A lithium slurry magnetic foreign matter online monitoring method based on high-frequency resonance spectroscopy

Abstract

The application discloses a kind of lithium electric slurry magnetic foreign matter on-line monitoring methods based on high-frequency resonance spectroscopy, belong to lithium ion battery slurry detection technical field, first construct magnetic focusing resonance sensing area, apply wideband excitation and obtain background impedance spectrum, real-time monitoring and extract multiple characteristic parameters, determine the flow rate of microparticle and carry out normalization processing to signal waveform, realize foreign matter identification by multiple characteristic fusion inversion, can be selected to carry out statistics and output.The application uses magnetic focusing resonance structure to improve signal-to-noise ratio, distinguishes interference through double-feature collaborative analysis, eliminates flow rate influence using double-sensing area layout, realizes the full-flow, real-time, on-line, accurate monitoring of ferromagnetic foreign matter above 25 μm in lithium battery slurry, can be applied to metal contaminant monitoring of a variety of high-cleanliness requirements fluid, and has important industrial application value.

Description

Technical Field

[0001] This invention belongs to the field of lithium-ion battery slurry detection technology, and in particular relates to an online monitoring method for magnetic foreign matter in lithium battery slurry based on high-frequency resonance spectroscopy. Background Technology

[0002] As a core component of clean energy storage, the safety of lithium-ion batteries is paramount. The purity of the positive electrode slurry (such as ternary materials NCM, lithium iron phosphate LFP, etc.) directly determines the safety and electrochemical consistency of the battery. During the preparation, transportation, and coating processes of the slurry, micron-sized (especially larger than 50μm) ferromagnetic metal particles (such as iron, nickel, cobalt, and their alloys) can easily be mixed in due to equipment wear, the introduction of raw material impurities, or improper operation. Because the electrode spacing inside the battery is extremely small (usually on the micron level), if these metal particles are not detected and enter the finished battery, they will puncture the separator or directly connect to the positive and negative electrodes, causing serious internal short circuits, which may lead to catastrophic safety accidents such as thermal runaway, fire, and explosion. Currently, the industry-wide standard for the lower limit of detection is 100μm, while the actual internal control requirements of mainstream battery manufacturers have been refined to 25μm, requiring simultaneous monitoring of particle size distribution and total content to achieve full-process quality control.

[0003] Currently, industrial detection of magnetic foreign matter in lithium battery slurries mainly relies on offline sampling methods, including strong magnetic adsorption and spectroscopic analysis. These methods all require manual sampling and complex sample pretreatment, resulting in low detection efficiency and poor real-time performance, making it impossible to achieve real-time monitoring and closed-loop control of the production process. Therefore, developing a magnetic foreign matter detection technology for lithium battery slurries that can achieve full-flow, real-time, online, and continuous monitoring has become an urgent problem to be solved in the industry.

[0004] Lithium-ion battery slurry is a complex colloidal system with high viscosity and high conductivity, containing components such as conductive carbon black, binders, and NMP solvents. Its high conductivity and complex dielectric properties generate huge background eddy current losses in the detection coil, forming strong background noise that drowns out the weak useful signal of the target particle, causing a sharp drop in the signal-to-noise ratio. For tiny ferromagnetic particles at the 25μm level, the change in effective permeability they cause is extremely weak. Conventional open-circuit hollow coil structures have divergent magnetic fields and low energy density, making it difficult to effectively capture this change. Actual slurries often contain interfering substances such as bubbles and conductive carbon black agglomerates. These interfering substances may produce signal characteristics similar to those of metal particles in conventional inductive detection, leading to serious false alarms. In industrial production, the flow rate of slurry in pipelines fluctuates greatly. The amplitude and waveform of the induced signal generated when particles of the same size pass through the detection area at different flow rates will change, causing serious distortion of the size inversion algorithm based on signal amplitude and inaccurate quantitative results.

[0005] Therefore, given the unique physicochemical properties of lithium battery slurry, which are characterized by high conductivity and high viscosity, there is an urgent need to develop a novel online monitoring method that can effectively penetrate background noise, accurately capture and quantitatively analyze ferromagnetic metal particles larger than 25 μm, in order to solve the core technical problems existing in the current technology, such as signal submersion, insufficient sensitivity, poor anti-interference ability, and large influence of flow rate. Summary of the Invention

[0006] Purpose of the invention: The purpose of this invention is to provide an online monitoring method for magnetic foreign matter in lithium battery slurry based on high-frequency resonant spectroscopy. This method aims to solve the key technical problems of existing offline detection methods being unable to monitor online in real time, and conventional fluid inductance detection methods being applied to high-conductivity lithium battery slurry, such as signal submersion due to slurry conductivity, insufficient sensitivity, inability to effectively distinguish interference such as bubbles, and severe impact of flow rate fluctuations on quantitative accuracy.

[0007] Technical solution: The present invention provides an online monitoring method for magnetic foreign matter in lithium battery paste based on high-frequency resonance spectroscopy, comprising the following steps:

[0008] S1. Constructing a magnetic focusing resonant sensing system: At a preset position in the lithium battery slurry conveying channel, at least two magnetic focusing resonant sensing units are set along the slurry flow direction. Each magnetic focusing resonant sensing unit uses a high-permeability magnetic core to confine the high-frequency electromagnetic field energy in a local space, forming a high-energy-density sensing area.

[0009] S2. Obtaining the background reference spectrum: Injecting a wideband high-frequency magnetic excitation signal into the magnetic focusing resonant sensing unit and collecting the background impedance spectrum data when a non-magnetic foreign object passes through. The background impedance spectrum data includes the resonant frequency reference value and the quality factor reference value.

[0010] S3. Real-time acquisition and feature extraction: When the lithium battery slurry flows through the sensing area, the impedance spectrum change data of each magnetic focusing resonant sensing unit is acquired in real time, and two feature parameters, resonant frequency shift and quality factor change, are extracted from the impedance spectrum change data. The resonant frequency shift reflects the change in magnetic permeability in the sensing area, and the quality factor change reflects the change in eddy current loss in the sensing area.

[0011] S4. Flow velocity measurement and signal normalization: Using the spatial layout of multiple magnetic focusing resonant sensing units, the time difference of the same magnetic foreign object passing through different sensing units in sequence is obtained. Combined with the fixed spacing between the sensing units, the instantaneous flow velocity of the magnetic foreign object is calculated. Then, the instantaneous flow velocity is used to normalize the resonant frequency offset signal to obtain normalized characteristic parameters related to the size of the magnetic foreign object.

[0012] S5. Foreign Object Identification and Quantitative Analysis: Using the normalized characteristic parameters and quality factor changes as inputs, and substituting them into a preset inversion model, the equivalent size and magnetic properties of the foreign object are decoupled to complete the identification and quantification of magnetic foreign objects.

[0013] Furthermore, in step S1, the high permeability magnetic core of the magnetic focusing resonant sensing unit is selected from C-type, E-type, or toroidal manganese-zinc ferrite cores or amorphous alloy cores, and a focusing air gap is formed at the opening of the high permeability magnetic core, the width of the focusing air gap being 0.5-5mm.

[0014] Furthermore, in step S1, the lithium battery slurry conveying channel is a bypass channel led out from the main conveying pipeline, or an insulating pipe section embedded in the main conveying pipeline; in addition to having insulation properties to avoid eddy current losses, the insulating pipe section is also made of non-ferromagnetic material to avoid interference with the magnetic field; it is made of polytetrafluoroethylene, borosilicate glass or polydimethylsiloxane material.

[0015] Furthermore, in step S2, the frequency range of the wideband high-frequency excitation signal is 100kHz to 10MHz, and the excitation signal is injected by frequency sweeping; the background impedance spectrum data is automatically updated when the system starts or at a preset cycle to adapt to the dynamic changes in slurry formulation, temperature and conductivity.

[0016] Furthermore, in step S3, the resonant frequency offset and quality factor change Calculated using the following formula:

[0017] ;

[0018] ;

[0019] in, To monitor the obtained resonant frequency in real time, The resonant frequency reference value obtained in step S2; To monitor the obtained quality factor in real time, The quality factor reference value obtained in step S2; the quality factor is defined as the ratio of the resonant peak frequency to the half-power bandwidth, or it can be directly measured by an impedance analyzer.

[0020] Furthermore, step S4 specifically involves recording the corresponding resonant frequency offset when the same magnetic foreign object passes through the upstream and downstream sensing areas of the magnetic focusing resonant sensing unit. Pulse signals, and calculate the time difference between the peak values ​​of two pulse signals. Based on the fixed distance between the upstream and downstream sensing areas and time difference Calculate the instantaneous flow velocity of magnetic foreign objects :

[0021] .

[0022] Furthermore, in step S4, the normalization process includes: based on the instantaneous flow velocity The pulse waveform of the resonant frequency offset is scaled proportionally, or the frequency shift amplitude is corrected according to the flow velocity using a lookup table method to obtain a normalized frequency shift amplitude that is only related to the size of the magnetic foreign object.

[0023] The calculation formula is as follows:

[0024] ;

[0025] in, For the normalized frequency shift amplitude, The measured pulse amplitude represents the original resonant frequency offset. To calculate the instantaneous flow velocity of the particles, A preset standard reference flow velocity (e.g., 1.0 m / s, which can vary) is used; this formula eliminates the amplitude distortion caused by flow velocity fluctuations, yielding a normalized frequency shift amplitude that is only related to particle size. .

[0026] Furthermore, in step S5, the inversion model is a classification regression model constructed based on a machine learning algorithm. The training samples of the classification regression model include normalized feature parameters and quality factor change feature pairs generated when magnetic foreign objects of different particle sizes and materials pass through the sensing unit; the materials include iron, non-ferromagnetic metal particles, and bubbles.

[0027] Furthermore, in step S5, the foreign object identification is achieved through the collaborative analysis of normalized feature parameters and quality factor changes: if the normalized feature parameter is greater than a first preset threshold and the quality factor change is less than a second preset threshold, it is determined to be a ferromagnetic metal particle; if the normalized feature parameter is less than a third preset threshold and the quality factor change is greater than a fourth preset threshold, it is determined to be a non-magnetic conductive impurity or bubble; if the normalized feature parameter is greater than the first preset threshold and the quality factor change is not significantly different, it is determined to be a special magnetic non-metallic impurity.

[0028] First preset threshold ( Lower limit): The corresponding target detection lower limit (e.g.) The normalized frequency shift amplitude of ferromagnetic particles.

[0029] Second preset threshold ( Upper limit): corresponding to ferromagnetic particles Range of variation.

[0030] The third preset threshold ( Upper limit): corresponding to non-magnetic conductive impurities Upper limit.

[0031] Fourth preset threshold ( (Lower limit): corresponding to non-magnetic conductive impurities Range of variation.

[0032] Furthermore, it also includes step S6: Statistics and Output: Accumulate the foreign object identification results per unit time, statistically analyze the foreign object size distribution and the total content of ferromagnetic foreign objects, and output the results; the output includes a histogram of foreign object size distribution, the total content of ferromagnetic foreign objects per unit time, and an alarm signal triggered when the detected foreign object size exceeds a preset threshold; the preset threshold is ≥25. .

[0033] The present invention also discloses a computer device, including a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the method of the present invention.

[0034] Beneficial effects: Compared with the prior art, the present invention has the following significant advantages:

[0035] (1) This invention addresses the significant background eddy current interference caused by the high conductivity of lithium battery slurry. Based on the principle of non-invasive sensing, it creatively employs a high-permeability double-c magnetic core (such as manganese-zinc ferrite) to construct a magnetic focusing resonant sensing structure, which is installed on the outer wall of the flow channel. This structure tightly confines the originally divergent magnetic lines of force within a centimeter-level air gap space, forming a high-energy-density local sensing area. Compared with conventional hollow coil solutions, this method not only enhances the interaction strength between the target particles and the detection magnetic field, improving the signal-to-noise ratio by more than two orders of magnitude, but also achieves continuous, real-time online monitoring of the entire slurry flow rate. No manual sampling or cumbersome sample pretreatment is required.

[0036] (2) This invention proposes to simultaneously extract and analyze the resonant frequency offset ( ) and the change in quality factor ( Two orthogonal characteristic parameters in physical sense. It is mainly dominated by the magnetic permeability of the particles (volume effect), while The conductivity of the particles (eddy current loss effect) is the primary factor. Through dual-feature synergistic analysis, ferromagnetic metal particles can be effectively distinguished at the physical mechanism level (feature combination: Significant+ (Descending), non-magnetic conductive impurities / bubbles (characteristic combination: Minimal+ This method effectively detects (obvious) and other types of interference, fundamentally solving the problem of false alarms caused by the inability of traditional single-parameter methods to distinguish interference, and improving the specificity and reliability of detection.

[0037] (3) This invention utilizes the spatial layout of dual sensing zones to accurately measure the instantaneous velocity of a particle by calculating the time difference between the two sensing zones. Based on this, the measured instantaneous velocity is used to normalize the detected pulse signal waveform, mathematically eliminating signal distortion caused by velocity variations. This technology overcomes the limitation of velocity fluctuations on the accuracy of online quantitative analysis, ensuring high consistency between online monitoring data and offline laboratory analysis results, and guaranteeing the stability and traceability of quantitative results under different operating conditions. Attached Figure Description

[0038] Figure 1 This is a schematic diagram of the magnetic focusing resonant sensing unit.

[0039] Figure 2 This is a schematic diagram of the overall process of the method of the present invention;

[0040] Figure 3 This is a schematic diagram showing the change in the resonance curve of the sensing coil when ferromagnetic particles pass through;

[0041] Figure 4 The resonant frequency offset output by the upstream and downstream sensing regions Pulse timing diagram.

[0042] Figure Labels

[0043] 1. C-type magnetic core, 2. Coil, 3. Insulating flow channel, 4. Focusing air gap, 5. Magnetic lines of force, 6. Ferromagnetic particles. Detailed Implementation

[0044] The technical solution of the present invention will be further described below with reference to the accompanying drawings.

[0045] In this embodiment, lithium iron phosphate (LFP) cathode slurry with a solid content of 50% is used as the detection object, and the target detection limit is iron particles with a diameter of 25μm.

[0046] like Figure 2 As shown, the present invention provides an online monitoring method for magnetic foreign matter in lithium battery paste based on high-frequency resonance spectroscopy, comprising the following steps:

[0047] Step S1: Construction of the magnetic focusing resonant sensing unit:

[0048] A section of corrosion-resistant, highly insulating polytetrafluoroethylene (PTFE) pipe was selected as the testing flow channel, namely insulating flow channel 3, with an inner diameter of 30 mm and a wall thickness of 2 mm. This PTFE pipe served as a bypass pipe, through which a portion of the slurry was drawn from the main pipe using an isodynamic sampling probe, and its flow velocity was controlled at approximately 1.5 m / s by a valve.

[0049] Two magnetic focusing resonant sensing units are installed on the outer wall of the PTFE pipe along the slurry flow direction, denoted as upstream sensing zone A and downstream sensing zone B, respectively. The geometric center distance L between the two sensing zones is set to 8 mm.

[0050] The specific structure of each magnetic focusing resonant sensing region is as follows: Figure 1 As shown, a manganese-zinc ferrite core is used, fabricated into a C-shaped core 1. The open end of the C-shaped core 1 faces the PTFE pipe, forming a focusing air gap 4 with a width of 5 mm and a depth of approximately 35 mm between the core and the outer wall of the pipe. This air gap is the high-energy-density sensing area. A coil 2 with 60-80 turns of Litz wire (0.1 mm × 50 strands) is wound on the C-shaped core 1 to form an LC resonant circuit. The high permeability of the C-shaped core 1 confines the high-frequency electromagnetic field energy to the focusing air gap 4, and the magnetic field lines 5 are densely distributed in the focusing air gap 4, significantly enhancing the effective magnetic field strength in this region.

[0051] Step S2, Obtaining the background impedance spectrum:

[0052] Connect a precision impedance analyzer (e.g., Agilent E4980A) to terminals 2 of the coils in both sensing areas. Simultaneously inject a sweep excitation signal into both sensing areas, with a frequency range of 500kHz to 3MHz, a step size of 10kHz, and an excitation voltage of 2.0V.

[0053] Figure 3 This shows the change in the resonance curve of the sensing coil when ferromagnetic particles pass through, with the horizontal axis representing the frequency of the excitation signal. (Unit: MHz), the vertical axis represents the coil impedance amplitude. (unit: The solid line represents the background resonance curve when no ferromagnetic particles pass through, at which point the sensing area is in the "reference state"; the dashed line represents the offset curve when ferromagnetic particles pass through. Since the permeability of the particles is much higher than that of the background slurry, it will significantly change the local permeability, causing the resonant frequency to shift to the left; at the same time, the conductivity of the particles introduces eddy current losses, resulting in a decrease in the quality factor.

[0054] like Figure 3 As shown, impedance spectrum data of coil 2 were collected when the pipe was filled with static LFP slurry to be tested. The initial resonant frequency of each sensing zone was extracted from the impedance spectrum. and quality factor For example, the initial resonant frequency of a certain sensing area is measured. Approximately 1.8MHz, quality factor Approximately 85. These background parameters serve as a dynamic baseline for subsequent real-time monitoring. The system can be set to automatically update this baseline periodically (e.g., hourly) to compensate for long-term, slow changes caused by slurry formulation fine-tuning or temperature drift.

[0055] Step S3: Real-time monitoring and feature extraction:

[0056] The slurry circulation system is activated, allowing the LFP slurry to flow through the insulating channel 3 at a set flow rate. The impedance analyzer continuously collects impedance data from each sensing zone at high speed and calculates the resonant frequency in real time. and quality factor .

[0057] When a standard ferromagnetic particle 6 with a diameter of 25μm flows with the slurry through the upstream sensing area This causes instantaneous changes in local permeability and eddy current losses. The impedance analyzer captures the resonant frequency in real time. The instantaneous drop, the calculated resonant frequency shift Approximately 280Hz. Meanwhile, the quality factor... A momentary decrease also occurs, and the calculated change in the quality factor is... Approximately 0.6. When the ferromagnetic particle 6 leaves the sensing area... Afterwards, the signal returned to the background level.

[0058] Subsequently, the same ferromagnetic particle 6 continued to flow with the slurry, passing through the downstream sensing area. Similarly, the sensing area The corresponding one was also detected. Pulse and pulse.

[0059] Step S4, Flow velocity measurement and waveform normalization:

[0060] Figure 4 This demonstrates the resonant frequency shift between the upstream and downstream sensing regions. Pulse timing sequence, with time as the horizontal axis. (Unit: ms), the vertical axis represents the resonant frequency offset. (Unit: Hz).

[0061] like Figure 4 As shown, the system records the upstream sensing area. The generated Peak moment of pulse signal and downstream sensing area The generated Peak moment of pulse signal Calculate the time difference Δ between the two peaks. For example, in this case, the measured... millisecond.

[0062] Based on the fixed distance between the two sensing zones Using the formula The instantaneous flow velocity of the ferromagnetic particle 6 was calculated. The velocity was consistent with the set pipeline flow rate, verifying the accuracy of the velocity measurement.

[0063] To eliminate the influence of flow velocity on signal amplitude, the calculated instantaneous flow velocity is used. right The pulse waveform is normalized. Specifically, this can be achieved by using a pre-calibrated "flow velocity-frequency shift amplitude attenuation curve" for lookup table correction, or by employing a scaling algorithm. For example, after normalization, the normalized frequency shift amplitude is obtained. This value characterizes the signal amplitude that a particle of this size should have at a standard flow rate (e.g., 1 m / s), and is independent of flow rate fluctuations.

[0064] Step S5, Inversion and Foreign Object Identification:

[0065] The normalized frequency shift amplitude value obtained in step S4 and the change in quality factor obtained in step S3 Feature pairs are formed and input into a pre-trained inversion model. This embodiment uses a Support Vector Machine (SVM) algorithm to construct the inversion model. The model's training sample library contains thousands of sets of standard particles of different sizes (20-200 μm, step size 10 μm) and materials (iron, copper, etc.) generated under the same sensing conditions. and Feature pairs and their known labels (size, material).

[0066] The output of the inversion model is as follows:

[0067] Equivalent diameter: 26.5 μm;

[0068] Material category confidence level: ferromagnetic particles 6, confidence level is 0.97.

[0069] The inversion results are in high agreement with the offline microscope measurements of the same particle (26 μm), demonstrating the accuracy of the quantitative and qualitative identification method of the present invention.

[0070] Step S6, Statistics and Output:

[0071] The system ran continuously for one hour, successfully detecting and identifying 152 iron particle events. The system automatically compiled and output a particle size distribution histogram, showing 89 instances of 25-50μm particles, 53 instances of 50-100μm particles, and 10 instances of >100μm particles; and calculated the total content of ferromagnetic foreign matter in the slurry during this period to be approximately 0.12ppm.

[0072] When the system detects a large particle event with a diameter exceeding 100μm, it immediately triggers an audible and visual alarm and can send a signal to the production line control system via the communication interface to drive the downstream electromagnetic demagnetizing valve or diversion valve to intercept and process the abnormal slurry, thereby achieving online closed-loop control of quality issues.

[0073] The above embodiments are merely preferred embodiments of the present invention. It should be noted that those skilled in the art can make several improvements and equivalent substitutions without departing from the principle of the present invention. All such improvements and equivalent substitutions to the claims of the present invention fall within the protection scope of the present invention.

Claims

1. A method for online monitoring of magnetic foreign matter in lithium battery paste based on high-frequency resonant spectroscopy, characterized in that, Includes the following steps: S1. Constructing a magnetic focusing resonant sensing system: At a preset position in the lithium battery slurry conveying channel, at least two magnetic focusing resonant sensing units are set along the slurry flow direction. Each magnetic focusing resonant sensing unit uses a high-permeability magnetic core to confine the high-frequency electromagnetic field energy in a local space, forming a high-energy-density sensing area. S2. Obtaining the background reference spectrum: Injecting a wideband high-frequency magnetic excitation signal into the magnetic focusing resonant sensing unit, and collecting the background impedance spectrum data when a non-magnetic foreign object passes through. The background impedance spectrum data includes at least the resonant frequency reference value and the quality factor reference value. In step S2, the frequency range of the wideband high-frequency excitation signal is 100kHz to 10MHz, and the excitation signal is injected through a frequency sweep method or a frequency division multiplexing method. The background impedance spectrum data is automatically updated when the system starts up or according to a preset cycle to adapt to the dynamic changes in slurry formulation, temperature, and conductivity. S3. Real-time Acquisition and Feature Extraction: When the lithium battery slurry flows through the sensing area, the impedance spectrum change data of each magnetic focusing resonant sensing unit is acquired in real time, and two feature parameters, resonant frequency shift and quality factor change, are extracted from the impedance spectrum change data. The resonant frequency shift reflects the change in permeability within the sensing area, and the quality factor change reflects the change in eddy current loss within the sensing area. In step S3, the resonant frequency shift... and quality factor change Calculated using the following formula: ； ； in, To monitor the obtained resonant frequency in real time, The resonant frequency reference value obtained in step S2; To monitor the obtained quality factor in real time, The quality factor reference value obtained in step S2; the quality factor is defined as the ratio of the resonant peak frequency to the half-power bandwidth, or it can be directly measured by an impedance analyzer. S4. Flow velocity measurement and signal normalization: Utilizing the spatial layout of multiple magnetic focusing resonant sensing units, the time difference between the same magnetic foreign object passing through different sensing units is obtained. The instantaneous flow velocity of the magnetic foreign object is calculated based on the fixed spacing between the sensing units. Then, the instantaneous flow velocity is used to normalize the resonant frequency offset signal to obtain normalized characteristic parameters related to the size of the magnetic foreign object. Specifically, step S4 involves recording the corresponding resonant frequency offset when the same magnetic foreign object passes through the upstream and downstream sensing areas of the magnetic focusing resonant sensing unit sequentially. Pulse signals, and calculate the time difference between the peak values ​​of two pulse signals. Based on the fixed distance between the upstream and downstream sensing areas and time difference Calculate the instantaneous flow velocity of magnetic foreign objects : ； S5. Foreign object identification and quantitative analysis: The normalized characteristic parameters and quality factor changes are used as inputs and substituted into a preset inversion model to decouple the equivalent size and magnetic properties of the foreign object, thereby completing the identification and quantification of magnetic foreign objects.

2. The method for online monitoring of magnetic foreign matter in lithium battery paste based on high-frequency resonance spectroscopy according to claim 1, characterized in that, In step S1, the high permeability magnetic core of the magnetic focusing resonant sensing unit is selected from C-type, E-type, or toroidal manganese-zinc ferrite cores or amorphous alloy cores. A focusing air gap is formed at the opening of the high permeability magnetic core, and the width of the focusing air gap is 0.5-5mm.

3. The method for online monitoring of magnetic foreign matter in lithium battery paste based on high-frequency resonance spectroscopy according to claim 1, characterized in that, In step S1, the lithium battery slurry conveying channel is a bypass channel led out from the main conveying pipeline, or an insulating pipe section embedded in the main conveying pipeline; in addition to having insulation properties to avoid eddy current losses, the insulating pipe section is also made of non-ferromagnetic material to avoid interference with the magnetic field, and is made of polytetrafluoroethylene, borosilicate glass or polydimethylsiloxane material.

4. The method for online monitoring of magnetic foreign matter in lithium battery paste based on high-frequency resonance spectroscopy according to claim 1, characterized in that, In step S4, the normalization process includes: based on the instantaneous flow velocity The resonant frequency offset pulse waveform is scaled proportionally, or the frequency shift amplitude is corrected according to the flow velocity using a lookup table method to obtain a normalized frequency shift amplitude that is only related to the size of the magnetic foreign object.

5. The method for online monitoring of magnetic foreign matter in lithium battery paste based on high-frequency resonance spectroscopy according to claim 1, characterized in that, In step S5, the inversion model is a classification regression model constructed based on a machine learning algorithm. The training samples of the classification regression model include normalized feature parameters and quality factor change feature pairs generated when magnetic foreign objects of different particle sizes and materials pass through the sensing unit; the materials include iron, non-ferromagnetic metal particles and bubbles.

6. The method for online monitoring of magnetic foreign matter in lithium battery paste based on high-frequency resonance spectroscopy according to claim 1, characterized in that, In step S5, the foreign object identification is achieved through the collaborative analysis of normalized feature parameters and quality factor changes: if the normalized feature parameter is greater than the first preset threshold and the quality factor change is less than the second preset threshold, it is determined to be a ferromagnetic metal particle; if the normalized feature parameter is less than the third preset threshold and the quality factor change is greater than the fourth preset threshold, it is determined to be a non-magnetic conductive impurity or bubble; if the normalized feature parameter is greater than the first preset threshold and the quality factor change is not significantly different, it is determined to be a special magnetic non-metallic impurity.

7. The method for online monitoring of magnetic foreign matter in lithium battery paste based on high-frequency resonance spectroscopy according to claim 1, characterized in that, The process also includes step S6: Statistics and Output: Accumulating foreign object identification results per unit time, statistically analyzing the foreign object size distribution and the total content of ferromagnetic foreign objects, and outputting the results; the output includes a histogram of foreign object size distribution, the total content of ferromagnetic foreign objects per unit time, and an alarm signal triggered when the detected foreign object size exceeds a preset threshold; the preset threshold is ≥25. .

Images