, effectively suppressing overshoot and oscillation, demonstrating stronger dynamic response capability and adaptability to operating conditions, and providing a reliable guarantee for high-precision motor control.

[0151] 2) Torque response curve analysis, from Figure 3 As can be seen from the torque response curves, the improved HBWO algorithm also exhibits excellent performance in torque control: Impact suppression and convergence speed: The HBWO curve has the smallest torque impact peak during the startup phase (approximately 11.5 Nm) and rapidly converges to a steady-state torque of 5 Nm within 0.2s; while the P&O algorithm has the largest torque impact peak (approximately 8 Nm) and still exhibits significant oscillations within 0.5s, with the slowest convergence speed. Steady-state fluctuations and accuracy: The HBWO curve shows minimal torque fluctuations in the steady-state phase, almost coinciding with the target torque; while the P&O, PSO, and other algorithms exhibit larger steady-state torque fluctuations, affecting the stability and accuracy of motor control. Disturbance resistance: Through the synergistic optimization of 3D compensation factors, the HBWO algorithm effectively suppresses torque fluctuations and impacts, demonstrating stronger disturbance resistance during startup and load changes, ensuring the smoothness and reliability of motor operation.

[0152] 5.2 Speed ​​Following Correlation Analysis

[0153] 1) Analysis of the curves between estimated and actual rotational speed

[0154] from Figure 4As can be seen, the estimated rotational speed (speedref, blue curve) and the actual rotational speed (speedact, orange curve) of this scheme rise rapidly and synchronously during the initial stage (0-0.05s), and stabilize at around 100 r / min after 0.05s. The two curves almost completely overlap, with only minimal fluctuations. A magnified view clearly shows that within the steady-state range of 0.0452-0.0454s, the difference between the two curves remains stable within 0.05 r / min, without significant hysteresis or shift. This indicates that the improved HBWO algorithm, through the synergistic optimization of the 3D compensation factor, effectively suppresses harmonic interference and parameter drift in flux linkage observation, enabling the rotational speed estimation to accurately reproduce the dynamic changes of the actual rotational speed, achieving "zero-delay" steady-state tracking of the target rotational speed.

[0155] 2) Analysis of the speed estimation error curve

[0156] from Figure 5 As shown in the speed estimation error curve, the error decays rapidly during the start-up phase (0-0.02s) and stabilizes within the range of 1×10⁻³-7×10⁻³ r / min after 0.05s. Even during subsequent minor fluctuations, the peak error does not exceed 9×10⁻³ r / min. This quantitative result directly verifies that the speed estimation accuracy of this scheme reaches the 10⁻³ r / min level, far lower than the error level of traditional schemes. The smooth fluctuation characteristics of the error curve further demonstrate the significant advantages of the improved HBWO algorithm in suppressing noise and improving observation stability, ensuring high-precision speed control of the motor under all operating conditions.

[0157] 5.3 Position-following correlation analysis

[0158] 1) Analysis of rotor position estimation versus actual value curves:

[0159] Figure 6 The dynamic response curve of the rotor position estimate directly characterizes the excellent position following performance of this invention through its core morphological features:

[0160] Steady-state linear following characteristic: During the steady-state operation phase of 0-0.3s, the position estimate curve exhibits a strictly linear growth trend with a constant slope. This characteristic indicates that the improved HBWO algorithm with optimized 3D compensation factor enables the flux linkage observer to accurately reproduce the uniform rotation characteristics of the motor rotor. The rate of change of the estimated value is completely consistent with the rate of change of the actual position, with no proportional misalignment or phase hysteresis, achieving "zero-delay" steady-state following of the actual position.

[0161] Dynamic tracking characteristic under sudden load changes: Under a load change condition of t=0.3s, the position estimate curve synchronously exhibits a step jump, and immediately recovers to the linear growth slope consistent with that before the change. This key feature proves that the iterative stagnation disturbance mechanism of this invention is instantaneously activated during sudden load changes, rapidly updating the three-dimensional compensation factors of harmonics, resistance, and inductance. This allows the flux linkage observer to instantly adapt to the sudden changes in parameters and load, ensuring that the dynamic changes in the position estimate and the actual position are completely synchronized, with no response lag or excessive oscillations. This fully demonstrates the algorithm's extremely strong dynamic tracking capability and robustness against disturbances.

[0162] 2) Analysis of position estimation error curve:

[0163] from Figure 7 It can be seen that the rotor position estimation error of this scheme shows a slow overall decreasing trend. In the initial stage (0-0.05s), the error fluctuates slightly around 0, then gradually converges towards the negative value, and stabilizes at -3×10⁻ after 0.05s. 4 Within the range of -0 rad, even with a sudden change in operating conditions at t=0.3s, the error did not show a significant jump, but only changed linearly with the curve. This is thanks to the improved HBWO algorithm's adaptive optimization of operating conditions and precise 3D factor compensation, which enabled the flux linkage observation accuracy to reach 10⁻⁻⁶. 4 The position estimation steady-state error is much lower than that of traditional solutions, fully meeting the stringent requirements of high-precision motor control for position tracking, while also possessing excellent anti-disturbance capabilities.

[0164] VI. Experimental Conclusions

[0165] 1) The improved HBWO algorithm of this invention significantly improves the accuracy of flux linkage observation through the synergistic optimization of the 3D compensation factor and the synergistic effect of four major improvement mechanisms, ultimately achieving a steady-state error of velocity following <2×10⁻³ and a steady-state error of position following <5×10⁻ ... 4 This method improves accuracy by more than 80% compared to traditional methods, fully meeting the requirements for high-precision control.

[0166] 2) The algorithm has excellent dynamic response and anti-disturbance capabilities, with a startup response time of <0.05s and a recovery time of <0.01s after a sudden load change, which is a significant improvement over the PSO algorithm and traditional observers, and can accurately adapt to changes in the dynamic operating conditions of the motor.

[0167] 3) The algorithm can still output the optimal compensation factor stably in a wide temperature range of -20℃ to 80℃ and under the condition of sudden load change from 0 to 100%, ensuring stable speed / position following accuracy. Its temperature robustness and adaptability to working conditions are significantly better than the comparison scheme.

[0168] 4) The improved HBWO algorithm has a simple engineering implementation logic and flexible parameter configuration. It can be directly embedded into the existing permanent magnet synchronous motor control model without major modifications to the system architecture, and has outstanding engineering application value.

[0169] In summary, this invention achieves adaptive optimization and precise embedding of 3D compensation factors by improving the HBWO algorithm, effectively solving the technical defects of traditional flux observers affected by harmonic interference and parameter drift, and significantly improving the accuracy and stability of motor speed / position control. It is suitable for high-precision servo control scenarios such as new energy vehicles and industrial robots.

Claims

1. A method for compensating parameters of a permanent magnet synchronous motor flux linkage observer based on an improved beluga optimization algorithm, characterized in that, Includes the following steps: S1: Construct a 3D compensation factor optimization objective for the flux linkage observer of a permanent magnet synchronous motor, including harmonic compensation factors. Resistance compensation factor Inductance compensation factor , respectively used to offset Shaft voltage 5th / 7th harmonic interference, correcting stator resistance drift caused by temperature, and adapting to inductor magnetic saturation caused by load fluctuations; S2: Design an improved beluga whale optimization algorithm HBWO for magnetic flux observation scenarios. The HBWO algorithm is used to achieve global optimal search for the 3D compensation factor. The HBWO algorithm adopts a hybrid population initialization improvement for population initialization, and introduces a weighted feeding strategy guided by factor importance weight and local neighborhood optimality. A chaotic perturbation strategy is designed to accurately identify the iteration stagnation state and apply controllable chaotic perturbation. S3: Embed the optimal 3D compensation factor output by the HBWO algorithm into the signal processing link of the flux linkage observer, through... magnetic flux observer The shaft voltage signal undergoes harmonic cancellation processing, through The stator resistance parameters of the observer are dynamically corrected by means of... Dynamically adapt the stator inductance parameters of the observer; S4: Collect real-time operating data of the permanent magnet synchronous motor, input it into the HBWO algorithm for iterative optimization, and update the 3D compensation factor in real time.

2. The method for compensating parameters of a permanent magnet synchronous motor flux linkage observer based on an improved beluga optimization algorithm according to claim 1, characterized in that, A dynamic optimization strategy is constructed using the HBWO algorithm to optimize the 3D compensation factor. , , It can autonomously adjust the value range and optimization weights according to the real-time operating conditions of the permanent magnet synchronous motor, achieving adaptive adaptation across the entire speed range; specifically, it includes the following processes: First, the dynamic range of the 3D compensation factor is defined, and an adaptive fitness weighting function for different working conditions is designed; the fitness weighting function is as follows: ； in, This represents the flux linkage estimation error; Standardized deviation for resistance drift; This refers to the standardized deviation of inductor drift. This is a quantification term for harmonic suppression effect, associated with the harmonic compensation factor. , used to characterize right Harmonic cancellation effect of shaft voltage; , These are the flux linkage error weight and the parameter drift deviation weight, respectively, which satisfy... .

3. The parameter compensation method for permanent magnet synchronous motor flux linkage observer based on the improved beluga optimization algorithm according to claim 1, characterized in that, The improved beluga whale optimization algorithm HBWO includes: Opposing learning strategies and It integrates chaotic initialization strategies and is specifically extended to adapt to 3D compensation factors and harmonic compensation factors. Resistance compensation factor Inductance compensation factor The population initialization process; 1) First, the population size is determined. To address the problem of limited coverage of basic random initialization parameters and the potential for missing optimal solutions at parameter boundaries in simulation models, an opposition learning strategy is introduced, and a formula is constructed in conjunction with upper and lower bound constraints: ； in, The parameters are those obtained after learning the opposite. The lower boundary; The upper boundary; The first one obtained by random initialization based on the base The first individual in the population Dimensional compensation factor parameters; By using parametric symmetric mapping, the coverage of the population is expanded, filling the distribution gap of basic random initialization in the parameter boundary region, and adapting to the optimization needs of the full parameter range in the simulation model. 2) Combining The traversal characteristics of a chaotic sequence can be constructed using the following formula: ； in, The initial value of the chaotic sequence; The updated chaotic sequence value; For chaotic parameters; Based on the updated chaotic sequence values ​​above The formula for mapping to compensation factor parameters is: ； in, The optimized first step for chaotic initialization The first individual in the population Dimensional compensation factor parameters; For the first The lower bound of the dimensional compensation factor under both simulation and engineering constraints; For the first The upper bound of the dual constraints of simulation and engineering for the dimensional compensation factor.

4. The parameter compensation method for permanent magnet synchronous motor flux linkage observer based on the improved beluga optimization algorithm according to claim 1, characterized in that, The weighted food-grabbing strategy is specifically as follows: ； in, For the first The individual The shrinkage coefficient of the next iteration; For the first Next iteration position; The iterative shrinkage coefficient; This is the factor importance weight vector; Use random numbers in the range [0,1] to balance exploration and convergence; To be the globally optimal individual, ensuring the global convergence direction; It is the optimal individual in the local neighborhood; This is the local neighborhood optimal guiding term, and 0.3 is the guiding coefficient; It achieves a dual innovative integration of factor weight allocation and global and local collaborative guidance; it improves the precision of optimization by guiding local neighborhood optimization, and focuses on optimizing core compensation factors by combining factor importance weights, thus adapting to the complex working conditions of motors.

5. The method for compensating parameters of a permanent magnet synchronous motor flux linkage observer based on an improved beluga optimization algorithm according to claim 1, characterized in that, The chaotic perturbation strategy is as follows: ； in, This is the new global optimal solution after the perturbation; The original global optimal solution before the perturbation; 0.05 is the perturbation coefficient; , These are the lower bound vector and upper bound vector of the 3D compensation factor, respectively. The perturbation is a uniform random number in the range of [0,1] to ensure the randomness and exploratory nature of the perturbation.

6. The parameter compensation method for permanent magnet synchronous motor flux linkage observer based on the improved beluga optimization algorithm as described in claim 1, characterized in that, The specific implementation method for embedding the optimal 3D compensation factor into the flux observer is as follows: The harmonic compensation factor output by the HBWO algorithm Resistance compensation factor Inductance compensation factor The core links for voltage signal processing and parameter configuration are embedded with magnetic flux observers. Optimal harmonic compensation factor Embedded magnetic flux observer In the shaft voltage signal preprocessing stage, harmonic cancellation is performed before the voltage signal is input to the core calculation module of the observer: ； ； in, , Raw data collected for motor control Shaft voltage signal; , For the process The compensated pure voltage signal; , Real-time output of the flux linkage observer shaft flux linkage estimate; Optimal resistance compensation factor A stator resistance parameter configuration module embedded with a flux linkage observer updates the observer's calculated resistance values ​​in real time. ； in, This refers to the rated value of the motor stator resistance; To improve the resistance compensation factor obtained by the HBWO optimization algorithm in real time; This is the stator resistance value after dynamic correction; Optimal inductance compensation factor A stator inductance parameter configuration module embedded with a flux linkage observer dynamically adjusts the inductance coefficient in flux linkage calculations. ； in, This refers to the rated value of the motor stator inductance. To improve the inductance compensation factor obtained by the HBWO optimization algorithm in real time; This is the stator inductance value after dynamic adaptation.

7. The method for compensating parameters of a permanent magnet synchronous motor flux linkage observer based on an improved beluga optimization algorithm according to claim 1, characterized in that, The real-time optimization and compensation update method in S4 is as follows: Real-time operating data of the permanent magnet synchronous motor, including mechanical speed, is collected through data acquisition. , Shaft flux estimate , , Shaft flux reference value , and stator resistance Stator inductor The running data is used as input to the HBWO algorithm to update the optimal value of the 3D compensation factor in real time. The updated optimal 3D compensation factor is synchronously applied to the magnetic flux observer through the embedding mechanism in step S3, forming a real-time closed-loop control of data acquisition, algorithm optimization, factor compensation, and accuracy assurance.

Images