A particulate filter self-cleaning method and system

Abstract

The application provides a kind of particulate filter self-cleaning method and system, the method comprises: based on the vibration response characteristics of particulate filter filter material and dust dielectric constant detection data, corresponding multi-field coupling model is constructed, corresponding vibration signal characteristic value is extracted simultaneously by Hilbert transform, to generate corresponding dust accumulation early warning threshold according to vibration signal characteristic value by multi-field coupling model;Based on dust accumulation early warning threshold, combined with the current operating load of particulate filter, the corresponding target cleaning parameter is matched in the preset parameter database;Target cleaning parameter is input to the preset cleaning device, so that the preset cleaning device cleans the inside of particulate filter;After cleaning, filter vibration response signal, air permeability recovery rate and exhaust particulate matter concentration are detected, and whether cleaning is up to standard is judged simultaneously, to complete corresponding cleaning process.The application can automatically and effectively complete the cleaning of particulate filter, and correspondingly improve the cleaning efficiency.

Description

Technical Field

[0001] This invention relates to the field of automotive technology, and in particular to a self-cleaning method and system for particulate filters. Background Technology

[0002] In the field of new energy vehicles, the electric drive system (referred to as "electric drive") and the power battery system are the core power components, and their heat dissipation stability under high power conditions is directly related to the vehicle's power performance and safety. Liquid cooling has become the mainstream heat dissipation method due to its high efficiency and precise temperature control. A typical liquid cooling system consists of a cooling pump, pipelines, heat exchanger, particulate filter, etc., among which the particulate filter is a key protective component to ensure the clean operation of the system.

[0003] In the liquid cooling cycle, the coolant is prone to mixing with impurities such as metal shavings and rubber fragments. The core function of the particulate filter is to intercept impurities through the filter element, prevent impurities from entering core components such as the cooling pump and heat exchanger, ensure the smooth operation of the liquid cooling system, and provide a guarantee for the stable operation of the electric drive and power battery system.

[0004] Furthermore, existing particulate filters lack self-cleaning capabilities, and after prolonged use, the filter element is easily clogged by particles, leading to decreased coolant flow, increased pressure differential, reduced heat dissipation efficiency, and even system malfunctions. Clogged filters require manual disassembly and cleaning or replacement, resulting in high maintenance costs, difficulty in timing the cleaning process, and disruption to normal vehicle operation. Therefore, developing a self-cleaning method and system for particulate filters is an urgent need to address the shortcomings of existing technologies. Summary of the Invention

[0005] Based on this, the purpose of the present invention is to provide a self-cleaning method and system for particulate filters, so as to solve the problem that existing particulate filters lack self-cleaning ability and are prone to system failure after long-term use.

[0006] The first aspect of the present invention proposes:

[0007] A self-cleaning method for particulate filters specifically includes the following steps:

[0008] Based on the vibration response characteristics of particulate filter media and dust dielectric constant detection data, a corresponding multi-field coupling model is constructed. Simultaneously, the corresponding vibration signal feature values ​​are extracted through Hilbert transform. The corresponding dust accumulation warning threshold is then generated based on the vibration signal feature values ​​through the multi-field coupling model.

[0009] Based on the dust accumulation warning threshold and the current operating load of the particulate filter, the corresponding target cleaning parameters are matched in the preset parameter database.

[0010] The target cleaning parameters are input into a preset cleaning device so that the preset cleaning device cleans the inside of the particulate filter.

[0011] After cleaning, the vibration response signal of the filter media, the air permeability recovery rate, and the concentration of exhaust particulate matter are detected to determine whether the cleaning meets the standards and complete the corresponding cleaning treatment.

[0012] The beneficial effects of this invention are as follows: This technical solution, through a multi-field coupling model combined with Hilbert transform, accurately identifies the dust accumulation state of particulate filters and generates reliable dust accumulation early warning thresholds, enabling early prediction of dust accumulation risks; by matching target cleaning parameters with operating load, it avoids filter media damage or incomplete cleaning caused by blind cleaning, improving cleaning targeting and efficiency; multi-dimensional detection after cleaning can verify the effectiveness in real time, ensuring that cleaning meets standards. It effectively solves the problem of filters lacking self-cleaning ability and prone to failure, extends equipment lifespan, stabilizes filtration efficiency, and ensures long-term reliable system operation.

[0013] Furthermore, the step of generating the corresponding dust accumulation warning threshold based on the vibration signal feature values ​​using the multi-field coupling model includes:

[0014] The vibration signal feature values ​​are separated by a blind source separation algorithm to generate corresponding vibration attenuation features, vibration phase shift features and vibration noise features. Simultaneously, the dust dielectric constant detection data and filter material porosity parameters are combined to construct a unique mapping relationship between the features and the dust accumulation state.

[0015] By introducing a cross-field coupling effect coefficient and combining it with the specific mapping relationship, the vibration attenuation feature, the vibration phase shift feature, and the vibration noise feature are weighted and fused to generate a fused feature vector characterizing the dust accumulation state.

[0016] The fused feature vector is input into the multi-field coupling model to perform dust accumulation state inversion to obtain actual dust accumulation data, and the dust accumulation warning threshold is generated simultaneously based on the actual dust accumulation data.

[0017] Furthermore, the step of synchronously generating the dust accumulation warning threshold based on the actual dust accumulation data includes:

[0018] Based on the material strength parameters, pore structure parameters and the specific mapping relationship of the filter material, the stress distribution state of the filter material under different actual dust accumulation is simulated by finite element simulation to determine the corresponding maximum dust accumulation tolerance. At the same time, combined with the preset minimum filtration conditions, the corresponding basic constraint interval is generated.

[0019] Multi-dimensional constraint correction coefficients and airflow disturbance correction coefficients are introduced, and the basic constraint interval is coupled and calculated to obtain the initial dust accumulation warning threshold.

[0020] The vibration attenuation feature, the vibration phase shift feature, and the initial dust accumulation warning threshold are correlated and matched. If the feature change rate does not exceed the preset stable range, the initial dust accumulation warning threshold is set as the corresponding dust accumulation warning threshold.

[0021] Furthermore, the step of matching the corresponding target cleaning parameters in the preset parameter database based on the dust accumulation warning threshold and the current operating load of the particulate filter includes:

[0022] Based on the gradient level of the dust accumulation warning threshold and the fluctuation characteristics of the current operating load, a corresponding two-dimensional parameter calibration matrix is ​​constructed. A dynamic correction factor is introduced simultaneously, and the gradient level and the fluctuation characteristics are coupled and calibrated through the least squares support vector machine algorithm to generate the corresponding core matching parameter set.

[0023] For the cleaning parameter samples in the preset parameter database, a preset clustering algorithm is used to divide them into several working condition feature clusters. The cosine similarity between the core matching dataset and each working condition feature cluster is calculated simultaneously to construct the corresponding multi-objective optimization matching function.

[0024] The multi-objective optimization matching function filters out candidate cleaning parameter groups from the preset parameter database, and the candidate cleaning parameter groups are simultaneously simulated to generate the target cleaning parameters.

[0025] Furthermore, the step of simultaneously simulating the candidate cleaning parameter group to generate the target cleaning parameters includes:

[0026] The preset cleaning process simulation model is brought up, the candidate cleaning parameter group is imported into the cleaning process simulation model, and the initial simulation conditions are set in combination with the vibration response characteristics of the filter material to simulate the dust stripping process and dynamic response of the filter material under different parameters, and the corresponding core simulation data is output synchronously.

[0027] Based on the core simulation data, the risk level of filter media damage caused by each candidate cleaning parameter group is evaluated, and intermediate cleaning parameter groups are selected simultaneously based on the risk level.

[0028] Based on the actual operating deviations of the particulate filter, a corresponding dynamic parameter adaptation function is constructed. Simultaneously, the intermediate cleaning parameter set is corrected through the dynamic parameter adaptation function to generate the target cleaning parameters.

[0029] Furthermore, the step of detecting the filter media vibration response signal, air permeability recovery rate, and exhaust particulate matter concentration, and simultaneously determining whether the cleaning standard is met, includes:

[0030] Based on the residual vibration attenuation characteristics of the cleaned filter media, the detection frequency and sampling duration are dynamically adapted, and the vibration response signals of the filter media at multiple measurement points are collected simultaneously through the laser Doppler vibration measurement module.

[0031] The vibration response signals from the multiple measurement points are subjected to Hilbert transform to extract the corresponding instantaneous frequency and damping ratio characteristic parameters. The corresponding four-dimensional feature vector is then constructed by combining the air permeability recovery rate and the exhaust particulate matter concentration.

[0032] Based on the dust accumulation characteristic correlation data output by the multi-field coupling model, the four-dimensional feature vector is fused and calculated using the random forest algorithm to obtain the corresponding comprehensive cleaning effect evaluation value, so as to determine whether the standard is met.

[0033] Furthermore, the step of using the random forest algorithm to fuse and calculate the four-dimensional feature vectors to obtain the corresponding comprehensive cleaning effect evaluation value, and to determine whether the standard is met, includes:

[0034] Based on the dust accumulation characteristic correlation data, a cross-enhanced feature set adapted to the four-dimensional feature vector is constructed, and the core features are simultaneously screened out through the mutual information entropy algorithm to generate the corresponding feature matrix.

[0035] The random forest algorithm is used to perform prior constraint processing on the feature matrix to calculate the contribution value corresponding to each feature.

[0036] The contribution values ​​are weighted and fused to obtain the comprehensive cleaning effect evaluation value, and the standard is determined based on the magnitude of the comprehensive cleaning effect evaluation value.

[0037] The second aspect of the present invention proposes:

[0038] A particulate filter self-cleaning system, wherein the system comprises:

[0039] The module is used to construct a corresponding multi-field coupling model based on the vibration response characteristics of the particulate filter media and the detection data of the dielectric constant of dust. Simultaneously, the corresponding vibration signal feature values ​​are extracted through Hilbert transform, so as to generate the corresponding dust accumulation warning threshold based on the vibration signal feature values ​​through the multi-field coupling model.

[0040] The matching module is used to match the corresponding target cleaning parameters in a preset parameter database based on the dust accumulation warning threshold and the current operating load of the particulate filter.

[0041] The processing module is used to input the target cleaning parameters into a preset cleaning device, so that the preset cleaning device cleans the inside of the particulate filter;

[0042] The detection module is used to detect the filter media vibration response signal, air permeability recovery rate and exhaust particulate matter concentration after cleaning is completed, and simultaneously determine whether the cleaning meets the standards, so as to complete the corresponding cleaning treatment.

[0043] Furthermore, the building module is specifically used for:

[0044] The vibration signal feature values ​​are separated by a blind source separation algorithm to generate corresponding vibration attenuation features, vibration phase shift features and vibration noise features. Simultaneously, the dust dielectric constant detection data and filter material porosity parameters are combined to construct a unique mapping relationship between the features and the dust accumulation state.

[0045] By introducing a cross-field coupling effect coefficient and combining it with the specific mapping relationship, the vibration attenuation feature, the vibration phase shift feature, and the vibration noise feature are weighted and fused to generate a fused feature vector characterizing the dust accumulation state.

[0046] The fused feature vector is input into the multi-field coupling model to perform dust accumulation state inversion to obtain actual dust accumulation data, and the dust accumulation warning threshold is generated simultaneously based on the actual dust accumulation data.

[0047] Furthermore, the building module is specifically used for:

[0048] Based on the material strength parameters, pore structure parameters and the specific mapping relationship of the filter material, the stress distribution state of the filter material under different actual dust accumulation is simulated by finite element simulation to determine the corresponding maximum dust accumulation tolerance. At the same time, combined with the preset minimum filtration conditions, the corresponding basic constraint interval is generated.

[0049] Multi-dimensional constraint correction coefficients and airflow disturbance correction coefficients are introduced, and the basic constraint interval is coupled and calculated to obtain the initial dust accumulation warning threshold.

[0050] The vibration attenuation feature, the vibration phase shift feature, and the initial dust accumulation warning threshold are correlated and matched. If the feature change rate does not exceed the preset stable range, the initial dust accumulation warning threshold is set as the corresponding dust accumulation warning threshold.

[0051] Furthermore, the matching module is specifically used for:

[0052] Based on the gradient level of the dust accumulation warning threshold and the fluctuation characteristics of the current operating load, a corresponding two-dimensional parameter calibration matrix is ​​constructed. A dynamic correction factor is introduced simultaneously, and the gradient level and the fluctuation characteristics are coupled and calibrated through the least squares support vector machine algorithm to generate the corresponding core matching parameter set.

[0053] For the cleaning parameter samples in the preset parameter database, a preset clustering algorithm is used to divide them into several working condition feature clusters. The cosine similarity between the core matching dataset and each working condition feature cluster is calculated simultaneously to construct the corresponding multi-objective optimization matching function.

[0054] The multi-objective optimization matching function filters out candidate cleaning parameter groups from the preset parameter database, and the candidate cleaning parameter groups are simultaneously simulated to generate the target cleaning parameters.

[0055] Furthermore, the matching module is specifically used for:

[0056] The preset cleaning process simulation model is brought up, the candidate cleaning parameter group is imported into the cleaning process simulation model, and the initial simulation conditions are set in combination with the vibration response characteristics of the filter material to simulate the dust stripping process and dynamic response of the filter material under different parameters, and the corresponding core simulation data is output synchronously.

[0057] Based on the core simulation data, the risk level of filter media damage caused by each candidate cleaning parameter group is evaluated, and intermediate cleaning parameter groups are selected simultaneously based on the risk level.

[0058] Based on the actual operating deviations of the particulate filter, a corresponding dynamic parameter adaptation function is constructed. Simultaneously, the intermediate cleaning parameter set is corrected through the dynamic parameter adaptation function to generate the target cleaning parameters.

[0059] Furthermore, the detection module is specifically used for:

[0060] Based on the residual vibration attenuation characteristics of the cleaned filter media, the detection frequency and sampling duration are dynamically adapted, and the vibration response signals of the filter media at multiple measurement points are collected simultaneously through the laser Doppler vibration measurement module.

[0061] The vibration response signals from the multiple measurement points are subjected to Hilbert transform to extract the corresponding instantaneous frequency and damping ratio characteristic parameters. The corresponding four-dimensional feature vector is then constructed by combining the air permeability recovery rate and the exhaust particulate matter concentration.

[0062] Based on the dust accumulation characteristic correlation data output by the multi-field coupling model, the four-dimensional feature vector is fused and calculated using the random forest algorithm to obtain the corresponding comprehensive cleaning effect evaluation value, so as to determine whether the standard is met.

[0063] Furthermore, the detection module is specifically used for:

[0064] Based on the dust accumulation characteristic correlation data, a cross-enhanced feature set adapted to the four-dimensional feature vector is constructed, and the core features are simultaneously screened out through the mutual information entropy algorithm to generate the corresponding feature matrix.

[0065] The random forest algorithm is used to perform prior constraint processing on the feature matrix to calculate the contribution value corresponding to each feature.

[0066] The contribution values ​​are weighted and fused to obtain the comprehensive cleaning effect evaluation value, and the standard is determined based on the magnitude of the comprehensive cleaning effect evaluation value.

[0067] The third aspect of the present invention proposes:

[0068] A computer includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, implements the particulate filter self-cleaning method as described above.

[0069] The fourth aspect of the present invention proposes:

[0070] A readable storage medium having a computer program stored thereon, wherein the program, when executed by a processor, implements the particulate filter self-cleaning method as described above.

[0071] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0072] Figure 1 A flowchart of a particulate filter self-cleaning method provided in the first embodiment of the present invention;

[0073] Figure 2 This is a structural block diagram of a particulate filter self-cleaning system provided in the third embodiment of the present invention.

[0074] The following detailed description, in conjunction with the accompanying drawings, will further illustrate the present invention. Detailed Implementation

[0075] To facilitate understanding of the present invention, a more complete description will be given below with reference to the accompanying drawings. Several embodiments of the invention are illustrated in the drawings. However, the invention can be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete.

[0076] It should be noted that when a component is said to be "fixed to" another component, it can be directly on the other component or there may be an intervening component. When a component is said to be "connected to" another component, it can be directly connected to the other component or there may be an intervening component. The terms "vertical," "horizontal," "left," "right," and similar expressions used in this document are for illustrative purposes only.

[0077] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0078] Please see Figure 1 The image shows a self-cleaning method for a particulate filter provided in the first embodiment of the present invention. The self-cleaning method for a particulate filter provided in this embodiment can automatically and accurately clean the particulate filter, thereby improving the cleaning efficiency.

[0079] Specifically, this embodiment provides:

[0080] A self-cleaning method for particulate filters specifically includes the following steps:

[0081] Step S10: Based on the vibration response characteristics of the particulate filter media and the dust dielectric constant detection data, a corresponding multi-field coupling model is constructed, and the corresponding vibration signal feature values ​​are extracted simultaneously through Hilbert transform, so as to generate the corresponding dust accumulation warning threshold based on the vibration signal feature values ​​through the multi-field coupling model.

[0082] It is important to note that, firstly, addressing the shortcomings of traditional cleaning timing determination methods that "rely on fixed time or pressure difference thresholds, without considering the differences in filter media characteristics and dust states," this paper constructs a multi-field coupling model (integrating the interaction of vibration field, dielectric field, and flow field) based on the vibration response characteristics of particulate filter media (the larger the dust accumulation, the more obvious the vibration attenuation and the greater the phase shift) and dust dielectric constant detection data (the dielectric constants of different types of dust differ significantly, which can help distinguish dust composition and accumulation state). Specifically, this model can accurately characterize the mapping relationship between "vibration signal - dust dielectric properties - dust accumulation state," avoiding the one-sidedness of single-physics field modeling. Simultaneously, it extracts the characteristic values ​​of vibration signals (such as instantaneous frequency, damping ratio, and phase difference) through Hilbert transform. Hilbert transform excels at handling non-stationary vibration signals (filter media vibration is non-stationary due to the influence of dust accumulation), effectively separating effective signals from noise and providing high-quality input for dust accumulation state inversion. Finally, the multi-field coupling model transforms the vibration signal characteristic values ​​into dust accumulation early warning thresholds, achieving "based on actual dust accumulation state." The cleaning timing is triggered to avoid energy waste caused by cleaning too early or filter clogging and damage caused by cleaning too late.

[0083] Step S20: Based on the dust accumulation warning threshold and the current operating load of the particulate filter, match the corresponding target cleaning parameters in the preset parameter database;

[0084] It should be noted that, secondly, to address the problem of traditional cleaning parameters being "fixed and not adapted to the operating load," target cleaning parameters are matched based on dust accumulation warning thresholds (quantifying the severity of dust accumulation, such as light, moderate, and heavy), combined with the current operating load of the particulate filter (such as exhaust gas volume, airflow velocity, and inlet / outlet pressure difference). These parameters are then matched with a preset parameter database (which stores cleaning parameter samples under different dust accumulation levels and loads, such as pulse pressure, cleaning duration, and pulse frequency). Specifically, the operating load directly affects the airflow stability during the cleaning process (e.g., high airflow velocity under high load requires increased pulse pressure to ensure dust removal effect while avoiding excessive impact that could damage the filter media). Therefore, parameter matching must consider both the dust accumulation state and load characteristics to ensure a balance between cleaning effectiveness and equipment safety.

[0085] Step S30: Input the target cleaning parameters into the preset cleaning device so that the preset cleaning device cleans the inside of the particulate filter;

[0086] It should be noted that, next, the target cleaning parameters are input into the preset cleaning device (such as a pulse jet cleaning device or a vibration cleaning device), and the driving device performs targeted cleaning of the inside of the filter (such as using low-pressure short-duration pulses for light dust accumulation, and using segmented high-pressure pulses + vibration for heavy dust accumulation), avoiding filter media damage or incomplete cleaning caused by the traditional "one-size-fits-all" cleaning method.

[0087] Step S40: After cleaning is completed, the vibration response signal of the filter material, the air permeability recovery rate and the concentration of exhaust particulate matter are detected, and it is simultaneously determined whether the cleaning meets the standards to complete the corresponding cleaning treatment.

[0088] It should be noted that, finally, to fully verify the cleaning effect, three core indicators are tested after cleaning: filter media vibration response signal (reflecting the amount of residual dust; the less residual dust, the closer the vibration response is to the initial cleaning state), air permeability recovery rate (reflecting the degree of filtration capacity recovery, which must reach a preset threshold such as ≥90%), and exhaust particulate matter concentration (reflecting emission compliance, which must meet national / industry emission standards). By using multiple indicators in synergy, it is determined whether the cleaning meets the standards. If it does, the cleaning is completed; if it does not, the cleaning parameters are adjusted and a second cleaning is performed based on the reasons for non-compliance (such as excessive residual dust or insufficient air permeability recovery), forming a closed-loop control to ensure the long-term stable operation of the filter.

[0089] Second Embodiment

[0090] Furthermore, the step of generating the corresponding dust accumulation warning threshold based on the vibration signal feature values ​​using the multi-field coupling model includes:

[0091] The vibration signal feature values ​​are separated by a blind source separation algorithm to generate corresponding vibration attenuation features, vibration phase shift features and vibration noise features. Simultaneously, the dust dielectric constant detection data and filter material porosity parameters are combined to construct a unique mapping relationship between the features and the dust accumulation state.

[0092] By introducing a cross-field coupling effect coefficient and combining it with the specific mapping relationship, the vibration attenuation feature, the vibration phase shift feature, and the vibration noise feature are weighted and fused to generate a fused feature vector characterizing the dust accumulation state.

[0093] The fused feature vector is input into the multi-field coupling model to perform dust accumulation state inversion to obtain actual dust accumulation data, and the dust accumulation warning threshold is generated simultaneously based on the actual dust accumulation data.

[0094] It should be noted that, firstly, since the vibration signal of the filter media contains interference components such as airflow disturbance and equipment operating noise, a blind source separation algorithm (such as the FastICA algorithm) is used to separate the field-induced response of the vibration signal feature values, accurately extracting three types of effective features directly related to the dust accumulation state: vibration attenuation features (the greater the amount of dust accumulation, the faster the vibration energy attenuates), vibration phase shift features (uneven dust distribution will cause vibration phase shift), and vibration noise features (friction between dust and filter media will generate noise at a specific frequency). Simultaneously, dust dielectric constant detection data (to distinguish dust types, such as the difference in dielectric constant between dry dust and sticky dust will affect the vibration response) and filter media porosity parameters (reflecting the structural characteristics of the filter media itself, the smaller the porosity, the easier it is for dust to accumulate) are combined to construct a specific mapping relationship between features and dust accumulation state. Specifically, this mapping relationship is not a general model, but is dynamically generated for specific filter media and dust types to ensure accurate correlation between features and dust accumulation amount and distribution.

[0095] Secondly, considering the cross-field coupling effect between the vibration field and the dielectric field (such as the change in the dielectric constant of dust affecting the polarization characteristics of the filter material, thus changing the vibration response), a cross-field coupling effect coefficient is introduced (to quantify the coupling strength between the two fields, obtained through experimental calibration). The three types of vibration features are weighted and fused (features with higher coupling coefficients are given higher weights) to generate a fused feature vector characterizing the dust accumulation state. Specifically, this vector integrates the dual information of vibration and dielectric, avoiding misjudgment of the dust accumulation state caused by a single physical field feature, and comprehensively reflecting key information such as dust accumulation amount, dust distribution, and dust type.

[0096] Finally, the fused feature vector is input into the multi-field coupling model for dust accumulation state inversion: the multi-field coupling model has been trained through finite element simulation and experimental data, and can calculate the actual dust accumulation data (such as dust mass per unit area and dust thickness) based on the fused feature vector; based on the actual dust accumulation data, a dust accumulation warning threshold is generated (such as a light dust accumulation threshold of 0.5 kg / m², a moderate dust accumulation threshold of 1.0 kg / m², and a heavy dust accumulation threshold of 1.5 kg / m²). This threshold provides a quantitative basis for subsequent cleaning timing determination, ensuring that cleaning is triggered neither too early nor too late.

[0097] Furthermore, the step of synchronously generating the dust accumulation warning threshold based on the actual dust accumulation data includes:

[0098] Based on the material strength parameters, pore structure parameters and the specific mapping relationship of the filter material, the stress distribution state of the filter material under different actual dust accumulation is simulated by finite element simulation to determine the corresponding maximum dust accumulation tolerance. At the same time, combined with the preset minimum filtration conditions, the corresponding basic constraint interval is generated.

[0099] Multi-dimensional constraint correction coefficients and airflow disturbance correction coefficients are introduced, and the basic constraint interval is coupled and calculated to obtain the initial dust accumulation warning threshold.

[0100] The vibration attenuation feature, the vibration phase shift feature, and the initial dust accumulation warning threshold are correlated and matched. If the feature change rate does not exceed the preset stable range, the initial dust accumulation warning threshold is set as the corresponding dust accumulation warning threshold.

[0101] It should be noted that, firstly, to avoid filter media damage due to excessive dust accumulation exceeding its tolerance limit (e.g., excessive dust accumulation increases airflow resistance and causes stress concentration), finite element simulation is used to simulate the stress distribution of the filter media under different actual dust accumulation levels, based on the material strength parameters (e.g., tensile strength, compressive strength), pore structure parameters (e.g., pore size, pore distribution), and specific mapping relationships of the filter media. Specifically, the simulation can accurately locate the stress concentration area caused by dust accumulation and determine the maximum dust accumulation that the filter media can withstand without damage (e.g., 2.0 kg / m²). At the same time, combined with preset minimum filtration conditions (e.g., filter inlet and outlet pressure difference ≤ 5 kPa, ensuring filtration efficiency not less than 95%), a corresponding basic constraint range is generated (e.g., 0.5 kg / m² ~ 1.5 kg / m², with the lower limit being the dust accumulation corresponding to the minimum filtration conditions and the upper limit being 75% of the maximum tolerable dust accumulation, reserving a safety redundancy).

[0102] Secondly, in practical applications, the operating environment of the filter media (such as temperature and humidity) and the airflow state (such as airflow disturbance and turbulence) will affect the dust accumulation state and the filter media response. Therefore, multi-dimensional constraint correction coefficients (such as temperature correction coefficient and humidity correction coefficient, which are obtained based on experimental data fitting) and airflow disturbance correction coefficients (quantifying the influence of airflow on vibration signals and dust accumulation distribution) are introduced. The basic constraint interval is coupled and calculated (such as the corrected upper limit of the threshold = basic upper limit × temperature correction coefficient × airflow disturbance correction coefficient) to obtain the initial dust accumulation warning threshold. Specifically, this correction process makes the threshold adaptable to complex operating environments and avoids threshold failure caused by environmental and airflow factors.

[0103] Finally, to ensure the stability of the threshold (avoiding frequent fluctuations caused by instantaneous vibration interference), the vibration attenuation characteristics, vibration phase shift characteristics, and initial dust accumulation warning threshold are correlated and matched, and the characteristic change rate (such as the change rate of vibration attenuation within 10 minutes) is calculated: if the characteristic change rate does not exceed the preset stable range (such as ±5%), it indicates that the dust accumulation state is stable and the initial warning threshold is reliable, and it is set as the final dust accumulation warning threshold; if it exceeds the stable range, it is determined to be an instantaneous interference (such as sudden airflow changes or instantaneous dust concentration peaks), and the data is re-collected to calculate the initial threshold and re-verify until the characteristic change rate is stable, so as to avoid false triggering of cleaning.

[0104] Furthermore, the step of matching the corresponding target cleaning parameters in the preset parameter database based on the dust accumulation warning threshold and the current operating load of the particulate filter includes:

[0105] Based on the gradient level of the dust accumulation warning threshold and the fluctuation characteristics of the current operating load, a corresponding two-dimensional parameter calibration matrix is ​​constructed. A dynamic correction factor is introduced simultaneously, and the gradient level and the fluctuation characteristics are coupled and calibrated through the least squares support vector machine algorithm to generate the corresponding core matching parameter set.

[0106] For the cleaning parameter samples in the preset parameter database, a preset clustering algorithm is used to divide them into several working condition feature clusters. The cosine similarity between the core matching dataset and each working condition feature cluster is calculated simultaneously to construct the corresponding multi-objective optimization matching function.

[0107] The multi-objective optimization matching function filters out candidate cleaning parameter groups from the preset parameter database, and the candidate cleaning parameter groups are simultaneously simulated to generate the target cleaning parameters.

[0108] It should be noted that, firstly, a two-dimensional parameter calibration matrix is ​​constructed based on the gradient levels of dust accumulation warning thresholds (e.g., light, moderate, and heavy, corresponding to different dust accumulation amounts) and the fluctuation characteristics of the current operating load (e.g., high-load stable operation, low-load fluctuating operation, and variable-load operation). Specifically, the rows of the matrix correspond to the warning threshold gradient, the columns correspond to the load fluctuation characteristics, and each element represents a set of initially matched cleaning parameter directions. Dynamic correction factors (e.g., aging correction factors based on filter runtime and viscosity correction factors based on dust adhesion) are introduced, and the gradient levels and fluctuation characteristics are coupled and calibrated using a least-squares support vector machine algorithm. Specifically, this algorithm has strong nonlinear fitting capabilities and can accurately capture the synergistic influence of dust accumulation status and load characteristics, generating a corresponding core matching parameter set (e.g., pulse pressure range, cleaning duration interval, and pulse frequency interval), providing precise constraints for subsequent parameter matching.

[0109] Secondly, for a large number of cleaning parameter samples in the preset parameter database (covering historical effective parameters under different working conditions), a preset clustering algorithm (such as K-means clustering) is used to divide them into several working condition feature clusters (such as the "light dust accumulation + high load" cluster and the "heavy dust accumulation + variable load" cluster). Each cluster contains cleaning parameter samples with similar working conditions. The cosine similarity between the core matching parameter set and each working condition feature cluster is calculated (quantifying the degree of working condition fit between the parameter set and the cluster; the higher the similarity, the stronger the fit). A multi-objective optimization matching function is constructed. Specifically, the optimization objective of the function is "maximizing cleaning effect (minimizing residual dust accumulation), minimizing filter material damage risk, and minimizing energy consumption", with the constraint being the range of the core matching parameter set.

[0110] Finally, candidate cleaning parameter groups (e.g., 3-5 groups) that meet the optimization objectives are selected from the preset parameter database through a multi-objective optimization matching function. The candidate parameter groups are then imported into a cleaning process simulation model (e.g., a CFD simulation model) for simulation, simulating the dust stripping process (e.g., the pulse impact force and stripping trajectory of the dust) and the dynamic response of the filter media (e.g., the stress change and vibration amplitude of the filter media), and outputting core simulation data (e.g., simulated residual dust amount, maximum stress of the filter media, and cleaning energy consumption). Based on the core simulation data, the parameter group with the best comprehensive performance (e.g., residual dust amount ≤ 0.1 kg / m², filter media stress ≤ 80% of the material yield strength, and energy consumption ≤ preset threshold) is selected as the target cleaning parameters, ensuring that the parameters are suitable for actual working conditions while also taking into account cleaning effect, equipment safety, and energy consumption control.

[0111] Furthermore, the step of simultaneously simulating the candidate cleaning parameter group to generate the target cleaning parameters includes:

[0112] The preset cleaning process simulation model is brought up, the candidate cleaning parameter group is imported into the cleaning process simulation model, and the initial simulation conditions are set in combination with the vibration response characteristics of the filter material to simulate the dust stripping process and dynamic response of the filter material under different parameters, and the corresponding core simulation data is output synchronously.

[0113] Based on the core simulation data, the risk level of filter media damage caused by each candidate cleaning parameter group is evaluated, and intermediate cleaning parameter groups are selected simultaneously based on the risk level.

[0114] Based on the actual operating deviations of the particulate filter, a corresponding dynamic parameter adaptation function is constructed. Simultaneously, the intermediate cleaning parameter set is corrected through the dynamic parameter adaptation function to generate the target cleaning parameters.

[0115] It should be noted that, firstly, the preset cleaning process simulation model is brought up (this model has been calibrated with experimental data and can accurately simulate the physical process under different cleaning parameters). The candidate cleaning parameter group (such as pulse air pressure 0.6MPa, cleaning time 0.5s, pulse frequency 5Hz) is imported into the model. Combined with the vibration response characteristics of the filter material (such as the natural frequency of the filter material and the damping coefficient), the initial simulation conditions (such as the initial dust accumulation and airflow velocity) are set. The dust peeling process under different parameters (for sticky dust, the process of pulse impact force overcoming adhesion force needs to be simulated, and for dry dust, the dust falling off due to vibration needs to be simulated) and the dynamic response of the filter material (such as avoiding the vibration frequency of the filter material from approaching the natural frequency and causing resonance) are simulated. The core simulation data (such as residual dust accumulation, maximum stress of the filter material, cleaning time, and energy consumption) are output.

[0116] Secondly, the risk level of filter media damage caused by each candidate cleaning parameter group is evaluated based on the core simulation data: a risk level classification standard is set (e.g., maximum stress of filter media ≥ yield strength is high risk, ≥80% yield strength is medium risk, <80% is low risk), and low-risk intermediate cleaning parameter groups are screened out. Specifically, this step can avoid filter media damage caused by improper parameters in actual application and reduce equipment maintenance costs. For medium-risk parameter groups, if their cleaning effect is significantly better than that of low-risk groups, they can be temporarily retained as alternatives, and the risk can be reduced later by parameter adjustment.

[0117] Finally, considering the discrepancies between the simulation model and actual operating conditions (such as the simulation not fully simulating airflow turbulence and the degree of filter media aging), and combining the actual operating condition deviations of the particulate filter (such as the actual airflow velocity being 10% higher than the simulation setting and the filter media aging causing a 5% decrease in elastic modulus), a dynamic parameter adaptation function is constructed (e.g., the adapted pulse pressure = simulation parameter × (actual airflow velocity / simulated airflow velocity) × aging correction coefficient). This function is used to correct the intermediate cleaning parameter set (e.g., increasing the pulse pressure to offset the effect of high airflow velocity and reducing the vibration amplitude to adapt to the aged filter media), generating the final target cleaning parameters. Specifically, the corrected parameters are more in line with the actual operating conditions, ensuring a balance between cleaning effect and equipment safety.

[0118] Furthermore, the step of detecting the filter media vibration response signal, air permeability recovery rate, and exhaust particulate matter concentration, and simultaneously determining whether the cleaning standard is met, includes:

[0119] Based on the residual vibration attenuation characteristics of the cleaned filter media, the detection frequency and sampling duration are dynamically adapted, and the vibration response signals of the filter media at multiple measurement points are collected simultaneously through the laser Doppler vibration measurement module.

[0120] The vibration response signals from the multiple measurement points are subjected to Hilbert transform to extract the corresponding instantaneous frequency and damping ratio characteristic parameters. The corresponding four-dimensional feature vector is then constructed by combining the air permeability recovery rate and the exhaust particulate matter concentration.

[0121] Based on the dust accumulation characteristic correlation data output by the multi-field coupling model, the four-dimensional feature vector is fused and calculated using the random forest algorithm to obtain the corresponding comprehensive cleaning effect evaluation value, so as to determine whether the standard is met.

[0122] It should be noted that, firstly, residual vibration will exist in the filter media after cleaning (the characteristics of residual vibration are related to the amount of residual dust). Based on the attenuation characteristics of residual vibration (e.g., the higher the damping ratio of residual vibration, the more residual dust), the detection frequency (e.g., high-frequency sampling is used when the residual vibration frequency is high, and low-frequency sampling is used when it is low) and sampling duration (e.g., the sampling duration is extended when the vibration attenuation is slow) are dynamically adapted. The vibration response signal of the filter media at multiple measurement points is collected by the laser Doppler vibration measurement module. Specifically, multi-point acquisition can avoid the randomness of single-point measurement (e.g., signal deviation caused by local residual dust). Laser Doppler technology has the advantages of high precision and non-contact measurement, and can accurately capture weak vibration signals.

[0123] Secondly, the vibration response signals from multiple measurement points are subjected to Hilbert transform to extract the corresponding instantaneous frequency (reflecting the dynamic changes in filter media vibration) and damping ratio characteristic parameters (reflecting the decay rate of vibration energy). Specifically, these two parameters can indirectly characterize the amount of residual dust (the more residual dust, the greater the damping ratio, and the more unstable the instantaneous frequency). Combined with the air permeability recovery rate (measured by the pressure difference method, air permeability recovery rate = air permeability after cleaning / initial air permeability before cleaning × 100%, reflecting the degree of filtration capacity recovery) and exhaust particulate matter concentration (measured by a particulate matter sensor, reflecting the emission compliance status), a four-dimensional feature vector is constructed. Specifically, this vector comprehensively characterizes the cleaning effect from three dimensions: "residual dust (vibration parameters), filtration capacity (air permeability), and emission compliance (particulate matter concentration)," avoiding misjudgments caused by a single indicator (such as air permeability meeting the standard but excessive residual dust, which may lead to re-clogging in the short term).

[0124] Finally, based on the dust accumulation characteristic correlation data (such as dust accumulation type, initial dust accumulation amount, and dust accumulation distribution) output by the multi-field coupling model, the four-dimensional feature vectors are fused and calculated using the random forest algorithm. Specifically, the random forest algorithm has strong anti-interference ability and high prediction accuracy, and can effectively handle the nonlinear correlation of multi-dimensional features. The algorithm calculates a comprehensive cleaning effect evaluation value (e.g., 0~100 points, with 80 points or above considered acceptable). If the evaluation value is ≥ the preset acceptable threshold, the cleaning is deemed acceptable, and the cleaning process is completed. If the standard is not met, the unacceptable indicators are analyzed (e.g., excessive exhaust particulate matter concentration may indicate incomplete cleaning, and insufficient air permeability recovery may indicate severe filter clogging). The target cleaning parameters are adjusted (e.g., increasing pulse air pressure and extending cleaning time), and the cleaning is repeated until the standard is met.

[0125] Furthermore, the step of using the random forest algorithm to fuse and calculate the four-dimensional feature vectors to obtain the corresponding comprehensive cleaning effect evaluation value, and to determine whether the standard is met, includes:

[0126] Based on the dust accumulation characteristic correlation data, a cross-enhanced feature set adapted to the four-dimensional feature vector is constructed, and the core features are simultaneously screened out through the mutual information entropy algorithm to generate the corresponding feature matrix.

[0127] The random forest algorithm is used to perform prior constraint processing on the feature matrix to calculate the contribution value corresponding to each feature.

[0128] The contribution values ​​are weighted and fused to obtain the comprehensive cleaning effect evaluation value, and the standard is determined based on the magnitude of the comprehensive cleaning effect evaluation value.

[0129] It should be noted that, firstly, based on dust accumulation characteristic correlation data (e.g., the cleaning effect evaluation of sticky dust needs to focus on the residual dust amount, while the air permeability recovery rate needs to be focused on dry dust), a cross-enhanced feature set adapted to the four-dimensional feature vector is constructed. Specifically, the four-dimensional features are cross-combined (e.g., "instantaneous frequency × air permeability recovery rate" and "damping ratio × exhaust particulate matter concentration") to improve the feature discrimination (e.g., the differences in cross features between samples with different cleaning effects are more significant). Simultaneously, the core features are screened out using the mutual information entropy algorithm (the higher the mutual information entropy value, the stronger the correlation between the feature and the cleaning effect), and redundant features (e.g., combinations of cross features that are weakly correlated with the cleaning effect) are eliminated to generate the corresponding feature matrix. Specifically, the screening of core features can reduce the computational load of the algorithm, improve the evaluation accuracy, and avoid interference from redundant features.

[0130] Secondly, a random forest algorithm is used to process the feature matrix with prior constraints: prior constraints are set based on the dust accumulation characteristic correlation data (e.g., in sticky dust samples, the weight of residual vibration parameters should be higher than that of air permeability parameters; in samples with high initial dust accumulation, the threshold for achieving the cleaning effect evaluation value can be appropriately reduced). During the calculation process, the algorithm will adjust the contribution distribution of features in combination with prior constraints to ensure that the evaluation results are consistent with the actual dust accumulation characteristics. The algorithm calculates the contribution value corresponding to each core feature (e.g., instantaneous frequency contribution value 30%, damping ratio 25%, air permeability recovery rate 25%, exhaust particulate matter concentration 20%). Specifically, the contribution value quantifies the degree of influence of each feature on the cleaning effect evaluation, avoiding evaluation bias caused by treating features equally.

[0131] Finally, the contribution values ​​of each feature are weighted and fused (comprehensive evaluation value = Σ(feature value × corresponding contribution value)) to obtain the final comprehensive cleaning effect evaluation value. A compliance threshold is set (e.g., 80 points according to industry standards and equipment requirements). If the evaluation value is ≥ the threshold, the cleaning is deemed to be compliant; if it is lower than the threshold, it is deemed not compliant and a secondary cleaning process is triggered. Specifically, the weighted fusion process ensures the objectivity and accuracy of the evaluation value, taking into account multi-dimensional indicators and highlighting the impact of key features, providing a reliable basis for judging the compliance of cleaning and completing the closed-loop control of the entire particulate filter self-cleaning process.

[0132] Please see Figure 2 The third embodiment of the present invention provides:

[0133] A particulate filter self-cleaning system, wherein the system comprises:

[0134] The module is used to construct a corresponding multi-field coupling model based on the vibration response characteristics of the particulate filter media and the detection data of the dielectric constant of dust. Simultaneously, the corresponding vibration signal feature values ​​are extracted through Hilbert transform, so as to generate the corresponding dust accumulation warning threshold based on the vibration signal feature values ​​through the multi-field coupling model.

[0135] The matching module is used to match the corresponding target cleaning parameters in a preset parameter database based on the dust accumulation warning threshold and the current operating load of the particulate filter.

[0136] The processing module is used to input the target cleaning parameters into a preset cleaning device, so that the preset cleaning device cleans the inside of the particulate filter;

[0137] The detection module is used to detect the filter media vibration response signal, air permeability recovery rate and exhaust particulate matter concentration after cleaning is completed, and simultaneously determine whether the cleaning meets the standards, so as to complete the corresponding cleaning treatment.

[0138] Furthermore, the building module is specifically used for:

[0139] The vibration signal feature values ​​are separated by a blind source separation algorithm to generate corresponding vibration attenuation features, vibration phase shift features and vibration noise features. Simultaneously, the dust dielectric constant detection data and filter material porosity parameters are combined to construct a unique mapping relationship between the features and the dust accumulation state.

[0140] By introducing a cross-field coupling effect coefficient and combining it with the specific mapping relationship, the vibration attenuation feature, the vibration phase shift feature, and the vibration noise feature are weighted and fused to generate a fused feature vector characterizing the dust accumulation state.

[0141] The fused feature vector is input into the multi-field coupling model to perform dust accumulation state inversion to obtain actual dust accumulation data, and the dust accumulation warning threshold is generated simultaneously based on the actual dust accumulation data.

[0142] Furthermore, the building module is specifically used for:

[0143] Based on the material strength parameters, pore structure parameters and the specific mapping relationship of the filter material, the stress distribution state of the filter material under different actual dust accumulation is simulated by finite element simulation to determine the corresponding maximum dust accumulation tolerance. At the same time, combined with the preset minimum filtration conditions, the corresponding basic constraint interval is generated.

[0144] Multi-dimensional constraint correction coefficients and airflow disturbance correction coefficients are introduced, and the basic constraint interval is coupled and calculated to obtain the initial dust accumulation warning threshold.

[0145] The vibration attenuation feature, the vibration phase shift feature, and the initial dust accumulation warning threshold are correlated and matched. If the feature change rate does not exceed the preset stable range, the initial dust accumulation warning threshold is set as the corresponding dust accumulation warning threshold.

[0146] Furthermore, the matching module is specifically used for:

[0147] Based on the gradient level of the dust accumulation warning threshold and the fluctuation characteristics of the current operating load, a corresponding two-dimensional parameter calibration matrix is ​​constructed. A dynamic correction factor is introduced simultaneously, and the gradient level and the fluctuation characteristics are coupled and calibrated through the least squares support vector machine algorithm to generate the corresponding core matching parameter set.

[0148] For the cleaning parameter samples in the preset parameter database, a preset clustering algorithm is used to divide them into several working condition feature clusters. The cosine similarity between the core matching dataset and each working condition feature cluster is calculated simultaneously to construct the corresponding multi-objective optimization matching function.

[0149] The multi-objective optimization matching function filters out candidate cleaning parameter groups from the preset parameter database, and the candidate cleaning parameter groups are simultaneously simulated to generate the target cleaning parameters.

[0150] Furthermore, the matching module is specifically used for:

[0151] The preset cleaning process simulation model is brought up, the candidate cleaning parameter group is imported into the cleaning process simulation model, and the initial simulation conditions are set in combination with the vibration response characteristics of the filter material to simulate the dust stripping process and dynamic response of the filter material under different parameters, and the corresponding core simulation data is output synchronously.

[0152] Based on the core simulation data, the risk level of filter media damage caused by each candidate cleaning parameter group is evaluated, and intermediate cleaning parameter groups are selected simultaneously based on the risk level.

[0153] Based on the actual operating deviations of the particulate filter, a corresponding dynamic parameter adaptation function is constructed. Simultaneously, the intermediate cleaning parameter set is corrected through the dynamic parameter adaptation function to generate the target cleaning parameters.

[0154] Furthermore, the detection module is specifically used for:

[0155] Based on the residual vibration attenuation characteristics of the cleaned filter media, the detection frequency and sampling duration are dynamically adapted, and the vibration response signals of the filter media at multiple measurement points are collected simultaneously through the laser Doppler vibration measurement module.

[0156] The vibration response signals from the multiple measurement points are subjected to Hilbert transform to extract the corresponding instantaneous frequency and damping ratio characteristic parameters. The corresponding four-dimensional feature vector is then constructed by combining the air permeability recovery rate and the exhaust particulate matter concentration.

[0157] Based on the dust accumulation characteristic correlation data output by the multi-field coupling model, the four-dimensional feature vector is fused and calculated using the random forest algorithm to obtain the corresponding comprehensive cleaning effect evaluation value, so as to determine whether the standard is met.

[0158] Furthermore, the detection module is specifically used for:

[0159] Based on the dust accumulation characteristic correlation data, a cross-enhanced feature set adapted to the four-dimensional feature vector is constructed, and the core features are simultaneously screened out through the mutual information entropy algorithm to generate the corresponding feature matrix.

[0160] The random forest algorithm is used to perform prior constraint processing on the feature matrix to calculate the contribution value corresponding to each feature.

[0161] The contribution values ​​are weighted and fused to obtain the comprehensive cleaning effect evaluation value, and the standard is determined based on the magnitude of the comprehensive cleaning effect evaluation value.

[0162] The fourth embodiment of the present invention provides a computer, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the particulate filter self-cleaning method as described above.

[0163] The fifth embodiment of the present invention provides a readable storage medium having a computer program stored thereon, wherein the program, when executed by a processor, implements the particulate filter self-cleaning method as described above.

[0164] In summary, the particulate filter self-cleaning method and system provided in the above embodiments of the present invention can automatically and accurately complete the cleaning process of particulate filters, thereby improving the cleaning efficiency.

[0165] It should be noted that the above modules can be functional modules or program modules, and can be implemented through software or hardware. For modules implemented through hardware, the above modules can reside in the same processor; or the above modules can be located in different processors in any combination.

[0166] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-including system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0167] More specific examples of computer-readable media (a non-exhaustive list) include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0168] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0169] In the description of this specification, references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0170] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the appended claims.

Claims

1. A self-cleaning method for a particulate filter, characterized in that, The method includes: Based on the vibration response characteristics of particulate filter media and dust dielectric constant detection data, a corresponding multi-field coupling model is constructed. Simultaneously, the corresponding vibration signal feature values ​​are extracted through Hilbert transform. The corresponding dust accumulation warning threshold is then generated based on the vibration signal feature values ​​through the multi-field coupling model. Based on the dust accumulation warning threshold and the current operating load of the particulate filter, the corresponding target cleaning parameters are matched in the preset parameter database. The target cleaning parameters are input into a preset cleaning device so that the preset cleaning device cleans the inside of the particulate filter. After cleaning, the vibration response signal of the filter media, the air permeability recovery rate, and the concentration of exhaust particulate matter are detected to determine whether the cleaning meets the standards and complete the corresponding cleaning treatment.

2. The self-cleaning method for a particulate filter according to claim 1, characterized in that, The step of generating the corresponding dust accumulation early warning threshold based on the vibration signal feature values ​​using the multi-field coupling model includes: The vibration signal feature values ​​are separated by a blind source separation algorithm to generate corresponding vibration attenuation features, vibration phase shift features and vibration noise features. Simultaneously, the dust dielectric constant detection data and filter material porosity parameters are combined to construct a unique mapping relationship between the features and the dust accumulation state. By introducing a cross-field coupling effect coefficient and combining it with the specific mapping relationship, the vibration attenuation feature, the vibration phase shift feature, and the vibration noise feature are weighted and fused to generate a fused feature vector characterizing the dust accumulation state. The fused feature vector is input into the multi-field coupling model to perform dust accumulation state inversion to obtain actual dust accumulation data, and the dust accumulation warning threshold is generated simultaneously based on the actual dust accumulation data.

3. The self-cleaning method for a particulate filter according to claim 2, characterized in that, The step of synchronously generating the dust accumulation warning threshold based on the actual dust accumulation data includes: Based on the material strength parameters, pore structure parameters and the specific mapping relationship of the filter material, the stress distribution state of the filter material under different actual dust accumulation is simulated by finite element simulation to determine the corresponding maximum dust accumulation tolerance. At the same time, combined with the preset minimum filtration conditions, the corresponding basic constraint interval is generated. Multi-dimensional constraint correction coefficients and airflow disturbance correction coefficients are introduced, and the basic constraint interval is coupled and calculated to obtain the initial dust accumulation warning threshold. The vibration attenuation feature, the vibration phase shift feature, and the initial dust accumulation warning threshold are correlated and matched. If the feature change rate does not exceed the preset stable range, the initial dust accumulation warning threshold is set as the corresponding dust accumulation warning threshold.

4. The self-cleaning method for a particulate filter according to claim 1, characterized in that, The step of matching the corresponding target cleaning parameters in the preset parameter database based on the dust accumulation warning threshold and the current operating load of the particulate filter includes: Based on the gradient level of the dust accumulation warning threshold and the fluctuation characteristics of the current operating load, a corresponding two-dimensional parameter calibration matrix is ​​constructed. A dynamic correction factor is introduced simultaneously, and the gradient level and the fluctuation characteristics are coupled and calibrated through the least squares support vector machine algorithm to generate the corresponding core matching parameter set. For the cleaning parameter samples in the preset parameter database, a preset clustering algorithm is used to divide them into several working condition feature clusters. The cosine similarity between the core matching parameter set and each working condition feature cluster is calculated simultaneously to construct the corresponding multi-objective optimization matching function. The multi-objective optimization matching function filters out candidate cleaning parameter groups from the preset parameter database, and the candidate cleaning parameter groups are simultaneously simulated to generate the target cleaning parameters.

5. The self-cleaning method for a particulate filter according to claim 4, characterized in that, The step of simulating the candidate cleaning parameter group to generate the target cleaning parameters includes: The preset cleaning process simulation model is brought up, the candidate cleaning parameter group is imported into the cleaning process simulation model, and the initial simulation conditions are set in combination with the vibration response characteristics of the filter material to simulate the dust stripping process and dynamic response of the filter material under different parameters, and the corresponding core simulation data is output synchronously. Based on the core simulation data, the risk level of filter media damage caused by each candidate cleaning parameter group is evaluated, and intermediate cleaning parameter groups are selected simultaneously based on the risk level. Based on the actual operating deviations of the particulate filter, a corresponding dynamic parameter adaptation function is constructed. Simultaneously, the intermediate cleaning parameter set is corrected through the dynamic parameter adaptation function to generate the target cleaning parameters.

6. The self-cleaning method for a particulate filter according to claim 1, characterized in that, The steps for detecting the filter media vibration response signal, air permeability recovery rate, and exhaust particulate matter concentration, and simultaneously determining whether the cleaning meets the standards, include: Based on the residual vibration attenuation characteristics of the cleaned filter media, the detection frequency and sampling duration are dynamically adapted, and the vibration response signals of the filter media at multiple measurement points are collected simultaneously through the laser Doppler vibration measurement module. The vibration response signals from the multiple measurement points are subjected to Hilbert transform to extract the corresponding instantaneous frequency and damping ratio characteristic parameters. The corresponding four-dimensional feature vector is then constructed by combining the air permeability recovery rate and the exhaust particulate matter concentration. Based on the dust accumulation characteristic correlation data output by the multi-field coupling model, the four-dimensional feature vector is fused and calculated using the random forest algorithm to obtain the corresponding comprehensive cleaning effect evaluation value, so as to determine whether the standard is met.

7. The self-cleaning method for a particulate filter according to claim 6, characterized in that, The step of using the random forest algorithm to fuse and calculate the four-dimensional feature vectors to obtain the corresponding comprehensive cleaning effect evaluation value, and to determine whether the standard is met, includes: Based on the dust accumulation characteristic correlation data, a cross-enhanced feature set adapted to the four-dimensional feature vector is constructed, and the core features are simultaneously screened out through the mutual information entropy algorithm to generate the corresponding feature matrix. The random forest algorithm is used to perform prior constraint processing on the feature matrix to calculate the contribution value corresponding to each feature. The contribution values ​​are weighted and fused to obtain the comprehensive cleaning effect evaluation value, and the standard is determined based on the magnitude of the comprehensive cleaning effect evaluation value.

8. A self-cleaning system for particulate filters, characterized in that, The system includes: The module is used to construct a corresponding multi-field coupling model based on the vibration response characteristics of the particulate filter media and the detection data of the dielectric constant of dust. Simultaneously, the corresponding vibration signal feature values ​​are extracted through Hilbert transform, so as to generate the corresponding dust accumulation warning threshold based on the vibration signal feature values ​​through the multi-field coupling model. The matching module is used to match the corresponding target cleaning parameters in a preset parameter database based on the dust accumulation warning threshold and the current operating load of the particulate filter. The processing module is used to input the target cleaning parameters into a preset cleaning device, so that the preset cleaning device cleans the inside of the particulate filter; The detection module is used to detect the filter media vibration response signal, air permeability recovery rate and exhaust particulate matter concentration after cleaning is completed, and simultaneously determine whether the cleaning meets the standards, so as to complete the corresponding cleaning treatment.

9. A computer comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the particulate filter self-cleaning method as described in any one of claims 1 to 7.

10. A readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the program implements the particulate filter self-cleaning method as described in any one of claims 1 to 7.

Images