Lithium ion battery mixed slurry viscosity detection method based on lithium iron phosphate

Abstract

The invention discloses a lithium ion battery mixed slurry viscosity detection method based on lithium iron phosphate. The method comprises the steps of slurry pretreatment and parameter initialization, multi-modal data acquisition, dynamic viscosity compensation model construction, rheological property prediction, abnormality diagnosis and online feedback control. The temperature, the shearing rate, the solid content and the rheological curve of the slurry are acquired in real time through a multi-sensor fusion technology, a dynamic viscosity compensation model is constructed in combination with the characteristics of the lithium iron phosphate material, and a rheological prediction algorithm based on machine learning is introduced, so that high-precision and nondestructive online detection is realized. The influence of particle size distribution and temperature drift is eliminated through material characteristic modeling, and the measurement precision is further improved; and an ultrasonic wave and near infrared combined detection mode is adopted, so that the slurry structure is prevented from being damaged by a traditional sampling method, viscosity abnormity self-diagnosis and process parameter real-time feedback are realized, and the rejection rate is reduced.

Description

Technical Field

[0001] The invention belongs to the technical field of lithium ion battery manufacturing, and in particular relates to a method for detecting viscosity of lithium iron phosphate-based lithium ion battery slurry. Background Art

[0002] Lithium iron phosphate batteries are lithium-ion batteries that use lithium iron phosphate (LiFePO4) as the positive electrode material and carbon as the negative electrode material. Their single-cell rated voltage is 3.2V, and the charge cutoff voltage is 3.6V to 3.65V. During charging, some lithium ions in the lithium iron phosphate are released, transferred through the electrolyte to the negative electrode, and embedded in the negative electrode's carbon material. Simultaneously, electrons are released from the positive electrode and travel through the external circuit to the negative electrode, maintaining the chemical reaction balance. During discharge, lithium ions are released from the negative electrode and travel through the electrolyte to the positive electrode. Simultaneously, electrons are released from the negative electrode and travel through the external circuit to the positive electrode, providing energy. Measuring the viscosity of lithium-ion battery slurry is a critical step in ensuring battery performance. Traditionally, viscometers have been used, including rotational and vibrational methods, suitable for measuring low-viscosity and high-viscosity slurries, respectively. Rheometers are also commonly used to test the viscosity of battery slurry, primarily to ensure slurry fluidity and stability, thereby optimizing production processes and product quality. Traditional rheometer testing has the following drawbacks: 1) Single-point measurement cannot reflect the dynamic changes in the slurry; 2) It fails to account for the adsorption effect caused by the high specific surface area of ​​lithium iron phosphate materials; and 3) It ignores viscosity drift caused by solvent evaporation during the slurry mixing process. While online testing solutions have been continuously improved with technological advancements, the problem of measurement distortion caused by particle agglomeration has not yet been resolved. Summary of the Invention

[0003] (1) Technical problems solved In response to the shortcomings of the existing technology, the present invention eliminates the influence of particle size distribution and temperature drift through rheological principles, multi-parameter dynamic compensation and material property modeling, and further improves measurement accuracy.

[0004] (2) Technical solution To achieve the above object, the present invention provides the following technical solution: a method for detecting viscosity of lithium iron phosphate-based lithium ion battery slurry, comprising the following steps: Step 1: Slurry pretreatment and parameter initialization Place the lithium iron phosphate slurry in a constant temperature sealed container, oscillate at a preset frequency to eliminate sedimentation and stratification, and simultaneously set the initial detection parameters, including temperature. , shear rate range , solid content threshold ; Step 2: Multimodal data collection The rheometer’s built-in torque sensor, temperature sensor, and ultrasonic probe collect the slurry’s apparent viscosity η, shear stress τ, temperature T, and particle distribution uniformity parameter U in real time. Step 3: Dynamic viscosity compensation model construction Based on the specific surface area SSA, particle size distribution D50 and conductive agent content C of lithium iron phosphate material, the viscosity compensation formula is established: ; in, 、 、 is the material coefficient calibrated by orthogonal experiment, is the actual particle size median, indicating the actual D50 value of the lithium iron phosphate particles in the current slurry being tested (obtained by a laser particle size analyzer); The median reference particle size represents the standard D50 value (e.g., 2.0 μm) set when calibrating the model, serving as a benchmark comparison value for the effect of particle size. Step 4: Rheological property prediction and abnormality diagnosis The compensated viscosity data is input into the pre-trained LSTM neural network model to output the thixotropic loop curve, yield stress prediction value and process defect probability of the slurry; Step 5: Online Feedback Control If the viscosity deviation is detected to be more than ±5% or the thixotropic ring area is abnormal, the stirring speed, vacuum degree or solvent addition amount of the slurry mixer will be automatically adjusted.

[0005] As a preferred solution, the oscillation frequency in the pretreatment of step 1 is 10-50 Hz, including: a low-frequency oscillation mechanism (10-20 Hz) to eliminate the sedimentation of particles >10 μm, lasting 60-90 seconds; a medium-frequency oscillation mechanism (20-40 Hz) to disperse soft agglomerates, lasting 30-60 seconds; a high-frequency pulse oscillation mechanism (40-50 Hz, duty cycle 30%) to refine the microstructure, lasting 20-30 seconds; after standing for 30 seconds, the thixotropic recovery rate of the slurry is tested. If the recovery rate is <5 mPa·s / min, the pretreatment is determined to be qualified, and the oscillation time t satisfies: ; Where ρ is the density, μ is the solvent viscosity, ρLFP is the true density of lithium iron phosphate, which refers to the density of the solid lithium iron phosphate material itself (typical value: ≈3.6 g / cm³), and is independent of the slurry concentration; ρsolvent is the density of the solvent, which refers to the density of the liquid phase solvent (such as NMP solvent ≈1.03 g / cm³), excluding solid particles.

[0006] As a preferred solution, the multimodal data collection in step 2 is based on a detection system, and the detection system includes: Intelligent rheological detection unit: a multi-stage temperature-controlled rheometer equipped with temperature adaptive adjustment function, which has a built-in laser particle size analyzer and an annularly distributed temperature sensor group, which can simultaneously measure slurry viscosity and particle distribution parameters; Dynamic compensation processor: An industrial control module equipped with a viscosity compensation algorithm calculates the impact of temperature and solid content on viscosity in real time and interacts with the LSTM prediction model for verification; Process feedback execution module: Connects to the enterprise MES system through the industrial Internet of Things gateway, dynamically adjusts the stirring speed and vacuum parameters according to the test results, and triggers the automatic liquid replenishment program and process adjustment instructions when the error exceeds the limit; Interactive display terminal: equipped with a digital process dashboard, which displays the dynamic curve of the thixotropic ring, process defect warning prompts and historical data comparison interface in real time; The rotor surface of the multi-stage temperature-controlled rheometer is coated with a polytetrafluoroethylene coating with a roughness Ra≤0.1 μm.

[0007] As a preferred solution, the detection system also includes a pulsed ultrasonic scanning module, which emits 5-10 MHz high-frequency pulse waves at intervals of 0.5-2 seconds to collect acoustic impedance change data of lithium iron phosphate particles in the slurry and calculate the particle dispersion index PDI value. When the PDI value exceeds 0.25, the medium-frequency oscillation mechanism is triggered.

[0008] As a preferred solution, the multi-stage temperature-controlled rheometer includes: The main detection chamber has a temperature control accuracy of ±0.3°C and is equipped with a coaxial cylindrical rotor; Pre-conditioning chamber with built-in ultrasonic dispersion module; Buffer transition chamber, equipped with a miniature near-infrared probe; The three chambers are connected by pneumatic valves to realize online circulation detection of slurry.

[0009] As a preferred solution, the conductive agent content C in step 3 is at least one of carbon black, CNT and graphene, and its content range is 0.5%-3.0%, and The value decreases exponentially with the increase of C.

[0010] As a preferred solution, the construction of the LSTM neural network model in step 4 includes: a) Input layer: contains 15-dimensional feature vectors, namely [temperature T, shear rate γ, apparent viscosity η, D10 / D50 / D90, thixotropic ring area A, yield stress σ_y, PVDF content, conductive agent type code, solid content S, solvent evaporation rate V, stirring power P, vacuum degree Z, time derivative dη / dt]; b) Hidden layer: 3-layer bidirectional LSTM structure, 128 neurons per layer, using Swish activation function; c) Output layer: Predict the viscosity change rate Δη, thixotropic loop offset ΔA, and process defect level (0 - 5 levels) within the next 30 s; Its model training adopts the transfer learning method. The pre-training dataset contains 2000 groups of slurry data of different cathode materials (NCM, LCO, LMO), and the fine-tuning dataset is 500 groups of special data for lithium iron phosphate.

[0011] As a preferred solution, step four further includes the evaluation of slurry timeliness: After the detection is completed, the slurry is left standing, and the viscosity value η(t) is collected every 10 minutes, and the stability index is calculated: ; where η0 is the initial viscosity at standing, the viscosity of the slurry at the start of standing (t = 0), that is, the reference value for timeliness evaluation; η60 is the viscosity after standing for 60 minutes, that is, the apparent viscosity measured when the slurry stands for 60 minutes; When SI > 15%, it is determined that the dispersion stability of the slurry is unqualified; When 5% < SI ≤ 15%, shorten the storage time to within 4 hours; When SI ≤ 5%, it is determined as high-quality slurry.

[0012] As a preferred solution, the online feedback control in step five further includes: Multi-threshold grading response mechanism: When the viscosity deviation is ±5% - 8%: Trigger a yellow warning and automatically record the fluctuation data; when the viscosity deviation is ±8% - 12%: Activate the online calibration program and reduce the stirring speed by 10%; when the viscosity deviation > ±12%: Emergency shutdown and generate a fault diagnosis report.

[0013] As a preferred solution, the detection method further includes setting a periodic self-check program. After every 10 detections, the standard sample verification process is automatically executed. The standard sample is a reference slurry with a viscosity value of 2500 ± 50 mPa·s. When the deviation of the standard sample detection exceeds ±1.5% for three consecutive times, trigger the detection system calibration alarm and lock the detection function until the sensor zero calibration is completed.

[0014] (III) Beneficial effects Compared with the prior art, the present invention provides a method for detecting the viscosity of the slurry for lithium-ion battery mixing based on lithium iron phosphate, having the following beneficial effects: The method of the present invention includes slurry pretreatment and parameter initialization, multimodal data acquisition, dynamic viscosity compensation model construction, rheological property prediction and abnormal diagnosis and online feedback control. It is intended to use multi-sensor fusion technology to collect the temperature, shear rate, solid content and rheological curve of the slurry in real time, build a dynamic viscosity compensation model in combination with the characteristics of lithium iron phosphate materials, and introduce a rheological prediction algorithm based on machine learning to achieve high-precision, non-destructive online detection. The present invention eliminates the influence of particle size distribution and temperature drift through material property modeling, further improving the measurement accuracy; it also avoids the traditional sampling method from destroying the slurry structure by adopting ultrasonic + near-infrared combined detection, and realizes viscosity abnormality self-diagnosis and real-time feedback of process parameters, reducing the scrap rate. The detection method simulates the actual working conditions by dynamically adjusting the detection parameters, solving the measurement error problem caused by the thixotropy of the slurry and particle sedimentation in the traditional method, and optimizing the feedback control of the slurry process parameters. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 Schematic diagram of the detection method of the present invention; Figure 2 It is a schematic diagram of the module structure and function of the detection system of the present invention. DETAILED DESCRIPTION

[0016] In order to better understand the purpose, structure and function of the present invention, the viscosity detection method of lithium iron phosphate-based lithium ion battery slurry will be further described below in conjunction with the accompanying drawings and specific embodiments.

[0017] Example 1

[0018] refer to Figure 1-2 The present invention provides a method for detecting viscosity of lithium iron phosphate-based lithium ion battery slurry, the method comprising the following steps: Step 1: Slurry pretreatment and parameter initialization Place the lithium iron phosphate slurry in a constant temperature sealed container, oscillate at a preset frequency to eliminate sedimentation and stratification, and simultaneously set the initial detection parameters, including temperature. , shear rate range , solid content threshold ; Step 2: Multimodal data collection The rheometer’s built-in torque sensor, temperature sensor, and ultrasonic probe collect the slurry’s apparent viscosity η, shear stress τ, temperature T, and particle distribution uniformity parameter U in real time. Step 3: Dynamic viscosity compensation model construction Based on the specific surface area SSA, particle size distribution D50 and conductive agent content C of lithium iron phosphate material, the viscosity compensation formula is established: ; in, 、 、 is the material coefficient calibrated by orthogonal experiment, is the actual particle size median, indicating the actual D50 value of the lithium iron phosphate particles in the current slurry being tested (obtained by a laser particle size analyzer); The median reference particle size represents the standard D50 value (e.g., 2.0 μm) set when calibrating the model, serving as a benchmark comparison value for the effect of particle size. Step 4: Rheological property prediction and abnormality diagnosis The compensated viscosity data is input into the pre-trained LSTM neural network model to output the thixotropic loop curve, yield stress prediction value and process defect probability of the slurry; Step 5: Online Feedback Control If the viscosity deviation is detected to be more than ±5% or the thixotropic ring area is abnormal, the stirring speed, vacuum degree or solvent addition amount of the slurry mixer will be automatically adjusted.

[0019] The detection method of the present invention also includes setting up a periodic self-test program, which automatically executes the standard sample verification process after every 10 tests. The standard sample is a reference slurry with a viscosity value of 2500±50mPa·s. If the deviation of the standard sample test exceeds ±1.5% for three consecutive times, the detection system calibration alarm is triggered and the detection function is locked until the sensor zero point calibration is completed.

[0020] Specifically, the specific process of slurry pretreatment and parameter initialization of the present invention is as follows: first, the lithium iron phosphate slurry (LFP solid content 48%-52%) after slurrying is transferred to a double-layer jacketed constant temperature container, and the slurry temperature is stabilized in the range of 25±1°C by a circulating water bath system (temperature control accuracy ±0.5°C) for 10-15 minutes to eliminate the thermal history effect; the oscillation frequency in the pretreatment is 10-50Hz, including: a low-frequency oscillation mechanism (10-20Hz) to eliminate the sedimentation of particles >10μm, which lasts for 60-90 seconds; a medium-frequency oscillation mechanism (20-40Hz) to disperse soft agglomerates, which lasts for 30-60 seconds; a high-frequency pulse oscillation mechanism (40-50Hz, duty cycle 30%) to refine the microstructure, which lasts for 20-30 seconds; after standing for 30 seconds, the slurry thixotropic recovery rate is detected. If the recovery rate is <5mPa·s / min, the pretreatment is determined to be qualified, and the oscillation time t satisfies: ; where ρ is the density and μ is the solvent viscosity.

[0021] This step matches the detection range according to the characteristics of the LFP material, and breaks the binding force of agglomerates of different sizes through frequency-amplitude coupling, thereby eliminating the problem of uneven viscosity distribution caused by local temperature rise during the slurry mixing process.

[0022] Specifically, the multimodal data acquisition method of the present invention uses a rheometer equipped with a coaxial cylindrical rotor and an ultrasonic detection module. By collecting backscattered signal intensity, the uniformity parameter U is calculated as: standard deviation / mean × 100% (threshold U ≤ 15%). Temperature, viscosity, and particle distribution data are then correlated using a three-dimensional grid to generate a slurry state cloud map.

[0023] Specifically, the dynamic viscosity compensation model of the present invention is constructed. After each batch of testing, the actual coating quality data (such as the thickness uniformity of the electrode) is fed back to the model, and the rolling optimization coefficient is calculated. The conductive agent content C is at least one of carbon black, CNT and graphene, and its content range is 0.5%-3.0%, and The value decays exponentially with the increase of C. Further, the construction of its dynamic viscosity compensation model also includes: Conductive agent synergistic compensation: Based on the compounding ratio of carbon black, CNT and graphene, a conductive network influencing factor database is established. When the total content of the conductive agent is detected to exceed 2.0%, exponential viscosity attenuation compensation is automatically enabled. Correction for solvation effect: Based on the specific surface area of ​​lithium iron phosphate and the slurry mixing time, the thickness of the solvent coating layer is calculated. For every 5m² / g increase in specific surface area, the compensation coefficient is reduced by 0.03, and for every 10 minutes of extended slurry mixing time, the compensation coefficient is increased by 0.015.

[0024] Specifically, in the rheological property prediction and abnormality diagnosis of the present invention, the construction of the LSTM neural network model includes: a) Input layer: Contains 15-dimensional feature vectors, namely [temperature T, shear rate γ, apparent viscosity η, D10 / D50 / D90, thixotropic ring area A, yield stress σ_y, PVDF content, conductive agent type code, solid content S, solvent evaporation rate V, stirring power P, vacuum degree Z, time derivative dη / dt]; b) Hidden layer: 3-layer bidirectional LSTM structure, 128 neurons per layer, using Swish activation function; c) Output layer: predicts the viscosity change rate Δη, thixotropic ring offset ΔA and process defect level (0-5) within the next 30 seconds; The model training adopts the transfer learning method. The pre-training data set contains 2000 sets of different cathode material (NCM, LCO, LMO) slurry data, and the fine-tuning data set is 500 sets of lithium iron phosphate special data. Furthermore, the slurry aging evaluation is also included: after the test is completed, the slurry is allowed to stand, and the viscosity value η(t) is collected every 10 minutes to calculate the stability index: ; Among them, η0 is the initial viscosity at rest, which is the viscosity of the slurry at the start of standing (t = 0), i.e., the reference value for timeliness evaluation; η60 is the viscosity after standing for 60 minutes, i.e., the apparent viscosity measured when the slurry stands for 60 minutes. When SI > 15%, it is determined that the dispersion stability of the slurry is unqualified. When 5% < SI ≤ 15%, shorten the storage time to within 4 hours. When SI ≤ 5%, it is determined as high-quality slurry. [[ID=,8]]

[0025] The online feedback control also includes: Multi-threshold hierarchical response mechanism: When the viscosity deviation is ±5% - 8%: Trigger a yellow warning and automatically record the fluctuation data; when the viscosity deviation is ±8% - 12%: Activate the online calibration program and reduce the stirring speed by 10%; when the viscosity deviation > ±12%: Emergency stop and generate a fault diagnosis report. Specifically, it includes:联动优化: When the detected viscosity η > η_target: Coating process: Increase the substrate preheating temperature (ΔT = 5 - 10°C); Rolling process: Reduce the pressure (ΔP = 10 - 20 kN / m); When the thixotropic index TI < 0.9: Coating process: Reduce the tape running speed (Δv = 0.2 - 0.5 m / min); Drying process: Increase the infrared radiation power (ΔW = 5 - 10%).

[0026] The online feedback control of the present invention adopts multi-parameter linkage control, and its control strategy is shown in Table 1 below: Table 1 Deviation range Control strategy ±5%-8% Yellow warning, record data and fine-tune stirring speed ±5% ±8%-12% Activate online calibration, refill at 0.1 mL / s and reduce speed by 10% >±12% Emergency stop, start ultrasonic dispersion module (300W / 30s)

[0027] Example 2

[0028] The multi-modal data acquisition of the present invention is based on a detection system, and the detection system includes: Intelligent rheological detection unit: A multi-stage temperature-controlled rheometer equipped with a temperature self-adaptive adjustment function, which is内置 with a laser particle size analyzer and a ring-shaped distributed temperature sensor group, and can synchronously measure the slurry viscosity and particle distribution parameters; Dynamic compensation processor: An industrial control module搭载 with a viscosity compensation algorithm, which calculates the influence of temperature and solid content on viscosity in real time and interacts with the LSTM prediction model for verification; Process feedback execution module: Connects to the enterprise MES system through an industrial Internet of Things gateway, dynamically adjusts the stirring speed and vacuum degree parameters according to the detection results, and triggers an automatic liquid replenishment program and a process adjustment instruction when the error exceeds the limit; Interactive display terminal: equipped with a digital process dashboard, which displays the dynamic curve of the thixotropic ring, process defect warning prompts and historical data comparison interface in real time; Specifically, the rotor surface of the multi-stage temperature-controlled rheometer of the present invention is coated with a polytetrafluoroethylene coating with a roughness Ra≤0.1μm, which includes: The main detection chamber has a temperature control accuracy of ±0.3°C and is equipped with a coaxial cylindrical rotor; Pre-conditioning chamber with built-in ultrasonic dispersion module for online dispersion of agglomerates; The buffer transition chamber is equipped with a micro near-infrared probe to monitor the solvent distribution in real time; The three chambers are connected by pneumatic valves, which can realize online circulation detection of slurry. The slurry circulation path is controlled by pneumatic valves to realize the integrated process of "pretreatment-detection-calibration".

[0029] In order to better understand the present invention, the following is an experimental process using a comparison experiment of detection accuracy under multiple working conditions as an example: Test conditions: Slurry formula: LFP (D50 = 2.0 μm), PVDF 3%, conductive agent (carbon black / CNT compound 1.5%); Variable control: temperature (20 / 25 / 30℃), solid content (45 / 50 / 55%), solvent volatility (0 / 3 / 5%).

[0030] The experimental results are shown in Table 2 below: Table 2 Working condition number Temperature (℃) Solid content (%) Solvent loss (%) Error of traditional method (%) Error of the present invention (%) System response action G1 20 45 0 +7.2 +0.5 none G2 25 50 3 +9.8 +1.1 Fluid infusion 0.1 mL / s G3 30 55 5 +15.3 +2.7 Stop + ultrasonic dispersion G4 22 48 5 +8.1 +0.9 Fine-tune stirring speed As can be seen from Table 2, the present invention can still control the error within ±3% under high temperature (30°C) and high solid content (55%) working conditions, which is 82% higher than the traditional method; the system response time is less than 5 seconds, meeting the real-time control requirements.

[0031] The method and system of the present invention form a closed data flow loop, including preprocessing-multimodal data acquisition-dynamic compensation-AI prediction-feedback control, wherein preprocessing provides homogenized slurry for data acquisition, multimodal data supports the construction of the compensation model, AI prediction results drive the feedback control strategy, and the control results feed back to the model optimization (such as the compensation coefficient for the correction of the fluid replenishment volume).

[0032] Specifically, this invention integrates multi-stage oscillation preprocessing, multimodal data fusion, and the collaborative design of a dynamic compensation model and LSTM predictive control to construct an intelligent viscosity measurement system that covers the entire "detection-compensation-prediction-control" process. This deep coupling of the system hardware (multi-cavity rheometer, sensor array) and the algorithm (compensation model, AI prediction) further improves measurement accuracy while significantly reducing measurement time.

[0033] It will be understood that the present invention is described by way of some embodiments, and it will be appreciated by those skilled in the art that various changes or equivalent substitutions may be made to these features and embodiments without departing from the spirit and scope of the present invention. In addition, under the teachings of the present invention, these features and embodiments may be modified to adapt to specific circumstances and materials without departing from the spirit and scope of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed herein, and all embodiments falling within the scope of the claims of this application are intended to be protected by the present invention.

Claims

1. A method for detecting viscosity of lithium-ion battery slurry based on lithium iron phosphate, characterized in that: The steps include: Step 1: Slurry pretreatment and parameter initialization Place the lithium iron phosphate slurry in a constant temperature sealed container, oscillate at a preset frequency to eliminate sedimentation and stratification, and simultaneously set the initial detection parameters, including temperature. , shear rate range , solid content threshold ; Step 2: Multimodal data collection The rheometer’s built-in torque sensor, temperature sensor, and ultrasonic probe collect the slurry’s apparent viscosity η, shear stress τ, temperature T, and particle distribution uniformity parameter U in real time. Step 3: Dynamic viscosity compensation model construction Based on the specific surface area SSA, particle size distribution D50 and conductive agent content C of lithium iron phosphate material, the viscosity compensation formula is established: ; in, 、 、 is the material coefficient calibrated by orthogonal experiment, is the actual particle size median, indicating the actual D50 value of the lithium iron phosphate particles in the current slurry being tested; is the median reference particle size, which represents the standard D50 value set when calibrating the model and serves as a benchmark comparison value for the effect of particle size; Step 4: Rheological property prediction and abnormality diagnosis The compensated viscosity data is input into the pre-trained LSTM neural network model to output the thixotropic loop curve, yield stress prediction value and process defect probability of the slurry; Step 5: Online Feedback Control If the viscosity deviation is detected to be more than ±5% or the thixotropic ring area is abnormal, the stirring speed, vacuum degree or solvent addition amount of the slurry mixer will be automatically adjusted.

2. The method for detecting viscosity of lithium iron phosphate-based lithium ion battery slurry according to claim 1, characterized in that: The oscillation frequency in the pretreatment of step 1 is 10-50 Hz, including: a low-frequency oscillation mechanism to eliminate the sedimentation of particles larger than 10 μm, lasting for 60-90 seconds; a medium-frequency oscillation mechanism to disperse soft agglomerates, lasting for 30-60 seconds; a high-frequency pulse oscillation mechanism to form the microstructure, lasting for 20-30 seconds; after standing for 30 seconds, the slurry thixotropic recovery rate is tested. If the recovery rate is less than 5 mPa·s / min, the pretreatment is determined to be qualified, and the oscillation time t meets the following requirements: ； Among them, ρ is density, μ is solvent viscosity, ρLFP is the true density of lithium iron phosphate, which refers to the density of solid lithium iron phosphate material itself and has nothing to do with slurry concentration; ρsolvent is the density of solvent, which refers to the density of liquid phase solvent without solid particles.

3. The method for detecting viscosity of lithium iron phosphate-based lithium ion battery slurry according to claim 1, characterized in that: The multimodal data collection in step 2 is based on a detection system, and the detection system includes: Intelligent rheological detection unit: a multi-stage temperature-controlled rheometer equipped with temperature adaptive adjustment function, which has a built-in laser particle size analyzer and an annularly distributed temperature sensor group, which can simultaneously measure slurry viscosity and particle distribution parameters; Dynamic compensation processor: An industrial control module equipped with a viscosity compensation algorithm calculates the impact of temperature and solid content on viscosity in real time and interacts with the LSTM prediction model for verification; Process feedback execution module: Connects to the enterprise MES system through the industrial Internet of Things gateway, dynamically adjusts the stirring speed and vacuum parameters according to the test results, and triggers the automatic liquid replenishment program and process adjustment instructions when the error exceeds the limit; Interactive display terminal: equipped with a digital process dashboard, which displays the dynamic curve of the thixotropic ring, process defect warning prompts and historical data comparison interface in real time; The rotor surface of the multi-stage temperature-controlled rheometer is coated with a polytetrafluoroethylene coating with a roughness Ra≤0.1 μm.

4. The method for detecting viscosity of lithium iron phosphate-based lithium ion battery slurry according to claim 3, characterized in that The detection system also includes a pulsed ultrasonic scanning module, which emits 5-10MHz high-frequency pulse waves at intervals of 0.5-2 seconds to collect acoustic impedance change data of lithium iron phosphate particles in the slurry and calculate the particle dispersion index PDI value. When the PDI value exceeds 0.25, the medium-frequency oscillation mechanism is triggered.

5. The method for detecting viscosity of lithium iron phosphate-based lithium ion battery slurry according to claim 3, characterized in that , the multi-stage temperature-controlled rheometer includes: The main detection chamber has a temperature control accuracy of ±0.3°C and is equipped with a coaxial cylindrical rotor; Pre-conditioning chamber with built-in ultrasonic dispersion module; Buffer transition chamber, equipped with a miniature near-infrared probe; The three chambers are connected by pneumatic valves to realize online circulation detection of slurry.

6. The method for detecting viscosity of lithium iron phosphate-based lithium ion battery slurry according to claim 1, characterized in that In step 3, the conductive agent content C is at least one of carbon black, CNT and graphene, and its content range is 0.5%-3.0%, and The value decreases exponentially with the increase of C.

7. The method for detecting viscosity of lithium iron phosphate-based lithium ion battery slurry according to claim 1, characterized in that ,The construction of the LSTM neural network model in step 4 includes: a) Input layer: It contains a 15-dimensional feature vector, namely [temperature T, shear rate γ, apparent viscosity η, D10 / D50 / D90, thixotropic loop area A, yield stress σ_y, PVDF content, conductive agent type code, solid content S, solvent evaporation rate V, stirring power P, vacuum degree Z, time derivative dη / dt]; b) Hidden layer: A 3-layer bidirectional LSTM structure with 128 neurons in each layer, and the Swish activation function is adopted; c) Output layer: Predict the viscosity change rate Δη, thixotropic loop offset ΔA and process defect level within the next 30 s; The model training adopts the transfer learning method. The pre-training dataset contains 2000 groups of data of different cathode material slurries, and the fine-tuning dataset is 500 groups of special data for lithium iron phosphate.

8. The method for detecting viscosity of lithium iron phosphate-based lithium ion battery slurry according to claim 7, characterized in that , the fourth step further includes the evaluation of slurry timeliness: After the detection is completed, the slurry is left standing, and the viscosity value η(t) is collected every 10 minutes, and the stability index is calculated: ； where η0 is the initial viscosity at the start of standing, the viscosity of the slurry at the start of standing, that is, the reference value for the timeliness evaluation; η60 is the viscosity after standing for 60 minutes, that is, the apparent viscosity measured when the slurry stands for 60 minutes; When SI > 15%, it is determined that the dispersion stability of the slurry is unqualified; 9. The method for detecting viscosity of lithium iron phosphate-based lithium ion battery slurry according to claim 1, characterized in that: When 5% < SI ≤ 15%, shorten the storage time to within 4 hours; When SI ≤ 5%, it is determined as high-quality slurry.

10. The method for detecting viscosity of lithium iron phosphate-based lithium ion battery slurry according to claim 3, characterized in that: The online feedback control in the fifth step further includes: Multi-threshold hierarchical response mechanism: When the viscosity deviation is ±5% - 8%: Trigger a yellow warning and automatically record the fluctuation data; When the viscosity deviation is ±8% - 12%: Activate the online calibration program and reduce the stirring speed by 10%; When the viscosity deviation > ±12%: Emergency stop and generate a fault diagnosis report. The detection method further includes setting a periodic self-check program. After every 10 detections, the standard sample verification process is automatically executed. The standard sample is a reference slurry with a viscosity value of 2500 ± 50 mPa·s. When the deviation of the standard sample detection exceeds ±1.5% for three consecutive times, trigger the detection system calibration alarm and lock the detection function until the zero calibration of the sensor is completed.

Images