Gain-adjustable neuron circuit and control method thereof

Abstract

The application relates to a gain-adjustable neuron circuit and a control method thereof, which comprises a memristor, a capacitor and a resistor, one end of the memristor is connected in parallel with the capacitor, then connected in series with the resistor, and is connected with a driving power supply; the memristor comprises a MIM structure, a heating module for providing a thermal field for the MIM structure, and a substrate for providing support for the whole memristor, the MIM structure comprises a top electrode, a functional layer and a bottom electrode which are sequentially distributed from top to bottom, the heating module comprises a heat-conducting layer and a heating layer which are sequentially distributed, and the heating layer is in contact with the MIM structure through the heat-conducting layer to perform heat transfer; the top electrode is connected with a driving-type voltage input, and the bottom electrode is grounded; one end of the heating layer is connected with a modulation-type voltage input, and the other end is grounded; and the port connected with the resistor of the memristor is a voltage output end. Compared with the prior art, the application not only can realize pulse response of the driving-type input, but also can realize dynamic adjustment of response gain.

Description

Technical Field

[0001] This invention relates to the field of neuronal circuits, and in particular to a neuronal circuit with adjustable gain and its control method. Background Technology

[0002] Spiking neural networks (SNNs) are considered the primary choice for building highly energy-efficient in-memory computing data processing units. Their core components include neurons and synapses, especially spiking neuron circuits with rich dynamic characteristics. These circuits simulate the process by which biological neurons generate electrical impulses (i.e., action potentials) when they receive external stimuli, achieving impulse response characteristics similar to those of biological neurons.

[0003] Traditional CMOS technology faces numerous challenges in realizing spiking neuron circuits, often requiring complex auxiliary circuits to simulate biodynamic characteristics, resulting in significantly lower energy consumption and integration density compared to biological units. However, with the end of Moore's Law, realizing neuronal circuits using novel principle devices, such as memristors, represents a completely new field. Novel neuromorphic devices, due to their small size, compactness, low power consumption, and rich dynamics, have become ideal choices for constructing neuronal circuits.

[0004] However, while existing neuronal circuits based on novel devices can achieve the basic threshold response characteristics of neurons, they still cannot effectively realize the gain-tunable characteristics of biological neurons. Gain refers to the neuron's response sensitivity to input (corresponding to the slope of the neuron's input-output curve). Gain modulation refers to the neuron's ability to accept additional input to adjust its response sensitivity to the current synaptic input, which is crucial for simulating the adaptability and flexibility of biological neurons under different physiological states. Currently, most neuronal circuits based on novel devices only simulate the basic characteristic of neurons firing pulses and have not yet been able to provide dynamically tunable gain responses. Therefore, future research needs to further optimize device characteristics and circuit design to achieve richer and more complex biological characteristics of neuronal circuits. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of the prior art by providing a gain-adjustable neuron circuit and its control method, which can not only realize the impulse response to the driving input, but also realize the dynamic adjustment of the response gain.

[0006] The objective of this invention can be achieved through the following technical solutions:

[0007] A gain-adjustable neuron circuit includes a memristor element, a capacitor, and a resistor. One end of the memristor element is connected in parallel with the capacitor and in series with the resistor, and is connected to a driving power supply.

[0008] The memristor includes a MIM structure, a heating module that provides a thermal field for the MIM structure, and a substrate that provides support for the entire memristor. The MIM structure includes a top electrode, a functional layer, and a bottom electrode arranged sequentially from top to bottom. The heating module includes a thermally conductive layer and a heating layer arranged sequentially. The heating layer transfers heat to the MIM structure through contact with the thermally conductive layer.

[0009] The top electrode is connected to a driving voltage input, and the bottom electrode is grounded; one end of the heating layer is connected to a modulation voltage input, and the other end is grounded; the port where the memristor element is connected to the resistor is the voltage output terminal.

[0010] Furthermore, the resistor, memristor, and capacitor constitute a charging circuit, and the memristor and capacitor constitute a discharging circuit.

[0011] Furthermore, when the driving power supply is working, one end of the functional layer receives a driving voltage input and charges the capacitor; when the voltage across the functional layer exceeds the threshold voltage of the functional layer material, the functional layer jumps from a high-resistivity state to a low-resistivity state, and the capacitor discharges through the memristor element until the voltage across the functional layer is lower than the holding voltage of the functional layer material, at which point the functional layer returns to a high configuration, thereby completing one pulse firing of the neuron circuit.

[0012] Furthermore, when a modulated voltage input is applied to the heating layer, the heating layer generates a thermal field, which is conducted to the functional layer through the thermally conductive layer, changing the threshold voltage and holding voltage of the functional layer to adjust the impulse response sensitivity of the neuron circuit.

[0013] Furthermore, the substrate is a silicon oxide substrate.

[0014] The present invention also provides a control method for a gain-adjustable neuron circuit as described above, comprising the following steps:

[0015] The driving power supply inputs voltage to the neuron circuit, applies a driving voltage input to the top electrode and charges the capacitor, which increases the voltage across the functional layer. When the voltage across the functional layer exceeds the threshold voltage, the material of the functional layer jumps from a high-resistivity state to a low-resistivity state and is discharged by the capacitor. When the voltage across the functional layer is lower than the holding voltage of the material, the material of the functional layer returns to a high-resistivity state, completing one pulse firing of the neuron circuit.

[0016] After applying a modulated voltage input to the heating layer, the thermal field generated by the heating layer is conducted to the functional layer, changing the threshold voltage and holding voltage of the functional layer material, and thus changing the pulse firing frequency of the neuron circuit.

[0017] The present invention also provides another gain-adjustable neuron circuit, including a memristor element, a capacitor and a resistor, wherein one end of the memristor element is connected in parallel with the capacitor and then in series with the resistor;

[0018] The memristor element includes a MIM structure, which includes a top electrode, a functional layer, and a bottom electrode arranged sequentially from top to bottom.

[0019] One end of the resistor is connected to a driving power supply, and the other end is connected to the top electrode and serves as a voltage output terminal. The top electrode receives the voltage input after being divided by the resistor; the bottom electrode receives a modulated voltage input.

[0020] Furthermore, by applying a modulated voltage input to the bottom electrode, the voltage input to the top electrode is adjusted accordingly, thereby achieving a linear translational adjustment of the threshold switching behavior of the neuron circuit along the input direction.

[0021] Furthermore, when the driving power supply is working, one end of the functional layer receives the voltage input after the top electrode is divided by a resistor, and the other end receives the modulated voltage input from the bottom electrode. When the potential difference between the two ends of the functional layer exceeds the threshold voltage of the functional layer material, the functional layer jumps from a high-resistivity state to a low-resistivity state, and the capacitor discharges through the memristor element until the voltage division of the functional layer is lower than the holding voltage of the functional layer material. Then, the functional layer returns to the high configuration state, thereby completing one pulse firing of the neuron circuit.

[0022] The present invention also provides a control method for a gain-adjustable neuron circuit as described above, comprising the following steps:

[0023] The driving power supply inputs voltage to the neuron circuit, which is then applied to the top electrode after being divided by a resistor. A modulated voltage input is applied to the bottom electrode and conducted to both ends of the functional layer. When the potential difference between the two ends of the functional layer exceeds the threshold voltage of the functional layer material, the functional layer jumps from a high-resistivity state to a low-resistivity state. The capacitor discharges through the memristor element until the voltage division of the functional layer is lower than the holding voltage of the functional layer material. At this point, the functional layer returns to the high configuration, thus completing one pulse firing of the neuron circuit.

[0024] By adjusting the magnitude of the modulated voltage input applied to the bottom electrode, the threshold switching behavior of the neuron circuit is linearly shifted along the input direction.

[0025] Compared with the prior art, the present invention has the following advantages:

[0026] (1) The gain-adjustable neuron circuit provided by the present invention generates a large current by setting a heating layer and accepting modulated input, thereby generating a thermal field that is conducted to the functional layer of the memristor element through a thermally conductive layer; the threshold voltage and holding voltage of the functional layer material change under the action of the thermal field, thereby changing the charging and discharging process of the neuron circuit, and thus causing the pulse firing frequency of the circuit to change. This change in the pulse response sensitivity of the neuron to the driving input caused by the modulated input realizes the gain modulation characteristics of the biological neuron.

[0027] (2) Another gain-adjustable neuron circuit provided by the present invention applies a modulated voltage input to the bottom electrode. The potential difference across the functional layer determines the threshold response behavior of the device. To ensure that the voltage division of the functional layer still exceeds the threshold voltage of the material, the voltage input to the top electrode of the device must be increased accordingly. The threshold switching behavior of the neuron only undergoes a linear shift along the input direction. Therefore, the modulated input received by this bottom electrode also changes the pulse firing behavior of the neuron circuit to the same driving voltage. It is a linear integration of modulated input and driving input, and is also considered to be a gain modulation characteristic of biological neurons. Attached Figure Description

[0028] Figure 1 This is a schematic diagram of the structure of a gain-adjustable neuron circuit provided in Embodiment 1 of the present invention;

[0029] Figure 2 This is a schematic diagram of a memristor element structure for a gain-adjustable neuron circuit provided in Embodiment 1 of the present invention;

[0030] Figure 3 This is a schematic diagram of the voltage input of a gain-adjustable neuron circuit provided in Embodiment 1 of the present invention;

[0031] Figure 4 The gain-adjustable neuron circuit provided in Embodiment 1 of the present invention is used in different modulation input V m Voltage oscillation waveform diagram;

[0032] Figure 5 The gain-adjustable neuron circuit provided in Embodiment 1 of the present invention uses a driving input V d The graph shows the change in pulse response frequency as a function of the modulated input.

[0033] Figure 6 This is a schematic diagram of the structure of a gain-adjustable neuron circuit provided in Embodiment 2 of the present invention;

[0034] Figure 7 This is a schematic diagram of the voltage input of a gain-adjustable neuron circuit provided in Embodiment 2 of the present invention;

[0035] Figure 8 This is an integral effect diagram of a gain-adjustable neuron circuit provided in Embodiment 2 of the present invention under a constant voltage input.

[0036] Figure 9 The gain-adjustable neuron circuit provided in Embodiment 2 of the present invention is used in different modulation input V b Voltage oscillation waveform diagram;

[0037] Figure 10 This is a graph showing the change in pulse response frequency of a gain-adjustable neuron circuit under a driving input as a function of a modulated input, as provided in Embodiment 2 of the present invention. Detailed Implementation

[0038] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0039] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0040] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0041] Example 1

[0042] like Figure 1 and Figure 2 As shown, this embodiment provides a gain-adjustable neuron circuit, including a memristor, a capacitor, and a resistor. One end of the memristor is connected in parallel with the capacitor and in series with the resistor, and is connected to a driving power supply.

[0043] The memristor includes a MIM structure, a heating module that provides a thermal field for the MIM structure, and a substrate that provides support for the entire memristor. The MIM structure includes a top electrode I6, a functional layer I5, and a bottom electrode I4 arranged sequentially from top to bottom. The heating module includes a thermally conductive layer I3 and a heating layer I2 arranged sequentially. The heating layer I2 transfers heat to the MIM structure through the thermally conductive layer I3.

[0044] The top electrode I6 is connected to a driving voltage input, and the bottom electrode I4 is grounded; one end of the heating layer I2 is connected to a modulation voltage input, and the other end is grounded; the port where the memristor element is connected to the resistor is the voltage output terminal.

[0045] Resistors, memristors, and capacitors form a charging circuit, while memristors and capacitors form a discharging circuit.

[0046] The substrate can be a silicon oxide substrate I1.

[0047] Working principle: such as Figure 3 As shown, when the driving power supply is working, one end of the functional layer receives the driving voltage input and charges the capacitor; when the voltage division across the functional layer exceeds the threshold voltage of the functional layer material, the functional layer jumps from a high-resistivity state to a low-resistivity state, and the capacitor discharges through the memristor element until the voltage division of the functional layer is lower than the holding voltage of the functional layer material, at which point the functional layer returns to the high configuration, thus completing one pulse firing of the neuron circuit.

[0048] When a modulated voltage input is applied to the heating layer, the heating layer generates a thermal field, which is conducted to the functional layer through the thermally conductive layer, changing the threshold voltage and holding voltage of the functional layer to adjust the impulse response sensitivity of the neuron circuit.

[0049] This embodiment also provides a control method for the above-described gain-adjustable neuron circuit, including the following steps:

[0050] The driving power supply inputs voltage to the neuron circuit, applies a driving voltage input to the top electrode and charges the capacitor, which increases the voltage across the functional layer. When the voltage across the functional layer exceeds the threshold voltage, the material of the functional layer jumps from a high-resistivity state to a low-resistivity state and is discharged by the capacitor. When the voltage across the functional layer is lower than the holding voltage of the material, the material of the functional layer returns to a high-resistivity state, completing one pulse firing of the neuron circuit.

[0051] After applying a modulated voltage input to the heating layer, the thermal field generated by the heating layer is conducted to the functional layer, changing the threshold voltage and holding voltage of the functional layer material, and thus changing the pulse firing frequency of the neuron circuit.

[0052] Specifically, in this embodiment, the neuron circuit consists of a threshold-adjustable memristor element, an external capacitor connected in parallel, and a resistor connected in series. The top electrode of the memristor element serves as the input terminal, receiving a driving input V. d Memristor and resistor R C The connected port serves as the voltage output terminal, with the bottom electrode of the memristor element grounded. The other port of the resistor is connected to the drive power supply V. DD One end of the heating layer of the memristor serves as the voltage input terminal, accepting a modulated input V. mOne end is grounded. The resistor, memristor, and capacitor form a charging circuit, and the memristor and capacitor form a discharging circuit.

[0053] The top electrode, functional layer, bottom electrode, thermally conductive layer, heating layer, and silicon oxide substrate are distributed sequentially from top to bottom.

[0054] In this neuron circuit, the driving power supply V DD After operation, due to the series voltage divider, one end of the functional layer of the memristor receives a driving input V. d The capacitor is charged, increasing the voltage across the functional layer. When the voltage drop across the functional layer exceeds the threshold voltage V... th When the material transitions from a high-resistivity state to a low-resistivity state, the voltage drop across the functional layer decreases, and the capacitor discharges through the memristor until the voltage drop across the functional layer falls below the material's holding voltage V. hold At this point, the material abruptly returns to a high-resistivity state, thus completing one pulse firing of the neuron circuit. When a modulated voltage V is applied to the heating layer of the memristor element... m Subsequently, the heating layer generates a large current, which in turn creates a thermal field that is conducted to the functional layer of the memristor element through the thermally conductive layer. The threshold voltage V of the material... th and holding voltage V hold The change in the thermal field alters the charging and discharging processes of the neuron circuit, leading to a change in the pulse firing frequency. This alteration of the neuron's pulse response sensitivity to the driving input, caused by the modulated input, achieves the gain modulation characteristic of biological neurons.

[0055] like Figure 1 As shown, this is a neuron circuit receiving modulated input V in the memristor element heating layer. m The diagram below illustrates this.

[0056] like Figure 4 As shown, this illustrates the neuron circuit under different modulation input V. m The voltage oscillation waveform under the given conditions. When V m As the heating layer of the device increases, the thermal field generated increases, the threshold voltage of the neuron decreases, and the oscillation window also decreases accordingly.

[0057] like Figure 5 As shown, this is a neuron circuit with a driving input V. d The graph shows the change in pulse response frequency as a function of the modulated input. When V m As the memristor's heating layer increases, the thermal field generated increases, the threshold voltage decreases, the integration process under the same drive input speeds up, and the pulse firing frequency increases. This can be seen in the figure as V m The modulation behavior of the response to the driving input is similar to the multiplication and division gain modulation characteristics in biology.

[0058] Example 2

[0059] like Figure 6 As shown, this embodiment provides a gain-adjustable neuron circuit, including a memristor element, a capacitor, and a resistor. One end of the memristor element is connected in parallel with the capacitor and then in series with the resistor.

[0060] Memristor devices include a MIM structure, which includes a top electrode, a functional layer, and a bottom electrode arranged sequentially from top to bottom.

[0061] One end of the resistor is connected to the driving power supply, and the other end is connected to the top electrode and serves as the voltage output terminal. The top electrode receives the voltage input after being divided by the resistor; the bottom electrode receives a modulated voltage input.

[0062] Working principle: By applying a modulated voltage input to the bottom electrode, the voltage input to the top electrode is adjusted accordingly, thereby realizing the linear translational adjustment of the threshold switching behavior of the neuron circuit along the input direction.

[0063] When the driving power supply is working, one end of the functional layer receives the voltage input after the voltage is divided by the resistor from the top electrode, and the other end receives the modulated voltage input from the bottom electrode. When the potential difference between the two ends of the functional layer exceeds the threshold voltage of the functional layer material, the functional layer jumps from a high-resistivity state to a low-resistivity state. The capacitor discharges through the memristor element until the voltage division of the functional layer is lower than the holding voltage of the functional layer material. Then the functional layer returns to the high configuration, thus completing one pulse firing of the neuron circuit.

[0064] This embodiment also provides a control method for the above-described gain-adjustable neuron circuit, including the following steps:

[0065] The driving power supply inputs voltage to the neuron circuit, which is then applied to the top electrode after being divided by resistors. A modulated voltage input is applied to the bottom electrode and conducted to both ends of the functional layer. When the potential difference between the two ends of the functional layer exceeds the threshold voltage of the functional layer material, the functional layer jumps from a high-resistivity state to a low-resistivity state. The capacitor discharges through the memristor element until the voltage division of the functional layer is lower than the holding voltage of the functional layer material. At this point, the functional layer returns to the high configuration, thus completing one pulse firing of the neuron circuit.

[0066] By adjusting the magnitude of the modulated voltage input applied to the bottom electrode, the threshold switching behavior of the neuron circuit is linearly shifted along the input direction.

[0067] Specifically, in this embodiment, the neuron circuit consists of a threshold-adjustable memristor element, an external capacitor connected in parallel, and a resistor connected in series. The resistor R... C One port is connected to the drive power supply V DDThe other terminal of the resistor is connected to the top electrode of the memristor and serves as the voltage output terminal. The top electrode of the memristor receives the voltage input after being divided by the resistor, while the bottom electrode receives a modulated voltage input, i.e., the potential of the bottom electrode is set to V. bias ,like Figure 7 As shown.

[0068] In this neuronal circuit, the potential difference across the functional layer of the memristor element determines the threshold response behavior of the device. A positive voltage bias V is applied to the bottom electrode of the memristor. bias To ensure that the voltage division of the functional layer still exceeds the material's threshold voltage V, th Therefore, the voltage input to the top electrode of the device must be increased accordingly. Furthermore, since the functional layer is not subjected to a thermal field, the switching voltage window of the functional layer material remains unchanged.

[0069] In summary, the threshold switching behavior of the neuron only involves a linear shift along the input direction. Therefore, the modulated input received by this bottom electrode also alters the pulse firing behavior of the neuronal circuit to the same driving voltage. This represents a linear integration of modulated and driving inputs and is considered a type of gain modulation characteristic of biological neurons.

[0070] like Figure 8 As shown, the voltage charges the capacitor until the voltage exceeds the memristor's threshold voltage V. th The neuron fires a pulse, and the capacitor discharges until the voltage drops below the memristor's holding voltage V. hold Let's start the next round of scoring. From Figure 8 Repeatable integral emission behavior can be observed. One integral emission is considered as one pulse.

[0071] Figure 6 This is a schematic diagram of a neuron circuit when the bottom electrode of the memristor element receives modulated input.

[0072] Figure 9 For neuronal circuits with different modulation inputs V b The voltage oscillation waveform under the given conditions. When V b As the source bias of the device increases, the drain voltage of the device also increases in order to maintain the same voltage oscillation window, which manifests as a linear upward shift in the voltage oscillation waveform of the neuron.

[0073] Figure 10This graph shows the change in pulse response frequency of a neuron circuit under a driving input as a function of a modulating input. As the modulating voltage increases, the driving voltage required for the memristor to reach its threshold also increases. The integration process slows down under the same driving input, and the pulse firing frequency decreases. However, this frequency reduction can be compensated for by increasing the driving voltage by the same amount as the modulating input; the input-output curve in the graph shifts to the right as the modulating voltage increases. This modulation behavior is similar to the gain-subtraction modulation characteristics observed in biological systems.

[0074] The preferred embodiments of the present invention have been described in detail above. It should be understood that those skilled in the art can make numerous modifications and variations based on the concept of the present invention without creative effort. Therefore, all technical solutions that can be obtained by those skilled in the art based on the concept of the present invention through logical analysis, reasoning, or limited experimentation on the basis of existing technology should be within the scope of protection defined by the claims.

Claims

1. A neuron circuit with adjustable gain, characterized in that, It includes a memristor, a capacitor, and a resistor. One end of the memristor is connected in parallel with the capacitor and in series with the resistor, and is connected to a driving power supply. The memristor includes a MIM structure, a heating module that provides a thermal field for the MIM structure, and a substrate that provides support for the entire memristor. The MIM structure includes a top electrode, a functional layer, and a bottom electrode arranged sequentially from top to bottom. The heating module includes a thermally conductive layer and a heating layer arranged sequentially. The heating layer transfers heat to the MIM structure through contact with the thermally conductive layer. The top electrode is connected to a driving voltage input, and the bottom electrode is grounded; one end of the heating layer is connected to a modulation voltage input, and the other end is grounded; the port where the memristor element is connected to the resistor is the voltage output terminal. The resistor, memristor, and capacitor form a charging circuit, and the memristor and capacitor form a discharging circuit. When the driving power supply is working, one end of the functional layer receives a driving voltage input and charges the capacitor. When the voltage across the functional layer exceeds the threshold voltage of the functional layer material, the functional layer jumps from a high-resistivity state to a low-resistivity state, and the capacitor discharges through the memristor element until the voltage across the functional layer is lower than the holding voltage of the functional layer material. Then the functional layer returns to the high configuration state, thus completing one pulse firing of the neuron circuit.

2. The gain-adjustable neuron circuit according to claim 1, characterized in that, When a modulated voltage input is applied to the heating layer, the heating layer generates a thermal field, which is conducted to the functional layer through the thermally conductive layer, changing the threshold voltage and holding voltage of the functional layer to adjust the impulse response sensitivity of the neuron circuit.

3. The gain-adjustable neuron circuit according to claim 1, characterized in that, The substrate is a silicon oxide substrate.

4. A control method for a gain-adjustable neuron circuit as described in any one of claims 1-3, characterized in that, Includes the following steps: The driving power supply inputs voltage to the neuron circuit, applies a driving voltage input to the top electrode and charges the capacitor, which increases the voltage across the functional layer. When the voltage across the functional layer exceeds the threshold voltage, the material of the functional layer jumps from a high-resistivity state to a low-resistivity state and is discharged by the capacitor. When the voltage across the functional layer is lower than the holding voltage of the material, the material of the functional layer returns to a high-resistivity state, completing one pulse firing of the neuron circuit. After applying a modulated voltage input to the heating layer, the thermal field generated by the heating layer is conducted to the functional layer, changing the threshold voltage and holding voltage of the functional layer material, and thus changing the pulse firing frequency of the neuron circuit.

5. A neuron circuit with adjustable gain, characterized in that, It includes a memristor, a capacitor, and a resistor, wherein one end of the memristor is connected in parallel with the capacitor and then in series with the resistor; The memristor element includes a MIM structure, which includes a top electrode, a functional layer, and a bottom electrode arranged sequentially from top to bottom. One end of the resistor is connected to a driving power supply, and the other end is connected to the top electrode and serves as a voltage output terminal. The top electrode receives the voltage input after being divided by the resistor. The bottom electrode receives a modulated voltage input. The resistor, memristor, and capacitor form a charging circuit, and the memristor and capacitor form a discharging circuit. When the driving power supply is working, one end of the functional layer receives a driving voltage input and charges the capacitor. When the voltage across the functional layer exceeds the threshold voltage of the functional layer material, the functional layer jumps from a high-resistivity state to a low-resistivity state, and the capacitor discharges through the memristor element until the voltage across the functional layer is lower than the holding voltage of the functional layer material. Then the functional layer returns to the high configuration state, thus completing one pulse firing of the neuron circuit.

6. The gain-adjustable neuron circuit according to claim 5, characterized in that, By applying a modulated voltage input to the bottom electrode, the voltage input to the top electrode is adjusted accordingly, thereby achieving a linear translational adjustment of the threshold switching behavior of the neuron circuit along the input direction.

7. The gain-adjustable neuron circuit according to claim 5, characterized in that, When the driving power supply is working, one end of the functional layer receives the voltage input after the voltage is divided by the resistor from the top electrode, and the other end receives the modulated voltage input from the bottom electrode. When the potential difference between the two ends of the functional layer exceeds the threshold voltage of the functional layer material, the functional layer jumps from a high-resistivity state to a low-resistivity state. The capacitor discharges through the memristor element until the voltage division of the functional layer is lower than the holding voltage of the functional layer material. Then the functional layer returns to the high configuration, thereby completing one pulse firing of the neuron circuit.

8. A control method for a gain-adjustable neuron circuit as described in any one of claims 5-7, characterized in that, Includes the following steps: The driving power supply inputs voltage to the neuron circuit, which is then applied to the top electrode after being divided by a resistor. A modulated voltage input is applied to the bottom electrode and conducted to both ends of the functional layer. When the potential difference between the two ends of the functional layer exceeds the threshold voltage of the functional layer material, the functional layer jumps from a high-resistivity state to a low-resistivity state. The capacitor discharges through the memristor element until the voltage division of the functional layer is lower than the holding voltage of the functional layer material. At this point, the functional layer returns to the high configuration, thus completing one pulse firing of the neuron circuit. By adjusting the magnitude of the modulated voltage input applied to the bottom electrode, the threshold switching behavior of the neuron circuit is linearly shifted along the input direction.

Images