Electrochemical directional circulation-based deintercalation and intercalation balance method and apparatus in lithium extraction, device, and medium

Abstract

Disclosed in the present invention are an electrochemical directional circulation-based deintercalation and intercalation balance method and apparatus in lithium extraction, a device, and a medium. The method comprises: performing deintercalation and intercalation in lithium extraction on the basis of working index parameters set for an electrolytic cell, and calculating a lithium deintercalation duration required by a lithium intercalation tank electrode plate; inputting the working index parameters into a pre-trained lithium intercalation duration calculation model, and outputting a lithium intercalation duration corresponding to complete lithium intercalation; calculating a duration difference on the basis of the lithium deintercalation duration and the lithium intercalation duration, inputting the duration difference and the working index parameters into a pre-trained reducing agent calculation model, and outputting a reducing agent dosage required by the lithium intercalation tank electrode plate; and controlling the balance of lithium intercalation and lithium deintercalation on the basis of a reducing agent of the required reducing agent dosage. The deintercalation and intercalation rate in lithium extraction can be balanced, and the lithium extraction efficiency is improved.

Description

Electrochemical directional cycle lithium extraction and deintercalation balance method, device, equipment and medium Technical Field

[0001] The present invention relates to the technical field of battery materials, and in particular to a method, device, equipment and medium for balancing lithium extraction and deintercalation in electrochemical directional circulation. Background Art

[0002] With the depletion of non-renewable energy, new energy vehicles, as a prime example of new energy development and utilization, have experienced rapid growth in recent years. The development and utilization of new energy vehicles is an inevitable trend that will eventually surpass the market share of traditional fuel vehicles and gradually replace them. As an essential energy metal for the power systems of new energy vehicles, lithium is also expected to see a sharp increase in market demand. Efficient, clean, and low-cost mining of lithium resources is crucial to the sustainable development of the new energy vehicle industry. With the mining of high-quality lithium ore, the availability of high-grade ore that meets current production processes is becoming increasingly scarce, and the cost of extracting lithium from ore is also relatively high. Compared to extracting lithium from ore, the abundant reserves of lithium in brine, where lithium exists in ionic form, offer a natural cost advantage in extraction.

[0003] As a green process technology, electrodialysis uses electrochemical directional cyclic deintercalation technology. An oxidation reaction occurs at the anode, lithium is deintercalated from lithium iron phosphate, and a reduction reaction occurs at the cathode, where lithium in the brine is intercalated into iron phosphate. For an ideal reaction, the cathode should also intercalate a lithium ion while the anode deintercalates a lithium ion, and the deintercalation rates of the two should be consistent. However, in the actual reaction process, the cathode lithium intercalation process is greatly affected by factors such as the viscosity of the brine and the ion concentration in the brine. This causes the cathode lithium intercalation process to be much slower than the anode lithium deintercalation process, resulting in an inconsistency in the working rates of the two, leading to a mismatch between the lithium deintercalation and lithium intercalation capacities of the cathode and anode. In addition, the two electrodes need to continuously exchange their polarity. As the number of times increases, the lithium absorption capacity will gradually decrease, and the lithium extraction efficiency will become slower and slower.

[0004] Therefore, it is urgent to calculate the time required for the material to be completely intercalated with lithium, balance the deintercalation rate according to the lithium extraction time, and solve the problem that as the number of deintercalation times increases, the lithium absorption capacity gradually decreases and the lithium extraction efficiency becomes slower and slower.

[0005] Summary of the Invention

[0006] In order to solve the above problems, the present invention proposes an electrochemical directional cycle lithium extraction and deintercalation balance method, device, equipment and medium, which can balance the lithium extraction and deintercalation rates and improve the lithium extraction efficiency.

[0007] An embodiment of the present invention provides a method for balancing lithium extraction and deintercalation in an electrochemical directional cycle, the method comprising:

[0008] Lithium extraction and deintercalation are performed according to the working index parameters set for the electrolytic cell, and the delithiation time required for the electrode plates of the lithium intercalation cell is calculated;

[0009] Inputting the working index parameters into a pre-trained lithium insertion time calculation model to output the lithium insertion time corresponding to complete lithium insertion;

[0010] Calculating a time difference according to the delithiation time and the lithium insertion time, inputting the time difference and the working index parameter into a pre-trained reducing agent calculation model, and outputting a reducing agent dosage required for the lithium insertion tank electrode plate;

[0011] The balance of lithium insertion and delithiation is controlled according to the required reducing agent dosage.

[0012] Preferably, the training process of the lithium insertion time calculation model or the reducing agent calculation model specifically includes:

[0013] Obtain a training sample set including input data and target values;

[0014] Inputting the training sample set into a neural network recognition model to be trained for training, adjusting model parameters using a gradient descent algorithm according to a preset training accuracy value, training speed, and number of cycles for training; and using the trained model as a target model, wherein the target model is specifically the lithium insertion time calculation model or the reducing agent calculation model;

[0015] The number of input layer nodes, hidden layer nodes and output layer nodes of the neural network recognition module is determined by the number of working indicator parameters in the input data.

[0016] As a preferred solution, before using the trained model as the target model, the method further includes:

[0017] Input the input data from the obtained verification sample set into the trained neural network recognition model to obtain the model output value corresponding to the input data;

[0018] Calculate the relative error between the model output value of the input data and the corresponding target value in the validation sample set;

[0019] The trained model is verified based on the calculated relative error, and the qualified trained model is output.

[0020] Preferably, the step of verifying the trained model based on the calculated relative error and outputting a qualified trained model specifically includes:

[0021] The qualified proportion of input data with a statistical relative error within a first preset range to all input data in the validation sample set;

[0022] When the statistically determined qualified ratio is not within the second preset range, the trained model is determined to be unqualified, the number of samples in the training sample set is increased, and the training sample set is input into the neural network recognition model to continue training;

[0023] When the statistically determined qualified ratio is within the second preset range, the trained model is determined to meet the requirements and the trained model is output.

[0024] Preferably, the step of inputting the training sample set into the neural network recognition model to be trained for training specifically includes:

[0025] A neural network recognition model is used to calculate the input data in the training sample set to obtain the output value of the model, and the weights of the neurons connected to each other in the neural network recognition model are corrected based on the error between the output value and the corresponding target value; the corrected weights are transferred to each layer of neurons layer by layer using a back propagation algorithm, and the error is corrected iteratively until the number of iterations reaches a preset value or the sum of the squares of the error between the output value of the model and the corresponding target value is within a preset range;

[0026] When the number of iterations reaches the preset value and the sum of squares of the errors between the output value of the model and the corresponding target value is not within a preset range, the number of hidden layers of the neural network recognition model is increased, and the training sample set is re-input into the neural network recognition model to be trained for training until the sum of squares of the errors between the output value of the model and the corresponding target value is within the preset range;

[0027] When the sum of squared errors between the model's output value and the corresponding target value is within the preset range, model training is completed.

[0028] Preferably, the neural network recognition model adopts a back propagation neural network;

[0029] The neural network recognition model uses the Sigmoid function as a training function;

[0030] The neural network recognition model uses the tansig function as the hidden layer transfer function;

[0031] The neural network recognition model adopts the Softmax function as the output layer transfer function.

[0032] Preferably, the working index parameters include temperature, lithium ion concentration in brine, capacity of lithium absorbing material coated on the electrode plate, particle diffusion coefficient of lithium absorbing material, constant current supply current and constant voltage supply voltage.

[0033] Preferably, when the target model is a lithium insertion time calculation model, the input layer of the neural network recognition model is set with 6 nodes, the hidden layer is set with 5 nodes, and the output layer is set with 1 node;

[0034] The process of obtaining the training sample set of the target model specifically includes:

[0035] Record the first indicator parameter used in each group of experiments and the corresponding complete lithium insertion time when using several lithium extraction electrode plates of different materials for single lithium extraction experiments;

[0036] Using the recorded first indicator parameter as input data of the sample set, and using the recorded complete lithium insertion time as the target value corresponding to the input data in the sample set;

[0037] The sample set is divided according to a preset ratio to obtain a training sample set and a verification sample set.

[0038] Furthermore, the lithium-extracting electrode plates made of different materials include lithium titanate material lithium-extracting electrode plates, lithium phosphate material lithium-extracting electrode plates, lithium silicate material lithium-extracting electrode plates and lithium manganate material lithium-extracting electrode plates;

[0039] The first index parameters recorded include temperature, lithium ion concentration in brine, capacity of lithium absorbing material coated on the electrode plate, particle diffusion coefficient of lithium absorbing material, constant current supply current and constant voltage supply voltage.

[0040] As a preferred solution, when the target model is the reducing agent calculation model, the input layer of the neural network recognition model is set with 4 nodes, the hidden layer is set with 3 nodes, and the output layer is set with 1 node;

[0041] The process of obtaining the training sample set of the target model specifically includes:

[0042] Record the second indicator parameters and corresponding reaction time used in each group of experiments when using several lithium extraction electrode plates with different reducing agent dosages for single lithium extraction experiments;

[0043] Using the recorded second indicator parameter and reaction time as input data of the sample set, and using the recorded reducing agent dosage as the target value corresponding to the input data in the sample set;

[0044] The sample set is divided according to a preset ratio to obtain a training sample set and a verification sample set.

[0045] Preferably, the recorded second indicator parameters include temperature, constant current supply current and constant voltage supply voltage.

[0046] The present invention also provides an electrochemical directional cycle lithium extraction and deintercalation balance device, which includes:

[0047] The data acquisition module is used to extract and deintercalate lithium according to the working index parameters set for the electrolytic cell, and calculate the deintercalation time required for the electrode plates of the lithium intercalation cell;

[0048] A duration calculation module is used to input the working index parameters into a pre-trained lithium insertion duration calculation model and output the lithium insertion duration corresponding to complete lithium insertion;

[0049] a dosage calculation module, configured to calculate a time difference based on the delithiation time and the lithium insertion time, input the time difference and the working index parameter into a pre-trained reducing agent calculation model, and output the reducing agent dosage required for the lithium insertion slot electrode plate;

[0050] The balance control module is used to control the balance of lithium insertion and lithium deintercalation according to the required reducing agent dosage.

[0051] An embodiment of the present invention also provides a terminal device, comprising a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein when the processor executes the computer program, it implements an electrochemical directional cycle lithium extraction and deintercalation balance method as described in any of the above embodiments.

[0052] An embodiment of the present invention also provides a computer-readable storage medium, which includes a stored computer program, wherein when the computer program is running, the device where the computer-readable storage medium is located is controlled to perform an electrochemical directional cycle lithium extraction and deintercalation balance method as described in any of the above embodiments.

[0053] The present invention provides an electrochemical directional cycle lithium extraction and deintercalation balance method, device, equipment, and medium. The method extracts and deintercalates lithium according to the working index parameters set for the electrolytic cell, calculates the lithium extraction and deintercalation time required for the lithium intercalation cell electrode plate, inputs the working index parameters into a pre-trained lithium intercalation time calculation model, and outputs the lithium intercalation time corresponding to complete lithium intercalation. The method calculates the time difference between the deintercalation time and the lithium intercalation time, inputs the time difference and the working index parameters into a pre-trained reducing agent calculation model, and outputs the reducing agent dosage required for the lithium intercalation cell electrode plate. The method controls the lithium insertion and deintercalation balance based on the reducing agent of the required reducing agent dosage. The method is capable of balancing the lithium extraction and deintercalation rates and improving lithium extraction efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] FIG1 is a schematic flow diagram of an electrochemical directional cycle lithium extraction and deintercalation equilibrium method provided by an embodiment of the present invention;

[0055] FIG2 is a schematic diagram of a process for training a target model according to an embodiment of the present invention;

[0056] FIG3 is a schematic diagram of a process for providing model verification according to an embodiment of the present invention;

[0057] FIG4 is a schematic diagram of the structure of a neural network recognition model provided by an embodiment of the present invention;

[0058] FIG5 is a schematic structural diagram of an electrochemical directional cycle lithium extraction and deintercalation balance device provided in an embodiment of the present invention;

[0059] FIG6 is a schematic structural diagram of a terminal device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0060] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0061] Example 1

[0062] 1 is a schematic flow chart of an electrochemical directional cycle lithium extraction and deintercalation equilibrium method provided by an embodiment of the present invention, wherein the method includes steps S1 to S4;

[0063] S1, extracting and deintercalating lithium according to the working index parameters set for the electrolytic cell, and calculating the time required for deintercalation of the lithium intercalation cell electrode plate;

[0064] S2, inputting the working index parameters into a pre-trained lithium insertion time calculation model, and outputting the lithium insertion time corresponding to complete lithium insertion;

[0065] S3, calculating a time difference according to the delithiation time and the lithium insertion time, inputting the time difference and the working index parameter into a pre-trained reducing agent calculation model, and outputting a reducing agent dosage required for the lithium insertion slot electrode plate;

[0066] S4, controlling the balance of lithium insertion and lithium removal according to the required reducing agent dosage.

[0067] During the specific implementation of this embodiment, it is necessary to pre-train a lithium insertion time calculation model and a reducing agent calculation model. Both models can be constructed using a neural network and obtained through training data.

[0068] The input of the lithium insertion time calculation model is the working index parameter of the electrolytic cell, and the output is the lithium insertion time corresponding to complete lithium insertion;

[0069] The input of the reducing agent calculation model is the working index parameters and reduction time of the electrolytic cell, and the output is the reducing agent dosage required for the lithium-embedded cell electrode plate;

[0070] It should be noted that the construction and training process of the lithium insertion time calculation model and the reducing agent calculation model can be obtained by using model training methods, including machine learning, neural network learning and other model training methods.

[0071] The reduction time of the reducing agent calculation model is obtained by the difference between the delithiation time required for the lithium insertion slot electrode plate and the lithium insertion time corresponding to complete lithium insertion. The difference between the two is caused by the imbalance between the delithiation rate and the lithium insertion rate. Therefore, it is necessary to balance the delithiation rate and the lithium insertion rate by adding a reducing agent to solve the problem that as the number of delithiation and insertion increases, the lithium absorption capacity gradually decreases, resulting in slower and slower lithium extraction efficiency, so as to improve the lithium extraction efficiency.

[0072] During specific execution, when the electrolytic cell is normally deintercalated, the electrolytic cell is used to extract and deintercalate lithium based on the working index parameters, and the delithiation time t2 required for the electrode plate of the lithium intercalation cell is calculated;

[0073] Obtaining the working index parameters of the electrolytic cell, inputting the working index parameters into the pre-trained lithium insertion time calculation model for calculation, and obtaining the lithium insertion time t1 required for complete lithium insertion;

[0074] The time difference Δt calculated based on the delithiation time t2 and the lithium insertion time t1, Δt = t1 - t2, is the reduction time of the reducing agent required to be added to the electrolytic cell;

[0075] Input the working index parameters of the electrolytic cell and the time difference Δt into the reducing agent calculation model for calculation to obtain the reducing agent dosage required for the lithium-embedded cell electrode plate;

[0076] The reducing agent is added to the electrolytic cell according to the reducing agent dosage to ensure that the lithium insertion rate and the lithium de-insertion rate are balanced, thereby avoiding the problem of slower and slower lithium extraction efficiency due to inconsistent de-insertion rates, thereby improving the lithium extraction efficiency.

[0077] Example 2

[0078] In another embodiment provided by the present invention, the training process of the lithium insertion time calculation model and the reducing agent calculation model is similar, and both adopt the training method of the target model, and the specific process includes:

[0079] A neural network recognition model is used as the target model to be trained;

[0080] Obtain a training sample set containing input data and target values, where the training sample set contains several groups of samples containing input data and corresponding target values;

[0081] Inputting the training sample set into a neural network recognition model to be trained for training, and adjusting the model parameters using a gradient descent algorithm according to a preset training accuracy value, training speed, and number of cycles for training; adjusting the model parameters using the gradient descent algorithm for model training, and training using input data and corresponding target values; the neural network recognition model approximates a function, and associates the output of the model with the input to obtain a corresponding relationship between the input data and the target value;

[0082] Through the construction of the model, the corresponding relationship between the working index parameters and the lithium insertion time can be established, and the corresponding relationship between the time difference and the working index parameters and the reducing agent dosage can be established. Combined with the high-speed computing power of the computer, the lithium insertion time and reducing agent dosage can be quickly and accurately obtained according to the model input;

[0083] The trained model is used as the target model.

[0084] It should be noted that the specific model training process model architecture can adopt convolutional neural networks, recurrent neural networks, etc., and the model training method can use training methods such as machine learning, transfer learning and reinforcement learning.

[0085] Preferably, Mini-batch, SGD, AdaGrad, momentum, RMSprop, adam, and learning rate decay methods can be used to improve the model training speed; by changing the number of features, changing the number of training sample sets, changing the model depth, changing the number of weights in a certain layer of the model, regularization, drop-out, early-stopping, and batch-norm.

[0086] It should be noted that the number of input layer nodes, hidden layer nodes and output layer nodes of the neural network recognition module is determined according to the number of work indicator parameters to improve the model accuracy.

[0087] Example 3

[0088] In another embodiment of the present invention, referring to FIG2 , which is a flow chart of target model training according to an embodiment of the present invention, the method includes:

[0089] Step S201, obtaining a training sample set including input data and target values;

[0090] Step S202: inputting the training sample set into the neural network recognition model to be trained for training;

[0091] Step S203, obtaining a validation sample set including input data and target values, wherein the validation sample set includes several groups of samples including input data and corresponding target values;

[0092] Step S204, verifying the trained neural network recognition model based on the verification sample set;

[0093] Step S205, determining whether the verification is passed;

[0094] If not, return to step S201;

[0095] If yes, go to step S206;

[0096] S206, outputting the verified qualified training model;

[0097] S207: Use the trained model as the target model.

[0098] Preferably, the trained neural network recognition model is verified based on the verification sample set, and the specific verification process includes:

[0099] Input the input data from the obtained verification sample set into the trained neural network recognition model to obtain the model output value t3 corresponding to the input data;

[0100] Calculate the error Δt2 between the model output value t3 of the input data and the corresponding target value t4 in the validation sample set, Δt2 = |t3-t4|, and the relative error is the ratio of the error to the target value, that is, relative error Δ = |t3-t4| / t4;

[0101] The relative error can be used to determine whether the trained neural network recognition model is qualified.

[0102] If the model fails to meet the requirements, continue model training;

[0103] If the model is qualified, the model is output as the lithium insertion time calculation model or the reducing agent calculation model.

[0104] The model trained with the training sample set is verified through the verification sample set, and the model that has passed the accuracy verification is output as the target model to improve the accuracy of the model.

[0105] Example 4

[0106] In another embodiment provided by the present invention, referring to FIG3 , which is a flow chart of model verification provided by an embodiment of the present invention; the model verification process includes the following steps:

[0107] Step S301, calculating the relative error between the model output value of the input data and the corresponding target value in the validation sample set, that is, calculating the error Δt2 between the model output value t3 of the input data and the corresponding target value t4 in the validation sample set, Δt2 = |t3-t4|, the relative error is the ratio of the error to the target value, that is, relative error Δ = |t3-t4| / t4;

[0108] Step S302, determining whether a relative error is within a preset range, that is, determining whether the relative error of a sample is greater than 3%;

[0109] If yes, execute step S303;

[0110] If not, execute step S304;

[0111] Step S303: The sample is determined to be an unqualified sample, and step S305 is executed; that is, if the relative error of a sample is greater than 3%, the sample is determined to be an unqualified sample;

[0112] Step S304: determine the sample as a qualified sample, and execute step S304; that is, if the relative error of a sample is not greater than 3%, then the sample is determined to be a qualified sample;

[0113] Step S305, counting the qualified ratio, that is, counting the ratio of qualified samples in all samples in the verification sample set;

[0114] Step S306, determining whether the qualified ratio is within a second preset range, that is, determining whether the qualified ratio is not less than 95%;

[0115] If not, execute step S307;

[0116] If yes, go to step S308;

[0117] Step S307: determine that the trained model does not meet the requirements, and jump to step S309;

[0118] Step S308: determine whether the trained model meets the requirements and output the trained model.

[0119] S309, inputting the sample set with increased sample quantity into the neural network recognition model to continue training, and executing step S202 in the previous embodiment;

[0120] It should be noted that, in this embodiment, the first preset range is [0, 0.03]. In other embodiments, the first preset range may be set to other values.

[0121] It should be noted that, in this embodiment, the second preset range is [0.95, 1]. In other embodiments, the second preset range may be set to other values.

[0122] The 30 groups of data that did not participate in the training in the total sample set are used to form a validation sample set. The qualified proportion of qualified samples in the validation sample set is counted. According to the qualified proportion, it is determined whether the trained model meets the requirements. The overall accuracy of the training model under multiple sample data can be judged to avoid sample errors.

[0123] Example 5

[0124] In another embodiment of the present invention, the training process in the neural network recognition model specifically includes:

[0125] First, 70 randomly selected training sample sets are input, and the neural network recognition model is used to calculate the input data to obtain the output value of the model; the purpose of neural network learning is to continuously correct the weights of the neurons connected to each other within the neural network through the error between the actual output value of the neural network and the target value of the inversion object, and to approach the target by continuously reducing the error.

[0126] At the same time, the back propagation algorithm is used to transfer the corrected weights to each layer of neurons layer by layer, and the errors are corrected iteratively to continuously reduce the sum of squared errors.

[0127] After the back propagation algorithm reaches the preset upper limit of the number of iterations, it is determined whether the sum of squared errors between the model's output value and the corresponding target value is within a preset range;

[0128] If not, it indicates that after the preset upper limit of the number of iterations, the sum of the squares of the errors between the model output value and the corresponding target value is not within the preset range, and the model accuracy still does not meet the requirements. At this time, the number of hidden layers of the neural network recognition model is increased, and the training sample set in the sample set is re-input into the neural network recognition model to be trained for training, and the sum of the squares of the errors is re-judged against the preset range until the sum of the squares of the errors between the model output value and the corresponding target value is within the preset range;

[0129] When the sum of squared errors between the model's output value and the corresponding target value is within the preset range, model training is completed.

[0130] During model training, if the sum of squared errors still does not meet the requirements after reaching the specified number of iterations, the number of hidden layers is increased and the network training is repeated until the sum of squared errors between the model's output value and the corresponding target value is within the preset range, the model meets the preset accuracy requirements, and the model training is completed.

[0131] Example 6

[0132] In another embodiment provided by the present invention, referring to FIG4 , which is a schematic diagram of the structure of a neural network recognition model provided by an embodiment of the present invention;

[0133] The neural network recognition model adopts the back-propagation neural network. The calculation principle of the back-propagation neural network model is to use function approximation to train the model. It has strong nonlinear mapping capabilities and flexible network structure. At present, the network theory is relatively mature and suitable for the construction of multi-factor input models.

[0134] Figure 3 includes the input layer LayerL1, the hidden layer LayerL2 and the output layer LayerL3; the input data x1, x2 and x3 of the input layer are transmitted through the hidden layer neurons to obtain the activation value a1 respectively. (2) 、a2 (2) and a3 (2) , and then output by the output layer.

[0135] Among them, when constructing the neural network recognition model, the Sigmoid function is used as the training function, the tansig function is used as the hidden layer transfer function, and the purelin function is used as the output layer transfer function.

[0136] The Sigmoid function expression is: Through the hidden layer and Sigmoid activation function, the nonlinear separable learning of the neural network recognition model can be completed.

[0137] Among them, x is the number of nodes, C is the number of output nodes, and the Softmax function can be used to convert the output values ​​of multiple categories into a probability distribution ranging from [0, 1] and 1.

[0138] Example 7

[0139] In another embodiment provided by the present invention, in the electrochemical lithium extraction process, the factors affecting the time taken for the electrode plate to be completely intercalated with lithium include temperature T, the concentration C of lithium ions in the brine, the capacity S of the lithium-absorbing material coated on the electrode plate, the ion diffusion coefficient f of the lithium-absorbing material, the constant current supply current I and the constant voltage supply voltage V. Obtaining the influencing factors as working indicator parameters can achieve precise control of lithium extraction and deintercalation in electrochemical directional cycles.

[0140] Example 8

[0141] In another embodiment provided by the present invention, when constructing the lithium insertion time calculation model, the input layer of the neural network recognition model is set with 6 nodes, the hidden layer is set with 5 nodes, and the output layer is set with 1 node; the process of obtaining the training sample set specifically includes:

[0142] When selecting lithium extraction electrode plates made of several different materials for single lithium extraction experiments, the index parameters used in each group of experiments and the corresponding complete lithium insertion time are recorded; the recorded first index parameter is used as the input data of the sample set, and the recorded complete lithium insertion time is used as the target value corresponding to the input data in the sample set; there are 100 groups of sample set data collected through the experiment, 70 groups are randomly selected from the sample set data to form a training sample set, and the remaining 30 groups of sample data in the sample set are used as verification sample sets.

[0143] The experimental results of lithium extraction electrode plates of different materials under different index parameters are selected as the sample set to improve the data breadth of the sample set, which can improve the calculation accuracy of the model under lithium extraction electrode plates of different materials.

[0144] Embodiment 9

[0145] In another embodiment provided by the present invention, 25 groups of lithium extraction electrode plates containing lithium titanate material, lithium phosphate material, lithium silicate material and lithium manganate material are selected for a single lithium extraction experiment. During the experiment, the temperature T, the concentration C of lithium ions in the brine, the capacity S of the lithium absorbing material coated on the electrode plate, the ion diffusion coefficient f of the lithium absorbing material, the constant current supply current I and the constant voltage supply voltage V and other index parameters in the lithium extraction process of each group are recorded, and the complete lithium insertion time used for complete lithium insertion is recorded; a sample set is formed with the temperature T, the concentration C of lithium ions in the brine, the capacity S of the lithium absorbing material coated on the electrode plate, the ion diffusion coefficient f of the lithium absorbing material, the current I and the voltage V as inputs and the complete lithium insertion time as output, and the recorded sample set is used to train a lithium insertion time calculation model.

[0146] Example 10

[0147] In another embodiment provided by the present invention, when the target model is the reducing agent calculation model, the input layer of the neural network recognition model is set with 6 nodes, the hidden layer is set with 5 nodes, and the output layer is set with 1 node. The process of obtaining the training sample set specifically includes:

[0148] 100 groups of lithium-extraction electrode plates with different reducing agent dosages were taken for single lithium extraction experiments. The 100 groups of experiments were divided into experimental groups with 0.1 mol, 0.2 mol, 0.3 mol...10 mol respectively. The index parameters and corresponding reaction time used in each group of experiments when several different reducing agent dosages were selected were recorded; the recorded index parameters and reaction time were used as input data of the sample set, and the recorded reducing agent dosage was used as the target value corresponding to the input data in the sample set; there were 100 groups of sample set data collected through the experiment, 70 groups were randomly selected from the sample set data to form a training sample set, and the remaining 30 groups of sample data in the sample set were used as a verification sample set.

[0149] The experimental results of lithium-extraction electrode plates with different reducing agent dosages under different index parameters are selected as the sample set to improve the data breadth of the sample set, which can improve the calculation accuracy of the model under lithium-extraction electrode plates of different materials.

[0150] When the solution of the present invention is specifically implemented, the temperature T, the concentration C of lithium ions in the brine, the capacity S of the lithium absorbing material coated on the electrode plate, the ion diffusion coefficient f of the lithium absorbing material, the constant current supply current I, and the constant voltage supply voltage V are obtained as working index parameters. The specific working index parameters are shown in Table 1:

[0151] Table 1 Working index parameters

[0152] The lithium insertion time calculation model calculates that the time for the electrode plate to be completely lithium-intercalated is t1 = 6.4h, and the time for the lithium-intercalation slot electrode plate to be delithiated is t2 = 4.9h. The calculated time difference Δt = 1.5h, and the amount of reducing agent required is calculated to be 2.13 mol by the reducing agent calculation model.

[0153] Based on the problem that the existing electrochemical lithium extraction technology has a lithium insertion rate that is slower than the lithium de-insertion rate, resulting in incomplete lithium insertion, and the problem that the capacity of the lithium insertion electrode plate becomes smaller and smaller after multiple lithium extraction operations, and the efficiency becomes slower and slower, the present invention proposes an electrochemical directional cycle lithium extraction and de-insertion balance method. Through model calculation, the lithium de-insertion time of the lithium de-insertion tank electrode plate corresponding to when the electrode plate of the lithium extraction tank is completely lithium-intercalated is quantified, so as to make targeted adjustments and achieve relative balance, thereby improving the lithium extraction efficiency.

[0154] Example 11

[0155] In another embodiment provided by the present invention, 100 groups of reducing agents are taken for experiments, and the recorded indicator parameters include the temperature T, constant current supply current I and constant voltage supply voltage V during the experimental reaction process. The second indicator parameter including temperature T, constant current supply current I and constant voltage supply voltage V and the reaction time of the experiment are used as model inputs, and the reducing agent dosage is used as the model output to perform reducing agent dosage calculation model training.

[0156] Example 12

[0157] 5 is a schematic diagram of the structure of an electrochemical directional cycle lithium extraction and deintercalation balance device provided by an embodiment of the present invention, wherein the device includes: a data acquisition module, a duration calculation module, a dosage calculation module and a balance control module;

[0158] The data acquisition module is used to extract and deintercalate lithium according to the working index parameters set for the electrolytic cell, and calculate the delithiation time required for the lithium intercalation cell electrode plate;

[0159] The duration calculation module is used to input the working index parameters into the pre-trained lithium insertion duration calculation model, and output the lithium insertion duration corresponding to complete lithium insertion;

[0160] The dosage calculation module is used to calculate the time difference according to the lithium de-lithiation time and the lithium insertion time, input the time difference and the working index parameter into the pre-trained reducing agent calculation model, and output the reducing agent dosage required for the lithium insertion slot electrode plate;

[0161] The balance control module is used to control the balance of lithium insertion and lithium removal according to the required reducing agent dosage.

[0162] The electrochemical directional cycle lithium extraction and deintercalation balance device provided in this embodiment can perform all the steps and functions of the electrochemical directional cycle lithium extraction and deintercalation balance method provided in any of the above embodiments. The specific functions of the device will not be described in detail here.

[0163] Example 13

[0164] Referring to Figure 6, it is a schematic diagram of the structure of a terminal device provided in an embodiment of the present invention. The terminal device includes: a processor, a memory, and a computer program stored in the memory and executable on the processor, such as a program for lithium extraction and deintercalation balance of an electrochemical directional cycle. When the processor executes the computer program, the steps in each of the above-mentioned embodiments of the method for lithium extraction and deintercalation balance of an electrochemical directional cycle are implemented, such as steps S1 to S4 shown in Figure 1. Alternatively, when the processor executes the computer program, the functions of each module in the above-mentioned device embodiments are implemented.

[0165] Exemplarily, the computer program can be divided into one or more modules, and the one or more modules are stored in the memory and executed by the processor to complete the present invention. The one or more modules can be a series of computer program instruction segments that can perform specific functions, and the instruction segments are used to describe the execution process of the computer program in the electrochemical directional cycle lithium extraction and deintercalation balance device. For example, the computer program can be divided into several modules, and the specific functions of each module have been described in detail in the electrochemical directional cycle lithium extraction and deintercalation balance method provided in any of the above embodiments, and the specific functions of the device will not be repeated here.

[0166] The electrochemical directional cycle lithium extraction and deintercalation balance device can be a computing device such as a desktop computer, a notebook, a PDA and a cloud server. The electrochemical directional cycle lithium extraction and deintercalation balance device can include, but is not limited to, a processor and a memory. Those skilled in the art will understand that the schematic diagram is only an example of an electrochemical directional cycle lithium extraction and deintercalation balance device and does not constitute a limitation of an electrochemical directional cycle lithium extraction and deintercalation balance device. It can include more or fewer components than shown in the figure, or combine certain components, or different components. For example, the electrochemical directional cycle lithium extraction and deintercalation balance device can also include input and output devices, network access equipment, buses, etc.

[0167] The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc. The processor is the control center of the electrochemical directional cycle lithium extraction and deintercalation balance device, and utilizes various interfaces and lines to connect the various parts of the entire electrochemical directional cycle lithium extraction and deintercalation balance device.

[0168] The memory can be used to store the computer program and / or module, and the processor realizes the various functions of the electrochemical directional cycle lithium extraction and deintercalation balance device by running or executing the computer program and / or module stored in the memory, and calling the data stored in the memory. The memory may mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, an application required for at least one function (such as a sound playback function, an image playback function, etc.); the data storage area can store data created according to the use of the mobile phone (such as audio data, a phone book, etc.). In addition, the memory may include a high-speed random access memory, and may also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), at least one disk storage device, a flash memory device, or other volatile solid-state storage device.

[0169] Wherein, if the module integrated in the electrochemical directional cycle lithium extraction and deintercalation balance device is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present invention implements all or part of the process in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. Wherein, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electrical carrier signal, telecommunication signal and software distribution medium, etc.

[0170] It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention, and these improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A method for balancing the extraction and deintercalation of lithium by electrochemical directional circulation, characterized in that: The method comprises: Lithium is extracted and deintercalated according to the working index parameters set for the electrolytic cell, and the time required for deintercalation of the electrode plates of the lithium intercalation cell is calculated; Inputting the working index parameter into a pre-trained lithium insertion time calculation model, and outputting the lithium insertion time corresponding to complete lithium insertion; Calculate the time difference according to the lithium de-lithiation time and the lithium insertion time, input the time difference and the working index parameter into the pre-trained reducing agent calculation model, and output the reducing agent dosage required for the lithium insertion slot electrode plate; The balance of lithium insertion and delithiation is controlled by the reducing agent according to the required reducing agent dosage.

2. The electrochemical directional cycle lithium extraction and deintercalation balance method according to claim 1, characterized in that: The training process of the lithium insertion time calculation model or the reducing agent calculation model specifically includes: Obtain a training sample set including input data and target values; The training sample set is input into the neural network recognition model to be trained for training, and the model parameters are adjusted by the gradient descent algorithm according to the preset training accuracy value, training speed and number of cycles for training; The trained model is used as a target model, wherein the target model is specifically the lithium insertion time calculation model or the reducing agent calculation model; Among them, the number of input layer nodes, the number of hidden layer nodes and the number of output layer nodes of the neural network recognition module are determined by the number of work indicator parameters in the input data.

3. The electrochemical directional cycle lithium extraction and deintercalation balance method according to claim 2, characterized in that: Before using the trained model as the target model, the method further includes: Input the input data in the obtained verification sample set into the trained neural network recognition model to obtain the model output value corresponding to the input data; Calculate the relative error between the model output value of the input data and the corresponding target value in the validation sample set; The trained model is verified based on the calculated relative error, and the qualified trained model is output.

4. The electrochemical directional cycle lithium extraction and deintercalation balance method according to claim 3, characterized in that: The step of verifying the trained model according to the calculated relative error and outputting a verified qualified training model specifically includes: The qualified proportion of input data within the first preset range of statistical relative errors to all input data in the verification sample set; When the statistically qualified ratio is not within the second preset range, it is determined that the trained model does not meet the requirements, the number of samples in the training sample set is increased, and the training sample set is input into the neural network recognition model to continue training; When the statistically qualified proportion is within the second preset range, it is determined that the trained model meets the requirements and the trained model is output.

5. The electrochemical directional cycle lithium extraction and deintercalation balance method according to claim 2, characterized in that: The step of inputting the training sample set into the neural network recognition model to be trained for training specifically includes: A neural network recognition model is used to calculate the input data in the training sample set to obtain the output value of the model, and the weights of the neurons connected inside the neural network recognition model are corrected based on the error between the output value and the corresponding target value; a back propagation algorithm is used to transfer the corrected weights to each layer of neurons layer by layer, and the error is corrected iteratively until the number of iterations reaches a preset value or the sum of the squares of the error between the output value of the model and the corresponding target value is within a preset range; When the number of iterations reaches the preset value, and the sum of squares of the errors between the output value of the model and the corresponding target value is not within the preset range, the number of hidden layers of the neural network recognition model is increased, and the training sample set is re-input into the neural network recognition model to be trained for training until the sum of squares of the errors between the output value of the model and the corresponding target value is within the preset range; When the sum of squared errors between the model's output value and the corresponding target value is within a preset range, model training is completed.

6. The electrochemical directional cycle lithium extraction and deintercalation balance method according to claim 2, characterized in that: The neural network recognition model adopts a back propagation neural network; The neural network recognition model uses the Sigmoid function as a training function; The neural network recognition model adopts the tansig function as the hidden layer transfer function; The neural network recognition model adopts purelin function as the output layer transfer function.

7. The electrochemical directional cycle lithium extraction and deintercalation balance method according to claim 1, characterized in that: The working index parameters include temperature, lithium ion concentration in brine, capacity of lithium absorbing material coated on electrode plates, particle diffusion coefficient of lithium absorbing material, constant current supply current and constant voltage supply voltage.

8. The electrochemical directional cycle lithium extraction and deintercalation balance method according to claim 2, characterized in that: When the target model is a lithium insertion time calculation model, the input layer of the neural network recognition model is set with 6 nodes, the hidden layer is set with 5 nodes, and the output layer is set with 1 node; The process of obtaining the training sample set of the target model specifically includes: Record the first index parameter used in each group of experiments and the corresponding complete lithium insertion time when using several lithium extraction electrode plates of different materials for single lithium extraction experiments; Using the recorded first indicator parameter as input data of the sample set, and using the recorded complete lithium insertion time as the target value corresponding to the input data in the sample set; The sample set is divided according to a preset ratio to obtain a training sample set and a verification sample set.

9. The electrochemical directional cycle lithium extraction and deintercalation balance method according to claim 8, characterized in that: Lithium-extracting electrode plates made of different materials include lithium titanate material lithium-extracting electrode plates, lithium phosphate material lithium-extracting electrode plates, lithium silicate material lithium-extracting electrode plates and lithium manganate material lithium-extracting electrode plates; The first index parameters recorded include temperature, lithium ion concentration in brine, capacity of lithium absorbing material coated on the electrode plate, particle diffusion coefficient of lithium absorbing material, constant current supply current and constant voltage supply voltage.

10. The electrochemical directional cycle lithium extraction and deintercalation balance method according to claim 2, characterized in that: When the target model is the reducing agent calculation model, the input layer of the neural network recognition model is set with 4 nodes, the hidden layer is set with 3 nodes, and the output layer is set with 1 node; The process of obtaining the training sample set of the target model specifically includes: Record the second index parameter and the corresponding reaction time used in each group of experiments when using several lithium extraction electrode plates with different reducing agent dosages for single lithium extraction experiments; Using the recorded second indicator parameter and reaction time as input data of the sample set, and using the recorded reducing agent dosage as the target value corresponding to the input data in the sample set; The sample set is divided according to a preset ratio to obtain a training sample set and a verification sample set.

11. The electrochemical directional cycle lithium extraction and deintercalation balance method according to claim 10, characterized in that: The second index parameters recorded include temperature, constant current supply current and constant voltage supply voltage.

12. An electrochemical directional cycle lithium extraction and deintercalation balance device, characterized in that: The device comprises: A data acquisition module is used to extract and deintercalate lithium according to the working index parameters set for the electrolytic cell, and calculate the time required for deintercalation of the electrode plates of the lithium intercalation cell; A duration calculation module, used to input the working index parameters into a pre-trained lithium insertion duration calculation model, and output the lithium insertion duration corresponding to complete lithium insertion; The dosage calculation module is used to calculate the time difference according to the lithium de-lithiation time and the lithium insertion time, input the time difference and the working index parameter into the pre-trained reducing agent calculation model, and output the reducing agent required for the lithium insertion slot electrode plate. dose; The balance control module is used to control the balance of lithium insertion and lithium deintercalation according to the required reducing agent dosage.

13. A terminal device, characterized in that: It comprises a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein when the processor executes the computer program, the electrochemical directional cycle lithium extraction and deintercalation balance method as described in any one of claims 1 to 11 is implemented.

14. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored computer program, wherein when the computer program is executed, the device where the computer-readable storage medium is located is controlled to execute the electrochemical directional cycle lithium extraction and deintercalation balance method as described in any one of claims 1 to 11.

Images