Dynamically coupled path controlled anti-EMI cooling fan filtering and noise suppression system

Abstract

The present invention relates to the technical field of cooling fan control, and discloses an anti-EMI cooling fan filtering and noise suppression system with dynamic coupling path control, wherein the system comprises: a modeling subsystem, used for performing electromagnetic energy transfer modeling on the cooling fan to obtain an EMI energy balance equation group; an extraction subsystem, used for extracting the electromagnetic interference transfer relationship between various components of the cooling fan; a calculation subsystem, used for calculating noise energy distribution data within a preset speed range; an analysis subsystem, used for performing frequency autocorrelation analysis on parameter space distribution data to obtain EMI and noise parameter timing characteristic data; a real-time adjustment subsystem, used for inputting the EMI and noise parameter timing characteristic data into a prediction model for characteristic operation, obtaining PWM control correction parameters and adjusting the output power and speed of the cooling fan, wherein the system realizes the coordinated optimization among EMI suppression, noise control and heat dissipation performance.

Description

Technical Field

[0001] The present invention relates to the technical field of cooling fan control, and in particular to an anti-EMI cooling fan filtering and noise suppression system with dynamic coupling path control. Background Art

[0002] The electromagnetic interference (EMI) generated by traditional cooling fans at high speeds not only affects the normal operation of surrounding electronic components, but may also cause the signal quality of the entire system to deteriorate, the bit error rate to increase, and even cause system failures. The EMI problem is particularly serious when the fan is running at a high speed of 8000-20000RPM. In addition, the acoustic performance of the cooling fan within the noise range of 34.0-60.1dB-A cannot be ignored, as it directly affects the user experience and working environment.

[0003] Existing technologies mainly use passive shielding measures or add filtering circuits to suppress fan EMI. These methods not only increase the size and cost of the system, but also have difficulty adapting to the dynamic changes in the EMI characteristics of the fan under different working conditions. For the noise problem, traditional methods often reduce noise by reducing the speed or changing the shape of the fan blades, but this will lead to the problem of reduced heat dissipation efficiency. Especially when the fan uses PBT material fan blades and a double ball bearing structure, the coupling path of EMI and noise is more complicated. The existing single, static suppression method is difficult to balance the relationship between heat dissipation efficiency, EMI suppression and noise control at the same time. Summary of the invention

[0004] The present invention provides an anti-EMI cooling fan filtering and noise suppression system with dynamic coupling path control. The present invention implements real-time closed-loop regulation of the output power and rotation speed of the cooling fan, which not only effectively solves the problem that it is difficult to balance EMI and noise suppression with cooling efficiency in traditional technologies, but also enables the system to have adaptive learning capabilities, and can automatically optimize control strategies according to the unique characteristics of cooling fans of different models.

[0005] In a first aspect, the present invention provides an anti-EMI cooling fan filtering and noise suppression system with dynamic coupling path control, the anti-EMI cooling fan filtering and noise suppression system with dynamic coupling path control comprising:

[0006] The modeling subsystem is used to model the electromagnetic energy transfer of the motor, fan blades, bearings and housing components in the cooling fan to obtain the EMI energy balance equations;

[0007] An extraction subsystem, used to construct an electromagnetic interference transmission network according to the EMI energy balance equation group, and extract the electromagnetic interference transmission relationship between the components of the cooling fan;

[0008] A calculation subsystem, for performing a time discrete sequence calculation on the vibration transmission between the PBT fan blade and the motor based on the electromagnetic interference transmission relationship, and obtaining noise energy distribution data within a preset speed range;

[0009] An analysis subsystem, used to determine the parameter space distribution within a preset noise range according to the noise energy distribution data, and perform frequency autocorrelation analysis on the parameter space distribution data to obtain EMI and noise parameter timing characteristic data;

[0010] The real-time adjustment subsystem is used to input the EMI and noise parameter timing characteristic data into the prediction model for characteristic calculation, obtain PWM control correction parameters, and adjust the output power and speed of the cooling fan in real time according to the PWM control correction parameters.

[0011] In combination with the first aspect, in a first implementation of the first aspect of the present invention, the modeling subsystem is specifically used for:

[0012] Calculating the efficiency of electromagnetic coupling between the motor winding and the rotor magnetic field of the cooling fan under the input voltage to obtain a power conversion efficiency function, and performing vector field calculation on the magnetic flux density distribution in the motor based on the power conversion efficiency function to obtain an electromagnetic field distribution function;

[0013] Based on the electromagnetic field distribution function, a spectrum analysis is performed on the electromagnetic radiation intensity generated by the motor to obtain electromagnetic interference spectrum data, and according to the electromagnetic interference spectrum data, a frequency domain decomposition is performed on the radiation and conduction EMI components to obtain an interference mode distribution function;

[0014] The EMI shielding characteristics of the PBT fan blades and the plastic shell of the cooling fan are attenuated and calculated to obtain an EMI attenuation coefficient matrix, and the propagation channel of the interference signal is analyzed according to the EMI attenuation coefficient matrix to obtain an EMI transfer matrix;

[0015] The power conversion efficiency function, the electromagnetic field distribution function, the electromagnetic interference spectrum data, the interference mode distribution function and the EMI transfer matrix are calculated simultaneously to obtain an EMI transfer equation group between components;

[0016] The EMI transfer equation group is subjected to eigenvalue decomposition and normalization processing to obtain an EMI energy balance equation group including input voltage, motor operating power, double ball bearing mechanical quality factor and electromagnetic coupling coefficient.

[0017] In combination with the first aspect, in a second implementation of the first aspect of the present invention, the extraction subsystem is specifically used for:

[0018] Based on the EMI energy balance equations, the power input terminal, motor control circuit, stator winding, rotor, double ball bearing, PBT fan blade and housing are set as network nodes, and the EMI propagation relationship between network nodes is mapped by directed graph to obtain the initial EMI network topology structure;

[0019] Performing a first channel identification on the EMI propagation path from the power supply to the motor winding in the initial EMI network topology structure to obtain a first EMI transfer path matrix, and performing a second channel identification on the mechanical vibration coupling path between the fan blade and the housing in the initial EMI network topology structure to obtain a second vibration coupling path matrix;

[0020] Based on the first EMI transfer path matrix, the electromagnetic interference transfer efficiency between adjacent components in the first channel is quantitatively calculated to obtain the first channel EMI coupling coefficient, and according to the second vibration coupling path matrix, the vibration transfer characteristics of the double ball bearing in the second channel are analyzed to obtain the second channel vibration loss coefficient;

[0021] The time characteristic analysis of the EMI transfer process in the first channel and the second channel is performed respectively to obtain the time delay parameters at different speeds, and the EMI coupling coefficient of the first channel, the vibration loss coefficient of the second channel and the time delay parameters are substituted into the transfer function model to perform quantitative analysis of the EMI flow process to obtain the electromagnetic interference transmission relationship between the components of the cooling fan.

[0022] In combination with the first aspect, in a third implementation of the first aspect of the present invention, the computing subsystem is specifically used for:

[0023] Determine a discrete time series sampling point set according to the electromagnetic interference transfer relationship, and perform piecewise linearization processing on the cooling fan PWM control signal according to the discrete time series sampling point set to obtain a motor control sequence;

[0024] Based on the motor control sequence, an electromagnetic field energy storage process is calculated for the current density distribution in the motor winding to obtain an electromagnetic field intensity distribution sequence;

[0025] According to the electromagnetic field intensity distribution sequence, the mechanical vibration equation is solved for the vibration force generated by the motor to obtain the motor output vibration sequence, and the vibration response of the mechanical resonance characteristics of the PBT fan blade is calculated to obtain the fan blade vibration response sequence;

[0026] Based on the wind blade vibration response sequence, the energy attenuation process of the double ball bearing is calculated to obtain the bearing vibration transfer sequence;

[0027] Performing time correlation analysis on the bearing vibration transmission sequence, and calculating the vibration transmission timing relationship between the components through time delay and phase difference, to obtain the time characteristic sequence of vibration transmission between the components;

[0028] The motor control sequence, the electromagnetic field intensity distribution sequence, the motor output vibration sequence, the fan blade vibration response sequence and the time characteristic sequence of the vibration transmission between the components are synchronized and time-aligned to obtain noise energy distribution data within a preset speed range.

[0029] In combination with the first aspect, in a fourth implementation of the first aspect of the present invention, the analysis subsystem further includes:

[0030] A spatial integration unit is used to perform spatial integration processing on the noise energy distribution data to obtain a sound field density distribution function, and perform temperature correlation analysis on the sound absorption coefficient and vibration damping ratio of the PBT material according to the sound field density distribution function to obtain a noise absorption model;

[0031] A nonlinear fitting unit is used to perform nonlinear fitting on the relationship between the noise suppression effect and the speed of the cooling fan based on the noise absorption model to obtain a speed-noise mapping relationship, and dynamically model the relationship between the PWM modulation degree and the current pulsation according to the speed-noise mapping relationship to obtain a PWM-EMI coupling function;

[0032] A multi-dimensional space mapping unit, used to perform a thermal balance analysis on the PWM-EMI coupling function to obtain a heat dissipation effect function, and based on the heat dissipation effect function, perform a multi-dimensional space mapping on EMI and noise parameters to obtain a parameter space response model of a preset noise range;

[0033] A solving unit, used for substituting the speed-noise mapping relationship and the PWM-EMI coupling function into the parameter space response model to solve the optimal parameter combination, obtain the spatial distribution matrix of EMI and noise parameters, and perform feature processing on the spatial distribution matrix of EMI and noise parameters to obtain parameter space distribution data;

[0034] The frequency autocorrelation analysis unit is used to perform frequency autocorrelation analysis on the input power, rotation speed and noise suppression amount parameters in the parameter spatial distribution data to obtain EMI and noise parameter timing characteristic data.

[0035] In combination with the first aspect, in a fifth implementation of the first aspect of the present invention, the frequency autocorrelation analysis unit is specifically used to:

[0036] Performing frequency window sampling on the parameter space distribution data to obtain a parameter sampling sequence;

[0037] According to the parameter sampling sequence, an autocorrelation function is calculated for the time series relationship between the input power and the noise suppression amount, and a power-noise correlation function is obtained by an integral operation;

[0038] Based on the parameter sampling sequence, an autocorrelation function is calculated for the time series relationship between the rotation speed and the noise suppression amount, and a rotation speed-noise correlation function is obtained by integral operation;

[0039] According to the power-noise correlation function and the speed-noise correlation function, correlation analysis is performed on the time lag effect between the parameters to obtain parameter response delay data;

[0040] Based on the parameter response delay data, the parameter cross-correlation strength of the input power, the rotation speed and the noise suppression amount is calculated to obtain parameter correlation strength data, and the parameter correlation strength data is subjected to eigenvalue decomposition to obtain a parameter main eigenvector;

[0041] Orthogonal projection is performed on the power-noise correlation function and the rotation speed-noise correlation function on the parameter main eigenvector to obtain EMI and noise parameter timing characteristic data.

[0042] In combination with the first aspect, in a sixth implementation of the first aspect of the present invention, the real-time adjustment subsystem is specifically used for:

[0043] Inputting the EMI and noise parameter timing characteristic data into the spatial feature extraction module of the prediction model to extract spatial correlation features to obtain a spatial feature mapping matrix;

[0044] Inputting the spatial feature mapping matrix into the time dependency analysis module of the prediction model to extract the timing dependency features, wherein the time dependency analysis module includes four layers of linear encoders adapted to a preset speed range, to obtain a time feature vector;

[0045] Performing feature fusion on the spatial feature mapping matrix and the temporal feature vector to obtain a deep feature representation, and inputting the deep feature representation into two layers of spatiotemporal graph convolution layers to perform dynamic feature extraction of EMI and noise parameters to obtain a suppression effect prediction value within a preset noise range;

[0046] Performing error calculation between the predicted value and the actual noise suppression effect to obtain a noise correction amount;

[0047] According to the noise correction amount, a PID controller is used to calculate a PWM duty cycle adjustment parameter of the cooling fan, and a dynamic adjustment parameter of the rotation speed is calculated based on the PWM adjustment parameter;

[0048] The PWM adjustment parameter and the dynamic adjustment parameter are combined into a PWM control correction parameter, and the output power and the rotation speed of the cooling fan are adjusted in real time in a closed loop according to the PWM control correction parameter.

[0049] In the technical solution provided by the present invention, the fan motor, fan blades, bearings and housing components are accurately modeled for electromagnetic energy transfer through the modeling subsystem, and the electromagnetic interference transmission network constructed by the extraction subsystem is combined to clearly identify the EMI transmission path and the vibration coupling path, so as to realize the visual analysis of the EMI propagation mechanism and provide an accurate theoretical basis for interference suppression. The calculation subsystem of the system performs time discrete sequence calculation on the vibration transmission between the PBT fan blades and the motor, and realizes the dynamic analysis of the EMI and noise characteristics under different working conditions, while the analysis subsystem realizes the coordinated optimization of EMI suppression, noise control and heat dissipation performance by establishing the speed-noise mapping relationship and the PWM-EMI coupling function. In particular, the real-time adjustment subsystem of the system performs accurate spectrum analysis on the parameters based on the frequency autocorrelation analysis unit, and combines the dynamic feature extraction capability of the multi-layer spatiotemporal graph convolution layer to implement real-time closed-loop regulation of the output power and speed of the cooling fan, which not only effectively solves the problem that EMI and noise suppression and heat dissipation efficiency are difficult to balance in traditional technologies, but also enables the system to have adaptive learning capabilities, and can automatically optimize the control strategy for the unique characteristics of different models of cooling fans, which is particularly suitable for application scenarios where the working environment and load change frequently, and greatly improves the versatility and scalability of the system.

[0050] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention are realized and obtained by the structures particularly pointed out in the description, claims and drawings.

[0051] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 Schematic diagram of an embodiment of an anti-EMI cooling fan filtering and noise suppression system with dynamic coupling path control in an embodiment of the present invention. DETAILED DESCRIPTION

[0053] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0054] The terms "including" and "having" and any variations thereof mentioned in the embodiments of the present invention are intended to cover non-exclusive inclusions. For example, a process, system, product, or device end including a series of steps or units is not limited to the listed steps or units, but may optionally include other steps or units that are not listed, or may optionally include other steps or units that are inherent to these processes, systems, products, or device ends.

[0055] To facilitate understanding of this embodiment, firstly, a dynamic coupling path controlled anti-EMI cooling fan filtering and noise suppression system disclosed in an embodiment of the present invention is described in detail. Figure 1 As shown, the system includes the following steps:

[0056] The modeling subsystem 101 is used to perform electromagnetic energy transfer modeling on the motor, fan blades, bearings and housing components in the cooling fan to obtain an EMI energy balance equation group;

[0057] It is understandable that the execution subject of the present invention can be an anti-EMI cooling fan filtering and noise suppression device with dynamic coupling path control, or a terminal or a server, which is not limited here. The embodiment of the present invention is described by taking a server as the execution subject as an example.

[0058] Specifically, the electromagnetic coupling efficiency between the motor winding and the rotor magnetic field of the cooling fan under the input voltage is calculated, and the interaction between the current in the motor winding and the rotor magnetic field is considered. By analyzing the change of the magnetic field and the response of the current, a power conversion efficiency function is established to reflect the electromagnetic energy conversion efficiency of the motor under different load and speed conditions. According to the power conversion efficiency function, the vector field calculation of the magnetic flux density distribution in the motor is performed, and the magnetic flux density intensity and direction at different positions inside the motor are described in the form of a vector field to obtain the electromagnetic field distribution function. This distribution function is a multidimensional function of space and time, and its form is ,in represents the magnetic flux density, is the spatial coordinate, The time dimension is the time dimension. According to the electromagnetic field distribution function, the electromagnetic radiation intensity generated by the motor is analyzed by spectrum analysis, and the complex time domain signal is converted into frequency domain information, so as to identify the main frequency components of the electromagnetic interference generated by the motor during operation. Through Fourier transform, the electromagnetic field change data in the time domain is converted into frequency domain signals to obtain electromagnetic interference spectrum data. The energy distribution of electromagnetic radiation at each frequency is displayed, and the characteristics of the interference source and the electromagnetic wave propagation characteristics within a specific frequency range are revealed. On this basis, according to the electromagnetic interference spectrum data, the radiated EMI and conducted EMI components generated by the motor are decomposed in the frequency domain. Through the decomposition process, the wide-band interference signal is separated into multiple independent interference modes. Each interference mode corresponds to a specific frequency component and propagation path, and the interference mode distribution function is obtained. The EMI shielding characteristics of the PBT fan blade and plastic shell of the cooling fan are attenuated and calculated. The electromagnetic shielding effect of the fan blade and shell materials is physically modeled, including the reflection, absorption and transmission characteristics of the material to electromagnetic waves. The EMI attenuation coefficient matrix is ​​obtained by calculating the attenuation coefficient of the material to electromagnetic waves of different frequencies. According to the EMI attenuation coefficient matrix, the propagation channel of the interference signal is analyzed. The components such as the motor, fan blades, and housing are regarded as network nodes, and the propagation path of electromagnetic interference is regarded as the edge in the network. The EMI transfer matrix is ​​constructed by calculating the attenuation and propagation characteristics on different paths. The power conversion efficiency function, electromagnetic field distribution function, electromagnetic interference spectrum data, interference mode distribution function, and EMI transfer matrix are calculated together to establish a mathematical model of the electromagnetic energy transfer relationship of each component of the cooling fan. The process of simultaneous calculation involves solving multiple linear and nonlinear equations. Through numerical calculation methods such as finite element analysis or time-domain finite difference method, the EMI transfer equation group between each component is solved. The EMI transfer equation group is subjected to eigenvalue decomposition and normalization. Through eigenvalue decomposition, the complex multidimensional system is decomposed into a group of relatively independent modes, each of which corresponds to a specific electromagnetic interference propagation mode. In the process of eigenvalue decomposition, the most significant EMI transfer path and influencing factors in the system are identified by calculating eigenvalues ​​and eigenvectors. At the same time, through normalization processing, the parameters of different dimensions are standardized, so that different physical quantities (such as input voltage, motor operating power, double ball bearing mechanical quality factor, electromagnetic coupling coefficient, etc.) can be compared and analyzed on the same scale. The EMI energy balance equation group is obtained.

[0059] An extraction subsystem 102 is used to construct an electromagnetic interference transmission network according to the EMI energy balance equation group, and extract the electromagnetic interference transmission relationship between the components of the cooling fan;

[0060] Specifically, based on the EMI energy balance equations, the power input terminal, motor control circuit, stator winding, rotor, double ball bearing, PBT fan blade and housing are set as network nodes, and the EMI propagation relationship between network nodes is mapped by directed graph to obtain the topological structure of the initial EMI network. Each node represents a physical component in the system, and the directed edges between nodes represent the specific path of electromagnetic interference propagation from one component to another. The direction of the path represents the direction of energy flow, and the weight of the path is related to the transmission efficiency and interference intensity. The first channel is identified for the EMI propagation path from the power supply to the motor winding in the initial EMI network topology. By analyzing how the high-frequency noise component in the power input signal enters the stator winding through the motor control circuit and then passes through the rotor to other components, the first EMI transfer path matrix is ​​obtained. The elements in the matrix represent the electromagnetic interference transfer coefficients between the components, thereby quantifying the efficiency of electromagnetic energy transfer at each step in the first channel. At the same time, the second channel identification is performed on the mechanical vibration coupling path between the PBT fan blade and the plastic shell in the initial EMI network topology. By analyzing how the air disturbance and mechanical vibration caused by the rotation of the fan blade are transmitted to the shell through the double ball bearing, the second vibration coupling path matrix is ​​obtained. Each element of this matrix represents the energy loss coefficient when the mechanical vibration is transmitted between different components. Based on the first EMI transmission path matrix, the electromagnetic interference transmission efficiency between adjacent components in the first channel is calculated. The quantification process analyzes the propagation characteristics of the electromagnetic field, the impedance matching relationship between components, and the electromagnetic absorption characteristics of the dielectric material to obtain the first channel EMI coupling coefficient, which represents the coupling strength of electromagnetic energy from one component to another at a specific frequency. For the mechanical vibration transmission path in the second channel, based on the second vibration coupling path matrix, the vibration transmission characteristics of the double ball bearing are analyzed. By calculating the mechanical damping, friction loss and structural resonance effect of the bearing at different vibration frequencies, the vibration loss coefficient of the second channel is obtained, which reflects the degree of energy attenuation of mechanical vibration when passing through the bearing, as well as the loss characteristics of vibration signals of different frequencies during the propagation process. The time characteristics of the EMI and vibration transmission processes in the first channel and the second channel are analyzed respectively to obtain the time delay parameters of electromagnetic interference and mechanical vibration at different fan speeds. When performing the time characteristics analysis, the propagation process of the interference signal is discretized into multiple time segments through the time domain analysis method, and the time required for the signal to propagate from one node to the next node is calculated. The time delay parameter depends not only on the physical distance between the components, but also on the electromagnetic wave propagation speed, mechanical vibration frequency and dielectric impedance. Combined with these time delay parameters, a dynamic transfer function model is constructed. The extraction subsystem substitutes the first channel EMI coupling coefficient, the second channel vibration loss coefficient and the time delay parameter into the transfer function model. Through quantitative analysis of the model, the electromagnetic interference transmission relationship between the components of the cooling fan is calculated.

[0061] The calculation subsystem 103 is used to perform a time discrete sequence calculation on the vibration transmission between the PBT fan blade and the motor based on the electromagnetic interference transmission relationship, and obtain the noise energy distribution data within a preset speed range;

[0062] Specifically, the discrete time series sampling point set is determined according to the electromagnetic interference transmission relationship, the continuous electromagnetic interference signal is discretized into a series of values ​​sampled at specific time points, and the EMI signal in the system is converted from the time domain to the frequency domain through the discrete Fourier transform or short-time Fourier transform method, and the phase and amplitude information of the key frequency components are extracted to construct a discrete time series sampling point set. In order to ensure the sampling accuracy, the sampling frequency is determined according to the minimum time delay and maximum frequency response of the system to avoid the aliasing effect caused by the low sampling rate, and ensure that the sampling point set accurately reflects the dynamic change characteristics of the EMI signal. According to the discrete time series sampling point set, the cooling fan PWM (pulse width modulation) control signal is piecewise linearized to obtain the motor control sequence. In the piecewise linearization process, the complex nonlinear PWM signal is decomposed into multiple linear intervals, and the actual PWM waveform is approximated by a linear function in each interval through the least squares method or linear interpolation method, thereby greatly simplifying the calculation complexity of the control signal and improving the response speed of the system in the real-time control process. Based on the motor control sequence, the electromagnetic field energy storage process is calculated for the current density distribution in the motor winding. By solving Maxwell's equations, especially Faraday's law of electromagnetic induction and Ampere's loop law, the electromagnetic field intensity distribution sequence inside the motor is calculated. In this calculation process, material properties (such as winding conductivity and magnetic permeability) and geometric structure parameters (such as winding arrangement and rotor shape) are introduced. The three-dimensional electromagnetic field distribution is discretized into specific values ​​in the space grid through the finite element method or the time-domain finite difference method, so that the system can depict the dynamic changes of the electromagnetic field with high precision. According to the electromagnetic field intensity distribution sequence, the mechanical vibration equation is solved for the vibration force generated by the motor to obtain the vibration sequence output by the motor. In the process of solving the mechanical vibration equation, consider how the electromagnetic attraction between the motor rotor and stator is converted into a vibration response in the mechanical structure. Apply Newton's second law combined with the motion equation of the resonant system (such as the simple harmonic vibration equation) to convert the transient changes of the electromagnetic field into acceleration, velocity and displacement signals of mechanical vibration. At the same time, the computing subsystem calculates the vibration response of the mechanical resonance characteristics of the PBT fan blade. By establishing a mechanical model of the fan blade, considering the material elastic modulus, geometric shape of the fan blade, and the relationship between the wind speed and the fan blade speed, the fan blade vibration response sequence is obtained, which reflects the dynamic response of the fan blade when it is subjected to motor vibration transmission, and reveals the resonance phenomenon of the fan blade at a specific frequency. Based on the fan blade vibration response sequence, the energy attenuation process of the double ball bearing is calculated to obtain the vibration transmission sequence of the bearing. The mechanical damping characteristics and friction effect of the double ball bearing are analyzed, and the loss of vibration energy in the bearing is calculated through the law of conservation of energy, and the attenuation of vibration energy from the fan blade to the housing is quantified.When calculating the vibration transmission sequence of the bearing, the mechanical quality factor (Q value) of the bearing is analyzed. This factor characterizes the energy loss characteristics of the bearing during the mechanical vibration transmission process. Combined with the actual structural parameters of the bearing (such as ball size and material friction coefficient), the vibration transmission sequence data is obtained. In order to analyze the timing relationship of vibration transmission between components, the bearing vibration transmission sequence is subjected to time correlation analysis. By calculating the time delay and phase difference, the time characteristic sequence of vibration transmission between components is constructed. In the time correlation analysis, the cross-correlation function is applied to analyze the timing matching degree between the vibration signals of different components, so as to identify the delay characteristics of the vibration signal when propagating between components, quantify the speed of mechanical vibration propagation in the system, and reveal the phase change of the vibration signal during the propagation process. The motor control sequence, electromagnetic field intensity distribution sequence, motor output vibration sequence, fan blade vibration response sequence and the time characteristic sequence of vibration transmission between components are synchronized and time-aligned. By constructing a unified time domain analysis framework, the data points in each sequence are mapped to the same time axis, the time deviation between different data sources is eliminated, and the noise energy distribution data within the preset speed range is obtained.

[0063] The analysis subsystem 104 is used to determine the parameter space distribution within a preset noise range according to the noise energy distribution data, and perform frequency autocorrelation analysis on the parameter space distribution data to obtain EMI and noise parameter timing characteristic data;

[0064] Specifically, the noise energy distribution data is spatially integrated by a spatial integration unit, and the discrete noise energy data is converted into a continuous sound field density distribution function. The sound pressure level distribution at different spatial positions is calculated by a three-dimensional spatial integration method (such as volume integration or surface integration), and a sound field model is constructed. Based on the sound field density distribution function, the sound absorption coefficient and vibration damping ratio of the PBT material are analyzed for temperature correlation. In the analysis process, by associating the physical properties of the material with the ambient temperature change, the change of the sound absorption coefficient of the material under different temperature conditions is calculated using experimental data or material models (such as the Delany-Bazley model or the JCA model). At the same time, the vibration damping ratio of the PBT material is analyzed, and the vibration energy loss characteristics of the material under different frequency and temperature conditions are obtained by a dynamic mechanical analysis method. These temperature correlation data are comprehensively processed to construct a noise absorption model. Based on the noise absorption model, the nonlinear fitting unit performs nonlinear fitting on the relationship between the noise suppression effect and the speed of the cooling fan. Since the relationship between the noise suppression effect and the fan speed presents complex nonlinear characteristics (for example, a resonance peak is generated at a specific speed), a nonlinear algorithm such as polynomial fitting, neural network fitting or support vector machine is used to construct a speed-noise mapping relationship. Through this mapping relationship, the noise level generated by the cooling fan at different speeds and the fan speed range required under specific noise suppression targets are quantified. According to the speed-noise mapping relationship, the relationship between PWM modulation and current pulsation is dynamically modeled. By analyzing the impact of the change in PWM signal duty cycle on the motor drive current and combining the electromagnetic coupling characteristics in the motor windings, the PWM-EMI coupling function is obtained. This function is used to describe how current pulsation affects the EMI level in the system through the motor control signal. The multidimensional space mapping unit performs a thermal balance analysis on the PWM-EMI coupling function to obtain the heat dissipation effect function. In the heat dissipation effect analysis, the heat generation, transfer and dissipation process during the operation of the cooling fan is modeled, and the energy balance equation in heat transfer is used to calculate the heat dissipation performance of the fan at different speeds and PWM control signals to obtain the heat dissipation effect function. On the basis of obtaining the heat dissipation effect function, the multi-dimensional space mapping unit performs multi-dimensional space mapping on the EMI and noise parameters, and establishes a parameter space response model by taking different parameters (such as input power, fan speed, PWM modulation, current pulsation, heat dissipation effect, etc.) as mapping dimensions. This model predicts the EMI and noise response characteristics of the system under specific input conditions through multi-dimensional interpolation or regression methods. Based on the solution unit, the speed-noise mapping relationship and PWM-EMI coupling function are substituted into the parameter space response model to solve the optimal parameter combination.In the solution process, by establishing an objective function (such as minimizing EMI and maximizing noise suppression), and introducing constraints (such as fan speed range, input power limit, system heat dissipation capacity, etc.), linear programming, nonlinear programming or particle swarm optimization algorithm is used to obtain the spatial distribution matrix of EMI and noise parameters. Each element in this matrix represents the EMI level and noise suppression effect of the system under specific input conditions. By performing feature processing on this matrix, such as singular value decomposition or principal component analysis, the most representative parameter combination is extracted to obtain parameter space distribution data. The frequency autocorrelation analysis unit performs frequency autocorrelation analysis on the input power, speed and noise suppression parameters in the parameter space distribution data. By calculating the autocorrelation function of different parameters in the time series, the characteristics of periodic noise and EMI in the system are analyzed. The autocorrelation analysis can reveal the periodic trend in parameter changes and quantify the contribution of noise sources at different frequencies, thereby identifying the main noise sources and electromagnetic interference sources in the system. By combining the frequency autocorrelation analysis results with the parameter space response model, the time series characteristic data of EMI and noise parameters are obtained.

[0065] The parameter spatial distribution data is sampled by frequency window. By selecting a suitable sampling window, the long time series data is divided into multiple short time period sampling sequences, thereby ensuring the frequency resolution while improving the accuracy of time domain analysis. In the frequency window sampling process, the appropriate window size is selected according to the operating frequency range and noise characteristics of the system, and the influence of the edge effect on the spectrum analysis results is reduced by smoothing to obtain the parameter sampling sequence. According to the parameter sampling sequence, the autocorrelation function of the time series relationship between the input power and the noise suppression amount is calculated. The autocorrelation analysis quantifies the influence of the input power change on the noise suppression effect by analyzing the correlation of the signal itself under different time lags of the time series, and identifies the potential periodic relationship between the power regulation and noise change in the system. By smoothing the time series of the input power and comparing it with the change curve of the noise suppression amount, a power-noise correlation function describing the relationship between the two is obtained. In this process, the signal changes at different time points are accumulated and calculated by integral operation to form a function model describing the dynamic relationship between the input power and the noise suppression amount. At the same time, based on the parameter sampling sequence, the autocorrelation function of the time series relationship between the speed and the noise suppression amount is calculated, and the speed-noise correlation function is obtained by integral operation. By analyzing the similarity between the speed change signal and the noise suppression amount signal, the influence of the speed adjustment on the noise suppression effect is obtained, and the noise resonance phenomenon generated by the system at a specific speed is identified. The speed-noise correlation function can quantify the noise level generated by the fan under different speed conditions, as well as the optimal control range of the fan speed under a specific noise suppression target. According to the power-noise correlation function and the speed-noise correlation function, the time lag effect between the parameters is correlated and analyzed to obtain the parameter response delay data. In the correlation analysis process, the time delay characteristics between the input power, speed change and the noise suppression amount response are identified by calculating the time lag characteristics between different parameters. The analysis results reveal the delayed response time of the noise suppression effect after the input power is adjusted, and the time interval of the noise suppression amount change caused by the speed change. Based on the parameter response delay data, the frequency autocorrelation analysis unit calculates the parameter cross-correlation strength of the input power, speed and noise suppression amount to obtain the parameter correlation strength data. By analyzing the contribution of different parameters to the system EMI and noise characteristics, it is identified which parameters have the greatest impact on the noise suppression effect under different operating conditions. In order to extract key features, the parameter correlation intensity data is decomposed through the eigenvalue decomposition method to obtain the main eigenvectors of the parameters. These main eigenvectors represent the most important parameter dimensions in the system. The key parameters in the system are projected into a low-dimensional feature space through the principal component analysis method to improve the efficiency and accuracy of the subsequent calculation process.The power-noise correlation function and the speed-noise correlation function are orthogonally projected on the parameter main eigenvector. By constructing an orthogonal basis, the data in the high-dimensional parameter space is projected into the feature space defined by the main eigenvector, so as to realize the extraction of the EMI and noise parameter time series feature data. In the orthogonal projection process, the projection coefficients of each eigenvector in the time series are obtained by calculating the inner product between the correlation function and the main eigenvector. These coefficients reflect the weight of each parameter in the dynamic response of the system and reveal the interaction relationship between different parameters.

[0066] The real-time adjustment subsystem 105 is used to input the EMI and noise parameter timing characteristic data into the prediction model for characteristic calculation, obtain the PWM control correction parameters, and adjust the output power and speed of the cooling fan in real time according to the PWM control correction parameters.

[0067] Specifically, the EMI and noise parameter time series feature data are input into the spatial feature extraction module of the prediction model. This module converts the high-dimensional time series data into a low-dimensional spatial feature mapping matrix by analyzing the distribution characteristics of the time series feature data in different spatial dimensions and using convolution operations and feature mapping methods. The spatial feature extraction module extracts the main interference components and noise transmission paths in the system by identifying the spatial correlation between different noise sources and EMI paths, so that the subsequent time series analysis can focus on the most representative parameter features, thereby effectively reducing the computational complexity and improving the generalization ability of the prediction model. The spatial feature mapping matrix is ​​input into the time dependency analysis module of the prediction model, which includes four layers of linear encoders adapted to the preset speed range. By performing multi-layer linear transformation on the spatial feature data, the complex time series features are decomposed into multi-dimensional time feature vectors. In this process, each layer of the linear encoder will perform in-depth mining of the time series characteristics of the input data through weighted calculation and activation function, thereby identifying the trend and periodic characteristics of the changes in noise and EMI parameters in different time periods in the system. The time dependency analysis module can provide the time dynamic characteristics of the input parameters and automatically adapt to the distribution characteristics of the time series data under different fan speed conditions, thereby ensuring that the prediction model can maintain stable time series analysis capabilities under different working conditions. Through the structural design of this multi-layer linear encoder, the time dependency analysis module converts the static features in the spatial feature mapping matrix into dynamic time feature vectors. After completing the time feature extraction, the real-time adjustment subsystem performs feature fusion on the spatial feature mapping matrix and the time feature vector. Through the feature fusion operation, the interference path information in the spatial dimension and the noise change trend in the time dimension are combined to obtain a deep feature representation. The deep feature representation constructs a multi-dimensional feature space by fusing spatial and temporal characteristics. The deep feature representation is input into the two-layer spatiotemporal graph convolution layer for dynamic feature extraction of EMI and noise parameters. The spatiotemporal graph convolution layer is a neural network structure for processing spatiotemporal data. By performing graph convolution operations in the spatial dimension, the electromagnetic interference and noise propagation paths between different nodes in the system (such as fan components, motors, housings, etc.) are identified. At the same time, through sequence convolution in the time dimension, the dynamic characteristics of the noise and EMI parameters in the system changing over time are extracted. The spatiotemporal graph convolution method can identify static interference paths and capture how the noise suppression effect of the system changes over time under a specific operating state, and obtain a predicted value of the suppression effect within a preset noise range. The predicted value represents the noise suppression level that the system is expected to achieve under the current input parameter conditions. The real-time adjustment subsystem calculates the error between the predicted value and the actual noise suppression effect, and obtains the noise correction by comparing the difference between the predicted noise suppression value and the actual measured value. The noise correction represents the deviation between the current state of the system and the ideal state. Positive and negative values ​​represent whether the system needs to increase or decrease the noise suppression strength.After obtaining the noise correction amount, the PID (proportional-integral-differential) controller is used to calculate the adjustment parameters of the PWM (pulse width modulation) duty cycle of the cooling fan. The PID controller dynamically feedbacks and adjusts the noise correction amount through three links: proportional, integral and differential, ensuring that the adjustment process is smooth and there is no obvious overshoot phenomenon, and obtaining an accurate PWM adjustment parameter. Based on the PWM adjustment parameter, the dynamic adjustment parameter of the fan speed is calculated. By analyzing the influence of the PWM signal duty cycle on the motor speed, combined with the current fan operation status and noise suppression requirements, a suitable speed adjustment value is calculated, so that the fan can effectively reduce the noise level during the speed adjustment process, while ensuring that the heat dissipation efficiency of the system is not affected. After completing the calculation of the PWM adjustment parameter and the dynamic adjustment parameter, the two parameters are combined into a comprehensive PWM control correction parameter. According to the PWM control correction parameter, the output power and speed of the cooling fan are adjusted in real time in a closed loop. In the closed-loop control process, by real-time monitoring of the fan's operating parameters (such as motor current, speed, noise level, etc.), the actual measurement value is fed back to the prediction model, the noise suppression effect prediction value is continuously updated, and the control parameters are dynamically adjusted to ensure that the system always works in the best operating state.

[0068] In the embodiment of the present invention, the modeling subsystem 101 accurately models the electromagnetic energy transfer of the fan motor, blades, bearings and housing components, and combines the electromagnetic interference transmission network constructed by the extraction subsystem 102 to clearly identify the EMI transmission path and the vibration coupling path, thereby realizing the visualization analysis of the EMI propagation mechanism and providing an accurate theoretical basis for interference suppression. The calculation subsystem 103 of the system performs a time discrete sequence calculation on the vibration transmission between the PBT blades and the motor, realizing the dynamic analysis of the EMI and noise characteristics under different working conditions, and the analysis subsystem 104 realizes the coordinated optimization of EMI suppression, noise control and heat dissipation performance by establishing a speed-noise mapping relationship and a PWM-EMI coupling function. In particular, the real-time adjustment subsystem 105 of the system performs precise spectral analysis of parameters based on the frequency autocorrelation analysis unit, and combines the dynamic feature extraction capability of the multi-layer spatiotemporal graph convolution layer to implement real-time closed-loop adjustment of the output power and speed of the cooling fan. This not only effectively solves the problem of balancing EMI and noise suppression with cooling efficiency in traditional technologies, but also enables the system to have adaptive learning capabilities, and can automatically optimize control strategies according to the unique characteristics of cooling fans of different models. It is particularly suitable for application scenarios where the working environment and load frequently change, greatly improving the versatility and scalability of the system.

[0069] In a specific embodiment, the modeling subsystem 101 is specifically used for:

[0070] The efficiency of the electromagnetic coupling between the motor winding and the rotor magnetic field of the cooling fan under the input voltage is calculated to obtain a power conversion efficiency function, and based on the power conversion efficiency function, a vector field calculation is performed on the magnetic flux density distribution in the motor to obtain an electromagnetic field distribution function;

[0071] Based on the electromagnetic field distribution function, the electromagnetic radiation intensity generated by the motor is analyzed spectrally to obtain electromagnetic interference spectrum data. Based on the electromagnetic interference spectrum data, the radiation and conduction EMI components are decomposed in the frequency domain to obtain the interference mode distribution function.

[0072] The EMI shielding characteristics of the PBT fan blades and plastic shell of the cooling fan are attenuated and calculated to obtain the EMI attenuation coefficient matrix. Based on the EMI attenuation coefficient matrix, the propagation channel of the interference signal is analyzed to obtain the EMI transfer matrix.

[0073] The power conversion efficiency function, electromagnetic field distribution function, electromagnetic interference spectrum data, interference mode distribution function and EMI transfer matrix are calculated simultaneously to obtain the EMI transfer equations between the components;

[0074] The EMI transfer equations are subjected to eigenvalue decomposition and normalization to obtain the EMI energy balance equations including input voltage, motor operating power, double ball bearing mechanical quality factor and electromagnetic coupling coefficient.

[0075] Specifically, an electromagnetic model of the motor is established. In this model, the stator winding of the motor generates an alternating magnetic field by applying an AC voltage. This magnetic field interacts with the permanent magnets or windings on the rotor through electromagnetic coupling, thereby realizing the conversion of electrical energy into mechanical energy. In order to quantify this conversion efficiency, the power conversion efficiency function is defined as the ratio of the output mechanical power to the input electrical power, and its expression is:

[0076]

[0077] in, represents the power conversion efficiency, is the mechanical power, is the electrical power. The input electrical power is and the current in the motor windings The calculation results are:

[0078]

[0079] in, is the phase difference between current and voltage, Indicates power factor. The output mechanical power is converted into torque by the motor. and speed The calculation results are:

[0080]

[0081] Substituting the above two formulas into the formula of power conversion efficiency, we get:

[0082]

[0083] In a motor, torque With magnetic flux and current Related, through the magnetic flux density The calculation results are:

[0084]

[0085] in, is the torque constant of the motor, and the magnetic flux is the magnetic flux density Integration over the rotor cross section:

[0086]

[0087] After obtaining the power conversion efficiency function, the magnetic flux density distribution in the motor is calculated using the vector field. Electromagnetic field distribution function Describes the magnitude and direction of the magnetic flux density in the internal space of the motor. , and are spatial coordinates. This distribution is obtained by solving Maxwell's equations:

[0088]

[0089]

[0090]

[0091]

[0092] in, is the electric field strength, is the magnetic field strength, is the current density, is the electric flux density, is the charge density. These equations describe the distribution characteristics of the electromagnetic field inside the motor, and the specific electromagnetic field distribution function is obtained through numerical calculation. Based on the electromagnetic field distribution function, the electromagnetic radiation intensity generated by the motor is analyzed by spectrum. Spectral analysis converts the time domain signal into frequency domain representation, and decomposes the time variation characteristics of the electromagnetic field distribution into different frequency components through Fourier transform to obtain the spectrum data of electromagnetic interference (EMI). Spectrum data Describe the different frequencies The intensity of electromagnetic radiation under the condition of low power consumption. Through the spectrum data, the radiated and conducted EMI components are decomposed in the frequency domain, and the wide-band interference signal is decomposed into multiple narrow-band signals to obtain the interference mode distribution function. This process reveals the source of electromagnetic interference of different frequency components and its propagation characteristics in the system. In order to analyze the EMI shielding characteristics of the PBT fan blades and plastic shell of the cooling fan, the shielding effect of these materials is attenuated. EMI attenuation coefficient matrix Describe electromagnetic signals from components Propagate to components The degree of attenuation when the material is reflected is calculated from the reflection, absorption and transmission characteristics of the material. For example, the absorption loss of PBT material is calculated from the material's loss factor. and material thickness The calculation results are:

[0093]

[0094] in, is the attenuation constant of the material, which is related to the conductivity of the material , magnetic permeability and frequency After obtaining the EMI attenuation coefficient matrix, the propagation channel of the interference signal is analyzed according to these coefficients to obtain the EMI transfer matrix. , where each element represents the efficiency of interference signal transmission from one component to another. After all models and functions are constructed, the power conversion efficiency function, electromagnetic field distribution function, electromagnetic interference spectrum data, interference mode distribution function and EMI transfer matrix are calculated jointly to obtain the EMI transfer equations between components. These equations integrate the electromagnetic coupling relationship between the motor, circuit, fan blade and shell in the system to form a comprehensive system description. By performing eigenvalue decomposition and normalization on the EMI transfer equations, the complex multidimensional system is simplified into a set of eigenmodes, each of which represents the EMI energy distribution state under specific conditions. Eigenvalue decomposition decomposes the EMI transfer matrix into eigenvalues ​​and eigenvectors. By analyzing the size of the eigenvalues, the most significant interference paths in the system are identified, and the eigenvectors describe the electromagnetic field distribution characteristics in these paths. Normalization converts data of different dimensions into dimensionless form, so that parameters such as input voltage, motor operating power, mechanical quality factor of double ball bearings and electromagnetic coupling coefficient can be compared on the same scale. The final EMI energy balance equations can describe the electromagnetic energy transfer process in the system and provide accurate model support for real-time control systems.

[0095] In a specific embodiment, the extraction subsystem 102 is specifically used for:

[0096] Based on the EMI energy balance equations, the power input terminal, motor control circuit, stator winding, rotor, double ball bearing, PBT fan blade and housing are set as network nodes, and the EMI propagation relationship between network nodes is mapped by directed graph to obtain the initial EMI network topology structure.

[0097] Performing a first channel identification on the EMI propagation path from the power supply to the motor winding in the initial EMI network topology structure to obtain a first EMI transfer path matrix, and performing a second channel identification on the mechanical vibration coupling path between the fan blade and the housing in the initial EMI network topology structure to obtain a second vibration coupling path matrix;

[0098] Based on the first EMI transfer path matrix, the electromagnetic interference transfer efficiency between adjacent components in the first channel is quantitatively calculated to obtain the first channel EMI coupling coefficient, and according to the second vibration coupling path matrix, the vibration transfer characteristics of the double ball bearings in the second channel are analyzed to obtain the second channel vibration loss coefficient;

[0099] The time characteristics of the EMI transfer process in the first channel and the second channel are analyzed respectively to obtain the time delay parameters at different speeds. The EMI coupling coefficient of the first channel, the vibration loss coefficient of the second channel and the time delay parameters are substituted into the transfer function model to quantitatively analyze the EMI flow process and obtain the electromagnetic interference transmission relationship between the components of the cooling fan.

[0100] Specifically, based on the EMI energy balance equations, the key components in the system, including the power input, motor control circuit, stator winding, rotor, double ball bearing, PBT fan blade and casing, are set as network nodes, and a directed graph mapping model is constructed through the electromagnetic interference (EMI) propagation relationship between these nodes to form an initial EMI network topology. In this topology, each network node represents a physical component, and the directed edges between the nodes represent the specific path of EMI propagation from one component to another. The direction of the path represents the direction of electromagnetic energy flow, and the weight of the path is related to the transmission efficiency of electromagnetic interference. The first channel is identified for the EMI propagation path from the power supply to the motor winding in the initial EMI network topology to obtain the first EMI transfer path matrix. , where 𝑖 and 𝑗 represent the starting and target nodes in the network respectively, and the elements in the matrix represent the slave nodes To Node EMI transfer efficiency. It is expressed as:

[0101]

[0102] in, Is electromagnetic interference from the node To Node The energy transmitted, Is a node The total input energy is quantified by this formula, which allows the system to identify key electromagnetic interference paths. For the mechanical vibration coupling path between the PBT fan blade and the plastic housing in the initial EMl network topology, the second vibration coupling path matrix is ​​obtained by analyzing how the mechanical vibration of the fan blade is transmitted to the housing through the double ball bearing. , where each element represents the energy loss coefficient of mechanical vibration during propagation. Vibration loss coefficient The calculation formula is:

[0103]

[0104] in, is the mechanical damping coefficient in the path, is the distance that vibration propagates. The vibration loss coefficient represents the energy attenuation of mechanical vibration when it is transmitted between components. These coefficients are used to analyze which paths in the system have the greatest impact on the attenuation of vibration energy. The electromagnetic interference transmission efficiency and mechanical vibration loss characteristics in the first EMI transfer path matrix and the second vibration coupling path matrix are quantitatively calculated. Based on the EMI transfer path matrix in the first channel, the EMI coupling coefficient of each path is calculated by analyzing the electromagnetic field coupling characteristics between adjacent components. EMI coupling coefficient The calculation method is:

[0105]

[0106] in, and Respectively represent nodes and The electric field strength at and is the electrical impedance of the corresponding node. The EMI coupling coefficient indicates the efficiency of electromagnetic field transmission from one component to another. A larger coefficient indicates a stronger coupling effect, which is prone to introduce higher electromagnetic interference. Similarly, according to the second vibration coupling path matrix, the vibration transmission characteristics of the double ball bearing in the second channel are analyzed to obtain the vibration loss coefficient , which represents the energy attenuation characteristics in the process of mechanical vibration being transmitted from the wind blade to the housing. In order to analyze the time characteristics in these paths, the time characteristics of the EMI and vibration transmission process in the first channel and the second channel are analyzed, and the time delay parameters at different speeds are calculated. :

[0107]

[0108] in, is the path length, It is the propagation speed of EMI or mechanical vibration in the path. This time delay parameter is used to describe the time required for the signal to be transmitted from one component to another. Under high speed conditions, the time delay parameter will significantly change the phase relationship between EMI and noise, so special attention should be paid to this characteristic when designing the control system. Substitute the first channel EMI coupling coefficient, the second channel vibration loss coefficient and the time delay parameter into the transfer function model to quantitatively analyze the flow process of EMI and vibration signals. Transfer function model Used to describe the dynamic relationship between input and output signals, where is a complex frequency variable:

[0109]

[0110] in, and are the Laplace transforms of the output and input signals of the system, respectively, is the transfer coefficient, is the time delay parameter. This model is used to analyze the propagation characteristics of signals of different frequencies in the system, as well as the response characteristics of the system to EMI and noise of different frequencies. By combining the EMI coupling coefficient, vibration loss coefficient and transfer function model, the electromagnetic interference transmission relationship between the components of the cooling fan is finally obtained. This relationship is expressed through the EMI transfer matrix express:

[0111]

[0112] in, is the angular frequency of the signal. This matrix describes the propagation path and transmission efficiency of electromagnetic interference and mechanical vibration between components in the system, and combines specific system parameters to calculate the EMI intensity and noise level of each component at a specific frequency.

[0113] In a specific embodiment, the computing subsystem 103 is specifically used for:

[0114] A discrete time series sampling point set is determined according to the electromagnetic interference transmission relationship, and a cooling fan PWM control signal is subjected to piecewise linearization processing according to the discrete time series sampling point set to obtain a motor control sequence;

[0115] Based on the motor control sequence, the electromagnetic field energy storage process is calculated for the current density distribution in the motor winding to obtain the electromagnetic field intensity distribution sequence;

[0116] According to the electromagnetic field intensity distribution sequence, the mechanical vibration equation is solved for the vibration force generated by the motor to obtain the motor output vibration sequence, and the vibration response of the mechanical resonance characteristics of the PBT fan blade is calculated to obtain the fan blade vibration response sequence;

[0117] Based on the vibration response sequence of the wind blade, the energy attenuation process of the double ball bearing is calculated, and the bearing vibration transfer sequence is obtained.

[0118] Perform time correlation analysis on the bearing vibration transmission sequence, and calculate the vibration transmission timing relationship between components through time delay and phase difference to obtain the time characteristic sequence of vibration transmission between components;

[0119] The motor control sequence, electromagnetic field intensity distribution sequence, motor output vibration sequence, fan blade vibration response sequence and the time characteristic sequence of vibration transmission between components are synchronized and time-aligned to obtain the noise energy distribution data within the preset speed range.

[0120] Specifically, an electromagnetic interference (EMI) transmission relationship model is constructed. Assume that in the frequency domain, the EMI signal Over time Change, its transfer relationship is expressed as a transfer function With input signal The product of:

[0121]

[0122] in, is the frequency domain transfer function, which represents the response characteristics of the system to EMI signals at different frequencies, and It is the spectrum representation of the input signal. By performing an inverse Fourier transform on the EMI transfer relationship, the EMI signal in the time domain is obtained. , and discretize it into a set of time series sampling points:

[0123]

[0124] The sampling point set is represented as ,in is a discrete time series index with a sampling frequency of , then the sampling point is is the sampling period. The discrete time series sampling point set is used to perform piecewise linearization processing on the cooling fan PWM control signal. In the piecewise linearization processing, the duty cycle change of the PWM signal is decomposed into multiple linear segments, and each segment is approximated by a linear function, thereby simplifying the complex PWM waveform analysis and obtaining the motor control sequence :

[0125]

[0126] in, and It is The linear fitting coefficient of the segment, and are the start and end sampling point indexes of this segment respectively. Based on the motor control sequence, the electromagnetic field energy storage process is calculated for the current density distribution in the motor winding to obtain the electromagnetic field intensity distribution sequence. and electromagnetic field strength The relationship between is described by Ampere's circuit law:

[0127]

[0128] in, is the magnetic field strength, is the electric flux density. In steady-state conditions, The term can be ignored, and the relationship between magnetic field intensity and current density is obtained:

[0129]

[0130] in, is the magnetic permeability of the material, is the conductivity, is the electric field strength, is the magnetic vector potential. Through numerical calculation methods (such as finite element method), the electromagnetic field intensity distribution sequence at different time points is obtained Based on the electromagnetic field intensity distribution sequence, the mechanical vibration equation is solved for the vibration force generated by the motor to obtain the motor output vibration sequence. The vibration force in the motor is mainly generated by electromagnetic force. and magnetic flux density And the current density The relationship between is described by the Lorentz force formula:

[0131]

[0132] Through Newton's laws of motion (in It's quality. is acceleration), convert the electromagnetic force into the acceleration signal of mechanical vibration, and then obtain the speed and displacement signals through integration operation to obtain the motor output vibration sequence After obtaining the motor output vibration sequence, the vibration response of the mechanical resonance characteristics of the PBT fan blade is calculated to obtain the fan blade vibration response sequence. The fan blade is an elastic body, and its mechanical response is modeled by the simple harmonic vibration equation:

[0133]

[0134] in, is the equivalent mass of the blade, is the damping coefficient, is the elastic modulus, is the displacement, is the vibration force acting on the fan blade. By numerically solving this equation (such as the Runge-Kutta method), the fan blade vibration response sequence is obtained Based on the vibration response sequence of the wind blade, the energy attenuation process of the double ball bearing is calculated, and the bearing vibration transmission sequence is obtained. The double ball bearing exhibits significant damping characteristics during the mechanical vibration transmission process, and its energy attenuation is described by an exponential decay model:

[0135]

[0136] in, is the vibration amplitude of the bearing output, is the input vibration amplitude, is the mechanical damping coefficient of the bearing. The bearing vibration transfer sequence is obtained by convolution operation of the fan blade vibration response sequence and the bearing transfer characteristic. The time correlation analysis of the bearing vibration transmission sequence is performed. By calculating the time delay and phase difference, the vibration transmission timing relationship between the components is obtained, and the time characteristic sequence of the vibration transmission between the components is obtained. In the timing relationship analysis, the time delay and phase difference This is achieved by calculating the cross-correlation function and phase spectrum analysis, thereby revealing the response time characteristics of each component in the system and the propagation characteristics of the vibration signal. , electromagnetic field intensity distribution sequence , motor output vibration sequence , wind blade vibration response sequence Time characteristic series of vibration transmission between components Perform data synchronization and time alignment. In the time alignment process, the timestamps of different sequences are aligned by unifying the sampling frequency and interpolation method to ensure that the data is analyzed on the same time axis. Through the joint analysis of these data, the noise energy distribution data within the preset speed range is obtained. ,in is the frequency component.

[0137] In a specific embodiment, the analysis subsystem 104 further includes:

[0138] A spatial integration unit is used to perform spatial integration processing on noise energy distribution data to obtain a sound field density distribution function, and to perform temperature correlation analysis on the sound absorption coefficient and vibration damping ratio of the PBT material based on the sound field density distribution function to obtain a noise absorption model;

[0139] A nonlinear fitting unit is used to perform nonlinear fitting on the relationship between the noise suppression effect and the speed of the cooling fan based on the noise absorption model to obtain a speed-noise mapping relationship, and dynamically model the relationship between the PWM modulation degree and the current pulsation according to the speed-noise mapping relationship to obtain a PWM-EMI coupling function;

[0140] A multi-dimensional space mapping unit is used to perform a thermal balance analysis on the PWM-EMI coupling function to obtain a heat dissipation effect function, and based on the heat dissipation effect function, perform a multi-dimensional space mapping on the EMI and noise parameters to obtain a parameter space response model of a preset noise range;

[0141] A solving unit is used to substitute the speed-noise mapping relationship and the PWM-EMI coupling function into the parameter space response model to solve the optimal parameter combination, obtain the spatial distribution matrix of EMI and noise parameters, and perform feature processing on the spatial distribution matrix of EMI and noise parameters to obtain parameter space distribution data;

[0142] The frequency autocorrelation analysis unit is used to perform frequency autocorrelation analysis on the input power, rotation speed and noise suppression parameters in the parameter space distribution data to obtain EMI and noise parameter timing characteristic data.

[0143] Specifically, the spatial integration unit calculates the noise energy distribution data Perform spatial integration processing to obtain the sound field density distribution function ,in , and are spatial coordinates, Represents the noise energy at a specific location. The spatial integration formula is:

[0144]

[0145] in, is the volume of the integrated space, Represents a volume element. The integration process accumulates the noise energy of each spatial point to obtain the sound field density distribution function in the entire space, reflecting the energy density of the sound field at different positions. Based on the sound field density distribution function, the spatial integration unit analyzes the sound absorption coefficient of the PBT material and vibration damping ratio The sound absorption coefficient describes the ability of a material to absorb sound energy, and the vibration damping ratio indicates the proportion of energy loss during vibration. Both parameters are affected by temperature. The sound absorption coefficient is expressed as:

[0146]

[0147] in, is the sound absorption coefficient at the reference temperature, is the temperature sensitivity coefficient. Similarly, the temperature dependence of the vibration damping ratio is expressed as:

[0148]

[0149] in, is the initial vibration damping ratio, is the attenuation coefficient related to material properties. Through these temperature correlation analyses, a noise absorption model is obtained, which can predict the noise suppression effect of PBT materials at different ambient temperatures. The nonlinear fitting unit is based on the noise absorption model and the noise suppression effect is related to the cooling fan speed. The relationship between the speed and noise is obtained by nonlinear fitting. Since there is a nonlinear relationship between the fan speed and the noise suppression effect, especially when mechanical resonance or airflow disturbance occurs at a specific speed, a polynomial fitting or exponential function is used for nonlinear fitting. For example, the cubic polynomial fitting model is expressed as:

[0150]

[0151] in, , , and is the fitting coefficient, obtained from the experimental data by the least square method. This mapping relationship provides a prediction of the fan's noise level at different speeds. According to the speed-noise mapping relationship, the PWM modulation index is With current pulsation The relationship between the dynamic modeling and the PWM-EMI coupling function are obtained. The PWM signal controls the motor speed and current waveform by adjusting the duty cycle. Its current pulsation is expressed as:

[0152]

[0153] in, is the maximum current, is the duty cycle of the PWM signal. This coupling function can describe the pulsation change of the motor current under different PWM modulation degrees. The multidimensional space mapping unit inputs the PWM-EMI coupling function into the thermal balance analysis model to calculate the heat dissipation effect function The heat dissipation effect function is used to describe the heat dissipation capacity of the cooling fan under different speeds and temperatures, through the heat transfer equation:

[0154]

[0155] in, is the heat transfer coefficient, is the heat dissipation area, is the ambient temperature. This function can quantify the cooling effect of the fan under specific conditions and help analyze the temperature change trend of the system during long-term operation. Based on the cooling effect function, multi-dimensional space mapping of EMI and noise parameters is performed to construct a parameter space response model within the preset noise range. ,in is the input power. The model maps multidimensional parameters such as input power, fan speed, PWM modulation degree, etc. to the response space of the system noise suppression effect through multidimensional interpolation or regression methods, thereby providing an analysis tool for multi-parameter joint optimization. The solution unit substitutes the speed-noise mapping relationship and PWM-EMI coupling function into the parameter space response model to solve the optimal parameter combination to obtain the spatial distribution matrix of EMI and noise parameters. . By constructing the objective function:

[0156]

[0157] Under constraints (such as power limit , speed range ), apply the gradient descent method or genetic algorithm to search for the optimal solution and obtain the optimal control parameters of EMI and noise under different operating conditions. The obtained spatial distribution matrix It includes the noise energy and EMI intensity distribution in the system under different operating conditions. Through eigenvalue decomposition or principal component analysis, the matrix data is processed to obtain simplified parameter space distribution data. These data provide the most important control variables and their influence, thereby simplifying the dynamic control model of the system. The frequency autocorrelation analysis unit performs frequency autocorrelation analysis on the input power, speed and noise suppression parameters in the parameter space distribution data to obtain the time series characteristic data of EMI and noise parameters. By calculating the autocorrelation function:

[0158]

[0159] in, is the autocorrelation function, is the time delay, is a time signal. Through frequency autocorrelation analysis, the time-delay response characteristics of the noise suppression effect caused by changes in input power and speed, as well as the phase relationship between different parameters, are identified.

[0160] In a specific embodiment, the frequency autocorrelation analysis unit is specifically used for:

[0161] Perform frequency window sampling on parameter space distribution data to obtain parameter sampling sequence;

[0162] According to the parameter sampling sequence, the autocorrelation function of the time series relationship between the input power and the noise suppression amount is calculated, and the power-noise correlation function is obtained by integration operation;

[0163] Based on the parameter sampling sequence, the autocorrelation function of the time series relationship between the speed and the noise suppression amount is calculated, and the speed-noise correlation function is obtained through integration operation;

[0164] According to the power-noise correlation function and the speed-noise correlation function, the time lag effect between parameters is analyzed to obtain the parameter response delay data.

[0165] Based on the parameter response delay data, the parameter cross-correlation strength of the input power, the rotation speed and the noise suppression amount is calculated to obtain the parameter correlation strength data, and the parameter correlation strength data is subjected to eigenvalue decomposition to obtain the parameter main eigenvector;

[0166] The power-noise correlation function and the speed-noise correlation function are orthogonally projected on the parameter principal eigenvector to obtain the EMI and noise parameter timing characteristic data.

[0167] Specifically, the parameter space distribution data is sampled by frequency window sampling technology. Segment processing, Over time Varying multi-dimensional parameter data, including input power , Fan speed and noise suppression amount The purpose of frequency window sampling is to improve the accuracy of time domain analysis while ensuring frequency resolution. (such as Hanning window or Blackman window), divide the original signal into several overlapping time periods, each of which is expressed as:

[0168]

[0169] in, is the windowed sampled signal, It is a window function that adjusts the balance between frequency and time resolution by controlling the window width. For example, 10 seconds of noise data is divided into 1 second windows to obtain a parameter sampling sequence including input power, speed and noise suppression amount. ,in is a discrete time index. The autocorrelation function is calculated based on the time series relationship between the input power and the noise suppression amount according to the parameter sampling sequence. For analyzing input power and noise suppression amount Between different time lags The similarity under , the calculation formula is:

[0170]

[0171] in, is the time delay, is the integration time length. Through the integration operation, the influence of input power change on noise suppression is expressed in the form of time correlation. Power-noise correlation function By integrating the autocorrelation function over time, we get:

[0172]

[0173] This correlation function can quantify the cumulative effect of input power changes on noise suppression and reveal the dynamic relationship between input power adjustment and noise response under a specific time delay. The same method is applied to the time series relationship analysis between speed and noise suppression. By calculating the speed signal and noise suppression amount The autocorrelation function :

[0174]

[0175] By integrating the autocorrelation function, the speed-noise correlation function is obtained:

[0176]

[0177] This function can analyze the impact of fan speed adjustment on the system noise level, and identify the delay characteristics of the system noise response under different speed changes. By performing a correlation analysis of the time lag effect on the power-noise correlation function and the speed-noise correlation function, the parameter response delay data is obtained. During the analysis process, by calculating the peak position of these correlation functions, it is determined at what time delay the change in input power or fan speed can have the greatest impact on the noise suppression effect. Based on the obtained parameter response delay data, the parameter cross-correlation strength of the input power, speed and noise suppression amount is calculated to obtain the parameter correlation strength data. Cross-correlation strength The similarity between different parameters at different time delays is calculated, which is defined as:

[0178]

[0179] in and Represent the input power, speed or noise suppression signal respectively, and calculate the cross-correlation function of different combinations (such as ), and the contribution of different parameters to the noise suppression effect is obtained. In order to extract key features, the cross-correlation intensity data is organized into a correlation matrix , and perform eigenvalue decomposition on it:

[0180]

[0181] in, is the eigenvector matrix, is the eigenvalue matrix, the eigenvalue represents the weight of each eigenvector, and the eigenvector represents the main parameter change mode in the system. and speed-noise correlation function Projecting onto the main eigenvector of the parameters, we can obtain the timing characteristic data of EMI and noise parameters. The orthogonal projection calculation process is:

[0182]

[0183] in, and They are the corresponding power and speed eigenvectors, and the projection results are and It is the main characteristic data of EMI and noise parameters in time series. These characteristic data reveal the dynamic change trend of EMI and noise under different operating conditions in the system.

[0184] In a specific embodiment, the real-time adjustment subsystem 105 is specifically used for:

[0185] Input the EMI and noise parameter timing feature data into the spatial feature extraction module of the prediction model to extract spatial correlation features and obtain a spatial feature mapping matrix;

[0186] The spatial feature mapping matrix is ​​input into the time dependency analysis module of the prediction model to extract the time dependency features. The time dependency analysis module includes four layers of linear encoders adapted to the preset speed range to obtain the time feature vector;

[0187] The spatial feature mapping matrix and the temporal feature vector are fused to obtain a deep feature representation, which is then input into two layers of spatiotemporal graph convolution layers to extract dynamic features of EMI and noise parameters, and obtain the suppression effect prediction value within the preset noise range.

[0188] The error between the predicted value and the actual noise suppression effect is calculated to obtain the noise correction amount;

[0189] According to the noise correction amount, the adjustment parameter of the PWM duty cycle of the cooling fan is calculated by the PID controller to obtain the PWM adjustment parameter, and the dynamic adjustment parameter of the speed is calculated based on the PWM adjustment parameter;

[0190] The PWM adjustment parameters and the dynamic adjustment parameters are combined into PWM control correction parameters, and the output power and the speed of the cooling fan are adjusted in real time in a closed loop according to the PWM control correction parameters.

[0191] Specifically, a high-dimensional time series feature dataset is constructed, which includes input power , Fan speed and noise suppression amount The changes at different time points are represented as a matrix The spatial feature extraction module analyzes the distribution characteristics of these data in different spatial dimensions and uses convolution operations to convert high-dimensional time series data into low-dimensional spatial feature mapping matrices. This process is done through the spatial convolution kernel Acting on the input matrix, we get each element of the feature map matrix:

[0192]

[0193] in, and is the coordinate index of the feature map matrix, and is the size of the convolution kernel. The convolution operation extracts the correlation between data in different dimensions through a sliding window method. For example, when analyzing the EMI propagation path, the spatial feature extraction module can identify the interaction between input power and noise suppression at a specific frequency, which helps to predict the impact of fan speed changes on noise suppression. The time dependency analysis module of the input prediction model is used to extract the time-series dependency features. The time dependency analysis module contains four layers of linear encoders adapted to the preset speed range. Each layer of encoders is used to process the time-series feature data at different speeds. The linear encoder converts the spatial feature mapping matrix into a time feature vector through linear transformation. , the output of each layer encoder is expressed as:

[0194]

[0195] in, is the linear transformation matrix, is the bias vector. Through the layer-by-layer processing of these four layers of linear encoders, the temporal feature vector retains the spatial characteristics of the original data and extracts the dynamic change characteristics of the system at different time points. The multi-layer structure of the linear encoder enables the prediction model to adaptively analyze the noise response characteristics of the fan in different speed ranges. For example, when running at low speed, the fan noise mainly comes from electromagnetic interference, while when running at high speed, it is significantly affected by mechanical vibration. The spatial feature mapping matrix and the time feature vector Perform feature fusion to obtain deep feature representation The feature fusion process is achieved through feature concatenation or feature weighting, combining the interference path information in the spatial dimension and the noise change trend in the temporal dimension. The feature fusion formula is:

[0196]

[0197] in, Indicates the splicing of spatial features and temporal features. It is a fusion weight matrix. Through feature fusion operation, a multi-dimensional feature space is constructed, so that the subsequent prediction model can quickly capture the key feature changes under different input conditions, thereby improving the accuracy of noise suppression effect prediction. The deep feature representation is input into two layers of spatiotemporal graph convolution layer to extract dynamic features of EMI and noise parameters. In the spatiotemporal graph convolution layer, a spatiotemporal graph is constructed. ,in is a set of nodes, representing different components in the system (such as motors, fan blades, housings, etc.). is an edge set, representing the propagation path of EMI and mechanical vibration. The spatiotemporal graph convolution layer unifies the static spatial relationship and dynamic time dependency in the system into one model through the combined operation of spatial convolution and temporal convolution. For each layer of convolution operation, the output feature is expressed as:

[0198]

[0199] in, is the activation function (such as ReLU), is the graph convolution weight matrix, It is an adjacency matrix that represents the connection relationship between the components in the system. Through two layers of spatiotemporal graph convolution, the most significant EMI and noise propagation paths in the system under specific fan speed and PWM signal adjustment conditions are identified, and the suppression effect prediction value within the preset noise range is obtained. In order to achieve dynamic control of the system, the predicted value The actual noise suppression effect Perform error calculation to obtain the noise correction amount :

[0200]

[0201] When the noise correction value is positive, it means that the current noise suppression effect is insufficient and the PWM signal adjustment amplitude needs to be increased; when it is negative, the PWM adjustment intensity needs to be reduced. Based on the noise correction value, the PID (proportional-integral-differential) controller calculates the adjustment parameters of the cooling fan PWM duty cycle. The PID control formula is:

[0202]

[0203] in, , , They are the proportional, integral and differential coefficients of the PID controller. By adjusting these coefficients, fast response and fine control of the noise suppression effect can be achieved. Based on the obtained PWM adjustment parameters , calculate the dynamic adjustment parameters of the speed , this parameter indicates the amplitude of fan speed adjustment to achieve the best noise suppression effect. The speed adjustment parameter calculation formula is:

[0204]

[0205] in, is the speed adjustment gain coefficient, is the current PWM duty cycle, which is obtained by combining the PWM adjustment parameter and the dynamic adjustment parameter into the PWM control correction parameter :

[0206]

[0207] Correction parameters according to PWM control Output power of cooling fan and speed Real-time closed-loop regulation. The closed-loop control system compares the actual measured noise suppression effect with the expected value output by the prediction model through a feedback loop, and dynamically adjusts the operating state of the fan based on the error calculation. For example, when the system detects that the actual noise level is higher than the predicted value, the PID controller increases the duty cycle of the PWM signal, thereby increasing the fan speed to enhance the noise suppression effect. Conversely, when the noise level is lower than the predicted value, the system's energy consumption is effectively reduced and electromagnetic interference caused by over-regulation is avoided by reducing the PWM signal strength and reducing the fan speed.

[0208] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, systems and units can refer to the corresponding processes in the aforementioned system embodiments and will not be repeated here.

[0209] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the system described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc., various media that can store program codes.

[0210] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features thereof may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A dynamic coupling path controlled anti-EMI cooling fan filtering and noise suppression system, characterized in that: The system comprises: The modeling subsystem is used to model the electromagnetic energy transfer of the motor, fan blades, bearings and housing components in the cooling fan to obtain the EMI energy balance equations; An extraction subsystem, used to construct an electromagnetic interference transmission network according to the EMI energy balance equation group, and extract the electromagnetic interference transmission relationship between the components of the cooling fan; A calculation subsystem, for performing a time discrete sequence calculation on the vibration transmission between the PBT fan blade and the motor based on the electromagnetic interference transmission relationship, and obtaining noise energy distribution data within a preset speed range; An analysis subsystem, used to determine the parameter space distribution within a preset noise range according to the noise energy distribution data, and perform frequency autocorrelation analysis on the parameter space distribution data to obtain EMI and noise parameter timing characteristic data; The real-time adjustment subsystem is used to input the EMI and noise parameter timing characteristic data into the prediction model for characteristic calculation, obtain PWM control correction parameters, and adjust the output power and speed of the cooling fan in real time according to the PWM control correction parameters.

2. The anti-EMI cooling fan filtering and noise suppression system with dynamic coupling path control according to claim 1, characterized in that: The modeling subsystem is specifically used for: Calculating the efficiency of electromagnetic coupling between the motor winding and the rotor magnetic field of the cooling fan under the input voltage to obtain a power conversion efficiency function, and performing vector field calculation on the magnetic flux density distribution in the motor based on the power conversion efficiency function to obtain an electromagnetic field distribution function; Based on the electromagnetic field distribution function, a spectrum analysis is performed on the electromagnetic radiation intensity generated by the motor to obtain electromagnetic interference spectrum data, and according to the electromagnetic interference spectrum data, a frequency domain decomposition is performed on the radiation and conduction EMI components to obtain an interference mode distribution function; The EMI shielding characteristics of the PBT fan blades and the plastic shell of the cooling fan are attenuated and calculated to obtain an EMI attenuation coefficient matrix, and the propagation channel of the interference signal is analyzed according to the EMI attenuation coefficient matrix to obtain an EMI transfer matrix; The power conversion efficiency function, the electromagnetic field distribution function, the electromagnetic interference spectrum data, the interference mode distribution function and the EMI transfer matrix are calculated simultaneously to obtain an EMI transfer equation group between components; The EMI transfer equations are subjected to eigenvalue decomposition and normalization processing to obtain an EMI energy balance equation including input voltage, motor operating power, double ball bearing mechanical quality factor and electromagnetic coupling coefficient.

3. The anti-EMI cooling fan filtering and noise suppression system with dynamic coupling path control according to claim 1, characterized in that: The extraction subsystem is specifically used for: Based on the EMI energy balance equations, the power input terminal, motor control circuit, stator winding, rotor, double ball bearing, PBT fan blade and housing are set as network nodes, and the EMI propagation relationship between network nodes is mapped by directed graph to obtain the initial EMI network topology structure; Performing a first channel identification on the EMI propagation path from the power supply to the motor winding in the initial EMI network topology structure to obtain a first EMI transfer path matrix, and performing a second channel identification on the mechanical vibration coupling path between the fan blade and the housing in the initial EMI network topology structure to obtain a second vibration coupling path matrix; Based on the first EMI transfer path matrix, the electromagnetic interference transfer efficiency between adjacent components in the first channel is quantitatively calculated to obtain the first channel EMI coupling coefficient, and according to the second vibration coupling path matrix, the vibration transfer characteristics of the double ball bearing in the second channel are analyzed to obtain the second channel vibration loss coefficient; The time characteristic analysis of the EMI transfer process in the first channel and the second channel is performed respectively to obtain the time delay parameters at different speeds, and the EMI coupling coefficient of the first channel, the vibration loss coefficient of the second channel and the time delay parameters are substituted into the transfer function model to perform quantitative analysis of the EMI flow process to obtain the electromagnetic interference transmission relationship between the components of the cooling fan.

4. The anti-EMI cooling fan filtering and noise suppression system with dynamic coupling path control according to claim 1, characterized in that: The computing subsystem is specifically used for: Determine a discrete time series sampling point set according to the electromagnetic interference transfer relationship, and perform piecewise linearization processing on the cooling fan PWM control signal according to the discrete time series sampling point set to obtain a motor control sequence; Based on the motor control sequence, an electromagnetic field energy storage process is calculated for the current density distribution in the motor winding to obtain an electromagnetic field intensity distribution sequence; According to the electromagnetic field intensity distribution sequence, the mechanical vibration equation is solved for the vibration force generated by the motor to obtain the motor output vibration sequence, and the vibration response of the mechanical resonance characteristics of the PBT fan blade is calculated to obtain the fan blade vibration response sequence; Based on the wind blade vibration response sequence, the energy attenuation process of the double ball bearing is calculated to obtain the bearing vibration transfer sequence; Performing time correlation analysis on the bearing vibration transmission sequence, and calculating the vibration transmission timing relationship between the components through time delay and phase difference, to obtain the time characteristic sequence of vibration transmission between the components; The motor control sequence, the electromagnetic field intensity distribution sequence, the motor output vibration sequence, the fan blade vibration response sequence and the time characteristic sequence of the vibration transmission between the components are synchronized and time-aligned to obtain noise energy distribution data within a preset speed range.

5. The anti-EMI cooling fan filtering and noise suppression system with dynamic coupling path control according to claim 1, characterized in that: The analysis subsystem also includes: A spatial integration unit, used for performing spatial integration processing on the noise energy distribution data to obtain a sound field density distribution function, and performing temperature correlation analysis on the sound absorption coefficient and vibration damping ratio of the PBT material according to the sound field density distribution function to obtain a noise absorption model; A nonlinear fitting unit is used to perform nonlinear fitting on the relationship between the noise suppression effect and the speed of the cooling fan based on the noise absorption model to obtain a speed-noise mapping relationship, and dynamically model the relationship between the PWM modulation degree and the current pulsation according to the speed-noise mapping relationship to obtain a PWM-EMI coupling function; A multi-dimensional space mapping unit, used to perform a thermal balance analysis on the PWM-EMI coupling function to obtain a heat dissipation effect function, and based on the heat dissipation effect function, perform a multi-dimensional space mapping on EMI and noise parameters to obtain a parameter space response model of a preset noise range; A solving unit, used for substituting the speed-noise mapping relationship and the PWM-EMI coupling function into the parameter space response model to solve the optimal parameter combination, obtain the spatial distribution matrix of EMI and noise parameters, and perform feature processing on the spatial distribution matrix of EMI and noise parameters to obtain parameter space distribution data; The frequency autocorrelation analysis unit is used to perform frequency autocorrelation analysis on the input power, rotation speed and noise suppression amount parameters in the parameter spatial distribution data to obtain EMI and noise parameter timing characteristic data.

6. The anti-EMI cooling fan filtering and noise suppression system with dynamic coupling path control according to claim 5, characterized in that: The frequency autocorrelation analysis unit is specifically used for: Performing frequency window sampling on the parameter space distribution data to obtain a parameter sampling sequence; According to the parameter sampling sequence, an autocorrelation function is calculated for the time series relationship between the input power and the noise suppression amount, and a power-noise correlation function is obtained by an integral operation; Based on the parameter sampling sequence, an autocorrelation function is calculated for the time series relationship between the rotation speed and the noise suppression amount, and a rotation speed-noise correlation function is obtained by integral operation; According to the power-noise correlation function and the speed-noise correlation function, correlation analysis is performed on the time lag effect between the parameters to obtain parameter response delay data; Based on the parameter response delay data, the parameter cross-correlation strength of the input power, the rotation speed and the noise suppression amount is calculated to obtain parameter correlation strength data, and the parameter correlation strength data is subjected to eigenvalue decomposition to obtain a parameter main eigenvector; Orthogonal projection is performed on the power-noise correlation function and the rotation speed-noise correlation function on the parameter main eigenvector to obtain EMI and noise parameter timing characteristic data.

7. The anti-EMI cooling fan filtering and noise suppression system with dynamic coupling path control according to claim 1, characterized in that: The real-time adjustment subsystem is specifically used for: Inputting the EMI and noise parameter timing characteristic data into the spatial feature extraction module of the prediction model to extract spatial correlation features to obtain a spatial feature mapping matrix; Inputting the spatial feature mapping matrix into the time dependency analysis module of the prediction model to extract the timing dependency features, wherein the time dependency analysis module includes four layers of linear encoders adapted to a preset speed range, to obtain a time feature vector; Performing feature fusion on the spatial feature mapping matrix and the temporal feature vector to obtain a deep feature representation, and inputting the deep feature representation into two layers of spatiotemporal graph convolution layers to perform dynamic feature extraction of EMI and noise parameters to obtain a suppression effect prediction value within a preset noise range; Performing error calculation between the predicted value and the actual noise suppression effect to obtain a noise correction amount; According to the noise correction amount, a PID controller is used to calculate a PWM duty cycle adjustment parameter of the cooling fan to obtain a PWM adjustment parameter, and a dynamic adjustment parameter of the rotation speed is calculated based on the PWM adjustment parameter; The PWM adjustment parameter and the dynamic adjustment parameter are combined into a PWM control correction parameter, and the output power and the rotation speed of the cooling fan are adjusted in real time in a closed loop according to the PWM control correction parameter.

Images