Time-synapse simulation system and method based on ltspice

Abstract

The application discloses a LTspice-based time fusion neuron simulation system, which comprises a time module, a comparison module, a state module and a neuron module. The time module is used for determining a starting time of neuron work according to an input voltage of an input time. The comparison module is used for calculating a third voltage of a current time according to a first voltage and a second voltage of a previous time. The state module is used for generating the first voltage of the current time according to the third voltage of the current time. The neuron module is used for calculating the second voltage of the current time according to the starting time of neuron work, the third voltage of the current time, an on-state resistance and an off-state resistance. The comparison module is further used for generating a third voltage of a next time based on the first voltage and the second voltage of the current time, so as to realize continuous simulation of the neuron. Through the above structure, the system can simulate the dynamic characteristics of the LIF neuron circuit affected by the parasitic capacitance, and the accuracy and practicability of LIF neuron modeling and simulation are significantly improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of electronic information, in particular to a neuron simulation system and method based on LTspice and time fusion. BACKGROUND

[0002] The present application relates to the field of neuron circuit modeling and simulation, in particular to a neuron simulation system based on LTspice and time fusion, belonging to the cross technical field of neuromorphic computing and circuit simulation.

[0003] The existing LIF (Leaky Integrate-and-Fire) neuron model is usually modeled and simulated by means of analog circuit structure in hardware implementation to study its discharge mechanism and pulse response behavior. For this purpose, operational amplifiers, capacitors and switch resistors and other analog devices are widely used to simulate the dynamic change process of neuron membrane potential. However, in traditional circuit simulation, parasitic capacitance effect exists universally, resulting in a time delay of neuron response that cannot be ignored, affecting the accuracy and simulation efficiency of the model.

[0004] In addition, the existing simulation method is mostly based on numerical integration calculation, lacking in-depth characterization of the "time-state" coupling relationship in the circuit, and it is difficult to effectively simulate the state evolution at each time caused by the characteristics of physical devices in the real neuron discharge process. At the same time, the mainstream neuron modeling tools mostly use pure software neural network framework, which cannot be directly integrated with LTspice and other circuit simulation platforms, limiting the verification and optimization ability at the hardware circuit level.

[0005] Therefore, there is an urgent need for a neuron modeling system based on a circuit simulation platform, with time advancing function and capable of simulating the dynamic evolution of voltage with time, to more realistically reflect the original neuron discharge process affected by parasitic capacitance, and to improve the guidance value and simulation accuracy of hardware implementation. SUMMARY

[0006] In order to solve the above problems existing in the prior art, the present application provides a neuron simulation system and method based on LTspice and time fusion.

[0007] According to a first aspect of an embodiment of the present application, a neuron simulation system based on LTspice and time fusion is provided, the system comprising: a time module, a comparison module, a state module and a neuron module; the time module comprising: a time power supply, a time measuring resistor R t1 , a first voltage dividing resistor R t2 and a time RC series network, the time RC series network comprising a time resistor R and a time capacitor C; the neuron module comprising a volatile memristor R1; the positive electrode of the time power supply is connected with the time measuring resistor R t1The first terminal is connected, and the negative terminal of the time power supply is grounded; the timing resistor R t1 The second terminal is connected to the first voltage divider resistor R t2 The first terminal is connected; the first voltage divider resistor R t2 The second terminal is grounded; the first terminal of the time RC series network is grounded, and the second terminal is connected to the first terminal of the volatile memristor R1;

[0008] The timing module, after acquiring the input voltage at the input moment, transfers charge to the timing RC series network through the first terminal of the volatile memristor R1 to achieve charging, and the voltage V across the timing capacitor C... t As time increases, the voltage V across the time capacitor C is determined based on the potential at the first terminal of the volatile memristor R1 during the initial stage of the simulation. t According to the voltage V t Calculate the time-measuring resistance R t1 The resistance value is based on the time-measuring resistor R. t1 The resistance value and the first voltage divider resistor R t2 The resistance value determines the first voltage divider resistor R. t2 The voltage V4 at both ends is used to determine the moment when the neuron starts working, based on the product of the voltage V4, the time resistance R, and the time capacitance C.

[0009] The comparison module is used to calculate the third voltage at the current moment based on the first voltage and the second voltage at the previous moment.

[0010] The state module is used to generate the first voltage at the current moment based on the third voltage at the current moment;

[0011] The neuron module is used to calculate the second voltage at the current moment based on the time when the neuron starts working, the third voltage at the current moment, the on-state resistance, and the off-state resistance; wherein the on-state resistance and the off-state resistance are determined according to the state of the volatile memristor R1;

[0012] The comparison module is further configured to calculate the third voltage at the next moment based on the first voltage at the current moment and the second voltage at the current moment.

[0013] Optionally, the time-measuring resistor R t1 The formula for calculating the resistance is as follows:

[0014] ;

[0015] Among them, V t It is the voltage across the time capacitor C in the time RC series network.

[0016] Optionally, when the voltage of the power supply is 1 volt, the first voltage divider resistor R t2 When the resistance is 1 ohm, the formula for calculating the time when the neuron starts working is as follows:

[0017] ;

[0018] Wherein, V4 is the first voltage divider resistor R t2 The voltage across the two ends, RC is the product of the time resistance R and the time capacitance C in the time-RC series network, and t is the moment when the neuron starts working.

[0019] Optionally, the formula for calculating the resistance value of the volatile memristor R1 at the current moment is as follows:

[0020] ;

[0021] Wherein, R1 is the resistance value of the volatile memristor R1 at the current moment. R is the voltage value of the third voltage at the current moment. on R is the resistance value of the on-state resistor. off The resistance value of the off-state resistor. The first voltage divider resistor R t2 The voltage at both ends.

[0022] Optionally, the neuron module further includes: a first power supply and a load resistor R. load And film capacitor C1; the positive terminal of the first power supply is connected to the first terminal of the volatile memristor R1, and the negative terminal of the first power supply is grounded; the second terminal of the volatile memristor R1 is connected to the load resistor R load The first terminal is connected to the load resistor R. load The second terminal is grounded; the film capacitor C1 is connected in parallel with the volatile memristor R1; the second voltage is the potential difference across the volatile memristor R1;

[0023] The neuron module is further configured to determine the resistance value of the volatile memristor R1 at the current moment based on the third voltage at the current moment, the on-state resistance, and the off-state resistance; the neuron module is further configured to determine the resistance value of the volatile memristor R1 at the current moment based on the time when the neuron starts working, the voltage of the first power supply, and the load resistance R. load The second voltage at the current moment is calculated from the resistance value of the volatile memristor R1 at the current moment.

[0024] Optionally, the comparison module comprises a second power supply, a third power supply, a comparison resistor R2, a second voltage dividing resistor R3, a voltage transmission resistor R4, a third voltage dividing resistor R5 and a capacitor C2; a positive electrode of the second power supply is connected with a first end of the second voltage dividing resistor R3, and a negative electrode of the second power supply is grounded; a second end of the second voltage dividing resistor R3 is connected with a first end of the comparison resistor R2, and a second end of the comparison resistor R2 is grounded; the voltage transmission resistor R4 is connected with the comparison resistor R2 in parallel, a first end of the voltage transmission resistor R4 is connected with a first end of the third voltage dividing resistor R5, and a second end of the voltage transmission resistor R4 is grounded; a second end of the third voltage dividing resistor R5 is connected with a first end of the capacitor C2, a second end of the capacitor C2 is connected with a positive electrode of the third power supply, a negative electrode of the third power supply is grounded, and the third voltage is a potential of the first end of the capacitor C2;

[0025] The comparison module is further configured to determine a resistance value of the comparison resistor R2 at a current moment according to the first voltage at the previous moment and the second voltage at the previous moment.

[0026] The comparison module is further configured to determine a voltage value across the comparison resistor R2 at the current moment according to a voltage value of the second power supply, a resistance value of the second voltage dividing resistor R3 and the resistance value of the comparison resistor R2 at the current moment.

[0027] The comparison module is further configured to determine a voltage value across the voltage transmission resistor R4 at the current moment according to the voltage value across the comparison resistor R2 at the current moment.

[0028] The comparison module is further configured to determine the potential of the first end of the capacitor C2 according to the voltage value across the voltage transmission resistor R4 at the current moment and the third voltage dividing resistor R5, so as to obtain the third voltage at the current moment.

[0029] Optionally, the resistance value of the comparison resistor R2 at the current moment is calculated according to the following formula:

[0030] ;

[0031] Wherein, R2 is the resistance value of the comparison resistor R2 at the current moment, V1 is the voltage value of the first voltage at the previous moment, and V2 is the voltage value of the second voltage at the previous moment.

[0032] Optionally, the state module comprises a fourth power supply, a fifth power supply, a state resistor R7 and a fourth voltage dividing resistor R6; a positive electrode of the fourth power supply is connected with a first end of the fourth voltage dividing resistor R6, a second end of the fourth voltage dividing resistor R6 is connected with a first end of the state resistor R7, a second end of the state resistor R7 is connected with a positive electrode of the fifth power supply, a negative electrode of the fourth power supply is connected with the positive electrode of the fifth power supply, and a negative electrode of the fifth power supply is grounded.

[0033] The status module is also used to determine the resistance value of the status resistor R7 at the current time based on the third voltage at the current time.

[0034] The state module is also used to calculate the first voltage at the current moment based on the resistance value of the state resistor R7, the fourth voltage divider resistor R6 and the fifth power supply, wherein the first voltage is the potential at the first end of the state resistor R7.

[0035] According to a second aspect of the present invention, a neuron simulation method based on LTspice is provided, applied to a time-fused neuron simulation system based on LTspice, the system comprising: a time module, a comparison module, a state module, and a neuron module; the time module comprising: a time power supply and a time-measuring resistor R. t1 First voltage divider resistor R t2 The time-controlled RC series network includes a time resistor R and a time capacitor C; the neuron module includes a volatile memristor R1; the positive terminal of the time power supply is connected to the time-measuring resistor R. t1 The first terminal is connected, and the negative terminal of the time power supply is grounded; the timing resistor R t1 The second terminal is connected to the first voltage divider resistor R t2 The first terminal is connected; the first voltage divider resistor R t2 The second terminal is grounded; the first terminal of the time RC series network is grounded, and the second terminal is connected to the first terminal of the volatile memristor R1; the method includes:

[0036] After acquiring the input voltage at the input moment, charge is transferred to the time RC series network through the first terminal of the volatile memristor R1 to achieve charging. The voltage V across the time capacitor C is... t As time increases, the voltage V across the time capacitor C is determined based on the potential at the first terminal of the volatile memristor R1 during the initial stage of the simulation. t According to the voltage V t Calculate the time-measuring resistance R t1 The resistance value is based on the time-measuring resistor R. t1 The resistance value and the first voltage divider resistor R t2 The resistance value determines the first voltage divider resistor R. t2 The voltage V4 at both ends is used to determine the moment when the neuron starts working, based on the product of the voltage V4, the time resistor R, and the time capacitor C.

[0037] Calculate the third voltage at the current moment based on the first voltage and the second voltage at the previous moment;

[0038] generate the first voltage at the current time according to the third voltage at the current time;

[0039] generate the second voltage at the current time according to the time when the neuron starts to work, the third voltage at the current time, the on-state resistance and the off-state resistance, wherein the on-state resistance and the off-state resistance are determined by the state of the volatile memristor R1;

[0040] generate the third voltage at the next time according to the first voltage at the current time and the second voltage at the current time, so as to realize the simulation of the neuron.

[0041] Optionally, the system further comprises a first power supply, a load resistance R load and a film capacitor C1; the positive pole of the first power supply is connected to the first end of the volatile memristor R1, and the negative pole of the first power supply is grounded; the second end of the volatile memristor R1 is connected to the first end of the load resistance R load , and the second end of the load resistance R load is grounded; the film capacitor C1 is connected in parallel with the volatile memristor R1; and the second voltage is the potential difference between the two ends of the volatile memristor R1;

[0042] The generating the second voltage at the current time according to the time when the neuron starts to work, the third voltage at the current time, the on-state resistance and the off-state resistance comprises:

[0043] determining the resistance value of the volatile memristor R1 at the current time according to the third voltage at the current time, the on-state resistance and the off-state resistance;

[0044] generating the second voltage at the current time according to the time when the neuron starts to work, the voltage of the first power supply, the resistance value of the load resistance R load and the resistance value of the volatile memristor R1 at the current time.

[0045] The technical scheme provided by the embodiment of the application can have the following beneficial effects:

[0046] In the above technical scheme, a neuron simulation system based on LTspice and time fusion is provided, which comprises a time module, a comparison module, a state module and a neuron module; the time module comprises a time power supply, a time measurement resistance R t1 , a first voltage division resistance R t2 and a time RC series network, the time RC series network comprising a time resistance R and a time capacitor C; the neuron module comprises a volatile memristor R1; the positive pole of the time power supply is connected to the first end of the time measurement resistance R t1 , and the negative pole of the time power supply is grounded; the second end of the time measurement resistance R t1 is connected to the first voltage division resistance R t2a first end of the first voltage dividing resistor R t2 a second end of the first voltage dividing resistor R t a voltage across the time capacitor C is determined according to the potential at the first end of the volatile memristor R1 in the initial stage of simulation t a voltage V across the time capacitor C is determined according to the potential at the first end of the volatile memristor R1 in the initial stage of simulation t a resistance value of the timing resistor R t1 a voltage V4 across the first voltage dividing resistor R t1 a voltage V4 across the first voltage dividing resistor R t2 a voltage V4 across the first voltage dividing resistor R t2 a time when the neuron starts to work is determined according to the product of the voltage V4, the time resistor R and the time capacitor C; the comparison module is configured to calculate a third voltage at a current time according to the first voltage and the second voltage at a previous time; the state module is configured to generate the first voltage at the current time according to the third voltage at the current time; the neuron module is configured to calculate a second voltage at the current time according to the time when the neuron starts to work, the third voltage at the current time, an on-state resistance and an off-state resistance; the on-state resistance and the off-state resistance are determined according to the state of the volatile memristor R1; the comparison module is further configured to calculate the third voltage at a next time according to the first voltage at the current time and the second voltage at the current time, so as to realize continuous simulation of the neuron. Through the technical scheme, the system deduces the voltage at the current time according to the voltage at the previous time, and then deduces the voltage at the next time according to the voltage at the current time, so as to form a cyclic recursive relationship. This mechanism is consistent with the characteristics of the LIF neuron circuit affected by the parasitic capacitance effect, realizes the neuron circuit simulation based on the computer and the time fusion, and effectively promotes the development of the neuron model research.

[0047] Other features and advantages of the present application will be described in detail in the following detailed description. BRIEF DESCRIPTION OF DRAWINGS

[0048] The accompanying drawings are included to provide a further understanding of the present application, and constitute a part of the specification, and are used together with the following detailed description to explain the present application, but do not constitute a limitation of the present application. In the drawings:

[0049] Figure 1 is a schematic diagram of a LTspice-based neuron simulation system with time fusion according to an exemplary embodiment;

[0050] Figure 2 is a schematic diagram of a time module according to an example embodiment;

[0051] Figure 3 is a schematic diagram of a neuron module according to an example embodiment;

[0052] Figure 4 is a simulation plot of a second voltage output by a neuron module according to an example embodiment;

[0053] Figure 5 is a plot of an actual measured waveform of a second voltage output by a neuron circuit according to an example embodiment;

[0054] Figure 6 is a schematic diagram of a comparison module according to an example embodiment;

[0055] Figure 7 is a schematic diagram of a state module according to an example embodiment;

[0056] Figure 8 is a flowchart of an LTspice-based neuron simulation method according to an example embodiment. DETAILED DESCRIPTION

[0057] The example embodiments will be described in detail herein with reference to the attached drawings. The following description is made with reference to the accompanying drawings in which like reference numerals refer to like elements, unless the context dictates otherwise. The following description of example embodiments is not representative of all embodiments consistent with the present application. Rather, it is merely an example of apparatus and methods consistent with some aspects of the present application as detailed in the appended claims.

[0058] In order to facilitate the understanding of the present application, the related prior art and the inventive concept of the present application will be briefly described first.

[0059] Brain-inspired computing is a computing method that simulates the electrical physiological collaboration process between neurons in the human brain, and is widely used in artificial intelligence, pattern recognition, and control. Unlike the traditional von Neumann architecture, brain-inspired computing does not need to frequently transfer a large amount of data, and can process complex information in a more flexible way. As the core component of brain-inspired computing, neurons have the ability to process a large amount of information in parallel, which makes brain-inspired computing have a significant advantage in simulating human brain information processing.

[0060] In code implementation, the dynamic behavior of LIF neuron is usually simulated by discretizing the continuous-time model (for example, using Euler method or modified Euler method), so as to convert the differential equation into an iterative form for calculation on a digital computer. In recent years, LIF neuron model and its variants are widely used in neuromorphic computing and spiking neural network field, because its structure is relatively simple and can effectively capture the key characteristics of biological neurons. For example, multi-chamber LIF neuron model and efficient neuromorphic learning system are extensions and optimizations of traditional LIF model, aiming to improve network performance and energy efficiency. However, the current research lacks a volatile memristor variable resistance model under the influence of parasitic capacitance, resulting in a lack of a method that can directly simulate the time-fused LIF neuron circuit using a computer, which limits the in-depth development of neuron models.

[0061] Figure 1 is a schematic diagram of an LTspice-based time-fused neuron simulation system according to an example embodiment, as shown in Figure 1 , the system 10 comprises a time module 101, a comparison module 102, a state module 103 and a neuron module 104; the neuron module 104 comprises a volatile memristor R1;

[0062] Figure 2 is a schematic diagram of a time module according to an example embodiment, as shown in Figure 2 , the time module 101 comprises a time power supply, a time resistance R t1 , a first voltage dividing resistance R t2 and a time RC series network, the time RC series network comprises a time resistance R and a time capacitance C; the positive electrode of the time power supply is connected with the first end of the time resistance R t1 , and the negative electrode of the time power supply is grounded; the second end of the time resistance R t1 is connected with the first end of the first voltage dividing resistance R t2 ; the second end of the first voltage dividing resistance R t2 is grounded; the first end of the time RC series network is grounded, and the second end is connected with the first end of the volatile memristor R1;

[0063] The time module 101 is used to deliver charge to the time RC series network through the first end of the volatile memristor R1 to realize charging after obtaining the input voltage at the input time, and the voltage V t across the time capacitance C increases with time, the voltage V t across the time capacitance C is determined according to the potential at the first end of the volatile memristor R1 in the initial stage of simulation, and the time resistance R t is calculated according to the voltage V t1a resistance value of the time resistance R t1 a resistance value of the first voltage dividing resistance R t2 a resistance value of the first voltage dividing resistance R t2 a voltage V4 across the first voltage dividing resistance R

[0064] The comparison module 102 is configured to calculate a third voltage at a current time according to a first voltage at a previous time and a second voltage at the previous time.

[0065] The state module 103 is configured to generate the first voltage at the current time according to the third voltage at the current time.

[0066] The neuron module 104 is configured to calculate a second voltage at the current time according to the time at which the neuron starts to work, the third voltage at the current time, an on-state resistance and an off-state resistance, wherein the on-state resistance and the off-state resistance are determined according to the state of the volatile memristor R1.

[0067] The comparison module 102 is further configured to calculate a third voltage at a next time according to the first voltage at the current time and the second voltage at the current time, so as to realize simulation of the neuron.

[0068] It can be understood that the present application simulates the emulated pulse behavior of the LIF neuron with a parasitic capacitor by using a volatile memristor. The volatile memristor is a passive electronic component similar to a resistor, which can generate and maintain a current through a device. However, unlike a resistor, the resistance value of the volatile memristor is determined by the voltage applied across it, so that it can exhibit different resistance values by applying different voltages, thus having a memory function. It is worth mentioning that the volatile memristor has two important parameters: a threshold voltage V th and a holding voltage V hold The threshold voltage refers to the voltage required for the volatile memristor to switch from one resistance state to another. When the applied voltage exceeds the threshold voltage, the volatile memristor will change its resistance state, for example, from a high resistance state to a low resistance state, and the resistance value will become lower. The holding voltage refers to the minimum voltage required to maintain the state after switching to a certain resistance state, or the state can be maintained without additional voltage. As long as no reverse voltage or current sufficient to trigger a state change is applied, the volatile memristor can maintain its resistance state at zero voltage or below the threshold voltage.

[0069] And, in the present application, the time module is used to determine the time when the neuron starts to work according to the input voltage of the input time; the input and output of the comparison module 102, the state module 103 and the neuron module 104 form a cycle, wherein the comparison module 102 is used to calculate the third voltage of the current time according to the first voltage of the last time and the second voltage of the last time; the state module 103 is used to generate the first voltage of the current time according to the third voltage of the current time; the neuron module 104 is used to calculate the second voltage of the current time according to the time when the neuron starts to work, the third voltage of the current time, the on-state resistance and the off-state resistance; wherein the on-state resistance and the off-state resistance are determined according to the state of the volatile memristor R1; finally, the first voltage and the second voltage of the current time are used by the comparison module 102 to obtain the third voltage of the next time, thereby forming a cycle. Due to the cycle relationship between the input and output of the comparison module 102, the state module 103 and the neuron module 104, the changes of the first voltage and the second voltage of the last time affect the third voltage of the current time, and the third voltage of the current time affects the first voltage and the second voltage of the current time, therefore, the voltage changes in the whole neuron simulation system present a certain rule, which conforms to the characteristics of the LIF neuron circuit with parasitic capacitance, and the effective simulation of the LIF neuron circuit is realized.

[0070] In the technical scheme, a neuron simulation system based on LTspice and time fusion is provided, which comprises a time module, a comparison module, a state module and a neuron module; the neuron module comprises a volatile memristor; the time module is used for determining a time when the neuron starts to work according to an input voltage at an input time; the comparison module is used for obtaining a third voltage at a current time according to a first voltage at a previous time and a second voltage at the previous time; the state module is used for obtaining the first voltage at the current time according to the third voltage at the current time; and the neuron module is used for obtaining the second voltage at the current time according to the third voltage at the current time, an on-state resistance and an off-state resistance, wherein the on-state resistance and the off-state resistance are determined by the volatile memristor; and the comparison module is further used for obtaining a third voltage at a next time according to the first voltage at the current time and the second voltage at the current time, so as to realize simulation of the neuron. Through the technical scheme, the system calculates the voltage at the current time according to the voltage at the previous time, and then calculates the voltage at the next time according to the voltage at the current time, so as to form a cyclic relationship among the voltages, the voltage at the previous time influences the voltage at the current time, and then influences the voltage at the next time, the cyclic process is consistent with the characteristics of the LIF neuron circuit, the LIF neuron circuit affected by the parasitic capacitor is effectively simulated by using the computer, and the development of the neuron model research is promoted. The input voltage, the volatile memristor potential, the charging and discharging process of the time RC series network and the neuron starting time are bound by the time module, so that the working state (such as on / off state switching of the volatile memristor R1) of the neuron directly depends on the time accumulation effect, is closer to the real behavior of the biological neuron “delayed discharge after stimulation” and “time integration of input”, reduces the energy loss caused by invalid resistance change (such as avoiding the false triggering of the volatile memristor in the non-working period), and can realize more accurate simulation of the time integration and delay response characteristics of the biological neuron; the network is endowed with time coding ability, supports time sequence dependent learning and dynamic signal processing; the resistance change regulation of the volatile memristor is optimized, the energy consumption is reduced, and the neuromorphic hardware is adapted; the system timing logic is improved, the difference between simulation and actual circuit is reduced, and the model is promoted from theoretical research to engineering application.

[0071] Optionally, the time measuring resistor R t1 The resistance calculation formula of the time measuring resistor R

[0072] The resistance calculation formula of the time measuring resistor R

[0073] The resistance calculation formula of the time measuring resistor R t The voltage V t The resistance of the time measuring resistor R t1 is accurately controlled, and the resistance of the time measuring resistor R The threshold voltage is 1.6V. The threshold voltage is 1.6V.

[0074] Optionally, in the case that the voltage of the time power supply is 1 volt and the resistance of the first voltage dividing resistor R t2 is 1 ohm, the calculation formula of the time when the neuron starts to work is shown in the following formula (2):

[0075] Formula (2);

[0076] Wherein, V4 is the voltage between the two ends of the first voltage dividing resistor R t2 , RC is the product of the size of the time resistance and the time capacitance in the time RC network, and t is the size of the time when the neuron starts to work. In the time module, the timing resistor R t1 and the first voltage dividing resistor R t2 constitute a series voltage dividing circuit, and the voltage V4 between the two ends of the first voltage dividing resistor R t1 needs to be regulated by the timing resistor R t2 .

[0077] The voltage of the time power supply can be represented as V cc . Since the timing resistor R t1 and the first voltage dividing resistor R t2 are in series voltage division, in the case that the voltage of the time power supply is 1 volt and the resistance of the first voltage dividing resistor R t2 is 1 ohm, the voltage V4 between the two ends of the first voltage dividing resistor R t2 can be represented by the following formula (3):

[0078] Formula (3);

[0079] Since , .

[0080] During the charging process of the time RC series network, the voltage V t between the two ends of the capacitor C changes with time t, which can be represented by the following formula (4):

[0081] Formula (4);

[0082] In the case that the voltage of the time power supply is 1 volt, the in formula (4) is substituted into formula (3), and formula (2) can be derived.

[0083] It can be understood that the size of R and C in the RC network is set according to actual conditions, the first end of the volatile memristor R1 charges the RC network, and the charging time constant is determined by the value of RC, so the voltage at both ends of C in the RC network is determined by the voltage at the first end of the memristor R1, thereby determining the resistance value of the measuring resistor R t1 , then determining the voltage V4 at both ends of the first voltage dividing resistor R t2 , and finally obtaining the size t of the starting working moment; wherein the value of the RC network can generally take R as 0.25 ohm and C as 20 microfarad, the resistance value of the first voltage dividing resistor R t2 is generally small, for example, it can be set as 1 ohm, and the voltage of the time power supply is generally small, for example, it can be set as 1 volt.

[0084] Alternatively, the resistance value of the volatile memristor R1 at the current moment is calculated according to the following formula (5):

[0085] Formula (5);

[0086] wherein R1 is the resistance value of the volatile memristor R1 at the current moment, is the voltage value of the third voltage at the current moment, R on is the resistance value of the on-state resistor, R off is the resistance value of the off-state resistor, is the voltage at both ends of the first voltage dividing resistor R t2 .

[0087] wherein, is the voltage regulation of the basic resistance state, and the resistance state is controlled by the voltage , the resistance state is switched between R on and R off , when tends to 0 (approaches the off state), the resistance state should tend to R on , and when tends to 1 (approaches the on state), the resistance state should tend to R off . the arctangent function (x) is introduced to simulate the smooth change of the resistance state in the transition interval, the fifth power is introduced to strengthen the nonlinearity, so that is more sensitive to the resistance state, and adapts to the demand of the neuron simulation for fast response, is used for normalization, and is adapted to the voltage , and is used for amplitude calibration, to ensure that the maximum value of the change of the resistance state does not exceed the off-state resistance , and the two parts are added to cover the full characteristics of the smooth change of the volatile memristor R1 from the basic resistance state to the transition state, and finally adapt to the demand of the dynamic change of the membrane potential in the neuron simulation.

[0088] It can be understood that according to the above formula, the resistance value of the volatile memristor R1 is first determined by the time module whether it is in a high resistance state for a long time, and secondly changes with the voltage value of the third voltage at the current time. The resistance value of the volatile memristor R1 will switch between the resistance value of the on-state resistance and the resistance value of the off-state resistance with the voltage value of the third voltage at the current time, to simulate the LIF neuron firing pulse behavior. The resistance value of the on-state resistance can be 2000 ohms, and the resistance value of the off-state resistance can be 80000 ohms.

[0089] It is worth mentioning that when the resistance value of the volatile memristor R1 is determined, the neuron module 104 can determine the potential difference between the two ends of the volatile memristor R1, that is, the potential difference between the first end of the membrane capacitor C1 and the lower plate (the second end), that is, the second voltage, according to the resistance value of the load resistance R load and the resistance value of the volatile memristor R1. And a resistance with a fixed resistance value can be added between the positive electrode of the first power supply and the first end of the volatile memristor R1, to avoid the direct connection of the membrane capacitor C1 and the volatile memristor R1 with the first power supply.

[0090] Optionally, Figure 3 is a schematic diagram of a neuron module according to an exemplary embodiment, as shown in Figure 3 The neuron module 104 further comprises: a first power supply, a load resistance R load and a membrane capacitor C1; the positive electrode of the first power supply is connected with the first end of the volatile memristor R1, the negative electrode of the first power supply is grounded, the second end of the volatile memristor R1 is connected with the first end of the load resistance R load , the second end of the load resistance R load is grounded, and the membrane capacitor C1 is connected in parallel with the volatile memristor R1; the second voltage is the potential difference between the two ends of the volatile memristor R1;

[0091] The neuron module 104 is further configured to determine the resistance value of the volatile memristor R1 at the current time according to the third voltage at the current time, the on-state resistance and the off-state resistance;

[0092] The neuron module 104 is further configured to calculate the second voltage at the current time according to the time when the neuron starts to work, the voltage of the first power supply, the resistance value of the load resistance R load and the resistance value of the volatile memristor R1 at the current time.

[0093] It can be understood that the first power supply is a voltage-adjustable power supply; the resistance value of the load resistance R load is usually between the resistance value of the on-state resistance and the resistance value of the off-state resistance, and the resistance value of the load resistance R load is adjustable. When the resistance value of the load resistance Rload The greater the resistance of the load resistor R load The smaller the resistance of the load resistor R

[0094] In an embodiment, Figure 4 is a schematic diagram of a second voltage output by a neuron module according to an example embodiment, as Figure 4 shown, the green curve represents the first voltage V(n004) applied, and the blue curve represents the second voltage V(vbot) output by the LTspice-based neuron simulation system, with time on the horizontal axis and voltage on the vertical axis. Figure 5 is a schematic diagram of a voltage output by a LIF neuron circuit with parasitic capacitance according to an example embodiment, as Figure 5 shown, the yellow curve represents the voltage output by the LIF neuron, with time on the horizontal axis and voltage on the vertical axis.

[0095] Optionally, Figure 6 is a schematic diagram of a comparison module according to an example embodiment, as Figure 6 shown, the comparison module 102 includes a second power supply, a third power supply, a comparison resistor R2, a second voltage dividing resistor R3, a voltage transmission resistor R4, a third voltage dividing resistor R5, and a capacitor C2; the positive electrode of the second power supply is connected to the first end of the second voltage dividing resistor R3, and the negative electrode of the second power supply is grounded; the second end of the second voltage dividing resistor R3 is connected to the first end of the comparison resistor R2, and the second end of the comparison resistor R2 is grounded; the voltage transmission resistor R4 is connected in parallel with the comparison resistor R2, the first end of the voltage transmission resistor R4 is connected to the first end of the third voltage dividing resistor R5, and the second end of the voltage transmission resistor R4 is grounded; the second end of the third voltage dividing resistor R5 is connected to the first end of the capacitor C2, the second end of the capacitor C2 is connected to the positive electrode of the third power supply, the negative electrode of the third power supply is grounded, and the third voltage is the potential at the first end of the capacitor C2;

[0096] The comparison module 102 is further configured to determine the resistance of the comparison resistor R2 at the current time according to the first voltage at the previous time and the second voltage at the previous time.

[0097] The comparison module 102 is further configured to determine the voltage across the comparison resistor R2 at the current time according to the voltage value of the second power supply, the resistance of the second voltage dividing resistor R3, and the resistance of the comparison resistor R2 at the current time.

[0098] The comparison module 102 is further configured to determine the voltage value across the voltage transmission resistor R4 at the current time according to the voltage value across the comparison resistor R2 at the current time;

[0099] The comparison module 102 is further configured to determine the first end potential of the capacitor C2 according to the voltage value across the voltage transmission resistor R4 at the current time and the third voltage dividing resistor R5, thereby obtaining the third voltage at the current time.

[0100] Optionally, the resistance value of the comparison resistor R2 at the current time is calculated according to the following formula (6):

[0101] Formula (6);

[0102] Wherein, R2 is the resistance value of the comparison resistor R2 at the current time, V1 is the voltage value of the first voltage at the previous time, and V2 is the voltage value of the second voltage at the previous time.

[0103] Wherein, The arctangent function is used to amplify the voltage difference (x) makes the resistance state change have sensitive response and stable saturation nonlinear characteristics, which is more suitable for neuron simulation, when x is very small, (x) approaches x, when x is very large, (x) approaches , which can limit the resistance value range, multiplied by After that, the output range is (-1, 1), plus 1 becomes (0, 2), and then divided by 2, finally makes the calculation result of R2 between (0, 1).

[0104] It can be understood that the sizes of the second power supply and the third power supply are set according to actual conditions, and the resistance value of the comparison resistor R2 at the current moment is related to the voltage value of the first voltage at the last moment and the voltage value of the second voltage at the last moment. That is, when the voltage value of the first voltage at the last moment is greater than the voltage value of the second voltage at the last moment, the resistance value of the comparison resistor R2 is 2 ohms; when the voltage value of the first voltage at the last moment is less than the voltage value of the second voltage at the last moment, the resistance value of the comparison resistor R2 is 0 ohm; when the voltage value of the first voltage at the last moment is equal to the voltage value of the second voltage at the last moment, the resistance value of the comparison resistor R2 is 1 ohm. According to the resistance value of the comparison resistor R2 at the current moment and the resistance value of the second voltage dividing resistor R3, the voltage value at the current moment between the two ends of the voltage transmission resistor R4 connected in parallel with the comparison resistor R2 can be obtained, and the potential of the first end of the capacitor C2 is obtained according to the voltage value at the current moment between the two ends of the voltage transmission resistor R4 and the third voltage dividing resistor R5. It is worth mentioning that the role of the third power supply in the comparison module 102 is to provide a bias voltage, so that the current moment third voltage obtained by the comparison voltage is inverted to change the timing of the third voltage output. Among them, the resistance value of the voltage transmission resistor R4 is usually large, and the resistance values of the first voltage dividing resistor R4 and the third voltage dividing resistor R5 are usually small, for example, which can be set to 1 ohm.

[0105] Optionally, Figure 7 is a schematic diagram of a state module according to an exemplary embodiment, as Figure 7 shown, the state module 103 includes: a fourth power supply, a fifth power supply, a state resistor R7 and a fourth voltage dividing resistor R6; the positive electrode of the fourth power supply is connected to the first end of the fourth voltage dividing resistor R6, the second end of the fourth voltage dividing resistor R6 is connected to the first end of the state resistor R7, the second end of the state resistor R7 is connected to the positive electrode of the fifth power supply, the negative electrode of the fourth power supply is connected to the positive electrode of the fifth power supply, and the negative electrode of the fifth power supply is grounded.

[0106] The state module 103 is also used to determine the resistance value of the state resistor R7 at the current moment according to the third voltage at the current moment.

[0107] The state module 103 is also used to calculate the first voltage at the current moment according to the resistance value of the state resistor R7 at the current moment, the fourth voltage dividing resistor R6 and the fifth power supply, and the first voltage is the potential of the first end of the state resistor R7.

[0108] It can be understood that the voltage value of the fourth power supply is set according to the threshold voltage V th and the holding voltage V hold of the volatile memristor, for example, the voltage value of the fourth power supply can be 3* (V th -V hold ). The voltage value of the fifth power supply can be the same as the holding voltage Vhold The voltage value of the first voltage is the same as the voltage value of the third voltage. Since the resistance value of the state resistance R7 at the current moment is the same in value as the third voltage at the current moment, the current in the state module 103 can be obtained after the resistance value of the state resistance R7 at the current moment is obtained, and the first voltage at the current moment can be obtained according to the voltage divided by the fifth power supply and the state resistance R7 at the current moment.

[0109] In an embodiment, when the LTspice-based time-fused neuron simulation system is in an initial state, the volatile memristor R1 is in a high-resistance state, and the resistance value is equal to the off-state resistance. At this time, the potential difference between the two ends of the volatile memristor R1, that is, the value of the second voltage, is high. When the second voltage at this time is greater than the threshold voltage, the first voltage is less than the second voltage. At this time, the resistance value of the comparison resistor R2 is 0 ohm, and the voltage divided by the voltage transmission resistor R4 is 0 V. Therefore, the third voltage at this time is 0 V, and the resistance value of the state resistance R7 in the state module 103 at this time is 0 ohm. Due to the change of the first voltage, the third voltage changes, and the resistance value of the volatile memristor R1 changes. The volatile memristor R1 changes to a low-resistance state. At this time, the voltage divided by the membrane capacitor C1 connected in parallel with the volatile memristor R1 decreases, that is, the second voltage at this time decreases. When the second voltage is less than the first voltage, the resistance value of the comparison resistor R2 changes to 1 ohm, the voltage divided by the voltage transmission resistor R4 increases, the third voltage at this time increases, and the resistance value of the state resistance R7 in the state module 103 is affected. The voltage value of the first voltage changes, which affects the third voltage, and further changes the resistance value of the volatile memristor R1 (the volatile memristor R1 switches to a high-resistance state or a low-resistance state), and further affects the voltage divided by the membrane capacitor C1, that is, the second voltage.

[0110] Through the above technical solution, the time module is used to determine the time when the system starts to work, the resistance value change caused by the voltage value change among the comparison module, the state module and the neuron module is used, and the voltage across the volatile memristor in the neuron module is used to simulate the neuron firing pulse behavior. It is consistent with the characteristics of the LIF neuron circuit affected by the parasitic capacitor, realizes the simulation of the time-fused LIF neuron circuit by using the computer, and promotes the development of the neuron model research to a certain extent.

[0111] Figure 8 is a flow chart of a LTspice-based time-fused neuron simulation method according to an example embodiment. The method is applied to a LTspice-based time-fused neuron simulation system, which includes a time module, a comparison module, a state module and a neuron module. The time module includes a time power supply, a time measurement resistor R t1 , a first voltage dividing resistor R t2and a time RC series network comprising a time resistance R and a time capacitance C; the neuron module comprises a volatile memristor R1; a positive pole of the time power supply is connected with a first end of the time resistance R t1 , and a negative pole of the time power supply is grounded; a second end of the time resistance R t1 is connected with a first end of the first voltage dividing resistance R t2 ; a second end of the first voltage dividing resistance R t2 is grounded; a first end of the time RC series network is grounded, and a second end of the time RC series network is connected with a first end of the volatile memristor R1; as shown in Figure 8 , the method comprises:

[0112] In S701, after obtaining an input voltage at an input time, a charge is transmitted from a first end of the volatile memristor R1 to the time RC series network to realize charging, and a voltage V t across the time capacitance C increases over time; the voltage V t across the time capacitance C is determined according to an electric potential at the first end of the volatile memristor R1 in an initial stage of simulation; the resistance value of the time resistance R t is calculated according to the voltage V t1 ; the voltage V4 across the first voltage dividing resistance R t1 is determined according to the resistance value of the time resistance R t2 and the resistance value of the first voltage dividing resistance R t2 ; and the time at which the neuron starts to work is determined according to a product of the voltage V4, the time resistance R and the time capacitance C.

[0113] In S702, a third voltage at a current time is obtained according to a first voltage at a previous time and a second voltage at the previous time.

[0114] In S703, a first voltage at the current time is generated according to the third voltage at the current time.

[0115] In S704, a second voltage at the current time is calculated according to the time at which the neuron starts to work, the third voltage at the current time, an on-state resistance and an off-state resistance, wherein the on-state resistance and the off-state resistance are determined by a state of the volatile memristor R1.

[0116] In S705, a third voltage at a next time is calculated according to the first voltage at the current time and the second voltage at the current time, so as to realize simulation of the neuron.

[0117] Optionally, the system further comprises a first power supply, a load resistance R loadand a film capacitor C1; a positive pole of the first power supply is connected with a first end of the volatile memristor R1, a negative pole of the first power supply is grounded, a second end of the volatile memristor R1 is connected with a first end of a load resistor R load , a second end of the load resistor R load is grounded, and the film capacitor C1 is connected with the volatile memristor R1 in parallel; the second voltage is a potential difference between two ends of the volatile memristor R1; S704 comprises:

[0118] determining the resistance value of the volatile memristor R1 at the current moment according to the third voltage at the current moment, the on-state resistance and the off-state resistance;

[0119] calculating the second voltage at the current moment according to the moment when the neuron starts to work, the voltage of the first power supply, the resistance value of the load resistor R load and the resistance value of the volatile memristor R1 at the current moment.

[0120] Optionally, the system further comprises a second power supply, a third power supply, a comparison resistor R2, a second voltage dividing resistor R3, a voltage transmission resistor R4, a third voltage dividing resistor R5 and a capacitor C2; a positive pole of the second power supply is connected with a first end of the second voltage dividing resistor R3, and a negative pole of the second power supply is grounded; a second end of the second voltage dividing resistor R3 is connected with a first end of the comparison resistor R2, and a second end of the comparison resistor R2 is grounded; the voltage transmission resistor R4 is connected with the comparison resistor R2 in parallel, a first end of the voltage transmission resistor R4 is connected with a first end of the third voltage dividing resistor R5, and a second end of the voltage transmission resistor R4 is grounded; a second end of the third voltage dividing resistor R5 is connected with a first end of the capacitor C2, a second end of the capacitor C2 is connected with a positive pole of the third power supply, a negative pole of the third power supply is grounded, and the third voltage is a potential at the first end of the capacitor C2; S702 comprises:

[0121] determining the resistance value of the comparison resistor R2 at the current moment according to the first voltage at the last moment and the second voltage at the last moment;

[0122] determining the voltage value between two ends of the comparison resistor R2 at the current moment according to the voltage value of the second power supply, the resistance value of the second voltage dividing resistor R3 and the resistance value of the comparison resistor R2 at the current moment;

[0123] determining the voltage value between two ends of the voltage transmission resistor R4 at the current moment according to the voltage value between two ends of the comparison resistor R2 at the current moment;

[0124] determining the potential at the first end of the capacitor C2 according to the voltage value between two ends of the voltage transmission resistor R4 at the current moment and the third voltage dividing resistor R5, so as to obtain the third voltage at the current moment.

[0125] Optionally, the system further comprises a fourth power supply, a fifth power supply, a state resistor R7 and a fourth voltage dividing resistor R6; a positive pole of the fourth power supply is connected to a first end of the fourth voltage dividing resistor R6, a second end of the fourth voltage dividing resistor R6 is connected to a first end of the state resistor R7, a second end of the state resistor R7 is connected to a positive pole of the fifth power supply, a negative pole of the fourth power supply is connected to the positive pole of the fifth power supply, and a negative pole of the fifth power supply is grounded; S703 comprises:

[0126] determining a resistance value of the state resistor R7 at the current time according to the third voltage at the current time;

[0127] calculating a first voltage at the current time according to the resistance value of the state resistor R7 at the current time, the fourth voltage dividing resistor R6 and the fifth power supply, the first voltage being a potential at a first end of the state resistor R7.

[0128] In the technical solution, a neuron simulation system based on LTspice and time fusion is provided, which comprises a time module, a comparison module, a state module and a neuron module; the neuron module comprises a volatile memristor; the time module is configured to determine a time at which the neuron starts to work according to an input voltage at an input time; the comparison module is configured to obtain a third voltage at a current time according to a first voltage at a previous time and a second voltage at the previous time; the state module is configured to obtain a first voltage at the current time according to the third voltage at the current time; the neuron module is configured to obtain a second voltage at the current time according to the third voltage at the current time, an on-state resistance and an off-state resistance, wherein the on-state resistance and the off-state resistance are determined according to the volatile memristor; and the comparison module is further configured to obtain a third voltage at a next time according to the first voltage at the current time and the second voltage at the current time, so as to realize simulation of the neuron. Through the technical solution, the system obtains a voltage at a current time according to a voltage at a previous time, and then obtains a voltage at a next time according to the voltage at the current time. The voltage at the previous time in the system influences the voltage at the current time, and then influences the voltage at the next time, forming a cycle. The cycle is consistent with the characteristics of the LIF neuron circuit, and thus the simulation of the LIF neuron circuit by using a computer is realized, and the development of the neuron model research is promoted to a certain extent.

[0129] As to the method in the above-mentioned embodiments, the specific manners of each step have been described in detail in the embodiments related to the system, and thus will not be described in detail here.

[0130] The preferred embodiments of the present application are described in detail above with reference to the drawings, but the present application is not limited to the specific details in the above-mentioned embodiments. Various simple modifications can be made to the technical solutions of the present application within the technical concept of the present application, and these simple modifications all belong to the protection scope of the present application.

[0131] It should be further noted that each of the various technical features described in the above embodiments can be combined with any other technical features in any suitable manner, and the present application shall be deemed to disclose all possible combinations thereof, without causing unnecessary repetition.

[0132] Furthermore, any combination of the various embodiments of the present application can be made, as long as it does not deviate from the spirit of the present application, and it shall be deemed to be disclosed by the present application.

Claims

1. An LTspice-based spiking neuron simulation system fused with time, characterized in that, The system comprises a time module, a comparison module, a state module and a neuron module; the time module comprises a time power supply, a time resistance R t1 , a first voltage division resistance R t2 and a time RC series network comprising a time resistance R and a time capacitor C; the neuron module comprises a volatile memristor R1; the positive pole of the time power supply is connected with the first end of the time resistance R t1 , and the negative pole of the time power supply is grounded; the second end of the time resistance R t1 is connected with the first end of the first voltage division resistance R t2 ; the second end of the first voltage division resistance R t2 is grounded; the first end of the time RC series network is grounded, and the second end is connected with the first end of the volatile memristor R1; the time module, for delivering charges to the time RC series network through the first end of the volatile memristor R1 to realize charging after obtaining the input voltage at the input time, the voltage V t across the time capacitor C is determined according to the potential at the first end of the volatile memristor R1 in the initial stage of simulation t , the voltage V t is determined according to the voltage V t1 , the resistance value of the time resistance R t1 , and the resistance value of the first voltage dividing resistor R t2 , the voltage V4 across the first voltage dividing resistor R t2 is determined according to the product of the voltage V4, the time resistance R and the time capacitor C, and the time when the neuron starts to work is determined. The comparison module is configured to calculate the third voltage at the current time according to the first voltage at the previous time and the second voltage at the previous time. The state module is configured to generate the first voltage at the current time according to the third voltage at the current time. The neuron module is configured to calculate the second voltage at the current time according to the time when the neuron starts to work, the third voltage at the current time, the on-state resistance and the off-state resistance, wherein the on-state resistance and the off-state resistance are determined according to the state of the volatile memristor R1. The comparison module is further configured to calculate the third voltage at the next time according to the first voltage at the current time and the second voltage at the current time.

2. The LTspice-based time fused neuron simulation system of claim 1, wherein, The time measuring resistor R t1 The resistance value calculation formula is as follows: ； where V t is the voltage across the time capacitor C in the time RC series network.

3. The LTspice-based time fused neuron simulation system of claim 2, wherein, In the case that the voltage of the time power supply is 1 volt and the resistance of the first voltage dividing resistor R t2 is 1 ohm, the calculation formula of the time when the neuron starts to work is as follows: ； Wherein, V4 is the first voltage division resistor R t2 The voltage at both ends, RC is the product of the size of the time resistance R and the time capacitance C in the time RC series network, and t is the time when the neuron starts to work.

4. The LTspice-based time fused neuron simulation system of claim 3, wherein, The resistance value of the volatile memristor R1 at the current time is calculated according to the following formula: ； wherein R1 is the resistance value of the volatile memristor R1 at the current time, is the voltage value of the third voltage at the current time, R on is the resistance value of the on-state resistance, R off is the resistance value of the off-state resistance, is the voltage value of the first voltage at the current time, R t2 is the voltage value of the second voltage at the current time, R 5. The LTspice-based time fused neuron simulation system of claim 1, wherein, The neuron module further comprises: a first power supply, a load resistor R load , and a film capacitor C1; a positive pole of the first power supply is connected with a first end of the volatile memristor R1, and a negative pole of the first power supply is grounded; a second end of the volatile memristor R1 is connected with a first end of the load resistor R load , and a second end of the load resistor R load is grounded; the film capacitor C1 is connected in parallel with the volatile memristor R1; the second voltage is a potential difference between two ends of the volatile memristor R1. The neuron module is further configured to determine the resistance value of the volatile memristor R1 at the current time according to the third voltage at the current time, the on-state resistance and the off-state resistance; and the neuron module is further configured to calculate the second voltage at the current time according to the time when the neuron starts to work, the voltage of the first power supply, the resistance value of the load resistance R load , and the resistance value of the volatile memristor R1 at the current time.

6. The LTspice-based time fused neuron simulation system of claim 1, wherein, The comparison module includes a second power supply, a third power supply, a comparison resistor R2, a second voltage dividing resistor R3, a voltage transmission resistor R4, a third voltage dividing resistor R5 and a capacitor C2. The positive electrode of the second power supply is connected to the first end of the second voltage dividing resistor R3, and the negative electrode of the second power supply is grounded. The second end of the second voltage dividing resistor R3 is connected to the first end of the comparison resistor R2, and the second end of the comparison resistor R2 is grounded. The voltage transmission resistor R4 is connected in parallel with the comparison resistor R2. The first end of the voltage transmission resistor R4 is connected to the first end of the third voltage dividing resistor R5, and the second end of the voltage transmission resistor R4 is grounded. The second end of the third voltage dividing resistor R5 is connected to the first end of the capacitor C2, the second end of the capacitor C2 is connected to the positive electrode of the third power supply, the negative electrode of the third power supply is grounded, and the third voltage is the potential at the first end of the capacitor C2. The comparison module is further configured to determine the resistance value of the comparison resistor R2 at the current time according to the first voltage at the previous time and the second voltage at the previous time. The comparison module is further configured to determine the voltage value across the comparison resistor R2 at the current time according to the voltage value of the second power supply, the resistance value of the second voltage dividing resistor R3 and the resistance value of the comparison resistor R2 at the current time. The comparison module is further configured to determine the voltage value across the voltage transmission resistor R4 at the current time according to the voltage value across the comparison resistor R2 at the current time. The comparison module is further configured to determine the potential at the first end of the capacitor C2 according to the voltage value across the voltage transmission resistor R4 at the current time and the third voltage dividing resistor R5, so as to obtain the third voltage at the current time.

7. The LTspice-based time fused neuron simulation system of claim 6, wherein, The resistance value of the comparison resistor R2 at the current time is calculated according to the following formula: ； wherein R2 is the resistance value of the comparison resistor R2 at the current time, V1 is the voltage value of the first voltage at the previous time, and V2 is the voltage value of the second voltage at the previous time.

8. The LTspice-based time fused neuron simulation system of claim 1, wherein, The state module comprises a fourth power supply, a fifth power supply, a fourth voltage dividing resistor R6 and a state resistor R7; a positive pole of the fourth power supply is connected to a first end of the fourth voltage dividing resistor R6, a second end of the fourth voltage dividing resistor R6 is connected to a first end of the state resistor R7, a second end of the state resistor R7 is connected to a positive pole of the fifth power supply, a negative pole of the fourth power supply is connected to the positive pole of the fifth power supply, and a negative pole of the fifth power supply is grounded; The state module is further configured to determine a resistance value of the state resistor R7 at the current time according to the third voltage at the current time; The state module is further configured to calculate the first voltage at the current time according to the resistance value of the state resistor R7 at the current time, the fourth voltage dividing resistor R6 and the fifth power supply, and the first voltage is a potential at a first end of the state resistor R7.

9. A method for simulating a neuron based on LTspice, characterized by, The application is applied to a LTspice-based time-fused neuron simulation system, which comprises a time module, a comparison module, a state module and a neuron module; the time module comprises a time power supply, a time resistance R t1 , a first voltage division resistance R t2 and a time RC series network comprising a time resistance R and a time capacitor C; the neuron module comprises a volatile memristor R1; the positive pole of the time power supply is connected with the first end of the time resistance R t1 , and the negative pole of the time power supply is grounded; the second end of the time resistance R t1 is connected with the first end of the first voltage division resistance R t2 ; the second end of the first voltage division resistance R t2 is grounded; the first end of the time RC series network is grounded, and the second end is connected with the first end of the volatile memristor R1; the method comprises the following steps: After acquiring the input voltage at the input instant, the charge is transferred through the first end of the volatile memristor R1 to the time RC series network to achieve charging, and the voltage V across the time capacitor C t With the increase of time, the voltage V across the time capacitor C is determined according to the potential of the first end of the volatile memristor R1 in the initial stage of simulation t , the voltage V t The resistance value of the timing resistor R t1 is calculated, and the voltage V4 across the first voltage dividing resistor R t1 is determined according to the resistance value of the timing resistor R t2 and the resistance value of the first voltage dividing resistor R t2 , and the time when the neuron starts to work is determined according to the product of the voltage V4, the time resistor R and the time capacitor C. The third voltage at the current time is calculated according to the first voltage at the previous time and the second voltage at the previous time; The first voltage at the current time is generated according to the third voltage at the current time; The second voltage at the current time is calculated according to the time when the neuron starts to work, the third voltage at the current time, an on-state resistance and an off-state resistance, wherein the on-state resistance and the off-state resistance are determined by the state of the volatile memristor R1; The third voltage at the next time is calculated according to the first voltage at the current time and the second voltage at the current time, so as to realize the simulation of the neuron.

10. The LTspice-based neuron simulation method of claim 9, wherein, The system further comprises a first power supply, a load resistor R load and a film capacitor C1; a positive pole of the first power supply is connected to a first end of a volatile memristor R1, and a negative pole of the first power supply is grounded; a second end of the volatile memristor R1 is connected to a first end of the load resistor R load , and a second end of the load resistor R load is grounded; the film capacitor C1 is connected in parallel with the volatile memristor R1; the second voltage is a potential difference between the two ends of the volatile memristor R1. The calculation of the second voltage at the current time according to the time when the neuron starts to work, the third voltage at the current time, the on-state resistance and the off-state resistance comprises: The resistance value of the volatile memristor R1 at the current time is determined according to the third voltage at the current time, the on-state resistance and the off-state resistance; The second voltage at the current time instant is calculated according to the time instant when the neuron starts to work, the voltage of the first power supply, the resistance value of the load resistor R load and the resistance value of the volatile memristor R1 at the current time instant.

Images