An adjustable electrochemical deintercalation and lithium extraction system and its application

Abstract

The present application provides an adjustable electrochemical deintercalation and lithium extraction system and its application, wherein the electrochemical deintercalation and lithium extraction system includes a voltage monitor, a neural network model and an automatic thermostat; the voltage monitor obtains the real-time voltage of the cathode plate, inputs the real-time voltage of the cathode plate into the neural network model, obtains the corresponding temperature value, inputs the temperature value into the automatic thermostat, and the automatic thermostat automatically adjusts to reach the required temperature so that the operating voltage of the anode electrode plate is consistent with the operating voltage of the cathode electrode plate.

Description

Technical Field

[0001] The present application belongs to the technical field of lithium resource extraction and relates to an adjustable electrochemical deintercalation and lithium extraction system and its application. Background Art

[0002] With the rapid expansion of the electric vehicle, information technology, and grid energy storage industries, demand for lithium raw materials has increased dramatically. Lithium resources primarily come from minerals such as montmorillonite, petalite, and spodumene; salt lake brines, including terrestrial brines, geothermal brines, and oilfield brines; and seawater. Lithium is abundant in salt lake brines, and extracting lithium from them is more cost-effective than extracting it from lithium ore. Therefore, in recent years, lithium salts produced from salt lake brines have accounted for over 85% of all lithium products.

[0003] Electrodialysis is a green process technology and a new research direction for lithium extraction from salt lakes in recent years. However, its practical application still faces certain difficulties. The method is relatively expensive, and the overall lithium recovery rate needs to be improved. In our actual electrochemical deintercalation process, an oxidation reaction occurs at the anode, deintercalating lithium from the lithium iron phosphate, while a reduction reaction occurs at the cathode, intercalating lithium from the brine into the iron phosphate. In an ideal reaction, the cathode intercalates lithium ions simultaneously with the anode's deintercalation of lithium ions, and the deintercalation rates of the two should be consistent. However, in actual reactions, the cathode intercalation process is significantly affected by the viscosity and ion concentration of the brine. This results in the cathode intercalation process being much slower than the anode deintercalation process, resulting in an inconsistency between the operating rates and the mismatch between the deintercalation and insertion capacities of the anode and cathode. Furthermore, because the two electrodes need to constantly switch polarity, the types of electrode materials used in the anode and cathode must be as consistent as possible. Therefore, during the lithium extraction process, it is impossible to design the anode and cathode to match their capacities, making this technical problem unsolvable by simply designing different capacities for the electrode plates.

[0004] CN115818801A discloses a method for extracting lithium from salt lake brine. The lithium extraction process is divided into two steps: insertion and removal. During insertion, lead is used as the anode, a conductive substrate coated with an ion sieve is used as the cathode, and an anion membrane separates the anode and cathode. A sulfate solution is added to the anode, and salt lake brine is added to the cathode. During removal, the polarity of the cathode and cathode electrodes in the lithium extraction process is reversed. At the same time, the solution on the lead electrode side of the previous cycle remains unchanged, and the brine is replaced with fresh sulfate solution as the supporting electrolyte. Then, power is applied again to release the lithium sulfate.

[0005] CN110442178A discloses an experimental salt lake lithium extraction power supply system including a controller, a switching power supply, an electronic commutation device, an electrolytic cell and an acquisition module; the controller is connected to the electronic commutation device and the acquisition module respectively, and controls the switching and / or commutation action of the electronic commutation device according to the acquisition parameters output by the acquisition module; the switching power supply is connected to the electronic commutation device, and changes the output power polarity of the switching power supply through the electronic commutation device; the acquisition module is used to obtain the acquisition parameters from the electrolytic cell.

[0006] The above scheme can improve the lithium recovery efficiency by adjusting the current to make the capacity of lithium removal and lithium insertion of the anode and cathode more matched. However, the method is cumbersome to operate and has poor continuity. Impurities are easily introduced during the application process, and the actual application efficiency is low. Summary of the Invention

[0007] The present application provides an adjustable electrochemical deintercalation and lithium extraction system and its application, which adopts intelligent intervention in electrochemical lithium extraction, so that the rate of ion adsorption of the electrode plates on both sides is balanced, fully ensuring that the adsorption capacity of the electrode is maximized and the efficiency of lithium extraction is improved.

[0008] This application adopts the following technical solutions:

[0009] In a first aspect, the present application provides an adjustable electrochemical deintercalation and lithium extraction system, the electrochemical deintercalation and lithium extraction system comprising a voltage monitor, a neural network model, and an automatic thermostat electrically connected in sequence;

[0010] The voltage monitor is used to obtain the real-time voltage of the cathode electrode plate of the electrolytic cell of the lithium extraction device, and input the real-time voltage of the cathode electrode plate into the neural network model to obtain the corresponding temperature value;

[0011] The neural network model is used to input the temperature value into the thermostat;

[0012] The automatic thermostat is used to automatically adjust the temperature to the required temperature so that the working voltage of the anode electrode plate of the lithium extraction device is consistent with the working voltage of the cathode electrode plate.

[0013] The system described in this application can accurately monitor the operating voltage of the electrode plate and adjust the anode electrode plate based on the real-time operating voltage of the cathode. By adjusting the temperature, the voltage of the anode electrode plate is changed to keep the operating voltage of the two electrodes consistent, thereby achieving a balance between the rates of lithium absorption and desorption. By adjusting the rates of lithium ion insertion and desorption at both ends of the anode and cathode to keep consistent, this application can fully ensure that the adsorption capacity of the electrode is maximized and the efficiency of lithium extraction is improved.

[0014] In one or more embodiments, the neural network model includes an input layer, a hidden layer, and an output layer.

[0015] In one or more embodiments, the input layer includes I nodes, where I≥1.

[0016] In one or more embodiments, the hidden layer includes J nodes, where J≥1.

[0017] In one or more embodiments, the output layer includes I nodes, where I≥1.

[0018] In one or more embodiments, the process of building the neural network model includes:

[0019] Take N groups of samples, input I training learning samples into the input layer, input the corresponding I output samples into the output layer, calculate the training model in the hidden layer, input the remaining NI output samples into the output layer through the training model, calculate the corresponding NI inversion learning samples, calculate the error between the inversion learning samples and the actual samples to continuously correct the weights of the neurons connected inside the neural network, and approach the target by continuously reducing the error. The error will be passed to each layer of neurons layer by layer through the backpropagation learning method, and the error will be continuously corrected and adjusted to minimize the sum of squared errors.

[0020] In one or more embodiments, the error = (T (反演) -T (实际) ) / T (实际) .

[0021] In constructing a neural network model, this application uses the "Sigmoid" function as the learning and training function, and the "Softmax" function as the output layer transfer function of the model. The function expression is as follows:

[0022]

[0023]

[0024] In one or more embodiments, the neural network model is trained before use.

[0025] In one or more embodiments, the training includes sequentially acquiring total sample set data, selecting part of the data to train the model, using the remaining data to verify the trained model, and obtaining a trained neural network model after the recognition error is less than a preset error.

[0026] In one or more embodiments, after determining the model structure, the input value of the model is determined to be voltage, the number of hidden layers is 1, and the output value is temperature. The neural model is trained until the sum of square errors between the neural network output value and the target value is less than 3%.

[0027] In a second aspect, the present application provides a method for electrochemical deintercalation and lithium extraction, which is carried out by the electrochemical deintercalation and lithium extraction system as described in the first aspect.

[0028] In one or more embodiments, the method comprises the following steps:

[0029] (1) Installing an adjustable system consisting of a voltage monitor, a device containing a trained neural network model, and an automatic thermostat at a suitable location on the electrolytic cell;

[0030] (2) The voltage monitor monitors the voltage of the cathode electrode plate of the electrolytic cell in real time and inputs the real-time data into the system model. The model calculates the temperature required for the anode electrode plate to reach this working voltage. The automatic temperature regulator receives the instruction and automatically adjusts to reach the required temperature so that the working voltage of the anode electrode plate is consistent with the working voltage of the cathode electrode plate.

[0031] In one or more embodiments, the present application adopts a LiFePO4-FePO4 electrode system for lithium extraction reaction, and the entire system consists of a voltage monitor, a neural network model and an automatic thermostat.

[0032] Specific usage methods include:

[0033] 1. Data Acquisition

[0034] The normal working state of the anode electrode plate is simulated, and a gradual temperature increase method is adopted to obtain the voltage change data with temperature during the entire working process; at the same time, the real-time voltage data in the voltage monitor is recorded.

[0035] 2. System model training

[0036] S1. Combining the voltage and temperature change data obtained above into a total sample set;

[0037] S2. Randomly select a preset number of training sample voltage and temperature data from the total sample set to form a training sample set, input the data of the training sample set into the neural network recognition model to be trained, and output the trained neural network recognition model;

[0038] S3, the remaining data of the total sample set that did not participate in the training are used to form a verification sample set, the voltage values ​​of the verification sample set are input into the trained neural network recognition model, and the corresponding temperature values ​​are obtained, and it is determined whether the recognition error of the output temperature relative to the corresponding temperature test value is less than a preset error value. If so, a verified neural network recognition model is obtained; otherwise, the preset number of training samples in step S2 is increased and step S2 is repeated;

[0039] 3. System application in electrolytic cells

[0040] The real-time data output by the voltage monitor on the cathode electrode plate is input into the system model. The model calculates the temperature required for the anode electrode plate to reach this working voltage. The automatic temperature regulator receives the instruction and automatically adjusts to reach the required temperature so that the working voltage of the anode electrode plate is consistent with the working voltage of the cathode electrode plate.

[0041] This application has the following effects:

[0042] (1) The system described in this application can accurately monitor the working voltage of the electrode plate and adjust the anode electrode plate based on the real-time working voltage of the cathode. By adjusting the temperature, the voltage of the anode electrode plate is changed to keep the working voltage of the two electrodes consistent, thereby achieving a balance between the rates of lithium absorption and desorption to a certain extent. By adjusting the rates of lithium ion insertion and desorption at both ends of the anode and cathode to keep consistent, this application can fully ensure that the adsorption capacity of the electrode is maximized and the efficiency of lithium extraction is improved.

[0043] (2) The present application adopts intelligent intervention electrochemical lithium extraction by establishing a relationship model between voltage and temperature, inputting the model according to the tested voltage, and adjusting the temperature through a temperature regulating device according to the temperature value obtained by the model. Adjusting the temperature can change the electrode resistance, and then change the voltage of lithium removal / lithium insertion, so that the anode and cathode voltages reach a balance, thereby improving the lithium extraction efficiency. Compared with chemical intervention, it can reduce the use of chemical raw materials and the introduction of impurities. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 This is a visualization diagram of the adjustable electrochemical deintercalation and lithium extraction system described in some embodiments of the present application.

[0045] Figure 2 This is a flowchart of neural network model training in some embodiments of the present application. DETAILED DESCRIPTION

[0046] The technical solution of the present application is further described below through specific implementation methods. Those skilled in the art should understand that the embodiments are only used to help understand the present application and should not be regarded as specific limitations of the present application.

[0047] Example 1

[0048] This embodiment provides an adjustable electrochemical deintercalation and lithium extraction system. The visualization diagram of the adjustable electrochemical deintercalation and lithium extraction system is as follows: Figure 1 As shown, the adjustable electrochemical deintercalation and lithium extraction system includes a voltage monitor, a neural network model and an automatic thermostat;

[0049] The voltage monitor is used to obtain the real-time voltage of the cathode electrode plate of the electrolytic cell of the lithium extraction device, and the real-time voltage of the cathode electrode plate is input into the neural network model to obtain the corresponding temperature value.

[0050] The neural network model is used to input temperature values ​​into a thermostat.

[0051] The automatic thermostat is used to automatically adjust the temperature to the required temperature so that the working voltage of the anode electrode plate of the lithium extraction device is consistent with the working voltage of the cathode electrode plate.

[0052] The system described in this application can accurately monitor the operating voltage of the electrode plate and adjust the anode electrode plate based on the real-time operating voltage of the cathode. By adjusting the temperature, the voltage of the anode electrode plate is changed to keep the operating voltage of the two electrodes consistent, thereby achieving a balance between the rates of lithium absorption and desorption. By adjusting the rates of lithium ion insertion and desorption at both ends of the anode and cathode to keep consistent, this application can fully ensure that the adsorption capacity of the electrode is maximized and the efficiency of lithium extraction is improved.

[0053] The neural network model training flow chart is as follows Figure 2 shown.

[0054] The neural network model described in this application includes an input layer, a hidden layer, and an output layer. The input layer includes I nodes, I≥1, the hidden layer includes J nodes, J≥1, and the output layer includes I nodes, I≥1. The construction process of the neural network model includes:

[0055] Take N groups of samples, input I training learning samples into the input layer, input the corresponding I output samples into the output layer, calculate the training model in the hidden layer, input the remaining NI output samples into the output layer through the training model, calculate the corresponding NI inversion learning samples, calculate the error between the inversion learning samples and the actual samples to continuously correct the weights of the neurons connected inside the neural network, and approach the target by continuously reducing the error. The error will be passed to each layer of neurons layer by layer through the backpropagation learning method, and the error will be continuously corrected and adjusted to minimize the sum of squared errors.

[0056] In one or more embodiments, the error = (T (反演) -T (实际) ) / T (实际) .

[0057] In constructing a neural network model, this application uses the "Sigmoid" function as the learning and training function, and the "Softmax" function as the output layer transfer function of the model. The function expression is as follows:

[0058]

[0059]

[0060] In one or more embodiments, the neural network model is trained before use.

[0061] In one or more embodiments, the training includes sequentially acquiring total sample set data, selecting part of the data to train the model, using the remaining data to verify the trained model, and obtaining a trained neural network model after the recognition error is less than a preset error.

[0062] In one or more embodiments, after determining the model structure, the input value of the model is determined to be voltage, the number of hidden layers is 1, and the output value is temperature. The neural model is trained until the sum of square errors between the neural network output value and the target value is less than 3%.

[0063] The construction and training methods of the neural network model are as follows:

[0064] Step S1: Install a real-time voltage monitor on a normally working anode electrode, increase the temperature of the electrode at a rate of 0.1°C / min, and record the change in the voltage of the electrode with temperature. The voltage and temperature data pairs are used as the total sample set required for the model. The sample set has a data volume of 200 data pairs.

[0065] Step S2, randomly selecting 130 preset training sample numbers of voltage and temperature data from the total sample set to form a training sample set, first inputting 130 voltage data, and knowing the corresponding 130 output temperature data, and continuously correcting the weights of each neuron connected inside the neural network by the error between the actual output temperature data of the neural network and the measured value of the inversion object (the measured value of the inversion object is the temperature value corresponding to each training sample voltage), and approaching the target by continuously reducing the error. At the same time, the back propagation learning method will be transmitted layer by layer to each layer of neurons, and the error will be continuously corrected and adjusted to minimize the sum of squared errors, thereby obtaining a neural network model, and using 130 random data in the total sample to train the neural network recognition model. After the model training is successful, the trained weights and thresholds need to be assigned to the recognition network as initial values ​​and incorporated into the system to achieve the purpose of adaptive control;

[0066] Step S3: 70 data pairs of the total sample set that did not participate in the training are used to form a validation sample set. The voltage data of the validation sample set is input into the trained neural network recognition model to obtain the corresponding temperature. At the same time, it is determined whether the recognition error of the output temperature relative to the corresponding temperature test value is less than a preset error value. If so, a validated neural network recognition model is obtained. Otherwise, the preset number of training samples in step S2 is increased and step S2 is repeated.

[0067] Error = (T (反演) -T (实际) ) / T (实际)

[0068] If the error of the trained neural network calculation model is less than 3%, the model is established and a fast calculation based on the voltage to obtain the corresponding temperature can be performed.

[0069] This embodiment provides a method for electrochemical deintercalation and extraction of lithium, the method comprising the following steps:

[0070] (1) Installing an adjustable system consisting of a voltage monitor, a device containing a trained neural network model, and an automatic thermostat at a suitable location on the electrolytic cell;

[0071] (2) The voltage monitor monitors the voltage of the cathode electrode plate of the electrolytic cell in real time and inputs the real-time data into the system model. The model calculates the temperature required for the anode electrode plate to reach this working voltage. The automatic temperature regulator receives the instruction and automatically adjusts to reach the required temperature so that the working voltage of the anode electrode plate is consistent with the working voltage of the cathode electrode plate.

[0072] Specifically include:

[0073] 1. Data Acquisition

[0074] The normal working state of the anode electrode plate is simulated, and a gradual temperature increase method is adopted to obtain the voltage change data with temperature during the entire working process; at the same time, the real-time voltage data in the voltage monitor is recorded.

[0075] 2. System model training

[0076] S1. Combining the voltage and temperature change data obtained above into a total sample set;

[0077] S2. Randomly select a preset number of training sample voltage and temperature data from the total sample set to form a training sample set, input the data of the training sample set into the neural network recognition model to be trained, and output the trained neural network recognition model;

[0078] S3, the remaining data of the total sample set that did not participate in the training are used to form a verification sample set, the voltage values ​​of the verification sample set are input into the trained neural network recognition model, and the corresponding temperature values ​​are obtained, and it is determined whether the recognition error of the output temperature relative to the corresponding temperature test value is less than a preset error value. If so, a verified neural network recognition model is obtained; otherwise, the preset number of training samples in step S2 is increased and step S2 is repeated;

[0079] 3. System application in electrolyzer

[0080] The real-time data output by the voltage monitor on the cathode electrode plate is input into the system model. The model calculates the temperature required for the anode electrode plate to reach this working voltage. The automatic temperature regulator receives the instruction and automatically adjusts to reach the required temperature so that the working voltage of the anode electrode plate is consistent with the working voltage of the cathode electrode plate.

[0081] System testing:

[0082] (1) A LiFePO4-FePO4 electrode system is used for lithium extraction reaction, and an adjustable system consisting of a voltage monitor, a device containing a trained neural network model, and an automatic thermostat is installed at a suitable position on the electrolytic cell;

[0083] (2) The voltage monitor monitors the voltage of the cathode electrode plate in real time and inputs the real-time data into the system model. The model calculates the temperature required for the anode electrode plate to reach this working voltage. The automatic temperature regulator receives the instruction and automatically adjusts to reach the required temperature so that the working voltage of the anode electrode plate reaches the same level as the working voltage of the cathode electrode plate.

[0084] Some of the results of this example are shown in the following table:

[0085] Cathode electrode plate real-time voltage (V) The temperature that the anode electrode plate needs to reach (℃) 0.3 25.3 0.4 25.9 0.5 26.7 0.6 28.0

[0086] When the electrochemical deintercalation method is working normally, the brine has a certain viscosity and is affected by the concentration of various ions in the brine, which will cause the rates of lithium absorption by the cathode and lithium desorption by the anode to be inconsistent; the electrochemical deintercalation and lithium extraction system described in the present application can accurately monitor the working voltage of the electrode plate and adjust the anode electrode plate based on the real-time working voltage of the cathode. The voltage of the anode electrode plate is changed by adjusting the temperature, and the working voltages of the two poles are controlled to remain consistent, thereby achieving a balance between the rates of lithium absorption and lithium desorption to a certain extent.

[0087] The applicant declares that the above is only a specific implementation method of the present application, but the protection scope of the present application is not limited thereto. Technical personnel in the relevant technical field should understand that any changes or substitutions that can be easily thought of by technical personnel in the relevant technical field within the technical scope disclosed in this application fall within the protection scope and disclosure scope of this application.

Claims

1. An adjustable electrochemical deintercalation and lithium extraction system, wherein: The electrochemical deintercalation and lithium extraction system includes a voltage monitor, a neural network model and an automatic thermostat which are electrically connected in sequence; The voltage monitor is used to obtain the real-time voltage of the cathode electrode plate of the electrolytic cell of the lithium extraction device, and input the real-time voltage of the cathode electrode plate into the neural network model to obtain the corresponding temperature value; The neural network model is used to input the temperature value into the thermostat; The automatic thermostat is used to automatically adjust the temperature to the required temperature so that the working voltage of the anode electrode plate of the lithium extraction device is consistent with the working voltage of the cathode electrode plate.

2. The electrochemical deintercalation and lithium extraction system according to claim 1, wherein: The neural network model includes an input layer, a hidden layer and an output layer.

3. The electrochemical deintercalation and lithium extraction system according to claim 2, wherein: The input layer includes I nodes, I≥1; Optionally, the hidden layer includes J nodes, J ≥ 1; Optionally, the output layer includes I nodes, where I≥1.

4. The electrochemical deintercalation and lithium extraction system according to any one of claims 1 to 3, wherein: The construction process of the neural network model includes: Take N groups of samples, input I training learning samples into the input layer, input the corresponding I output samples into the output layer, calculate the training model in the hidden layer, input the remaining NI output samples into the output layer through the training model, calculate the corresponding NI inversion learning samples, calculate the error between the inversion learning samples and the actual samples to continuously correct the weights of the neurons connected inside the neural network, and approach the target by continuously reducing the error. The error will be passed to each layer of neurons layer by layer through the backpropagation learning method, and the error will be continuously corrected and adjusted to minimize the sum of squared errors.

5. The electrochemical deintercalation and lithium extraction system according to claim 4, wherein: The error = (T (反演) -T (实际) ) / T (实际) .

6. The electrochemical deintercalation and lithium extraction system according to any one of claims 1 to 5, wherein: The neural network model is trained before use.

7. The electrochemical deintercalation and lithium extraction system according to claim 6, wherein: The training includes sequentially acquiring data pairs from the total sample set, selecting part of the data to train the model, using the remaining data to verify the trained model, and obtaining a trained neural network model after the recognition error is less than a preset error.

8. The electrochemical deintercalation and lithium extraction system according to claim 7, wherein: After determining the model structure, the input value of the model is determined to be the number of remaining input cycles, the number of hidden layers is 1, the output value is temperature, and the neural model is trained until the sum of square errors between the neural network output and the target value is less than 3%.

9. A method for electrochemical deintercalation and lithium extraction, the method being carried out by the electrochemical deintercalation and lithium extraction system according to any one of claims 1 to 8.

10. The method according to claim 9, comprising the steps of: (1) Installing an adjustable system consisting of a voltage monitor, a device containing a trained neural network model, and an automatic thermostat at a suitable location on the electrolytic cell; (2) The voltage monitor monitors the voltage of the cathode electrode plate of the electrolytic cell in real time and inputs the real-time data into the system model. The model calculates the temperature required for the anode electrode plate to reach this working voltage. The automatic temperature regulator receives the instruction and automatically adjusts to reach the required temperature so that the working voltage of the anode electrode plate is consistent with the working voltage of the cathode electrode plate.

Images