Waste lithium battery crushing and sorting treatment method

Abstract

The invention discloses a waste lithium battery crushing and sorting treatment method which comprises the following steps: S1, automatically classifying waste lithium batteries, and removing battery shells by utilizing a non-destructive laser cutting technology; s2, electrolyte residues are removed through a low-temperature plasma processing device, and cleaning parameters are dynamically adjusted; s3, feeding the battery unit without electrolyte residue into crushing equipment, and optimizing parameters of the crushing equipment by using an octopus algorithm; s4, the crushed fragments are graded according to the particle size through a micro-vibration screening device; s5, classifying and identifying the graded fragments by using a graph convolutional network; s6, inputting the metal fragments and the non-metal fragments into a sorting module, and dynamically optimizing magnetic separation and winnowing parameters by adopting an octopus algorithm; and S7, metal resources enter a hydrometallurgy process, and non-metal resources are physically separated and then sent to a composite material recovery process. According to the method, the low-temperature plasma, the octopus algorithm, the multi-modal data fusion technology and the like are combined, and intelligent crushing and sorting treatment of the waste lithium batteries is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of waste lithium battery treatment, and in particular to a method for crushing and sorting waste lithium batteries. Background Art

[0002] With the widespread application of lithium batteries in consumer electronics, electric vehicles and energy storage, the global production and consumption of lithium batteries continue to rise. However, a large number of lithium batteries are discarded after their service life ends, becoming a special type of solid waste. Waste lithium batteries contain high-value metal materials such as lithium, cobalt, and nickel, as well as other components such as aluminum foil, copper foil, and plastic diaphragms. They also contain chemical substances such as electrolytes. If they are not properly handled, they will not only cause waste of resources, but may also pose a serious threat to the environment and human health. Therefore, the recycling and treatment of waste lithium batteries has become an important issue in the field of global resource recycling.

[0003] Existing waste lithium battery processing technologies are mainly concentrated in the two core links of crushing and sorting, usually using physical treatment, hydrometallurgy or pyrometallurgy and other processes. Physical treatment technologies include mechanical crushing and screening, which aim to deconstruct batteries into fragments of different particle sizes by mechanical means, and then separate metal and non-metal components by screening, magnetic separation or air separation equipment. However, traditional physical crushing processes are difficult to accurately adapt to different types of waste lithium batteries, which can easily cause excessive crushing or loss of valuable metal materials. The hydrometallurgical process extracts metals such as lithium, cobalt, and nickel from battery fragments through acid and alkali leaching, but its processing process requires a large amount of chemical reagents, which is prone to secondary pollution. At the same time, it has high requirements for operating conditions. If the parameters are not set properly, the metal recovery rate will be significantly reduced. In addition, pyrometallurgical technology directly treats waste lithium batteries at high temperatures to recover metals, but high energy consumption and emission problems limit its large-scale application.

[0004] In the crushing and sorting links, the key problem of existing technologies is the lack of accuracy and automation in the processing process. Existing mechanical crushing equipment usually adopts a fixed parameter mode and lacks the ability to dynamically adapt to the characteristics of battery materials (such as hardness and toughness), resulting in uneven distribution of fragment particle size, affecting the subsequent sorting effect. At the same time, in the sorting stage, traditional magnetic separation and air separation equipment mostly use fixed parameters or manual adjustment modes, which are difficult to cope with complex fragment combinations and dynamically changing processing requirements. In addition, the identification of different components in the sorting process usually relies on a single characteristic (such as conductivity or density), which is prone to misclassification when processing complex mixed materials, reducing resource recovery rates.

[0005] Therefore, how to provide a method for crushing and sorting waste lithium batteries is an urgent problem that technicians in this field need to solve. Summary of the invention

[0006] One purpose of the present invention is to propose a method for crushing and sorting waste lithium batteries. The present invention combines low-temperature plasma cleaning technology, octopus algorithm optimization and multimodal data fusion technology to propose an intelligent method for crushing and sorting waste lithium batteries. Through non-destructive laser cutting, dynamic parameter optimization and graph convolutional network classification model, the whole process from battery classification, cleaning, crushing to resource recovery is intelligently processed. This method not only improves the sorting accuracy of metal and non-metal components, but also significantly improves the resource recovery efficiency, while reducing energy consumption and environmental pollution. It has the significant advantages of strong environmental protection, high resource utilization and high degree of automation, and provides an efficient solution for the green recovery of waste lithium batteries.

[0007] A method for crushing and sorting waste lithium batteries according to an embodiment of the present invention comprises the following steps:

[0008] S1. Automatically sort waste lithium batteries, distinguish battery types and remove battery shells using non-destructive laser cutting technology;

[0009] S2. Removing electrolyte residues through a low-temperature plasma treatment device and dynamically adjusting cleaning parameters to obtain a battery cell without electrolyte residues;

[0010] S3, sending the battery cells without electrolyte residue to the crushing equipment, optimizing the parameters of the crushing equipment by using the octopus algorithm, dynamically adjusting the crushing mode according to the material characteristics, and obtaining crushed fragments with different particle sizes;

[0011] S4, classifying the crushed fragments according to particle size through a micro-seismic screening device to obtain classified fragments;

[0012] S5, collect spectral data, thermal imaging data and electromagnetic property data, and use graph convolutional networks to classify and identify graded fragments, including metal fragments and non-metal fragments;

[0013] S6. Input the metal fragments and non-metal fragments into the sorting module, and use the octopus algorithm to dynamically optimize the magnetic separation and wind separation parameters to achieve real-time optimization of the fragment path and obtain the sorted metal resources and non-metal resources;

[0014] S7. The metal resources are sent into the hydrometallurgical process, and the non-metallic resources are physically separated and sent into the composite material recycling process.

[0015] Optionally, the S2 specifically includes:

[0016] S21, placing the sorted waste lithium batteries into a low-temperature plasma treatment device, starting a plasma source, setting an initial power and a gas flow rate, and using nitrogen as the working gas;

[0017] S22, real-time monitoring of the residual electrolyte concentration on the battery surface, and building an optimization model for cleaning efficiency and energy consumption:

[0018]

[0019] Wherein, η(t) represents the cleaning efficiency at time t, C(t) represents the residual concentration of electrolyte at time t, C(0) represents the initial residual concentration of electrolyte, γ represents the efficiency dynamic adjustment factor, E represents the total energy consumption during the treatment process, T represents the total cleaning time, α represents the weight coefficient of power consumption, P(t) represents the low-temperature plasma treatment power at time t, β represents the weight coefficient of gas flow rate consumption, F(t) represents the gas flow rate at time t, δ represents the weight coefficient of gas flow resistance consumption, and R(t) represents the resistance loss caused by gas flow at time t;

[0020] S23, dynamically adjusting the processing power and gas flow rate based on multi-objective optimization, with the optimization objectives being to maximize cleaning efficiency and minimize energy consumption;

[0021] S24. Real-time monitoring of the maximum temperature of the battery cell surface by thermal imaging sensors:

[0022] T max (t)≤T lim ;

[0023] Among them, T max (t) represents the maximum temperature of the battery cell surface at time t during the cleaning process, T lim Indicates the critical temperature range set according to the characteristics of battery materials, ranging from 100°C to 200°C;

[0024] S25. Continuously collect residual concentration data during the cleaning process. When the cleaning efficiency η(t)≥η threshold and energy consumption limit E≤E budget If the cleaning condition is met, the cleaning is judged to be completed, otherwise the cleaning parameters are readjusted, where η threshold represents the cleaning efficiency threshold, E budget represents the energy consumption budget;

[0025] S26. Finally, a battery cell with no electrolyte residue is obtained.

[0026] Optionally, the S3 specifically includes:

[0027] S31, sending the battery cells without electrolyte residue into the crushing equipment, starting the crushing equipment and setting the initial crushing parameters, including the blade speed and crushing force, and collecting the initial physical properties of the battery material, including strength, toughness and deformation coefficient, through real-time sensors;

[0028] S32. Introduce the octopus algorithm to simulate the dynamic exploration mechanism of multiple tentacles, take the blade speed, crushing force and crushing mode as multi-dimensional optimization variables, and simulate a parameter combination path for each tentacle:

[0029]

[0030] Wherein, O represents the crushing optimization target value, κ represents the deformation sensitivity coefficient, Δd(t) represents the particle size change rate, w1 and w2 represent the optimization weights, μ represents the power consumption weight coefficient, ν represents the force consumption weight coefficient, N(t) represents the blade speed at time t, and F(t) represents the crushing force at time t;

[0031] S33, through the multi-tentacle distributed search strategy of the octopus algorithm, each tentacle dynamically explores the optimal path in the parameter space, and real-time feedback is given on the impact of parameter combinations on the target value of fragmentation optimization. Each tentacle path is adaptively adjusted through the local update formula after each iteration:

[0032]

[0033] Among them, θ t+1 represents the updated tentacle state, θ t represents the current tentacle state, η represents the learning rate, represents the gradient of the fragmentation optimization objective value, ξ represents the exploration expansion coefficient, and rand represents the random disturbance term;

[0034] S34. In each round of iteration, all tentacles are ranked according to the real-time feedback data of particle size change, energy consumption and crushing efficiency, the optimal tentacle path is retained, and the search range of the remaining tentacles is adjusted;

[0035] S35. Combined with the historical memory mechanism of the octopus algorithm, the parameter combination and the corresponding optimization results in the crushing process are stored as empirical data, and the tentacle gives priority to trying the historical optimal solution in the empirical data in each iteration;

[0036] S36, when the optimization results satisfy I≥I threshold ,d min ≤d≤d max ,E≤E budget Under the three conditions, the output fragments with different particle sizes are broken, among which I threshold Indicates the lower limit of material integrity, d min and d max Indicates the particle size range limit, E budget Represents the energy budget.

[0037] Optionally, the S4 specifically includes:

[0038] S41, sending the crushed fragments into the micro-seismic screening device, setting the sieve aperture range according to the initial particle size distribution of the crushed fragments, and starting the micro-seismic screening device;

[0039] S42, setting the initial parameters of the micro-vibration screening device, including screening frequency, vibration amplitude and screen inclination, and starting the particle size classification operation;

[0040] S43, monitoring the operating status of the microseismic screening device in real time through sensors, including the quality of crushed fragments passing through the screen, the screen blockage rate and the screening efficiency, and recording the real-time monitoring data during the screening process;

[0041] S44, adjusting the parameters of the microseismic screening device according to the real-time monitoring data, and optimizing the screening effect to adapt to fragments of different particle sizes;

[0042] S45, classifying and collecting the powdered particles that have passed through the sieve and the large particle fragments that have not passed through the sieve, and storing them as powdered fragments and large particle fragments respectively;

[0043] S46: Output the classified fragments after screening. If the screening efficiency does not reach the preset target, repeat steps S41 to S45 until the preset target is reached.

[0044] Optionally, the S5 specifically includes:

[0045] S51, collecting spectral data, thermal imaging data and electromagnetic characteristic data of each classified fragment to form a multimodal feature set of the classified fragments;

[0046] S52, recording the spectral reflectance characteristics of the graded fragments in different wavelength ranges by a spectral imaging device to generate a spectral feature vector; collecting the temperature distribution on the surface of the graded fragments by a thermal imaging sensor to generate a thermal imaging feature matrix; measuring the conductivity and magnetic response intensity by an electromagnetic induction device to generate an electromagnetic characteristic scalar;

[0047] S53. Input the multimodal feature set of the graded fragments into the graph convolutional network model, construct a comprehensive feature vector for each graded fragment, and calculate the probability that the graded fragment belongs to metal:

[0048]

[0049] Among them, P metal (i) represents the probability that the i-th classified fragment belongs to metal, F i represents the comprehensive feature vector of the i-th hierarchical fragment, W represents the weight matrix;

[0050] S54. Calculate the metal property score of the graded fragments based on the spectral reflectance characteristics, thermal imaging thermal conductivity characteristics and electromagnetic characteristics:

[0051] Fmetal (i) = γ1·S reflect (i)+γ2·T heat (i)+γ3·E cond (i);

[0052] Among them, F metal (i) represents the metal property score of the i-th classified fragment, S reflect (i) represents the spectral reflectance characteristic score of the i-th graded fragment, T heat (i) represents the thermal imaging thermal conductivity score of the i-th graded fragment, E cond (i) represents the conductivity characteristic score of the i-th classified fragment, γ1, γ2 and γ3 represent dynamic weighting coefficients;

[0053] S55. Calculate the uncertainty of the classification of graded fragments:

[0054]

[0055] Among them, H uncertainty (i) represents the uncertainty of the classification of the i-th graded fragment, P k (i) represents the probability that the i-th classified fragment belongs to the k-th category;

[0056] S56. Classifying the graded fragments:

[0057]

[0058] Among them, C i represents the classification result of the i-th classified fragment, 1 represents metal fragment, 0 represents non-metal fragment, ζ represents the adjustment factor, and Γ represents the classification decision threshold.

[0059] Optionally, the S6 specifically includes:

[0060] S61, inputting the classified output metal fragments and non-metal fragments into the sorting module respectively, and setting the initial sorting parameters, including the magnetic field strength of the magnetic separation device and the wind speed of the wind separation device;

[0061] S62, monitoring the sorting movement path and sorting status of the fragments in real time through sensors, including the magnetic response speed of metal fragments and the wind-driven speed of non-metal fragments, and recording the real-time positions;

[0062] S63. Combine the octopus algorithm to dynamically optimize the sorting motion path, and model the sorting parameters and fragment motion trajectory as a multi-objective optimization problem:

[0063] L sort =α1·Q metal (t)-α2·(μ1·B 2 (t)+μ2·v3 (t));

[0064] Among them, L sort represents the sorting objective function, α1 and α2 represent weight coefficients, Q metal (t) represents the probability that the metal fragments at time t are successfully separated by the magnetic separation equipment, μ1 and μ2 represent the equipment energy consumption weight factors, B(t) represents the magnetic field strength of the magnetic separation equipment at time t, and v(t) represents the wind speed of the wind separation equipment at time t;

[0065] S64. Through the multi-tentacle dynamic decision-making mechanism of the octopus algorithm, the sorting effect of the tentacles under different magnetic field strength and wind speed combinations is simulated. Each tentacle corresponds to a set of parameter paths, and the operating state of the sorting module is adjusted in real time;

[0066] S65. In each round of iteration, the fragment sorting efficiency and sorting energy consumption are calculated based on the real-time feedback data. When the condition η is met, sort (t)≥η threshold ,R sort (t)≤R budget Stop optimization and lock the current parameters when η sort (t) represents the real-time sorting efficiency, η threshold Represents the sorting efficiency threshold, R sort (t) represents the sorting energy consumption, R budget Indicates the energy consumption limit of sorting;

[0067] S66. Obtain the sorted metal resources and non-metal resources.

[0068] The beneficial effects of the present invention are:

[0069] First, the present invention safely disassembles the waste lithium battery shell through non-destructive laser cutting technology, avoiding the pollution of the environment and equipment by electrolyte leakage, while ensuring the integrity of the internal materials, providing a pure processing environment for subsequent cleaning and crushing. Combined with low-temperature plasma cleaning technology, it can dynamically adjust the cleaning power and gas flow rate to achieve efficient removal of the electrolyte. Compared with traditional chemical cleaning, this technology not only reduces the use of acid and alkali solvents and reduces environmental pollution during the cleaning process, but also protects the physical properties of the material by real-time monitoring of the battery surface temperature.

[0070] Secondly, in the crushing stage, the present invention optimizes the parameters of the crushing equipment through the octopus algorithm, dynamically adjusts the crushing mode according to the characteristics of the battery material, and makes the particle size distribution of the fragments more uniform. Combined with the use of the micro-seismic screening device, it can efficiently separate powdered particles and large particle fragments, avoiding the impact of fragments with too large particle size on the accuracy of subsequent sorting. This optimized design not only improves the efficiency of crushing and screening, but also significantly reduces the energy consumption and operating costs of the equipment.

[0071] In addition, the present invention combines multimodal data fusion and graph convolutional network technology in the sorting process, and constructs a high-dimensional feature representation of the fragments by collecting spectral, thermal imaging and electromagnetic characteristic data, which significantly improves the classification accuracy of complex mixed fragments. The octopus algorithm dynamically optimizes the sorting path and adjusts the operating parameters of the magnetic separation and wind separation equipment in real time, making the separation process of metal fragments and non-metal fragments more accurate, solving the problem of insufficient classification ability of complex fragments in traditional fixed parameter sorting methods. BRIEF DESCRIPTION OF THE DRAWINGS

[0072] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0073] Figure 1 This is an overall flow chart of a method for crushing and sorting waste lithium batteries proposed by the present invention. DETAILED DESCRIPTION

[0074] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, which only illustrate the basic structure of the present invention in a schematic manner, and therefore only show the components related to the present invention.

[0075] refer to Figure 1 A method for crushing and sorting waste lithium batteries comprises the following steps:

[0076] S1. Automatically sort waste lithium batteries, distinguish battery types and remove battery shells using non-destructive laser cutting technology;

[0077] S2. Removing electrolyte residues through a low-temperature plasma treatment device and dynamically adjusting cleaning parameters to obtain a battery cell without electrolyte residues;

[0078] S3, sending the battery cells without electrolyte residue to the crushing equipment, optimizing the parameters of the crushing equipment by using the octopus algorithm, dynamically adjusting the crushing mode according to the material characteristics, and obtaining crushed fragments with different particle sizes;

[0079] S4, classifying the crushed fragments according to particle size through a micro-seismic screening device to obtain classified fragments;

[0080] S5, collect spectral data, thermal imaging data and electromagnetic property data, and use graph convolutional networks to classify and identify graded fragments, including metal fragments and non-metal fragments;

[0081] S6. Input the metal fragments and non-metal fragments into the sorting module, and use the octopus algorithm to dynamically optimize the magnetic separation and wind separation parameters to achieve real-time optimization of the fragment path and obtain the sorted metal resources and non-metal resources;

[0082] S7. The metal resources are sent into the hydrometallurgical process, and the non-metallic resources are physically separated and sent into the composite material recycling process.

[0083] In this implementation, S2 specifically includes:

[0084] S21, placing the sorted waste lithium batteries into a low-temperature plasma treatment device, starting a plasma source, setting an initial power and a gas flow rate, and using nitrogen as the working gas;

[0085] S22, real-time monitoring of the residual electrolyte concentration on the battery surface, and building an optimization model for cleaning efficiency and energy consumption:

[0086]

[0087] Wherein, η(t) represents the cleaning efficiency at time t, C(t) represents the residual concentration of electrolyte at time t, C(0) represents the initial residual concentration of electrolyte, γ represents the efficiency dynamic adjustment factor, E represents the total energy consumption during the treatment process, T represents the total cleaning time, α represents the weight coefficient of power consumption, P(t) represents the low-temperature plasma treatment power at time t, β represents the weight coefficient of gas flow rate consumption, F(t) represents the gas flow rate at time t, δ represents the weight coefficient of gas flow resistance consumption, and R(t) represents the resistance loss caused by gas flow at time t;

[0088] S23, dynamically adjusting the processing power and gas flow rate based on multi-objective optimization, with the optimization objectives being to maximize cleaning efficiency and minimize energy consumption;

[0089] S24. Real-time monitoring of the maximum temperature of the battery cell surface by thermal imaging sensors:

[0090] T max (t)≤T lim ;

[0091] Among them, T max (t) represents the maximum temperature of the battery cell surface at time t during the cleaning process, T lim Indicates the critical temperature range set according to the characteristics of battery materials, ranging from 100°C to 200°C;

[0092] S25. Continuously collect residual concentration data during the cleaning process. When the cleaning efficiency η(t)≥η threshold and energy consumption limit E≤E budget If the cleaning condition is met, the cleaning is judged to be completed, otherwise the cleaning parameters are readjusted, where η threshold represents the cleaning efficiency threshold, E budget represents the energy consumption budget;

[0093] S26. Finally, a battery cell with no electrolyte residue is obtained.

[0094] In this implementation, S3 specifically includes:

[0095] S31, sending the battery cells without electrolyte residue into the crushing equipment, starting the crushing equipment and setting the initial crushing parameters, including the blade speed and crushing force, and collecting the initial physical properties of the battery material, including strength, toughness and deformation coefficient, through real-time sensors;

[0096] S32. Introduce the octopus algorithm to simulate the dynamic exploration mechanism of multiple tentacles, take the blade speed, crushing force and crushing mode as multi-dimensional optimization variables, and simulate a parameter combination path for each tentacle:

[0097]

[0098] Wherein, O represents the crushing optimization target value, κ represents the deformation sensitivity coefficient, Δd(t) represents the particle size change rate, w1 and w2 represent the optimization weights, μ represents the power consumption weight coefficient, ν represents the force consumption weight coefficient, N(t) represents the blade speed at time t, and F(t) represents the crushing force at time t;

[0099] S33, through the multi-tentacle distributed search strategy of the octopus algorithm, each tentacle dynamically explores the optimal path in the parameter space, and real-time feedback is given on the impact of parameter combinations on the target value of fragmentation optimization. Each tentacle path is adaptively adjusted through the local update formula after each iteration:

[0100]

[0101] Among them, θ t+1 represents the updated tentacle state, θ t represents the current tentacle state, η represents the learning rate, represents the gradient of the fragmentation optimization objective value, ξ represents the exploration expansion coefficient, and rand represents the random disturbance term;

[0102] S34. In each round of iteration, all tentacles are ranked according to the real-time feedback data of particle size change, energy consumption and crushing efficiency, the optimal tentacle path is retained, and the search range of the remaining tentacles is adjusted;

[0103] S35. Combined with the historical memory mechanism of the octopus algorithm, the parameter combination and the corresponding optimization results in the crushing process are stored as empirical data, and the tentacle gives priority to trying the historical optimal solution in the empirical data in each iteration;

[0104] S36, when the optimization results satisfy I≥I threshold ,d min ≤d≤d max ,E≤Ebudget Under the three conditions, the output fragments with different particle sizes are broken, among which I threshold Indicates the lower limit of material integrity, d min and d max Indicates the particle size range limit, E budget Represents the energy budget.

[0105] In this implementation, S4 specifically includes:

[0106] S41, sending the crushed fragments into the micro-seismic screening device, setting the sieve aperture range according to the initial particle size distribution of the crushed fragments, and starting the micro-seismic screening device;

[0107] S42, setting the initial parameters of the micro-vibration screening device, including screening frequency, vibration amplitude and screen inclination, and starting the particle size classification operation;

[0108] S43, monitoring the operating status of the microseismic screening device in real time through sensors, including the quality of crushed fragments passing through the screen, the screen blockage rate and the screening efficiency, and recording the real-time monitoring data during the screening process;

[0109] S44, adjusting the parameters of the microseismic screening device according to the real-time monitoring data, and optimizing the screening effect to adapt to fragments of different particle sizes;

[0110] S45, classifying and collecting the powdered particles that have passed through the sieve and the large particle fragments that have not passed through the sieve, and storing them as powdered fragments and large particle fragments respectively;

[0111] S46: Output the classified fragments after screening. If the screening efficiency does not reach the preset target, repeat steps S41 to S45 until the preset target is reached.

[0112] In this implementation manner, S5 specifically includes:

[0113] S51, collecting spectral data, thermal imaging data and electromagnetic characteristic data of each classified fragment to form a multimodal feature set of the classified fragments;

[0114] S52, recording the spectral reflectance characteristics of the graded fragments in different wavelength ranges by a spectral imaging device to generate a spectral feature vector; collecting the temperature distribution on the surface of the graded fragments by a thermal imaging sensor to generate a thermal imaging feature matrix; measuring the conductivity and magnetic response intensity by an electromagnetic induction device to generate an electromagnetic characteristic scalar;

[0115] S53. Input the multimodal feature set of the graded fragments into the graph convolutional network model, construct a comprehensive feature vector for each graded fragment, and calculate the probability that the graded fragment belongs to metal:

[0116]

[0117] Among them, P metal (i) represents the probability that the i-th classified fragment belongs to metal, F i represents the comprehensive feature vector of the i-th hierarchical fragment, W represents the weight matrix;

[0118] S54. Calculate the metal property score of the graded fragments based on the spectral reflectance characteristics, thermal imaging thermal conductivity characteristics and electromagnetic characteristics:

[0119] F metal (i) = γ1·S reflect (i)+γ2·T heat (i)+γ3·E cond (i);

[0120] Among them, F metal (i) represents the metal property score of the i-th classified fragment, S reflect (i) represents the spectral reflectance characteristic score of the i-th graded fragment, T heat (i) represents the thermal imaging thermal conductivity score of the i-th graded fragment, E cond (i) represents the conductivity characteristic score of the i-th classified fragment, γ1, γ2 and γ3 represent dynamic weighting coefficients;

[0121] S55. Calculate the uncertainty of the classification of graded fragments:

[0122]

[0123] Among them, H uncertainty (i) represents the uncertainty of the classification of the i-th graded fragment, P k (i) represents the probability that the i-th classified fragment belongs to the k-th category;

[0124] S56. Classifying the graded fragments:

[0125]

[0126] Among them, C i represents the classification result of the i-th classified fragment, 1 represents metal fragment, 0 represents non-metal fragment, ζ represents the adjustment factor, and Γ represents the classification decision threshold.

[0127] In this implementation manner, S6 specifically includes:

[0128] S61, inputting the classified output metal fragments and non-metal fragments into the sorting module respectively, and setting initial sorting parameters, including the magnetic field strength of the magnetic separation device and the wind speed of the wind separation device;

[0129] S62, monitoring the sorting movement path and sorting status of the fragments in real time through sensors, including the magnetic response speed of metal fragments and the wind-driven speed of non-metal fragments, and recording the real-time positions;

[0130] S63. Combine the octopus algorithm to dynamically optimize the sorting motion path, and model the sorting parameters and fragment motion trajectory as a multi-objective optimization problem:

[0131] L sort =α1·Q metal (t)-α2·(μ1·B 2 (t)+μ2·v 3 (t));

[0132] Among them, L sort represents the sorting objective function, α1 and α2 represent weight coefficients, Q metal (t) represents the probability that the metal fragments at time t are successfully separated by the magnetic separation equipment, μ1 and μ2 represent the equipment energy consumption weight factors, B(t) represents the magnetic field strength of the magnetic separation equipment at time t, and v(t) represents the wind speed of the wind separation equipment at time t;

[0133] S64. Through the multi-tentacle dynamic decision-making mechanism of the octopus algorithm, the sorting effect of the tentacles under different magnetic field strength and wind speed combinations is simulated. Each tentacle corresponds to a set of parameter paths, and the operating state of the sorting module is adjusted in real time;

[0134] S65. In each round of iteration, the fragment sorting efficiency and sorting energy consumption are calculated based on the real-time feedback data. When the condition η is met, sort (t)≥η threshold ,R sort (t)≤R budget Stop optimization and lock the current parameters when η sort (t) represents the real-time sorting efficiency, η threshold represents the sorting efficiency threshold, R sort (t) represents the sorting energy consumption, R budget Indicates the energy consumption limit of sorting;

[0135] S66. Obtain the sorted metal resources and non-metal resources.

[0136] Embodiment 1:

[0137] In order to verify the feasibility of the present invention in practice, the present invention was applied to the actual production line of a large battery recycling company to carry out resource processing on a batch of retired lithium-ion batteries. The test site was located in an environmental protection technology industrial park in a certain city, and the experimental data was collected by the on-site real-time monitoring system and the recycling statistics system.

[0138] This batch of waste lithium batteries mainly comes from electric vehicles and energy storage equipment, totaling 1 ton, including different types of batteries, including lithium iron phosphate batteries and nickel-cobalt-manganese ternary batteries. When processing such mixed batteries, traditional treatment methods usually lead to waste of resources and environmental pollution due to insufficient classification, inaccurate crushing parameters and large sorting errors. The processing flow of the present invention introduces intelligent equipment and optimization algorithms to solve problems such as low classification accuracy, high energy consumption and insufficient recovery rate.

[0139] In actual application, non-destructive laser cutting technology was first used to sort and remove the shells of 1 ton of waste lithium batteries. The automatic sorting equipment identified the type and appearance characteristics of the battery, with a classification accuracy of 98.5%, achieving rapid sorting and avoiding manual sorting errors. At the same time, laser cutting technology effectively prevented electrolyte leakage, processing batteries weighing 200 kilograms per hour, and it took 5 hours to remove the shells of 1 ton of batteries.

[0140] The classified batteries are then cleaned of electrolyte by a low-temperature plasma device. In the test, the initial electrolyte residual concentration was 950ppm. After low-temperature plasma treatment, the residual concentration dropped to below 10ppm, and the cleaning efficiency reached 98.95%. Combined with the dynamic optimization parameters of the Octopus algorithm, the cleaning energy consumption is 0.45 kWh / kg, which reduces energy consumption by 30% compared to traditional chemical cleaning methods, while eliminating the generation of chemical waste liquid.

[0141] The cleaned battery cells are sent to the crushing equipment, and the crushing parameters, including blade speed and crushing force, are optimized through the Octopus algorithm. The particle size distribution of the fragments is concentrated in the range of 2-5 mm, and the standard deviation is reduced to 0.8 mm, which is 60% less than that of traditional equipment. The energy consumption is monitored in real time during the crushing process, and the average power consumption is 2.1 kWh / ton, which is 15% lower than that of traditional crushing equipment.

[0142] The broken fragments are graded by microseismic screening device, and the separation efficiency of large particles and powdery fragments reaches 95.2%. The graded fragments are sent to the graph convolution network classification module. Through multimodal data fusion and classification algorithm, the classification accuracy of metal fragments and non-metal fragments is as high as 99%, achieving accurate identification and classification.

[0143] In the sorting module, the octopus algorithm dynamically optimizes the magnetic separation and wind separation parameters, and the sorting rate of metal fragments reaches 97.8%, and the separation rate of non-metallic fragments is 96.5%. Among them, the metal resources including lithium, cobalt, nickel and other elements have high content, and then enter the hydrometallurgical process for purification. The experimental results show that the recovery rate of lithium is 92.5%, cobalt is 94.8%, and nickel is 91.7%. After the non-metallic fragments are processed by composite material recycling technology, the reuse rate of plastic diaphragms reaches 88%, further reducing waste generation.

[0144] By comparing with the traditional process, the present invention has shown significant advantages in processing time, energy consumption, recovery rate, etc. The processing time of 1 ton of waste lithium batteries is shortened to 12 hours, the total energy consumption is reduced by more than 20%, and the resource recovery rate is increased by about 15%.

[0145] Table 1 Experimental data of crushing and sorting of waste lithium batteries

[0146] Procedure Traditional process data Data of the present invention Improvement Classification Accuracy 85％ 98.5％ +13.5％ Electrolyte removal efficiency 85％ 98.95％ +13.95％ Cleaning energy consumption (kWh / kg) 0.65 0.45 -30％ Crushed particle size distribution range 1-8 mm 2-5 mm More concentrated particle size Crushing energy consumption (kWh / ton) 2.5 2.1 -15％ Classification efficiency 85％ 95.2％ +10.2％ Classification accuracy 90％ 99％ +9％ Metal separation rate 87％ 97.8％ +10.8％ Non-metal separation rate 85％ 96.5％ +11.5％ Lithium recovery rate 75％ 92.5％ +17.5％ Cobalt recovery rate 80％ 94.8％ +14.8％ Nickel recovery rate 78％ 91.7％ +13.7％ Plastic diaphragm recycling rate 65％ 88％ +23％ Total processing time (hours) 18 12 -33％ Total energy consumption (kWh) 750 600 -20％

[0147] By analyzing the data in Table 1 above, we can see that, first, in the classification stage, the present invention improves the classification accuracy to 98.5% through non-destructive laser cutting and automatic classification technology, which is 13.5% higher than the traditional process. This high-precision classification ensures accurate identification of battery types and lays a solid foundation for subsequent processing. In traditional processes, due to reliance on manual or simple mechanical classification, the classification accuracy is only 85%, which easily leads to parameter mismatch problems in subsequent processes.

[0148] In the electrolyte removal stage, the removal efficiency of low-temperature plasma cleaning technology reached 98.95%, significantly higher than the 85% of the traditional process. This green cleaning technology not only improves efficiency, but also reduces cleaning energy consumption by 30%, from 0.65 kWh / kg of the traditional process to 0.45 kWh / kg, completely eliminating the generation of chemical waste liquid.

[0149] In terms of crushing, the present invention uses the octopus algorithm to optimize the crushing equipment parameters, making the particle size distribution of the fragments more uniform, concentrated in the range of 2-5 mm, and reducing the standard deviation of the particle size by 60%. At the same time, the crushing energy consumption is reduced from the traditional 2.5 kWh / ton to 2.1 kWh / ton, and the energy consumption is reduced by 15%. This optimization process avoids excessive crushing of fragments and waste of metal.

[0150] During the screening stage, the micro-vibration screening device of the present invention increases the classification efficiency to 95.2% through dynamic parameter adjustment, which is 10.2% higher than the traditional process. This efficient classification ensures that fragments of different particle sizes are accurately separated, providing ideal conditions for subsequent sorting.

[0151] In terms of classification and sorting, the present invention uses multimodal data fusion and graph convolutional network technology to achieve a classification accuracy of 99%, which is 9% higher than the traditional process. At the same time, the octopus algorithm is used to dynamically optimize the sorting path in the sorting module, so that the sorting rates of metals and non-metals reach 97.8% and 96.5%, respectively, which are 10.8% and 11.5% higher than the traditional process. This efficient sorting greatly improves the purity and quality of resource recovery.

[0152] In the resource recovery phase, the hydrometallurgical process has increased the recovery rates of lithium, cobalt and nickel to 92.5%, 94.8% and 91.7% respectively, which are 17.5%, 14.8% and 13.7% higher than the traditional process. The reuse rate of non-metallic resources through composite material recycling technology has reached 88%, which is 23% higher than the traditional process, achieving maximum utilization of resources.

[0153] Overall, the present invention has made significant progress in processing time, energy consumption, recovery rate, etc. The total processing time is shortened from 18 hours in the traditional process to 12 hours, and the total energy consumption is reduced from 750 kWh to 600 kWh, which is a 20% reduction in energy consumption. These data fully demonstrate the superiority of the present invention in terms of resource utilization efficiency, environmental friendliness and economy, and provide strong technical support and practical basis for the efficient and green recovery of waste lithium batteries.

[0154] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A method for crushing and sorting waste lithium batteries, characterized in that: The steps include: S1. Automatically sort waste lithium batteries, distinguish battery types and remove battery shells using non-destructive laser cutting technology; S2. Removing electrolyte residues through a low-temperature plasma treatment device and dynamically adjusting cleaning parameters to obtain a battery cell without electrolyte residues; S3, sending the battery cells without electrolyte residue to the crushing equipment, optimizing the parameters of the crushing equipment by using the octopus algorithm, dynamically adjusting the crushing mode according to the material characteristics, and obtaining crushed fragments with different particle sizes; S4, classifying the crushed fragments according to particle size through a micro-seismic screening device to obtain classified fragments; S5, collect spectral data, thermal imaging data and electromagnetic property data, and use graph convolutional networks to classify and identify graded fragments, including metal fragments and non-metal fragments; S6. Input the metal fragments and non-metal fragments into the sorting module, and use the octopus algorithm to dynamically optimize the magnetic separation and wind separation parameters to achieve real-time optimization of the fragment path and obtain the sorted metal resources and non-metal resources; S7. The metal resources are sent into the hydrometallurgical process, and the non-metallic resources are physically separated and sent into the composite material recycling process.

2. A method for crushing and sorting waste lithium batteries according to claim 1, characterized in that: The S2 specifically includes: S21, placing the sorted waste lithium batteries into a low-temperature plasma treatment device, starting a plasma source, setting an initial power and a gas flow rate, and using nitrogen as the working gas; S22. Real-time monitoring of the residual electrolyte concentration on the battery surface and construction of an optimization model for cleaning efficiency and energy consumption: Wherein, η(t) represents the cleaning efficiency at time t, C(t) represents the residual concentration of electrolyte at time t, C(0) represents the initial residual concentration of electrolyte, γ represents the efficiency dynamic adjustment factor, E represents the total energy consumption during the treatment process, T represents the total cleaning time, α represents the weight coefficient of power consumption, P(t) represents the low-temperature plasma treatment power at time t, β represents the weight coefficient of gas flow rate consumption, F(t) represents the gas flow rate at time t, δ represents the weight coefficient of gas flow resistance consumption, and R(t) represents the resistance loss caused by gas flow at time t; S23, dynamically adjusting the processing power and gas flow rate based on multi-objective optimization, with the optimization objectives being to maximize cleaning efficiency and minimize energy consumption; S24. Real-time monitoring of the maximum temperature of the battery cell surface by thermal imaging sensors: T max (t)≤T lim ; Among them, T max (t) represents the maximum temperature of the battery cell surface at time t during the cleaning process, T lim Indicates the critical temperature range set according to the characteristics of battery materials, ranging from 100°C to 200°C; S25. Continuously collect residual concentration data during the cleaning process. When the cleaning efficiency η(t)≥η threshold and energy consumption limit E≤E budget If the cleaning condition is met, the cleaning is judged to be completed, otherwise the cleaning parameters are readjusted, where η threshold represents the cleaning efficiency threshold, E budget represents the energy consumption budget; S26. Finally, a battery cell with no electrolyte residue is obtained.

3. The method for crushing and sorting waste lithium batteries according to claim 1, characterized in that: The S3 specifically includes: S31, sending the battery cells without electrolyte residue into the crushing equipment, starting the crushing equipment and setting the initial crushing parameters, including the blade speed and crushing force, and collecting the initial physical properties of the battery material, including strength, toughness and deformation coefficient, through real-time sensors; S32. Introduce the octopus algorithm to simulate the dynamic exploration mechanism of multiple tentacles, take the blade speed, crushing force and crushing mode as multi-dimensional optimization variables, and simulate a parameter combination path for each tentacle: Wherein, O represents the crushing optimization target value, κ represents the deformation sensitivity coefficient, Δd(t) represents the particle size change rate, w1 and w2 represent the optimization weights, μ represents the power consumption weight coefficient, ν represents the force consumption weight coefficient, N(t) represents the blade speed at time t, and F(t) represents the crushing force at time t; S33, through the multi-tentacle distributed search strategy of the octopus algorithm, each tentacle dynamically explores the optimal path in the parameter space, and real-time feedback is given on the impact of parameter combinations on the target value of fragmentation optimization. Each tentacle path is adaptively adjusted through the local update formula after each iteration: Among them, θ t+1 represents the updated tentacle state, θ t represents the current tentacle state, η represents the learning rate, represents the gradient of the fragmentation optimization objective value, ξ represents the exploration expansion coefficient, and rand represents the random disturbance term; S34. In each round of iteration, all tentacles are ranked according to the real-time feedback data of particle size change, energy consumption and crushing efficiency, the optimal tentacle path is retained, and the search range of the remaining tentacles is adjusted; S35. Combined with the historical memory mechanism of the octopus algorithm, the parameter combination and the corresponding optimization results in the crushing process are stored as empirical data, and the tentacle gives priority to trying the historical optimal solution in the empirical data in each iteration; S36, when the optimization results satisfy I≥I threshold ,d min ≤d≤d max ,E≤E budget Under the three conditions, the output fragments with different particle sizes are broken, among which I threshold Indicates the lower limit of material integrity, d min and d max Indicates the particle size range limit, E budget Represents the energy budget.

4. The method for crushing and sorting waste lithium batteries according to claim 1, characterized in that: The S4 specifically includes: S41, sending the crushed fragments into the micro-seismic screening device, setting the sieve aperture range according to the initial particle size distribution of the crushed fragments, and starting the micro-seismic screening device; S42, setting the initial parameters of the micro-vibration screening device, including screening frequency, vibration amplitude and screen inclination, and starting the particle size classification operation; S43, monitoring the operating status of the microseismic screening device in real time through sensors, including the quality of crushed fragments passing through the screen, the screen blockage rate and the screening efficiency, and recording the real-time monitoring data during the screening process; S44, adjusting the parameters of the microseismic screening device according to the real-time monitoring data, and optimizing the screening effect to adapt to fragments of different particle sizes; S45, classifying and collecting the powdered particles that have passed through the sieve and the large particle fragments that have not passed through the sieve, and storing them as powdered fragments and large particle fragments respectively; S46: Output the classified fragments after screening. If the screening efficiency does not reach the preset target, repeat steps S41 to S45 until the preset target is reached.

5. The method for crushing and sorting waste lithium batteries according to claim 1, characterized in that: The S5 specifically includes: S51, collecting spectral data, thermal imaging data and electromagnetic characteristic data of each classified fragment to form a multimodal feature set of the classified fragments; S52, recording the spectral reflectance characteristics of the graded fragments in different wavelength ranges by a spectral imaging device to generate a spectral feature vector; collecting the temperature distribution on the surface of the graded fragments by a thermal imaging sensor to generate a thermal imaging feature matrix; measuring the conductivity and magnetic response intensity by an electromagnetic induction device to generate an electromagnetic characteristic scalar; S53. Input the multimodal feature set of the graded fragments into the graph convolutional network model, construct a comprehensive feature vector for each graded fragment, and calculate the probability that the graded fragment belongs to metal: Among them, P metal (i) represents the probability that the i-th classified fragment belongs to metal, F i represents the comprehensive feature vector of the i-th hierarchical fragment, W represents the weight matrix; S54. Calculate the metal property score of the graded fragments based on the spectral reflectance characteristics, thermal imaging thermal conductivity characteristics and electromagnetic characteristics: F metal (i)=γ1·S reflect (i)+γ2·T heat (i)+γ3·E cond (i); Among them, F metal (i) represents the metal property score of the i-th classified fragment, S reflect (i) represents the spectral reflectance characteristic score of the i-th graded fragment, T heat (i) represents the thermal imaging thermal conductivity score of the i-th graded fragment, E cond (i) represents the conductivity characteristic score of the i-th classified fragment, γ1, γ2 and γ3 represent dynamic weighting coefficients; S55. Calculate the uncertainty of the classification of graded fragments: Among them, H uncertainty (i) represents the uncertainty of the classification of the i-th graded fragment, P k (i) represents the probability that the i-th classified fragment belongs to the k-th category; S56. Classifying the graded fragments: Among them, C i represents the classification result of the i-th classified fragment, 1 represents metal fragment, 0 represents non-metal fragment, ζ represents the adjustment factor, and Γ represents the classification decision threshold.

6. The method for crushing and sorting waste lithium batteries according to claim 1, characterized in that: The S6 specifically includes: S61, inputting the classified output metal fragments and non-metal fragments into the sorting module respectively, and setting initial sorting parameters, including the magnetic field strength of the magnetic separation device and the wind speed of the wind separation device; S62, monitoring the sorting movement path and sorting status of the fragments in real time through sensors, including the magnetic response speed of metal fragments and the wind-driven speed of non-metal fragments, and recording the real-time positions; S63. Combine the octopus algorithm to dynamically optimize the sorting motion path, and model the sorting parameters and fragment motion trajectory as a multi-objective optimization problem: L sort =α1·Q metal (t)-α2·(μ1·B 2 (t)+μ2·v 3 (t)); Among them, L sort represents the sorting objective function, α1 and α2 represent weight coefficients, Q metal (t) represents the probability that the metal fragments at time t are successfully separated by the magnetic separation equipment, μ1 and μ2 represent the equipment energy consumption weight factors, B(t) represents the magnetic field strength of the magnetic separation equipment at time t, and v(t) represents the wind speed of the wind separation equipment at time t; S64. Through the multi-tentacle dynamic decision-making mechanism of the octopus algorithm, the sorting effect of the tentacles under different magnetic field strength and wind speed combinations is simulated. Each tentacle corresponds to a set of parameter paths, and the operating state of the sorting module is adjusted in real time; S65. In each round of iteration, the fragment sorting efficiency and sorting energy consumption are calculated based on the real-time feedback data. When the condition η is met, sort (t)≥η threshold ,R sort (t)≤R budget Stop optimization and lock the current parameters when η sort (t) represents the real-time sorting efficiency, η threshold Represents the sorting efficiency threshold, R sort (t) represents the sorting energy consumption, R budget Indicates the energy consumption limit of sorting; S66. Obtain the sorted metal resources and non-metal resources.

Images