Patents
Literature
Patsnap Eureka AI that helps you search prior art, draft patents, and assess FTO risks, powered by patent and scientific literature data.

1696 results about "Memristor" patented technology

A memristor (/ˈmɛmrɪstər/; a portmanteau of memory resistor) is a hypothetical non-linear passive two-terminal electrical component relating electric charge and magnetic flux linkage. It was envisioned, and its name coined, in 1971 by circuit theorist Leon Chua. According to the characterizing mathematical relations, the memristor would hypothetically operate in the following way: the memristor's electrical resistance is not constant but depends on the history of current that had previously flowed through the device, i.e., its present resistance depends on how much electric charge has flowed in what direction through it in the past; the device remembers its history—the so-called non-volatility property. When the electric power supply is turned off, the memristor remembers its most recent resistance until it is turned on again.

Resistor disc defect online detection system and grading method based on machine vision

The invention discloses a machine vision-based resistor disc defect online detection system and a grading method, relates to the technical field of industrial machine vision detection, and solves the defect problems in the aspects of multi-scale defect dynamic perception, cross-level feature interaction and process adaptive optimization in the prior art. According to the scheme, metal reflection interference is inhibited through Retinex illumination correction and a combined denoising model; adopting a deformable convolution kernel and cavity space pyramid pooling to realize gradient entropy driving dynamic sensing of the multi-scale defect; constructing a bidirectional cross-layer attention network to realize early fusion of high-resolution details and high-level semantics; modeling local-global feature physical association based on a graph attention network and a self-supervised message passing mechanism; integrating reinforcement learning and a memristor random calculation unit to form a closed-loop parameter optimization system; according to the method, the multi-scale defect detection precision, the cross-modal feature fusion efficiency and the system adaptive capacity under complex working conditions are remarkably improved.
Owner:NANYANG GOLDEN CROWN IND CO LTD

Multistage compression collaborative optimization neural network deployment method and device based on memristor and storage medium

The invention relates to the field of artificial intelligence hardware acceleration, and discloses a memristor-based multilevel compression collaborative optimization neural network deployment method and device, and a storage medium. The method comprises the following steps: carrying out progressive structured pruning on a pre-training model based on a dual-drive scoring mechanism of an L1 norm and gradient sensitivity and hardware feedback, and generating a hardware-friendly sparse weight structure; the characteristics of the memristor are simulated through a micro-nonlinear conductance modeling function, and network weight and conductance parameters are synchronously optimized to reduce errors by adopting mixed precision quantification of four bits of a convolutional layer and two bits of a full-connection layer; a conductance drift and read-write noise model is injected, and the anti-interference capability of the model is improved in combination with adaptive noise enhancement and KL divergence loss; and mapping the optimized model to a memristor memory architecture to complete weight coding and reasoning. Through collaborative optimization of pruning, quantification and distillation, the problems of insufficient storage density, non-ideal characteristic interference and algorithm and hardware mismatch in memristor deployment are solved.
Owner:YANGTZE DELTA REGION INST (QUZHOU) UNIV OF ELECTRONIC SCI & TECH OF CHINA

Biomimetic moving target sensing, storage and calculation integrated photoelectric detection system based on continuous photoconduction

The invention discloses a bionic moving target sensing, storage and calculation integrated photoelectric detection system based on continuous photoconduction, and relates to the field of bionic intelligent vision and the field of brain-like calculation, and the system comprises a synaptic bionic photoelectric detector array and a brain-like memristor array which are interconnected. When the synaptic-like bionic photoelectric detector array detects visual information of a moving target, a fixed bias voltage is applied to a device in each pixel, space illumination information corresponding to each pixel is converted into a continuously attenuated photocurrent signal, illumination information of a multi-frame dynamic target is continuously received in the continuous attenuation process of the photocurrent, and the visual information of the moving target is detected by the synaptic-like bionic photoelectric detector array. After the last frame of illumination information is input, a light current signal of each pixel is read, and a feature map memorizing all historical moment information is determined; the brain-like memristor array is based on the Ohm's law and the Kirchhoff's law, performs deep neural network operation on the feature map, outputs a current corresponding to a task result label, identifies a dynamic target, and can meet the requirements of a deep neural network and a complex dynamic visual task.
Owner:SHANGHAI INSTITUTE OF TECHNICAL PHYSICS CHINESE ACADEMY OF SCIENCES

Fast Fourier transform calculation circuit, chip and electronic equipment

The invention belongs to the technical field of in-memory computing, and particularly discloses a fast Fourier transform computing circuit, a chip and electronic equipment. The circuit comprises a data input module and a multi-stage butterfly calculation module, each stage of butterfly calculation module is constructed based on a memristor array; the data input module is connected with the first-stage butterfly calculation module, and is used for grouping an input data sequence to be calculated and sequentially inputting each data group obtained by grouping into the first-stage butterfly calculation module for butterfly operation processing; each intermediate-stage butterfly calculation module performs butterfly calculation on the calculation result output by the previous-stage butterfly calculation module, and transmits the calculated data result to the next-stage butterfly calculation module for processing; and the last-stage butterfly calculation module performs butterfly calculation on the calculation result output by the previous-stage butterfly calculation module, and outputs the Fourier transform calculation result of the data sequence to be calculated. According to the invention, the power consumption of the calculation circuit can be greatly reduced, and the data transmission delay is reduced.
Owner:HUAZHONG UNIV OF SCI & TECH

Fault diagnosis and remote monitoring system and method for solar power supply system

The invention discloses a fault diagnosis and remote monitoring system and method for a solar power supply system, and relates to the technical field of fault diagnosis of a solar system, and the system comprises a heterogeneous multi-mode sensing module which collects the multi-dimensional information of an assembly through a plurality of sensors; the memristor storage and calculation integrated unit is used for realizing data filtering and feature extraction; the multi-scale causal diagnosis engine is used for diagnosing faults by fusing deep learning and causal diagrams; a self-adaptive topology communication network ensures data transmission; a digital twinborn monitoring platform and visual operation and maintenance are adopted, and in addition, an intelligent evolution decision and self-reconfiguration sensor network module is further arranged, so that the intelligence and reliability of the system are improved. Through cooperation of multiple modules, accurate fault diagnosis and positioning are realized, the diagnosis time is shortened, stable data transmission is ensured, self-repairing and autonomous learning capabilities are provided, the operation and maintenance cost can be reduced, the power generation efficiency can be improved, and the reliability and economic benefits of a solar power supply system can be enhanced.
Owner:CHANGZHOU DATANG PHOTOVOLTAICTECHNOLOGY CO LTD

Monomolecular electric memristor based on dipole control and preparation method thereof

The invention relates to the technical field of electronic devices, in particular to a monomolecular electric memristor based on dipole control and a preparation method thereof.The monomolecular electric memristor comprises a molecular heterojunction, and the molecular heterojunction is formed by connecting dipole molecules and graphene point electrodes through amido bonds; the dipole molecule comprises an amide structure based on a carbazole nitrogen-containing heterocyclic ring unit, under the action of an external electric field, a side chain chemical bond of the dipole molecule can rapidly rotate, the size and direction of an intramolecular dipole moment can be changed, the molecule can be rapidly and stably switched among different resistance states, and the read-write speed of the memristor is remarkably improved; the storage state is stably kept under the power-off condition, and the long-term stability and reliability of the memristor function are ensured. The preparation method comprises the processes of introducing dipole molecules and forming amido bonds between the dipole molecules and the graphene point electrode, and the preparation method is simple and convenient to operate, mild in condition and beneficial to large-scale production.
Owner:NANKAI UNIV

Memristive neural network circuit for generating multi-direction multi-double-scroll chaotic attractors

The invention discloses a memristor neural network circuit for generating a multi-direction multi-double-scroll chaotic attractor, which comprises a first integral channel circuit, a second integral channel circuit, a third integral channel circuit, a first memristor, a second memristor, a third memristor, a first switch, a second switch and a third switch, and is characterized in that the first integral channel circuit comprises three input ends and one output end; the second integral channel circuit comprises three input ends and one output end; the third integral channel circuit comprises two input ends and one output end. The memristive neural network circuit can generate a multi-direction multi-scroll chaotic attractor by controlling the on-off state of the memristive channels connected with the input ends of the integral channel circuits, and the circuit has a discharge characteristic closer to an interactive neuron in a real nervous system; the method can provide new possibility for physically constructing an intelligent neuromorphic system, and has important value in the fields of chaotic secret communication, image encryption, complex signal processing and the like.
Owner:NANJING UNIV OF INFORMATION SCI & TECH

A decoder, an encoder and methods for decoding and encoding wireless communications signals

PCT designated stage expiredWO2025119523A1Multiple-port networksTransversal filtersComputer hardwareCrossbar switch
An electronic decoder for decoding of block coded wireless communications signals. The electronic decoder comprises at least two electrically interconnected analogue crossbar arrays (1110, 1120) of memristors. An electrical interconnection (1115) between a column (1111) of an analogue crossbar array of the at least two electrically 5 interconnected analogue crossbar arrays (1110, 1120) and a row (1121) of a following analogue crossbar array of the at least two electrically interconnected analogue crossbar arrays (1110, 1120) comprises a current-to-voltage converter (1101). A first analogue crossbar array (1110) of the at least two analogue crossbar arrays (1110, 1120) comprises at least a first row of memristors (1112a) with a single input and a 10 respective output for each column of memristors.
Owner:TELEFONAKTIEBOLAGET LM ERICSSON (PUBL)

Signal detection circuit and method based on memristor, receiver and equipment

The invention discloses a memristor-based signal detection circuit and method, a receiver and equipment, and relates to the field of wireless communication, a voltage source module of the circuit is connected with a voltage input end of a memristor array module, and a current source module is connected with a current input end of the memristor array module; the current output end of the memristor array module is connected with the reverse input end of the operational amplifier module and one end of the feedback resistor module, the other end of the feedback resistor module and the output end of the operational amplifier module are connected with the input end of the multiplier module, and the output end of the multiplier module is connected with one end of the resistor module. The voltage of the voltage source module is set according to the emission vector set; the current of the current source module is set according to the received radio frequency signal, and the conductance value of the memristor is set according to the channel matrix, so that a relatively low bit error rate of signal detection can be realized on the basis of greatly reducing calculation time and power consumption, and meanwhile, the requirement on programmable bit precision of the memristor is also reduced.
Owner:BEIJING UNIV OF POSTS & TELECOMM

Optimization method of resistive three-input majority logic circuit

The invention discloses an optimization method of a resistive three-input majority logic circuit, which comprises the following steps of: calculating an optimal resistive three-input majority-inverter graph RM3IG of all four-input NPN Boolean functions by utilizing an accurate comprehensive algorithm based on Boolean satisfiability SAT and NPN equivalent classification, and establishing an offline database; converting the input circuit into an RM3IG structure through a graph mapping algorithm and an offline database; based on the algebraic property of the RM3 logic, an inverter transmission rule in the RM3IG structure is provided, the number of inverters in the RM3IG structure is optimized, finally, the inverters in the RM3IG structure are represented by using the RM3 logic, and a logic circuit only taking the RM3 logic as a basic operation set is obtained. According to the method, the area and delay of a logic circuit taking RM3 logic as a basic operation set are effectively reduced, and particularly, when the method is applied to a memristor circuit, the number of required memristors can be remarkably reduced, and the design cost is reduced.
Owner:NINGBO UNIV

In-memory computing array and operation method thereof

The invention discloses an in-memory computing array and an operation method thereof, and belongs to the technical field of semiconductors and CMOS hybrid integrated circuits. A 2T1R in-memory computing array based on subthreshold current is designed, each unit in the array is composed of a memristor, a gate tube T1 and a reading tube T2, write-in and erase operations are similar to a 1T1R array architecture, but read-out is carried out through the transistor T2 working in a subthreshold region, the reading current generated by the reading tube is small, and the read-in and erase operations can be carried out through the memristor, the gate tube T1 and the reading tube T2. And the parallelism degree of in-memory calculation is increased, and ultrahigh parallel calculation of 100-1000 orders of magnitude can be realized. And meanwhile, the reading current of the reading tube T2 is in exponential correlation with the gate voltage, so that a larger switching current ratio can be obtained, and a reading window of the in-memory calculation unit is amplified. Compared with a 1T1R array architecture, more complex in-memory calculation operation can be realized, and an operation strategy can be flexibly selected, so that more efficient multiply-accumulate operation is realized.
Owner:PEKING UNIV

Optical memristor

The invention discloses an optical memristor, and relates to the field of photon integrated circuits and optical computing, and the optical memristor comprises a waveguide layer, a ferroelectric film layer, an electrode structure and a substrate. The ferroelectric film layer is located on the upper surface of the substrate; the waveguide layer and the electrode structure are located on the upper surface of the ferroelectric film layer; the waveguide layer is used for propagating an incident light signal and performing amplitude modulation on the incident light signal; the electrode structure is used for applying an electric field to the ferroelectric film layer; and the ferroelectric film layer is used for carrying out phase modulation on optical signals transmitted in the waveguide layer under the action of the electric field. The ferroelectric material is used as the optical switch of the optical memristor, efficient modulation of the optical phase is achieved through the non-volatility and the high electro-optical effect of the ferroelectric material, and meanwhile the optical memristor has the advantages of being low in energy consumption and rapid in response.
Owner:NANKAI UNIV

Ternary compound for memristor array and preparation method thereof

The invention relates to the technical field of electronic elements, in particular to a ternary compound for a memristor array and a preparation method of the ternary compound. The ternary complex comprises A, B and C, A is a guest molecule of a pyridine derivative with an amino terminal, B is a bridging molecule, and C is a host molecule with a cucurbituril structure. A, B and C are self-assembled in an aqueous solution through Van der Waals force and hydrophobic interaction among molecules to form a ternary complex; the host molecule C has a barrel-shaped cavity structure, and the guest molecule A and the bridging molecule B are included in the barrel-shaped cavity structure. The ternary complex forms a supramolecular chain junction at the end part of the graphene electrode, so that the graphene electrode with different gaps can be adapted, and the molecular connection success rate of the array is greatly improved. The preparation method of the ternary complex is simple, convenient, easy to operate and suitable for large-scale production.
Owner:NANKAI UNIV

Hardware emulator and emulation system including hardware emulator

A hardware emulator and an emulation system including the hardware emulator are provided. The hardware emulator includes an artificial neural network-based reconstruction model configured to reconstruct dynamics of a dynamical system based on input data and a memristor-based circuit configured to emulate state space representation of the dynamical system based on the reconstruction model.
Owner:SAMSUNG ELECTRONICS CO LTD

Motion recognition method and system based on redox photoelectric memristor, terminal and storage medium

The invention discloses a motion recognition method and system based on a redox photoelectric memristor, a terminal and a storage medium, and the method comprises the steps: selecting classical motions based on a human body motion data set, extracting time sequence data, and coding the time sequence data into an optical pulse sequence; constructing a reservoir array composed of a plurality of photoelectric memristors, and expanding an optical pulse sequence signal into a high-dimensional state vector; constructing a supervised training model, solving a weight matrix, and constructing a memristive cross array; and outputting an action classification result through simulation domain operation based on the output current multi-path light current signals in combination with the memristor cross array. According to the method, the motion features can be directly fed into a rear-end classification network for action recognition without depending on a complex digital feature extraction algorithm, so that the transmission and processing overhead of redundant data is fundamentally eliminated; a high-efficiency, low-delay and high-robustness hardware solution is provided for real-time and anti-noise motion recognition in scenes such as intelligent monitoring and man-machine interaction.
Owner:SHENZHEN UNIV

Brain-computer interface signal identification circuit based on memristor neural network

The invention discloses a brain-computer interface signal identification circuit based on a memristor neural network. The brain-computer interface signal identification circuit is composed of three neuron circuits. Each neuron circuit is composed of nine parallel branches and comprises a signal input module, a weight adjusting module, a first summing module, a memristor, a second summing module, a sampling module and a comparison module. In a single neuron circuit, after nine paths of input signals Vin1-Vin9 are respectively processed by signal input modules, weight adjusting modules, summation modules (1) and memristors of nine parallel branches, the nine paths of signals are combined together and pass through a single summation module (2), a sampling module and a comparison module, and a single identification signal of the neuron circuit is output. According to the designed circuit, a memristor neural network circuit is built by using resistance plasticity and threshold characteristics of a memristor, the meaning of a brain-computer interface signal can be preliminarily identified through a self-learning mechanism of the memristor, and the identification precision of the brain-computer interface signal can be remarkably enhanced.
Owner:SHENZHEN BAIYU IND CO LTD

MIMO communication system signal detection method, circuit, device and equipment based on memristor array

The invention relates to a memristor array-based MIMO communication system signal detection method, circuit, device and equipment, which are used in the field of wireless communication signal processing. In order to solve the problem that an existing MIMO signal detection calculation circuit based on a memristor array is very sensitive to conductance errors of memristors, an MIMO channel matrix is expressed as a product of a small-scale fading coefficient matrix and a large-scale fading coefficient matrix, and a matrix calculation expression of ZF signal detection and MMSE signal detection is converted into a new equivalent form. A matrix calculation circuit based on the memristor array is used for calculating the converted equivalent expression, and signal detection is achieved; the invention provides two kinds of matrix calculation circuit design schemes based on the memristor array, two kinds of corresponding MIMO communication system signal detection devices and three kinds of MIMO communication system signal detection equipment based on the memristor array. According to the invention, the sensitivity to the conductivity error of the memristor is lower, and the signal detection performance is better.
Owner:BEIJING UNIV OF POSTS & TELECOMM

Fractional order memristor Hopfield neural network circuit implementation method based on FPGA

The invention relates to an FPGA (Field Programmable Gate Array)-based fractional order memristor Hopfield neural network circuit implementation method, which adopts a Grunwald-Letnikov fractional order derivative definition and short-term memory principle to finish the implementation of an FPGA circuit of a fractional order memristor Hopfield model through a multi-segment linear fitting hyperbolic tangent function. The method comprises the specific steps of processing a fractional order integral and a hyperbolic tangent function, compiling through Verilog codes and simulating in Xilinx ISE. According to the method, the short-time memory principle and the resource-efficient fitting algorithm are combined, high-precision periodic and chaotic signals can be generated on the FPGA, and the method has a wide application prospect.
Owner:NANJING NORMAL UNIVERSITY

TiO2 / ReS2 heterojunction-based self-rectification memristor array and preparation method and application thereof

PendingCN119095472AWith self rectifying effectHighly integratedHeterojunctionCrossbar switch
The invention provides a preparation method of a self-rectification memristor array based on a TiO2 / ReS2 heterojunction and an application of the self-rectification memristor array. The structure of a single device in the memristor array sequentially comprises an FTO conductive substrate, a TiO2 / ReS2 heterojunction functional layer and an Au top electrode from bottom to top. Titanium dioxide / rhenium disulfide is used as the functional layer of the memristor for the first time, and TiO2 is used as a classic material of the functional layer of the memristor and has good resistance change and synaptic characteristics. After the heterojunction is formed by compounding with ReS2, the energy band structure of the device is changed, and the self-rectification performance of the device is improved. The characteristic can inhibit the sneak current effect in a large-scale crossbar switch array, greatly improves the integration level of the array, and greatly reduces the power consumption of the array.
Owner:HANGZHOU DIANZI UNIV

Image chaotic encryption method based on magnetic control memristor and electronic equipment

The invention relates to the technical field of image encryption, in particular to an image chaotic encryption method based on a magnetic control memristor and electronic equipment, and the method comprises the steps: obtaining an input image; calculating a hash value or generating a random challenge value according to the input image to determine input excitation; judging whether the magnetic field intensity is in a set legal magnetic field domain; if so, calling a real response path of the magnetic control memristor to generate a real response code stream at least according to the input excitation and the magnetic field intensity; constructing a chaotic parameter and a disturbance source according to the real response code stream, and injecting the chaotic parameter and the disturbance source into the fractional order hyperchaotic system; generating a dynamic key stream through a fractional order hyper-chaotic system according to the chaotic parameters; the dynamic secret key stream is combined with a mapping algorithm to globally scramble pixels of an input image, the dynamic secret key stream and a disturbance source are used to carry out bidirectional diffusion on the scrambled image, and a chaotic encryption image is obtained. According to the invention, the magnetic control memristor PUF and the fractional order hyper-chaotic system are deeply fused, and a hardware and algorithm collaborative image encryption mechanism is realized.
Owner:NANTONG XINGRUIXIN TECHNOLOGY CO LTD

Mixing fitting FPGA technology chaotic signal generator realization method

According to the hybrid fitting FPGA technology chaotic signal generator implementation method, a memristive chaotic system with complex nonlinear characteristics such as a hyperbolic function and a natural logarithm can be efficiently generated on a resource-limited FPGA. According to the scheme, target function slope distribution is analyzed, binomial fitting is adopted for slope steep intervals, hardware architecture is optimized to reduce resource occupation, and linear fitting is adopted for slope gentle intervals to simplify implementation logic. According to the memristor chaotic system model constructed based on the hybrid strategy, an iterative operation module is integrated on a chip. Experimental verification shows that the method obviously reduces the consumption of storage resources while keeping high precision, and is particularly suitable for chaotic signal generation scenes with low power consumption and high real-time requirements.
Owner:NANJING NORMAL UNIVERSITY

Balanced three-valued dynamic logic gate implementation circuit based on memristor

The invention discloses a balanced three-valued dynamic logic gate implementation circuit based on memristors. The balanced three-valued dynamic logic gate implementation circuit comprises a balanced three-valued dynamic NOT gate circuit, a balanced three-valued dynamic NOR gate circuit and a balanced three-valued dynamic NAND gate circuit which are implemented according to the memristors. The dynamic logic circuit provided by the invention is widely applied to the design of a super-large-scale integrated circuit, and can reduce the circuit area and improve the circuit speed.
Owner:HANGZHOU DIANZI UNIV

Heterojunction memristor based on two-dimensional material CrPS4 and preparation method thereof

The invention provides a heterojunction memristor based on a two-dimensional material CrPS4 and a preparation method of the heterojunction memristor, and relates to the field of microelectronic semiconductor material photoelectric memristors. The heterojunction memristor comprises a substrate, a two-dimensional material CrPS4-based heterojunction and an active metal electrode which are stacked in sequence. According to the vertical heterojunction memristor disclosed by the invention, the prepared novel two-dimensional material CrPS4 is adopted as a resistive function layer material, and the vertical heterojunction memristor shows excellent non-volatile resistive property.
Owner:HEBEI UNIV OF SCI & TECH

Direct current charging pile detection system and method applied to field

The invention discloses a direct current charging pile detection system and method applied to the field, and relates to the technical field of power equipment on-line monitoring, and the method comprises the steps: inputting a three-dimensional data set into a Toeplitz cyclic measurement matrix for compressed sampling, reconstructing the three-dimensional data set into a high-resolution signal through employing an improved orthogonal matching pursuit algorithm, and carrying out the detection of the high-resolution signal; analyzing and calculating a local variable coefficient through a sliding window, marking abnormal positions, and integrating the abnormal positions into an abnormal position set; inputting the abnormal position set into a pre-trained long and short term memory prediction model, evaluating the insulation residual life and the fault risk level, and calculating the maximum allowable output current of the DC charging pile in real time; the load impedance is dynamically adjusted and detected through the memristor array, the output current of the direct current charging pile is collected in real time, and synchronous compression wavelet transform is adopted to analyze a harmonic ridge line and extract a harmonic component; the accuracy of fault early warning is remarkably improved, rapid dynamic adjustment of the output current is achieved, and the technical problem of multi-physics field coupling monitoring of the rapid charging pile is effectively solved.
Owner:浙江三辰电器股份有限公司

Three-dimensional storage and calculation integrated system and calculation method thereof

The invention belongs to the technical field of storage and calculation integration, and particularly discloses a three-dimensional storage and calculation integration system and an operation method thereof, and the system comprises an input module which has an input end connected to an output end of a main processor and is used for converting to-be-processed data into an analog voltage signal; the input end of the memristor array is connected to the output end of the input module, and the memristor array is used for carrying out matrix vector multiplication or matching operation on pre-stored operation data and the analog voltage signal and outputting a first operation result of the data to be processed; the input end of the sensing module is connected to the output end of the memristor array, and the sensing module is used for converting the first operation result into a corresponding digital signal based on the signal form of the first operation result; and the main processor is used for analyzing the digital signal and determining a second operation result of the data to be processed. According to the invention, the storage area and power consumption of the system can be effectively reduced, and the operation speed and efficiency of the system are improved.
Owner:HUAZHONG UNIV OF SCI & TECH

Circuits with Charge Trapping Transistors

There is described a two-terminal multi-level memristor element synthesised from binary memristors, which is configured to implement a variable resistance based on unary or binary code words. There is further described a circuit such as a synapse circuit implemented using a multi-level memristor element. There is further described programmable resistance cells using charge-trapping-transistors (CTTs) and analog signal processing circuits using CTTs to provide tuneability.
Owner:CIRRUS LOGIC INT SEMICON LTD

Heterogeneous neural network acceleration device based on cooperation of DSP and memristor

The invention relates to a heterogeneous neural network acceleration device based on cooperation of a DSP (Digital Signal Processor) and a memristor, which adopts a hybrid computing architecture design of a DSP and memristor cross array, such as dynamic precision switching, a real-time closed-loop feedback mechanism based on a Lyapunov stability theory, a self-adaptive cross-domain quantization module, hardware instruction set optimization and the like. The hybrid architecture has good balance performance and universality, stability is remarkably improved through conductivity monitoring, self-adaptive PWM pulse generation and quantization table updating, a higher signal-to-noise ratio is achieved at lower cost through combination of a self-adaptive cross-domain quantization module and noise reduction, hardware level optimization is achieved through customization of an instruction set, the hybrid architecture is more suitable for an HNN calculation task, and the method is suitable for the HNN calculation task. The energy efficiency ratio and the anti-drifting capability of the HNN acceleration circuit are improved, and the acceleration performance is improved.
Owner:HUNAN GREAT WALL GALAXY TECH CO LTD

Neural network training method suitable for memristor type brain-like chip hardware deployment

The invention discloses a neural network training method suitable for memristor type brain-like chip hardware deployment. According to the method, a cross array equivalent circuit is extracted through a partial element equivalent circuit method in a training stage, and a time domain transient simulation solving process of the equivalent circuit of the cross array is integrated into gradient propagation, so that the network can learn and adapt to the output characteristics of the cross array in the training process; furthermore, an unstructured dynamic pruning method based on a weight threshold is introduced, the memristor proportion in a low resistance state is effectively reduced, sneaking paths in the cross array are inhibited, and meanwhile, the operation cost of the cross array is reduced. Compared with a traditional training method, the method provided by the invention can significantly enhance the stability and precision of the neural network during hardware deployment, has wide applicability, and can be used for neural network training and deployment optimization of various advanced memory devices.
Owner:ZHEJIANG UNIV

Equivalent circuit and chaotic circuit of second-order magnetic control memristor

The invention discloses an equivalent circuit of a second-order magnetic control memristor and a chaotic circuit, and belongs to the field of memristors, a first resistor in the equivalent circuit can simulate the resistance characteristic of the second-order magnetic control memristor, a power supply can provide voltage, an integrator can integrate the output voltage of the power supply and generate a magnetic flux characterization signal, and the magnetic flux characterization signal is used for representing the magnetic flux. The multiplier can multiply the output voltage of the power supply by the magnetic flux characterization signal, so that the current meeting the target characteristic (the voltage and current characteristic of the second-order magnetic control memristor) is generated in the equivalent circuit, namely, the equivalent circuit of the second-order magnetic control memristor has the resistance characteristic of the second-order magnetic control memristor. The equivalent circuit is simple in structure, so that the size and the implementation cost of the equivalent circuit are reduced.
Owner:SHANDONG YUNHAI GUOCHUANG CLOUD COMPUTING EQUIP IND INNOVATION CENT CO LTD

Image recognition device and method

According to the image recognition device and method, a current resistance relational expression is obtained according to the resistance values of a plurality of second memristors, the resistance value of a second column selection switch and a plurality of test currents, and the resistance value when the column selection switch is conducted is considered; after the weight voltage obtained according to the conductance value of the first memristor and the current resistance relational expression is input to the row signal end of the memristor sub-array, accurate mapping of the conductance value of the first memristor and the weight of the neural network can be realized. The plurality of column currents output from the plurality of column signal ends of the memristor sub-array are multiplication accumulation of the to-be-identified voltage and the conductance value of the memristor sub-array on the column, so that the accuracy of the image identification result determined according to the plurality of column currents is very high; switch control and data control are separated to different FPGAs, so that the reliability, flexibility and isolation of the device can be improved; and the serial-parallel conversion module is arranged, so that the IO interface pressure of the FPGA can be reduced, and the expandability is improved.
Owner:SHENZHEN TIANYI TIMES TECHNOLOGY CO LTD