Battery slurry detection device and method

Abstract

The invention discloses a battery slurry detection device and method, and the device comprises an electromagnetic induction sensor which comprises at least two excitation coils reversely wound on the outer side of a slurry pipeline and at least one induction coil wound on the outer sides of the excitation coils; the electric signal sensor is connected with the induction coil and is used for acquiring an electric signal change value of the induction coil under the condition that the battery slurry flows through the electromagnetic induction sensor from the slurry pipeline; the processor is connected with the electric signal sensor and is used for detecting the content of the magnetic metal particles in the battery slurry according to the electric signal change value; and the power supply is connected with the electromagnetic induction sensor, the electric signal sensor and the processor. According to the method, the electromagnetic induction sensor is arranged on the outer side of the slurry pipeline, and when the battery slurry flows through the electromagnetic induction sensor, the content of the magnetic metal particles in the battery slurry can be obtained according to the electric signal change value.

Description

Technical Field

[0001] The present application relates to the field of batteries, and in particular to a battery slurry detection device and method. Background Art

[0002] The performance of lithium-ion batteries is closely related to the quality of the positive electrode material. When there are more magnetic metal particles in the positive electrode material, it will seriously affect the safety performance of the lithium-ion battery. In order to ensure the stable safety performance of lithium-ion batteries and reduce the risk of introducing metal foreign matter into the positive electrode material.

[0003] Currently, the content level of magnetic metal particles is mainly determined by randomly sampling the positive electrode materials and subjecting the samples to inductively coupled plasma testing.

[0004] However, this method currently requires destructive pre-treatment of the sample to extract the magnetic metal impurities in the sample, which leads to sample loss and increases the cost of testing metal foreign matter in the positive electrode material. Summary of the invention

[0005] The present application provides a battery slurry detection device and method, which can improve and reduce the detection cost of the magnetic metal particle content in the battery slurry.

[0006] In a first aspect, the present application provides a battery slurry detection device, the device comprising: An electromagnetic induction sensor, comprising at least two excitation coils reversely wound on the outside of a slurry pipe and at least one induction coil wound on the outside of the excitation coils; an electrical signal sensor connected to the induction coil and used to obtain an electrical signal change value of the induction coil when the battery slurry flows from the slurry pipe through the electromagnetic induction sensor; A processor, connected to the electrical signal sensor, for detecting the content of magnetic metal particles in the battery slurry according to the received electrical signal change value; A power supply connected to the electromagnetic induction sensor, the electrical signal sensor and the processor; An excitation capacitor, wherein the excitation capacitor is connected in parallel with the excitation coil, and after being connected in parallel with the excitation coil, the excitation capacitor is connected to an AC power supply in the power supply to form a second resonant circuit; the excitation capacitor is used to put the second resonant circuit in a resonant state.

[0007] In the present technical solution, an electromagnetic induction sensor is arranged on the outside of the slurry pipe. When the battery slurry flows through the electromagnetic induction sensor, a changing electrical signal will be output. The content of magnetic metal particles in the battery slurry is determined by the change of the electrical signal. In this way, there is no need for sampling and detection, which avoids sample loss and reduces the cost of detecting the magnetic metal particle content.

[0008] In a possible implementation, the method further includes: the power supply includes at least a direct current power supply and an alternating current power supply; The DC power supply is connected to the electrical signal sensor and the induction coil; The AC power source is connected to at least two excitation coils.

[0009] In this technical solution, the induction coil is powered by a DC power supply, which can eliminate the influence of the internal resistance of the induction coil, and the AC power supply is used to generate an AC excitation signal to the excitation coil, thereby improving the sensitivity of the electromagnetic induction sensor.

[0010] In a possible implementation, it also includes an inductive capacitor; The inductive capacitor is connected to the electrical signal sensor, the DC power supply and the inductive coil to form a first resonant circuit, and the inductive capacitor is used to put the first resonant circuit in a resonant state.

[0011] In this technical solution, the induction coil is put into a resonant state by means of the induction capacitor, which can improve the accuracy and efficiency of data collection.

[0012] In a possible implementation manner, the induction coil is wound around a middle portion of at least two excitation coils.

[0013] In this technical solution, when no slurry flows through, the magnetic field strength in the middle part is exactly zero, and the induction coil will not generate an induced electromotive force at this time, which facilitates the subsequent detection of the induced electromotive force change.

[0014] In a possible implementation, the electrical signal sensor includes a signal generator and a signal processor; The signal generator is connected to the excitation coil and is used to output an AC signal to the induction coil; The signal processor is connected to the excitation coil and is used to detect a first voltage across the induction coil under the excitation of the DC power supply and a second voltage across the induction coil under the excitation of the AC signal when the battery slurry flows from the slurry pipe through the electromagnetic induction sensor; The signal processor is further used to determine the electrical signal change value according to the first voltage and the second voltage.

[0015] In this technical solution, the signal processor uses the same channel to time-share detect the voltage across the induction coil under DC excitation and the voltage across the coil under AC excitation, wherein the DC voltage can eliminate the influence of the internal resistance of the induction coil, while the measurement of the AC voltage can detect the inductance value of the induction coil.

[0016] In a possible implementation, it further includes a first switch and a second switch; The first switch is connected in series with the induction coil; The second switch is connected in series with at least two excitation coils.

[0017] In this technical solution, by controlling the access or cutoff of the excitation signal through the switch, the sensitivity amplification effect of the electromagnetic induction sensor can be sensed.

[0018] In a possible implementation, the device further includes a display; The display is connected to the electrical signal sensor and is used to display the change value of the electrical signal of the induction coil.

[0019] In this technical solution, the display can show the real-time change value of the electrical signal. Since there is a positive correlation between the change size of the electrical signal and the content of magnetic metal particles, the content of metal particles can be initially and immediately grasped directly through the displayed change value of the electrical signal, which is conducive to the staff to respond quickly and promptly.

[0020] In a second aspect, the present application provides a battery slurry detection method, comprising: Obtaining a prediction result output by a magnetic metal particle content prediction model to be trained according to a sample value of an electric signal change of the electromagnetic induction sensor when the sample battery slurry flows through the electromagnetic induction sensor, wherein the sample value of the electric signal change includes a magnetic resistance of an induction coil in the electromagnetic induction sensor and an inductance value under the magnetic resistance; Based on the prediction result and the content of magnetic metal particles in the sample battery slurry, adjusting the network parameters in the magnetic metal particle content prediction model to be trained to obtain a trained magnetic metal particle content prediction model; In response to battery slurry flowing through an electromagnetic induction sensor disposed outside the slurry pipeline, obtaining a change value of an electrical signal output by an induction coil of the electromagnetic induction sensor; The content of the magnetic metal particles in the battery slurry is determined according to the electrical signal change value and a pre-trained magnetic metal particle content prediction model. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] The features, advantages and technical effects of exemplary embodiments of the present application will be described below with reference to the accompanying drawings.

[0022] Figure 1 A schematic cross-sectional view of an electromagnetic induction sensor provided in an embodiment of the present application; Figure 2 A schematic diagram of the change of magnetic field strength provided in an embodiment of the present application; Figure 3 A schematic diagram of the structure of a battery slurry detection device provided in an embodiment of the present application; Figure 4 A schematic diagram of the circuit structure of a battery slurry detection device provided in another embodiment of the present application; Figure 5 A schematic diagram of the circuit structure of a battery slurry detection device provided in another embodiment of the present application; Figure 6 A schematic diagram of the circuit structure of a battery slurry detection device provided in another embodiment of the present application; Figure 7 A schematic diagram of the circuit structure of a battery slurry detection device provided in another embodiment of the present application; Figure 8 A schematic diagram of the structure of a signal acquisition unit provided in an embodiment of the present application; Fig. 9 A schematic diagram of the structure of a battery slurry detection system provided in an embodiment of the present application; Fig.10 A schematic diagram of the structure of a magnetic metal particle content prediction model provided in an embodiment of the present application; Fig.11 A schematic diagram of a battery slurry detection process provided in an embodiment of the present application; Fig.12 A schematic diagram of the training process of the magnetic metal particle content prediction model provided in an embodiment of the present application; Fig.13 A schematic diagram of the training steps of the magnetic metal particle content prediction model provided in an embodiment of the present application; Fig.14 A schematic diagram of the algorithm processing flow of the target value network provided in an embodiment of the present application; Fig.15 A schematic diagram of a sample data construction process provided in an embodiment of the present application; Fig.16 Schematic diagram of the model training framework of an embodiment of the present application.

[0023] In the drawings, the drawings are not necessarily drawn to scale. DETAILED DESCRIPTION

[0024] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0025] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by technicians in the technical field to which this application belongs; the terms used herein are only for the purpose of describing specific embodiments and are not intended to limit this application; the terms "including" and "having" in the specification and claims of this application and the above-mentioned figure descriptions and any variations thereof are intended to cover non-exclusive inclusions.

[0026] In the description of the embodiments of the present application, the technical terms "first", "second", etc. are only used to distinguish different objects, and cannot be understood as indicating or implying relative importance or implicitly indicating the number, specific order or primary and secondary relationship of the indicated technical features. In the description of the embodiments of the present application, the meaning of "multiple" is more than two, unless otherwise clearly and specifically defined.

[0027] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0028] In the description of the embodiments of the present application, the term "and / or" is only a description of the association relationship of the associated objects, indicating that there may be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship.

[0029] In the description of the embodiments of the present application, the term "multiple" refers to more than two (including two). Similarly, "multiple groups" refers to more than two groups (including two groups), and "multiple pieces" refers to more than two pieces (including two pieces).

[0030] Battery slurry is an active material used to coat the positive and negative electrodes of batteries. Due to incoming materials, equipment wear and tear, environment, and human operation, there may be magnetic substances (metal particles, Fe, Ni, Zn, Cr, Cu, etc., or alloys or oxides) in the battery slurry. The quality of the battery slurry determines the performance of the final battery. Taking lithium-ion batteries as an example, the performance of lithium batteries is closely related to the quality of the positive electrode materials. When there are more magnetic metal particles in the positive electrode materials, the magnetic metal will first be oxidized at the positive electrode of the battery and then reduced to the negative electrode. When the metal element at the negative electrode accumulates to a certain extent, the hard edges and corners of the deposited metal will pierce the diaphragm, causing the battery to self-discharge. Self-discharge will seriously affect the safety performance of lithium-ion batteries. In addition, the magnetic metal particles in the cathode powder may also affect the safety of the battery cell. Therefore, in order to ensure the stable safety performance of lithium batteries, it is necessary to effectively control the source of metal foreign matter in battery materials and reduce the risk of introducing metal foreign matter. In the related art, there are two main methods for detecting metal foreign bodies: (1) manual sampling (for example, taking out 1 kg of test sample from 4,000 kg of material), and evaluating the level of magnetic substances in the sample through inductively coupled plasma testing; (2) after manual sampling, the number of magnetic metal particles in the sample is evaluated through cleanliness microscope testing. Both tests require manual sampling and offline testing, and destructive pre-treatment is required before the test to extract the magnetic metal impurities in the sample. The test efficiency is relatively poor, the abnormal response speed is slow, and the test samples cannot be recovered, resulting in high test costs.

[0031] In view of the problems existing in the detection of magnetic metals in the above-mentioned battery slurry, this application proposes a detection scheme for magnetic metals in electromagnetic slurry. In combination with the actual battery production process, since the battery slurry usually flows in the slurry pipeline, this scheme sets a highly sensitive electromagnetic induction sensor on the outer wall of the slurry pipeline. When the battery slurry flows in the pipeline and flows through the electromagnetic induction sensor, based on the electromagnetic induction principle, the magnetic metals present in the battery slurry can be predicted. The whole process does not require manual sampling, which reduces sample loss, and can achieve real-time detection, greatly shortening the time period required for detection. Different from the above-mentioned sampling detection method, this scheme can conduct comprehensive detection of all battery slurries flowing through the pipeline, avoiding the problem of inaccurate sampling detection.

[0032] The technical solution of the present application is described in detail below through specific embodiments. It should be noted that the following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described in detail in some embodiments.

[0033] Figure 1This is a cross-sectional schematic diagram of an electromagnetic induction sensor provided in an embodiment of the present application. It should be noted that the structure shown in the figure is only an example of the present application and does not limit other solutions of the present application. Figure 1 As shown, the electromagnetic induction sensor comprises at least two excitation coils 12 reversely wound outside the slurry pipe 11 and at least one induction coil 13 wound outside the excitation coils. The radius of the slurry pipe 11 is r, and the coil radius of the excitation coil is D.

[0034] For example, Figure 1 As shown, when the battery slurry flows in the slurry pipe 11, the slurry flows in a vertically downward direction. This is mainly because the battery slurry can automatically flow in the slurry pipe under its own gravity without the need for external force. In other embodiments, the slurry flow direction can also be set to vertically upward. In this case, a reaction force needs to be applied to overcome the gravity of the battery slurry itself. Whether it is vertically downward or vertically upward, when the battery slurry contains magnetic metal particles and flows through the electromagnetic induction sensor, due to the principle of electromagnetic induction, an induced electromotive force change will be generated at the induction coil.

[0035] Among them, the slurry flow direction is usually consistent with the extension direction of the slurry pipeline. The electromagnetic induction sensor can obtain better detection effect when it is arranged perpendicular to the slurry pipeline, ensuring that the material transportation direction can cut the magnetic induction coil and generate electromotive force.

[0036] In this embodiment, after two excitation coils are reversely wound on the outside of the slurry pipe 11, a group of induction coils 13 are wound on the two excitation coils, thereby forming two layers of inner and outer coils to obtain a screw-type electromagnetic induction sensor.

[0037] The excitation capacitor is connected in parallel with the excitation coil, and after the excitation capacitor and the excitation coil are connected in parallel, the AC power supply in the power supply is connected to form a second resonant circuit; the excitation capacitor is used to make the second resonant circuit in a resonant state In which, taking the slurry pipe extending in a first direction as an example, the first excitation coil is wound on the slurry pipe along the first direction, and the second excitation coil is wound on the slurry pipe along a second direction opposite to the first direction. Figure 1 , the slurry pipe is vertical, at this time the first excitation coil can be wound vertically upward on the slurry pipe, and the second excitation coil can be wound vertically downward on the slurry pipe.

[0038] Among them, when the same alternating current (such as sinusoidal alternating current) is passed through the two excitation coils, dynamically changing magnetic fields of equal magnitude and opposite direction will be generated inside the two excitation coils. Due to the existence of two magnetic fields of equal magnitude and opposite direction, a region with zero magnetic field strength will be formed in the middle of the two excitation coils. Further, in some embodiments, in order to facilitate the measurement of electrical signal changes, such as induced electromotive force, the induction coil can be wound around the middle of the two excitation coils. When there is no battery slurry flowing through, the magnetic field generated by the two reversely wound excitation coils cancels each other, so the magnetic field strength in the middle of the two excitation coils is zero, that is, the magnetic field strength at the position point of the induction coil is zero, so no induced electromotive force is output. When the battery slurry containing magnetic metal particles flows through, the magnetic field strength of a certain excitation coil will be changed, making the magnetic field of the excitation coil unbalanced, which will cause the magnetic flux at the induction coil to be non-zero, and the induction coil outputs an induced electromotive force. Usually the magnitude of the induced electromotive force increases with the increase of the content of magnetic metal particles in the battery slurry.

[0039] For example, Figure 2 The schematic diagram of the change of magnetic field strength provided in the embodiment of the present application is as follows: Figure 2 As shown, the center point O of the two excitation coils (reference Figure 2 , located on the slurry pipe 11) as the origin, establish a spatial coordinate system, at this time in a static state (that is, when no battery slurry flows through), the magnetic field strength of the position point above the center position point O and the position point below the center position point O is calculated by the following formula: H=

[0040] R=(D+r) / 2 In the above formula, I is the average current of the excitation coil, W is the number of turns of the excitation coil, L is the length of the excitation coil, R is the average radius of the excitation coil, x is the coordinate axis with the center point O as the origin, the positive half axis is above the coordinate axis, and the negative half axis is below.

[0041] In this embodiment, by calculating the magnetic field strength at each position point above and below the center position point O, the winding position of the induction coil can be easily located. By accurately winding the induction coil, it can be ensured that the magnetic field strength at the induction coil is zero in a static state, so that the induction coil does not output an induced electromotive force in a static state.

[0042] In this embodiment, the induction coil is connected to an electric signal sensor, which is used to obtain the electric signal change value generated by the induction coil when the battery slurry flows through. At the same time, the electric signal change value is sent to the processor, and the processor determines the content of magnetic metal particles in the battery slurry based on the electric signal change value. The electric signal change value may refer to the inductance change value and / or the induced electromotive force change value.

[0043] The purpose of providing the electrical signal sensor is to accurately obtain the electrical signal change value of the induction coil, thereby improving the accuracy of subsequent detection of magnetic metal particles in the battery slurry.

[0044] In addition, in order to ensure the normal operation of the electromagnetic induction sensor, the electrical signal sensor and the processor, all three need to be powered by a power supply. For example, the above mentioned connection of alternating current to the two excitation coils.

[0045] The power supplies of the three may be independent of each other, for example, a separate first power supply may be configured for the electromagnetic induction sensor, a separate second power supply may be configured for the electrical signal sensor, and a separate third power supply may be configured for the processor. Exemplarily, the electrical signal sensor may include a single chip microcomputer.

[0046] In this embodiment, the electromagnetic induction sensor uses two excitation coils wound in opposite directions, which can further improve its inductance strength and the sensitivity of the electromagnetic induction sensor. Specifically, when the same alternating current is passed through the two excitation coils wound in opposite directions, since the direction and magnitude of the alternating current change periodically over time, according to Faraday's law of electromagnetic induction, the changing current will generate a changing magnetic field, which will stimulate a stronger induced electromotive force in the induction coil when the battery slurry flows through, thereby improving the induction sensitivity and the detection accuracy of the magnetic metal particles in the battery slurry.

[0047] In this embodiment, when the battery slurry flows in the slurry pipe and passes through the electromagnetic induction sensor, the different contents of magnetic metal particles in the battery slurry will cause changes in the magnetic resistance in the magnetic field leakage path of the induction coil. The change in magnetic resistance will directly affect the inductance change of the induction coil. The inductance change can be used to convert it into a digital signal that is easy to detect and process, and then the content of magnetic metal particles in the battery slurry can be deduced.

[0048] In the embodiment of the present application, a highly sensitive electromagnetic induction sensor is directly arranged on the outside of the slurry pipeline. When the battery slurry flows in the pipeline and passes through the electromagnetic induction sensor, the induced electrical signal change can be obtained based on the electromagnetic induction principle, thereby realizing the detection of the content of magnetic metal particles in the battery slurry, without the need for manual sampling, reducing sample loss, and without the need for subsequent manual detection and analysis, which greatly improves the detection efficiency and reduces the detection cost. Finally, the electromagnetic induction sensor can be in working condition at all times until all battery slurries have flowed through the slurry pipeline, realizing comprehensive detection of all battery slurries, and its detection accuracy is higher than the current sampling detection.

[0049] The circuit structure of the entire battery slurry detection device is described in detail below through some embodiments. For example, Figure 3 A schematic diagram of the structure of a battery slurry detection device provided in an embodiment of the present application is shown in FIG. Figure 3 As shown, the electrical signal sensor 31, the processor 32 and the power supply 33 can all be connected to the electromagnetic induction sensor 30 through long wires, so that the electrical signal sensor 31, the processor 32 and the power supply 33 can be separated from the slurry pipe independently, and only the electromagnetic induction sensor 30 is left on the outside of the slurry pipe. This can improve the convenience of installation in practical applications. Even in a very short section of the slurry pipe, the electromagnetic induction sensor 30 can be set without occupying a large space.

[0050] In other embodiments, during the actual use of the battery slurry detection device, the battery slurry detection device may malfunction. In order to facilitate rapid troubleshooting of circuit faults, the electrical signal sensor 31, the processor 32, the induction coil 13, and the excitation coil 12 may be powered separately.

[0051] For example, Figure 4 A circuit diagram of a battery slurry detection device provided in another embodiment of the present application is shown in FIG. Figure 4 As shown, the electromagnetic induction sensor 30 is arranged outside the slurry pipe 11. At this time, the electric signal sensor 31 and the induction coil 13 can be powered by a DC power supply 331, and the two excitation coils can be powered by an AC power supply 332. In this way, the two excitation coils are separately isolated from the power supply of the electric signal sensor 31 and the induction coil 13.

[0052] Among them, the AC power supply 332 can be a sinusoidal AC power, and the AC excitation signal generated by it is passed into the two excitation coils. Since the two excitation coils are wound in opposite directions, dynamically changing magnetic fields of equal magnitude and opposite directions will be generated inside the two excitation coils. In the middle part of the two excitation coils, the opposite magnetic fields will cancel each other out, making the magnetic field strength in the middle part zero. If the induction coil is wound here, the induction coil will not generate an induced electromotive force at this time. When the battery slurry flows through the electromagnetic induction sensor, it will enhance the magnetic field inside the excitation coil, causing the magnetic field of the two excitation coils to be unbalanced. At this time, the induced electromotive force output by the induction coil can be enhanced, that is, the sensitivity of the electromagnetic induction sensor can be improved. In addition, using a DC power supply to power the electrical signal sensor and the induction coil can also eliminate high-order harmonics and improve detection stability.

[0053] The resistor r1 may refer to the internal resistance of the excitation coil, and the resistor r2 may refer to the internal resistance of the induction coil.

[0054] Further, Figure 5 A circuit diagram of a battery slurry detection device provided in another embodiment of the present application is shown in FIG. Figure 5As shown, the battery slurry detection device also includes an induction capacitor 51. One end of the induction capacitor 51 is connected to the electrical signal sensor 31, and the other end of the induction capacitor 51 is connected to the induction coil 13 and then to the DC power supply 331. The other end of the electrical signal sensor 31 is connected to the induction coil 13 and then to the DC power supply 331.

[0055] In this embodiment, after the inductive capacitor 51 is connected to the electrical signal sensor 31, the DC power supply 331 and the inductive coil 13, a first resonant circuit is formed, wherein the capacitance of the inductive capacitor needs to be pre-configured so that the entire circuit is in a resonant state.

[0056] For example, assuming that the resonance condition is that the resonance frequency is f0, the capacitance of the inductive capacitor needs to be determined by the following formula: f0=

[0057] In the above formula, L1 is the inductance of the induction coil, r1 is the internal resistance of the induction coil, and C1 is the capacitance required to be configured for the induction capacitor.

[0058] In this embodiment, the value of the resonant frequency f0 can be flexibly adjusted according to different actual needs, and it is not a single fixed value.

[0059] In the embodiment of the present application, by configuring the induction capacitor, the entire circuit works in a resonant state, so that the electric field energy in the induction capacitor and the magnetic field energy in the induction coil can be converted to each other, one increases and the other decreases, and the energy transmission is fully compensated, making the energy transmission more efficient. At the same time, through the first resonant circuit formed, signal filtering can be achieved, signals other than specific frequencies can be suppressed or filtered out, and the required electrical signals can be selected from many electrical signals, thereby improving data acquisition efficiency and accuracy, thereby further improving the accuracy of detecting the content of magnetic metal particles in battery slurry.

[0060] Further, Figure 6 A circuit diagram of a battery slurry detection device provided in another embodiment of the present application is shown in FIG. Figure 6 As shown, the battery slurry detection device also includes an excitation capacitor 61. One end of the excitation capacitor 61 is connected to one end of the excitation coil 12 and then to one end of the AC power supply 332, and the other end of the excitation capacitor 61 is connected to the other end of the excitation coil 12 and then to the other end of the AC power supply 332.

[0061] In this embodiment, the excitation capacitor 61 is connected in parallel with at least two excitation coils to form an AC power supply, so as to form a second resonant circuit for making at least two excitation coils in a resonant state. Assuming that the resonant frequency required to make at least two excitation coils in a resonant state is f1, the capacitance of the excitation capacitor needs to be configured by the following formula: f1=

[0062] In the above formula, L2 is the inductance of the excitation coil, r2 is the internal resistance of the excitation coil, and C2 is the capacitance required to be configured for the excitation capacitor.

[0063] In this embodiment, the value of the resonant frequency f1 can be flexibly adjusted according to different actual needs, and it is not a single fixed value.

[0064] In this embodiment, the induced electromotive force output by the induction coil will change according to the actual situation. Specifically, the induced electromotive force output by the induction coil is related to the induction coil parameters (including the number of coil turns and inductance value), signal parameters (including resonant frequency and current flowing through the coil) and the magnetism of the battery slurry (such as positive electrode material) itself. In addition, metal particles with different properties have different effects on the magnetic field. For example, ferromagnetic particles will strengthen the external magnetic field, and non-ferromagnetic particles will weaken the external magnetic field. Therefore, when metal particles with different properties pass through the electromagnetic induction sensor, the electromagnetic induction sensor will output induced electromotive forces of different phases.

[0065] In the embodiment of the present application, by configuring the excitation capacitor to form a second resonant circuit, the excitation coil works in a resonant state, so that the electric field energy in the excitation capacitor and the magnetic field energy in the excitation coil are converted to each other, one increases while the other decreases, and the energy transmission is fully compensated, making the energy transmission more efficient. At the same time, by forming a resonant circuit with the excitation capacitor and the excitation coil, signal filtering can be achieved, and signals other than specific frequencies can be suppressed or filtered out.

[0066] Figure 7 A circuit diagram of a battery slurry detection device provided in another embodiment of the present application is shown in FIG. Figure 7 As shown, the battery slurry detection device further includes a first switch K1 and a second switch K2, wherein the first switch K1 is connected in series with the induction coil 13, and the second switch K2 is connected in series with at least two excitation coils 12.

[0067] The first switch K1 can cut off the connection between the induction coil 13 and the DC power supply 331, and cut off the connection between the induction coil 13 and the induction capacitor 51 and the electrical signal sensor. The second switch K2 can cut off the connection between the excitation coil 12 and the AC power supply 332, and cut off the connection between the excitation coil 12 and the excitation capacitor 61.

[0068] In this embodiment, a first switch K1 and a second switch K2 are configured. When there is no battery slurry flowing in the slurry pipeline, the first switch K1 and the second switch K2 can be turned on, so that both the excitation coil and the induction coil are in a power-off state, thereby preventing the coil from heating up and causing the slurry pipeline to have an excessive temperature rise, thereby affecting the normal service life of the slurry pipeline.

[0069] In addition, the power supply to the excitation coil 12 can be cut off by the second switch K2, and the excitation signal is no longer output to the excitation coil 12. Therefore, according to the change of the magnetic field before and after the second switch K2 is cut off, the enhanced sensitivity of the excitation coil to the electromagnetic induction sensor can be known, so as to improve the detection accuracy of magnetic metal particles in the battery slurry.

[0070] Figure 8 A schematic diagram of the structure of a signal acquisition unit provided in an embodiment of the present application is shown in FIG. Figure 8 As shown, the signal acquisition unit at least includes an electric signal sensor 81. The electric signal sensor 81 can be powered by a power supply 80 (such as the above-mentioned DC power supply), and the electric signal sensor 81 can include a signal generator 810 and a signal processor 811, one end of the signal generator 810 is connected to the induction coil, and the other end serves as one end of the electric signal sensor 81, and one end of the signal processor 811 is connected to the induction coil, and the other end serves as the other end of the electric signal sensor 81.

[0071] The signal generator 810 and the signal processor 811 are divided into two paths, one of which outputs an AC excitation signal (such as a controllable sine wave signal) to the induction coil, and the other collects the voltage across the induction coil and detects the inductance value of the induction coil by measuring the voltage change. The inductance value is then analyzed by the processor to determine the content of magnetic metal particles in the battery slurry.

[0072] The signal processor may be an analog-to-digital converter (ie, A / D converter), which converts an analog signal into a digital signal and then sends the digital signal to a processor for analysis to determine the content of magnetic metal particles in the battery slurry.

[0073] In this embodiment, during the signal acquisition process of the signal processor, the voltage across the induction coil under DC excitation (provided by the DC power supply mentioned above) and the voltage across the coil under AC excitation (provided by the AC excitation signal output by the signal generator) are detected in a time-sharing manner using the same channel. The detection of the DC voltage is used to eliminate the influence of the internal resistance of the induction coil, and the measurement of the AC voltage is used to detect the inductance value of the induction coil.

[0074] In this embodiment, when no battery slurry flows through the electromagnetic induction sensor, the static inductance value of the induction coil is for: φ

[0075] =

[0076] =

[0077] In the above formula, μ 0 is the vacuum magnetic permeability, N is the number of turns of the induction coil, d 1 is the average diameter of the induction coil, is the width of the induction coil, , and is the intermediate parameter.

[0078] When the battery slurry flows through the electromagnetic induction sensor, the inductance value of the induction coil for: [ ] In the above formula, μ 1 is the effective magnetic permeability of air, W is the number of turns of the excitation coil, l is the length of the excitation coil, r is the radius of the slurry pipe, and r c is the radius of the induction coil, u k is the effective magnetic permeability of the battery slurry.

[0079] In this embodiment, the inductance value is The effective magnetic permeability u of the battery slurry k There is a direct relationship between the change in inductance value and the change in inductance value of the induction coil from the static inductance value. , get the inductance value , we can deduce the effective magnetic permeability u of battery slurry (such as iron ore slurry) k , and further deduce the content of magnetic metal particles in the battery slurry.

[0080] In this embodiment, the induced electromotive force output by the induction coil will change according to the actual situation. Specifically, the induced electromotive force output by the induction coil is related to the induction coil parameters (including the number of coil turns and inductance value), signal parameters (including resonant frequency and current flowing through the coil) and the magnetism of the battery slurry (such as positive electrode material) itself. In addition, metal particles with different properties have different effects on the magnetic field. For example, ferromagnetic particles will strengthen the external magnetic field, and non-ferromagnetic particles will weaken the external magnetic field. Therefore, when metal particles with different properties pass through the electromagnetic induction sensor, the electromagnetic induction sensor will output induced electromotive forces of different phases.

[0081] Exemplarily, in some embodiments, continue to refer to the above Figure 8A display 83 may also be provided, and the display 83 may be connected to the electric signal sensor to display the electric signal change value of the induction coil. Thus, by finding the electric signal change value of the induction coil, the time or duration of the battery slurry flowing through the slurry pipeline may be recorded. At the same time, the content level of the magnetic metal particles in the battery slurry may be instantly estimated through the size of the electric signal change value displayed on the display, so as to adjust the production strategy in a timely manner.

[0082] Fig. 9 A schematic diagram of the structure of the battery slurry detection system provided in the embodiment of the present application is shown in FIG. Fig. 9 As shown, the battery slurry detection system 90 includes an information acquisition unit 91 , a signal conversion unit 92 , and a signal analysis unit 93 .

[0083] For the signal acquisition unit 91, please refer to the above Figure 8 Among them, according to the different contents of magnetic metal particles in the battery slurry flowing in the slurry pipeline, the influence of magnetic resistance in the magnetic field leakage path of the induction coil is different, which causes the magnetic flux generated by it to change. The magnetic flux change information is transmitted to the signal acquisition unit through the wire for unified collection; For the signal conversion unit 92, the magnetic anisotropy detection method is originally applied. Under the condition that the magnetic anisotropy sensor can provide a steady magnetic excitation, the change of magnetic resistance in the magnetic circuit will cause the change of magnetic flux. The change of induced electromotive force induced by the induction coil will reflect this change, and then the unmeasurable stress can be converted into a measurable voltage signal, thereby realizing the detection signal conversion.

[0084] For the signal analysis unit 93, the voltage signal is introduced into the computer processor, and deep reinforcement learning is performed through the deep Q-Learning (DQN) network to output the content prediction result of the final target of the magnetic metal particles.

[0085] Among them, the signal analysis unit performs deep reinforcement learning on the electrical signal output by the signal conversion unit through DQN, and its learning process can be described as: ① Collect the electrical signal output by the signal conversion unit, including the inductance value; ② Establish a magnetic metal particle content prediction model based on inductance value; ③Training a deep reinforcement learning neural network to predict the content of magnetic metal particles; ④ Iteratively upgrade the magnetic metal particle content prediction model of deep reinforcement learning neural network; ⑤ Use deep reinforcement learning neural network to output the prediction results of the magnetic metal particle content prediction model.

[0086] For example, Fig.10The schematic diagram of the structure of the magnetic metal particle content prediction model provided in the embodiment of the present application is as follows: Fig.10 As shown, the magnetic metal particle content prediction model includes a current value network 1002 and a target value network 1001. It specifically includes the following steps: Step 1010, input s; Step 1020, output ArgMaxaQ(s, a, Ø-); Step 1030, store (s, a, r, s'); Step S1040, extract (s, a); Step S1050, extract s'; Step S1060, input MaxaQ (s', a', Ø-); Step S1070, input Q (s, a, Ø); Step S1080, input r; Step S1090, outputting the error function gradient; Step S10110, network parameter copying.

[0087] Among them, parameter s represents the state, parameter a represents the action, parameter r represents the reward, and parameter represents the discount factor used to penalize uncertainty in future returns, represents the action-value function. Represents the output of the current value network, which is used to evaluate the value function of the current state-current action; Represents the output of the target value network.

[0088] Among them, you can use Approximately represents the optimization target of the value function, that is, the target Q value. The first network parameter θ of the current value network is updated in real time. After every N rounds of iteration (N is a positive integer, which is a pre-set threshold), the updated first network parameter of the current value network is copied to the target value network, that is, the second network parameter of the target value network is consistent with the first network parameter of the current value network at this time step.

[0089] Among them, the first network parameter of the current value network is updated by minimizing the mean square error between the current Q value (i.e., the output of the current value network) and the target Q value, and the error function of the calculated error is:

[0090] In the above formula, is the error function, is the state, action, and reward corresponding to a time step in a trajectory, For the next state, is the second network parameter of the target network after the i-th iteration, is the first network parameter of the current value network.

[0091] Among them, the partial derivative of the parameter θ is obtained, and the following gradient is obtained:

[0092] In this embodiment, after introducing the target value network, the target Q value remains unchanged for a period of time (because the second network parameter of the target value network is not updated in real time), which reduces the correlation between the current Q value and the target Q value to a certain extent and improves the stability of the algorithm.

[0093] Among them, the DQN algorithm is described as:

[0094] in, is the mean square error function, is a greedy strategy, is the discount factor.

[0095] In this embodiment, the state s may refer to the magnetic resistance of the electromagnetic induction sensor, the action a may refer to the inductance value of the electromagnetic induction sensor, and r may refer to the content of magnetic metal particles in the battery slurry flowing through the electromagnetic induction sensor.

[0096] After the introduction Fig.10 After the functions and meanings of the various parameters and functions in Fig.10 Steps ① to ⑤ in the figure introduce the entire magnetic metal particle content prediction model training process.

[0097] Specifically, for step ①, first the environment 1003 gives a first state s to the current value network 1002.

[0098] For step ②, the current value network outputs the Q value corresponding to each action in the first state based on the first state and each action corresponding to the first state.

[0099] Among them, MaxaQ (s, a) represents the optimal Q value among the Q values ​​of each action corresponding to the first state, that is, the current value network can output the optimal Q value based on the first state s and the action a. The current value network 1002 can determine the next second state s' corresponding to the first state by further interacting with the environment 1003, and then the environment 1003 stores the sample data (s, a, r, s') in the playback memory unit 1000. Among them, the number of sample data stored in the playback memory unit 1000 is not limited.

[0100] For step ③, after obtaining the sample data, the current value network takes (s, a) in the sample data as input and outputs the current Q value, i.e. Q(s, a, φ).

[0101] For step ④, the target value network takes the second state s' in the sample data and the second action a' corresponding to the second state s' (the determination of the second action a' can refer to the above step ②) as input, and outputs the target Q value (i.e., Q(s, a, φ-).

[0102] For step ⑤, the first network parameter of the current value network 1002 is adjusted and optimized through the error function, playback based on the reward r in the unit, the target Q value and the current Q value.

[0103] In addition, after N time steps, the first network parameter of the current value network 1002 is copied to the target value network 1001.

[0104] The battery slurry detection process is described in detail below through some examples.

[0105] Fig.11 The battery slurry detection process diagram provided in the embodiment of the present application can be applied to a processor, taking the processor as the execution subject as an example, such as Fig.11 As shown, the method comprises the following steps: Step S1100: in response to the battery slurry flowing through the electromagnetic induction sensor disposed outside the slurry pipeline, obtaining a change value of an electrical signal output by an induction coil of the electromagnetic induction sensor; Step S1110: Determine the content of magnetic metal particles in the battery slurry according to the change value of the electrical signal and the pre-trained magnetic metal particle content prediction model.

[0106] In this embodiment, the structure and location of the electromagnetic induction sensor can refer to the above description of the structure of the battery slurry detection device, which will not be repeated here.

[0107] In this embodiment, a magnetic metal particle content prediction model is used to predict the content of magnetic metal particles in the battery slurry.

[0108] In an embodiment of the present application, a magnetic metal particle content prediction model is used to convert the electrical signal change value and then indirectly detect the magnetic metal particle content in the battery slurry. The detection speed is fast and no material consumption is required, which greatly improves the detection efficiency and reduces the detection cost.

[0109] The change value of the electrical signal may include the magnetic resistance of the induction coil and the inductance value under the magnetic resistance. The magnetic metal particle content prediction model can be obtained by training through the following steps, for example, Fig.12 A schematic diagram of the training process of the magnetic metal particle content prediction model provided in the embodiment of the present application, such as Fig.12 As shown, it includes the following steps: Step S1210: obtaining a prediction result output by a magnetic metal particle content prediction model to be trained according to a sample value of an electrical signal change of the electromagnetic induction sensor when the sample battery slurry flows through the electromagnetic induction sensor.

[0110] The electrical signal change sample value includes the magnetic resistance of the induction coil in the electromagnetic induction sensor and the inductance value under the magnetic resistance.

[0111] Step S1220: Based on the prediction result and the magnetic metal particle content in the sample battery slurry, the network parameters in the magnetic metal particle content prediction model to be trained are adjusted to obtain a trained magnetic metal particle content prediction model.

[0112] In this embodiment, the magnetic metal particle content prediction model may include an input layer, a convolution layer, and a fully connected layer. After the battery slurry detection device is arranged in the slurry pipeline, the battery slurry detection device is used to obtain the electric signal change value and the magnetic metal particle content in the battery slurry flowing through the battery slurry detection device, thereby obtaining training data.

[0113] The training data can be divided into a training set and a test set. The training set and the test set contain the same data, but their uses are different. The training set is used to train the magnetic metal particle content prediction model, and the test set is used to test the accuracy of the magnetic metal particle content prediction model.

[0114] For example, both the training set and the test set contain the magnetic resistance of the induction coil, the inductance value under the magnetic resistance, and the content of magnetic metal particles in the battery slurry. First, the magnetic resistance of the induction coil and the inductance value under the magnetic resistance in the training set are used as the input of the magnetic metal particle content prediction model, and the magnetic metal particle content in the battery slurry in the training set is used as the target value. The network parameters of the magnetic metal particle content prediction model are optimized and adjusted so that the output prediction value is as close to the target value as possible.

[0115] After completing multiple training sessions, the magnetic resistance of the induction coil in the test set and the inductance value under the magnetic resistance are used as the input of the magnetic metal particle content prediction model to obtain the prediction result output by the magnetic metal particle content prediction model, and determine whether the prediction result is close to the magnetic metal particle content in the battery slurry in the test set. If the two are very different, it means that the prediction accuracy of the trained magnetic metal particle content prediction model does not meet the requirements. At this time, new training data needs to be added to the training set to continue training the magnetic metal particle content prediction model.

[0116] In an embodiment of the present application, by converting the change value of the electrical signal into the magnetic resistance of the induction coil and the inductance value under the magnetic resistance, and using it as training data, a magnetic metal particle content prediction model is trained, and the magnetic metal particle content in the battery slurry can be detected using the inductance data. The detection speed is fast and no material consumption is required, which greatly improves the detection efficiency and reduces the detection cost.

[0117] Furthermore, in some embodiments, Fig.13 A schematic diagram of the training steps of the magnetic metal particle content prediction model provided in the embodiment of the present application, such as Fig.13 As shown, it includes the following steps: Step S1310: Initialize the first network parameter of the current value network and the second network parameter of the target value network; Step S1320: Obtain at least one sample data from the experience replay memory unit, the sample data including a first state, a first action corresponding to the first state, a reward corresponding to the first state and the first action, and a second state next to the first state.

[0118] Among them, the state is used to characterize the magnetic resistance of the electromagnetic induction sensor, the action is used to characterize the inductance value of the electromagnetic induction sensor, the feedback is used to characterize the content of magnetic metal particles, and the experience playback memory unit includes multiple sample data.

[0119] Step S1330: input the first state and the first action in the sample data into the current value network to obtain the first output of the current value network; Step S1340: inputting the second state in the sample data into the target value network to obtain a second output of the target value network; Step S1350: According to the first output, the second output and the feedback, the first network parameter of the current value network and the second network parameter of the target value network are adjusted to obtain a magnetic metal particle content prediction model through training.

[0120] In this embodiment, the magnetic metal particle content prediction model can be the above Fig.10 The entire training process can refer to the above embodiment for the prediction model of magnetic metal particle content. Fig.10 The description of the magnetic metal particle content prediction model in will not be repeated here.

[0121] In this embodiment, a plurality of sample data can be used to repeatedly train the magnetic metal particle content prediction model. Among them, the first network parameter of the current value network will be updated at each time step, that is, training is performed once at each time step, and the first network parameter of the current value network is adjusted once according to the first output, the second output and the feedback. The second network parameter of the target value network will not be updated until the set N time steps. For example, from the first time step to the N-1 time step, the second network parameter will not be updated. At the Nth time step, the target value network will copy the first network parameter of the current value network at the Nth time step to update the second network parameter.

[0122] Furthermore, in some other embodiments, an accuracy threshold may be set, and after the first network parameter of the current value network is continuously updated, the data in the test set may be used to verify the prediction accuracy of the current value network.

[0123] Among them, when the prediction accuracy of the current value network is less than the accuracy threshold, it is necessary to continue to take new sample data from the experience playback memory unit, and continue to repeat the above steps S1320 to S1350, until the prediction accuracy of the current value network is found to be greater than or equal to the accuracy threshold through the test set test, and finally the trained magnetic metal particle content prediction model is obtained. In this way, the training effect of the magnetic metal particle content prediction model can be guaranteed, and the prediction accuracy of the magnetic metal particle content prediction model can be further improved.

[0124] In an embodiment of the present application, the above-mentioned Q network is used as a magnetic metal particle content prediction model, which can process high-dimensional, continuous state space. In addition, the magnetic metal particle content prediction model uses a current value network and a target value network, which can improve the stability and convergence speed of the model.

[0125] In some embodiments, Fig.14 The following is a schematic diagram of the algorithm processing flow of the target value network provided in the embodiment of the present application, such as Fig.14 As shown, it includes the following steps: Step S1410: inputting the second state input and at least one action corresponding to the second state into the target value network, and obtaining at least one first result output by the target value network; Step S1420: Determine, according to the at least one first result, a second action corresponding to the second state from the at least one action; Step S1430: Input the second state and the second action into the target value network to obtain the second output of the target value network.

[0126] In this embodiment, the output of the target value network is used to further determine the optimization target (ie, target Q value) of the value function in the current value network.

[0127] in, ) represents the action-value function of the current value network. Represents the output of the current value network, which is used to evaluate the value function of the current state-current action; ) represents the output of the target value network. Approximately represents the optimization objective of the value function, that is, the target Q value.

[0128] In this embodiment, after the target value network obtains the next state s' from the sample data, the predicted Q value corresponding to each action in all actions executed in the next state s' can be obtained, and the best Q value can be determined from the predicted Q values ​​corresponding to each action.

[0129] For example, taking the next state s' corresponding to action a1, action a2 and action a3 as an example, the target value network can output multiple first results according to the next state s' and a1, action a2 and action a3 , and .

[0130] Among them, assuming If the set conditions are met and the Q value is optimal, then a1 is the second action corresponding to the second state. That is, the second output of the target value network.

[0131] In an embodiment of the present application, the target value network can learn the optimal strategy and obtain the optimal Q value by utilizing the next second state s' and combining all feasible actions under the second state s' through interaction with the environment and continuous trial and error, thereby improving the accuracy of model training.

[0132] Furthermore, Fig.15 The sample data construction process diagram provided in the embodiment of the present application is as follows: Fig.15 As shown, it includes the following steps: Step S1510: randomly selecting a first state from at least one state; Step S1520: inputting the first state and at least one action corresponding to the first state into the current value network, and obtaining at least one second result output by the current value network; Step S1530: Determine a first action from at least one action according to at least one second result; Step S1540: Obtain a second state and a reward according to the first state and the first action; Step S1550: Construct sample data based on the first state, the first action, the reward and the second state.

[0133] In this embodiment, the magnetic metal particle content prediction model can be similar to an intelligent agent, which is in an initial state (i.e., the first state, such as the state where the battery slurry has not yet flowed through the electromagnetic induction sensor) when it first enters the environment. This initial state is the starting point of the environment, and the intelligent agent begins to explore and learn from this starting point. The intelligent agent can use the current value network to estimate the Q value of each possible action in the current state. This Q value represents the reward (i.e., the second result) that may be obtained after performing a certain action in the current state.

[0134] Based on these Q values, the agent can use a preset strategy (such as a greedy strategy) to select the best action (i.e., the first action) from all possible actions. For example, according to the preset strategy, the action with the highest current Q value is selected as the first action.

[0135] After the agent performs the first action it chooses, the environment will give new feedback, including: reward (i.e., the return r mentioned above), which is the reward signal obtained immediately after performing the first action in the first state, which can be a positive number (indicating a reward) or a negative number (indicating a penalty). The new next state (i.e., the second state s'), which is the new state reached by the environment after the action is performed, becomes the starting point for the next decision.

[0136] In this embodiment, the experience of the interaction between the agent and the environment can be saved in the experience replay memory unit in the form of a tuple, which usually contains the following elements: (s, a, r, s'). The purpose is to provide diversified data samples in the subsequent model training process. After accumulating enough experience in the experience replay memory unit, the agent will periodically randomly extract a batch from the experience replay memory unit as sample data for training.

[0137] In an embodiment of the present application, through interaction with the environment and continuous trial and error, the first action corresponding to the first state and the next second state are obtained, and stored in the experience replay memory unit as experience replay values. The subsequent use of the experience replay values ​​as sample data to train the model can improve the stability and convergence speed of the model.

[0138] The following is a complete example to introduce the steps involved in the training process of the entire magnetic metal particle content prediction model. Fig.16 This is a schematic diagram of the model training framework of the embodiment of the present application, such as Fig.16 As shown, the training process can be simplified into the following steps: Step 1: Get structured data; Step 2: Network initialization; Step 3: Initialize the experience replay cache; Step 4: Construct training set and test set, and establish magnetic metal particle content prediction model; Step 5: Training the magnetic metal particle content prediction model; Step 6: Iterate the magnetic metal particle content prediction model; Step 7: Verify the accuracy of the magnetic metal particle content prediction model; Step 8: Output the prediction results of magnetic metal particles; For the above step 1, the structured data is the inductance value of the induction coil.

[0139] For the above step 2, the network initialization is used to initialize the first parameter of the current value network and the second parameter of the target value network in the magnetic metal particle content prediction model.

[0140] For the above step 3, the initialization experience replay cache is used to store the network initialization data into the experience replay memory bank; For the above step 4: construct training set and test set, the original data set is divided into training set and test set according to the set ratio (for example, 0.7:0.3), and the characteristic data corresponding to the training set is used to train the magnetic metal particle content prediction model. Among them, the training set and the test set contain the same type of data, but the data values ​​are different.

[0141] For the above step 5: In the process of training the magnetic metal particle content prediction model, first initialize the state s, and for each time step t, perform the following steps: ① Initialization: Initialize the parameters of the current value network and the target value network. The current value network is used for updating each iteration, while the target value network is used to calculate the target value; ②Select action: According to the state s (i.e. magnetic resistance) in the current environment, the current value network uses a greedy strategy to select action a (i.e. inductance value); ③Perform actions and observe rewards: The current value network takes action a and the above Fig.10 The sensor interacts with the environment 1003 (i.e., the magnetic field in which the electromagnetic induction sensor is located) and observes the next state s' and the reward r.

[0142] ④Store experience: store (s, a, r, s') as a set of sample data in the experience playback memory library; ⑤ Randomly sample a batch of sample data from the experience playback memory bank; ⑥ Use the target value network to calculate the target Q value, that is, max(Q(s',a,θ-)); ⑦Update the current value network: Update the first parameter of the current value network according to the loss function L(θ).

[0143] ⑧Update the target network: The target value network will copy the first parameter of the current value network at a certain frequency (the first parameter is updated at this time).

[0144] Regarding the above step six: when the prediction accuracy of the magnetic metal particle content prediction model does not meet the expected target (such as the accuracy threshold), repeat ①-⑥ in the above step five until the expected target is met.

[0145] For the above step 7: output the predicted value of the model after the parameter update in this round, verify the trained model with the feature data obtained from the validation set, and compare the predicted value with the true value to obtain the prediction accuracy of the current value network; For the above step eight: input the data to be predicted into the trained model to obtain the final output magnetic metal particle results.

[0146] In the embodiment of the present application, the DQN algorithm can process high-dimensional, continuous state space. In addition, through the interaction of the current value network with the environment and continuous trial and error, the DQN algorithm can learn better strategies and have certain planning capabilities, as well as high algorithm stability and convergence speed. The DQN algorithm can improve the stability and convergence speed of the magnetic metal particle content prediction model through the experience playback memory unit and the parameters in the fixed target value network.

[0147] For the detailed description of the method steps in the above embodiment, reference may be made to the above description of the battery slurry detection device, which will not be repeated here.

[0148] Unless otherwise specified, all embodiments and optional embodiments of the present application can be combined with each other to form a new technical solution.

[0149] Unless otherwise specified, all technical features and optional technical features of this application can be combined with each other to form a new technical solution.

[0150] Unless otherwise specified, all steps of the present application may be performed sequentially or randomly, preferably sequentially. For example, a method includes steps (a) and (b), which means that the method may include steps (a) and (b) performed sequentially, or may include steps (b) and (a) performed sequentially. For example, a method may also include step (c), which means that step (c) may be added to the method in any order, for example, the method may include steps (a), (b) and (c), or may include steps (a), (c) and (b), or may include steps (c), (a) and (b), etc.

[0151] Although the present application has been described with reference to preferred embodiments, various modifications may be made thereto and parts thereof may be replaced with equivalents without departing from the scope of the present application. In particular, the various technical features mentioned in the various embodiments may be combined in any manner as long as there is no structural conflict. The present application is not limited to the specific embodiments disclosed herein, but includes all technical solutions falling within the scope of the claims.

Claims

1. A battery slurry detection device, characterized in that: The device comprises: An electromagnetic induction sensor, comprising at least two excitation coils reversely wound on the outside of a slurry pipe and at least one induction coil wound on the outside of the excitation coils; an electrical signal sensor connected to the induction coil and used to obtain an electrical signal change value of the induction coil when the battery slurry flows from the slurry pipe through the electromagnetic induction sensor; A processor, connected to the electrical signal sensor, for detecting the content of magnetic metal particles in the battery slurry according to the received electrical signal change value; A power supply connected to the electromagnetic induction sensor, the electrical signal sensor and the processor; An excitation capacitor, wherein the excitation capacitor is connected in parallel with the excitation coil, and after being connected in parallel with the excitation coil, the excitation capacitor is connected to an AC power supply in the power supply to form a second resonant circuit; the excitation capacitor is used to put the second resonant circuit in a resonant state.

2. The device according to claim 1, characterized in that The power supply at least includes a DC power supply and an AC power supply; The DC power supply is connected to the electrical signal sensor and the induction coil; The AC power source is connected to at least two of the excitation coils.

3. The device according to claim 2, characterized in that Also includes inductive capacitance; The inductive capacitor is connected to the electrical signal sensor, the DC power supply and the inductive coil to form a first resonant circuit, and the inductive capacitor is used to put the first resonant circuit in a resonant state.

4. The device according to any one of claims 1 to 3, characterized in that The induction coil is wound around the middle part of at least two of the excitation coils.

5. The device according to any one of claims 1 to 3, characterized in that The electrical signal sensor includes a signal generator and a signal processor; The signal generator is connected to the induction coil and is used to output an AC signal to the induction coil; The signal processor is connected to the induction coil and is used to detect a first voltage across the induction coil under the excitation of a DC power supply and a second voltage across the induction coil under the excitation of the AC signal when the battery slurry flows from the slurry pipe through the electromagnetic induction sensor; The signal processor is further used to determine the electrical signal change value according to the first voltage and the second voltage.

6. The device according to any one of claims 1 to 3, characterized in that Also includes a first switch and a second switch, The first switch is connected in series with the induction coil; The second switch is connected in series with at least two of the excitation coils.

7. The device according to any one of claims 1 to 3, characterized in that Also includes a display; the display is connected to the electrical signal sensor; The display is used to display the change value of the electrical signal of the induction coil.

8. A battery slurry detection method, characterized in that: include: Obtaining a prediction result output by a magnetic metal particle content prediction model to be trained according to a sample value of an electric signal change of the electromagnetic induction sensor when the sample battery slurry flows through the electromagnetic induction sensor, wherein the sample value of the electric signal change includes a magnetic resistance of an induction coil in the electromagnetic induction sensor and an inductance value under the magnetic resistance; Based on the prediction result and the content of magnetic metal particles in the sample battery slurry, adjusting the network parameters in the magnetic metal particle content prediction model to be trained to obtain a trained magnetic metal particle content prediction model; In response to battery slurry flowing through an electromagnetic induction sensor disposed outside the slurry pipeline, obtaining a change value of an electrical signal output by an induction coil of the electromagnetic induction sensor; The content of the magnetic metal particles in the battery slurry is determined according to the electrical signal change value and a pre-trained magnetic metal particle content prediction model.

9. The method according to claim 8, characterized in that The step of obtaining the prediction result output by the magnetic metal particle content prediction model to be trained includes: According to the sample value of the electrical signal change, at least one sample data is obtained from the experience playback memory unit, the sample data including a first state, a first action corresponding to the first state, a report corresponding to the first state and the first action, and a second state next to the first state, the state is used to characterize the magnetic resistance of the electromagnetic induction sensor, the action is used to characterize the inductance value of the electromagnetic induction sensor, and the report is used to characterize the magnetic metal particle content of the sample battery slurry; The sample data is input into the magnetic metal particle content prediction model to be trained, and a prediction result output by the magnetic metal particle content prediction model to be trained is obtained.

10. The method according to claim 9, characterized in that The magnetic metal particle content prediction model includes a current value network and a target value network; the step of inputting the sample data into the magnetic metal particle content prediction model to be trained to obtain the prediction result output by the magnetic metal particle content prediction model to be trained includes: Input the first state and the first action in the sample data into the current value network to obtain a first output of the current value network; Inputting the second state in the sample data into the target value network to obtain a second output of the target value network; The prediction result is determined according to the first output and the second output.

11. The method according to claim 10, characterized in that The adjusting the network parameters in the magnetic metal particle content prediction model to be trained based on the prediction result and the magnetic metal particle content in the sample battery slurry includes: Initializing a first network parameter of the current value network and a second network parameter of the target value network; According to the first output, the second output and the reward, a first network parameter of the current value network and a second network parameter of the target value network are adjusted.

12. The method according to claim 10, characterized in that The step of inputting the second state in the sample data into the target value network to obtain a second output of the target value network includes: Inputting the second state and at least one action corresponding to the second state into the target value network, and obtaining at least one first result output by the target value network; Determining, according to the at least one first result, a second action corresponding to the second state from among the at least one action; The second state and the second action are input into the target value network to obtain a second output of the target value network.

13. The method according to claim 10, characterized in that Also includes: randomly selecting the first state from at least one state; Inputting the first state and at least one action corresponding to the first state into the current value network, and obtaining at least one second result output by the current value network; determining the first action among the at least one action according to the at least one second result; According to the first state and the first action, obtain the second state and the reward; The sample data is constructed based on the first state, the first action, the reward and the second state.

14. The method according to claim 11, characterized in that The adjusting, according to the first output, the second output and the reward, a first network parameter of the current value network and a second network parameter of the target value network comprises: When the prediction accuracy of the current value network after the first network parameter is updated is less than the accuracy threshold, continue to acquire new sample data; According to the new sample data, update the first output, the second output and the return; According to the updated first output, the updated second output and the updated reward, the first network parameter of the current value network and the second network parameter of the target value network continue to be adjusted.

Images