A wind direction recognition device and method based on humidity-sensitive memristive neurons

Abstract

The present invention discloses a wind direction identification device and method based on humidity-sensitive memristor neurons. The device comprises: a humidity-sensitive memristor, a LIF circuit, an FPGA, and a display. The humidity-sensitive memristor is disposed in the middle of the device and is surrounded by a circle of pure water. The LIF circuit is disposed at the bottom of the device and is connected to the humidity-sensitive memristor via a wire. The FPGA is connected to each LIF circuit, each LIF circuit is connected to a comparator, and the output results of each LIF circuit are transmitted to a pulse neural network in the FPGA via the comparator for wind direction identification. The display is used to output the wind direction identification results. The present invention realizes wind direction identification based on the high sensitivity and low power consumption characteristics of the humidity-sensitive memristor, giving it broad application prospects in smart cities, precision agriculture, and other fields, and is conducive to promoting the development of wind direction identification technology towards a more intelligent and efficient direction.

Description

Technical Field

[0001] The present invention relates to the field of electronic science and technology, and in particular to a wind direction recognition device and method based on humidity-sensitive memristive neurons. Background Art

[0002] Wind direction identification is crucial in areas such as industrial and agricultural safety, environmental assessment, and meteorological monitoring. In agriculture, wind direction data can guide pesticide spraying and reduce pollution; in urban planning, it can optimize ventilation design and mitigate the heat island effect. Accurate wind direction data not only provides critical decision-making support for wind power generation, aviation, and navigation, but can also be used to predict pollutant dispersion and provide disaster warnings.

[0003] Currently, the most widely used ultrasonic anemometer is based on the time-of-flight (TOF) principle. It calculates wind speed and direction by measuring the time difference between ultrasonic waves traveling downwind and upwind. This method offers the advantages of simplicity and high measurement accuracy. However, this method requires extremely high time measurement accuracy, especially in low wind speed conditions, where a reduced signal-to-noise ratio can severely impact the accuracy of time delay measurements. Furthermore, traditional TOF methods require complex signal processing circuits and algorithms, resulting in high system power consumption and difficulty in miniaturization and distributed deployment.

[0004] Therefore, the prior art still has defects. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to provide a wind direction identification device and method based on humidity-sensitive memristive neurons to address the above-mentioned defects of the prior art. The technical solution adopted by the present invention is as follows:

[0006] In a first aspect, the present invention provides a wind direction recognition device based on humidity-sensitive memristive neurons, the device comprising:

[0007] A humidity-sensitive memristor, the humidity-sensitive memristor being disposed in the middle of the device and surrounded by a circle of pure water;

[0008] A LIF circuit, wherein the LIF circuit is disposed at the bottom of the through-hole device and is connected to the humidity sensitive memristor through a wire;

[0009] FPGA, wherein the FPGA is connected to each LIF circuit, wherein each LIF circuit is respectively connected to a comparator, and the output result of each LIF circuit is respectively transmitted to the pulse neural network in the FPGA through the comparator for wind direction recognition;

[0010] A display is used to output wind direction recognition results.

[0011] In one implementation, nine LIF circuits and nine humidity-sensitive memristors are provided.

[0012] In one implementation, the spiking neural network consists of 27 input nodes in the input layer, 2 hidden layers, and 8 nodes in the output layer.

[0013] In one implementation, three comparators are provided, and the thresholds are respectively set to: 3.5 volts, 2.5 volts, and 1.5 volts.

[0014] In a second aspect, an embodiment of the present invention further provides a method for wind direction recognition based on humidity-sensitive memristive neurons, wherein the method is applied to the wind direction recognition device based on humidity-sensitive memristive neurons described in any of the above solutions, and the method includes:

[0015] Prepare humidity-sensitive memristors and build a LIF neuron system for wind direction monitoring based on humidity-sensitive memristors;

[0016] After using an air blowing device to simulate northwest wind input, the output results of the LIF circuit on the humidity-sensitive memristor are detected. The output results of each LIF circuit are transmitted to the pulse neural network in the FPGA after passing through three comparators for wind direction recognition.

[0017] The wind direction identification result is outputted via a display.

[0018] In one implementation, the step of preparing a humidity-sensitive memristor includes:

[0019] Take 1 ml of 5% Nafion solution and dilute it in 3.6 ml of anhydrous ethanol, shake for 5 minutes to mix thoroughly, and obtain a Nafion mixed solution;

[0020] A glass substrate with a 185 nm thick ITO electrode was ultrasonically cleaned with deionized water, blown dry with nitrogen, and dried at 120 °C for 30 min.

[0021] 200 μL of the nafion mixed solution was taken and spin-coated on a cleaned ITO glass substrate to obtain a 30 nm thick nafion polymer film, which was then annealed at 100 degrees Celsius for 45 minutes.

[0022] Electrodes are prepared using a metal mask with a pre-fabricated pattern to obtain a humidity-sensitive memristor.

[0023] In one implementation, the spin coating speed condition is 500 rpm for 5 seconds and then 1500 rpm for 40 seconds.

[0024] In one implementation, the method of preparing an electrode using a metal mask having a pre-fabricated pattern to obtain a humidity-sensitive memristor includes:

[0025] The mask plate is covered on the spin-coated glass substrate, and 50 nanometers of silver is evaporated on the glass substrate by thermal evaporation. After removing the mask plate, a humidity-sensitive memristor is obtained.

[0026] In one implementation, the evaporation rate is controlled at 0.08-0.12 angstroms / second.

[0027] In a third aspect, an embodiment of the present invention further provides a terminal, wherein the terminal includes a memory, a processor, and a wind direction recognition program based on humidity-sensitive memristive neurons stored in the memory and runnable on the processor. When the processor executes the wind direction recognition program based on humidity-sensitive memristive neurons, the steps of the wind direction recognition method based on humidity-sensitive memristive neurons in any one of the above-mentioned schemes are implemented.

[0028] In a fourth aspect, an embodiment of the present invention also provides a computer-readable storage medium, wherein a wind direction recognition program based on humidity-sensitive memristive neurons is stored on the computer-readable storage medium, and the wind direction recognition program based on humidity-sensitive memristive neurons implements the steps of the wind direction recognition method based on humidity-sensitive memristive neurons described in any one of the above-mentioned schemes on the computer-readable storage medium.

[0029] Beneficial effects: Compared with the prior art, the present invention provides a wind direction identification device and method based on humidity-sensitive memristor neurons, the device comprising: a humidity-sensitive memristor, a LIF circuit, an FPGA, and a display. The humidity-sensitive memristor is arranged in the middle of the device, and a circle of pure water is covered around the humidity-sensitive memristor; the LIF circuit is arranged at the bottom of the device and connected to the humidity-sensitive memristor via a wire; the FPGA is connected to each LIF circuit, wherein each LIF circuit is respectively connected to a comparator, and the output result of each LIF circuit is respectively transmitted to the pulse neural network in the FPGA through the comparator for wind direction identification; the display is used to output the wind direction identification result. The present invention realizes wind direction identification based on the high sensitivity and low power consumption characteristics of the humidity-sensitive memristor, which has broad application prospects in the fields of smart cities, precision agriculture, etc., and is conducive to promoting the development of wind direction identification technology in a more intelligent and efficient direction. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 This is a working diagram of a wind direction recognition device based on humidity-sensitive memristive neurons provided by an embodiment of the present invention.

[0031] Figure 2 The threshold voltage and current changes of humidity-sensitive memristor under different humidity conditions.

[0032] Figure 3The effects of humidity and voltage on the turn-on speed of humidity-sensitive memristors.

[0033] Figure 4 Humidity field programming for humidity-sensitive memristors.

[0034] Figure 5 Humidity-sensitive neuron circuit and spike pulse output under different humidity conditions.

[0035] Figure 6 This is the preprocessing process and results of the wind direction recognition device based on humidity-sensitive memristive neurons provided by an embodiment of the present invention under northwest wind.

[0036] Figure 7 A schematic diagram of a spiking neural network provided by an embodiment of the present invention.

[0037] Figure 8 This is a flow chart of a preferred embodiment of a method for wind direction recognition based on humidity-sensitive memristive neurons provided by an embodiment of the present invention.

[0038] Figure 9 This is a functional block diagram of a terminal provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0039] In order to make the purpose, technical solution and effect of the present invention clearer and more specific, the present invention is further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0040] The flowcharts shown in the accompanying drawings are for illustrative purposes only and do not necessarily include all contents, operations, or steps, nor must they be executed in the order described. For example, some operations or steps may be decomposed, combined, or partially merged, so the actual execution order may vary depending on the actual situation.

[0041] It should be understood that the terms used in the present specification are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the present specification and the appended claims, the singular forms "a", "an" and "the" are intended to include the plural forms unless the context clearly indicates otherwise.

[0042] It should be understood that, to facilitate a clear description of the technical solutions of the embodiments of the present invention, in the embodiments of the present invention, terms such as "first" and "second" are used to distinguish between identical or similar items with substantially the same functions and effects. For example, the first control information and the second control information are merely used to distinguish different control information and do not limit their order.

[0043] Those skilled in the art can understand that words such as "first" and "second" do not limit the quantity and execution order, and words such as "first" and "second" do not necessarily limit them to be different.

[0044] It will also be understood that the term "and / or" used in the present description and appended claims refers to and includes any and all possible combinations of one or more of the associated listed items.

[0045] Traditional humidity sensing systems use a serial architecture with separate sensors, memory, and computing units, resulting in high energy consumption and delays during signal conversion and transmission. Currently, nanowire-based humidity-sensitive memristors face integration difficulties, while the vertically structured humidity-sensitive memristors currently under study are unable to meet the requirements for ultra-fast response. In response to these challenges, the present invention innovatively designs a sensing-computing integrated circuit based on ultra-fast humidity-sensitive memristors. Through material and structural optimization, it significantly improves the response speed while achieving high integration. The present invention not only breaks through the speed and power consumption limitations of the von Neumann architecture, and resolves the contradiction between the integration and response speed of existing memristor technology, but also pioneers the application of humidity-sensitive neurons in the field of wind direction monitoring, filling this technological gap.

[0046] This embodiment uses a humidity-sensitive polymer material to fabricate a two-terminal threshold-shift vertical memristor. This memristor is then used to construct a neuron circuit to sense and process humidity signals. Increased humidity causes the memristor's turn-on threshold voltage to decrease, shortening its turn-on time and increasing its device current. By leveraging the differences in the memristor's electrical properties at different humidity levels, the neuron circuit constructed using it produces differentiated pulse discharge behavior under varying humidity conditions, with pulse frequency and amplitude increasing at high humidity and decreasing at low humidity. By establishing a relationship between humidity and pulse amplitude and frequency, humidity signal sensing and processing can be achieved. Advantages include: a simple preparation process for humidity-sensitive memristors and low difficulty in large-scale integration; fast response speed, low threshold voltage, and correspondingly reduced power consumption. By directly processing analog signals through a memristor array, this invention achieves a high degree of integration of sensing, storage, and computing functions, providing an efficient, low-latency hardware solution for intelligent environmental monitoring.

[0047] This embodiment provides a wind direction recognition device based on humidity-sensitive memristive neurons, such as Figure 1As shown in . The wind direction identification device based on humidity-sensitive memristor neurons of this embodiment includes: a humidity-sensitive memristor, a LIF (Leaky Integrity Fire) circuit, an FPGA (Field-Programmable Gate Array) and a display. The humidity-sensitive memristor is arranged in the middle of the device, and a circle of pure water is covered around the humidity-sensitive memristor; the LIF circuit is arranged at the bottom of the device and is connected to the humidity-sensitive memristor through a wire; the FPGA is connected to each LIF circuit, wherein each LIF circuit is respectively connected to a comparator, and the output results of each LIF circuit are respectively transmitted to the pulse neural network in the FPGA through the comparator for wind direction identification; the display is used to output the wind direction identification result, such as Figure 1 The middle display shows "West Wind".

[0048] Specifically, this example employs a vertical memristor structure, with a humidity-sensitive Nafion polymer as the dielectric layer and silver as the electrode material. To prepare the humidity-sensitive memristor, 1 ml of a 5% Nafion solution was diluted in 3.6 ml of anhydrous ethanol and shaken for 5 minutes to thoroughly mix, yielding a Nafion mixed solution. A glass substrate with a 185-nanometer-thick ITO (Indium Tin Oxide) electrode was ultrasonically cleaned with deionized water, then blown dry with nitrogen and dried at 120°C for 30 minutes. 200 μL of the Nafion mixed solution was spin-coated onto the cleaned ITO glass substrate, yielding a 30-nanometer-thick Nafion polymer film. The spin-coating speed was 500 rpm for 5 seconds, followed by 1500 rpm for 40 seconds, and then annealed at 100°C for 45 minutes. Electrodes were formed using a pre-patterned metal mask to produce the humidity-sensitive memristor. Specifically, in this embodiment, a mask is placed over a spin-coated glass substrate. 50 nanometers of silver are deposited on the glass substrate via thermal evaporation, with the deposition rate controlled at 0.08-0.12 angstroms / second, specifically 0.1 angstroms / second. The mask is then removed, yielding a humidity-sensitive memristor.

[0049] The performance research of humidity-sensitive memristors focuses on two main aspects. First, the study of the threshold transition behavior of humidity-sensitive memristors under different humidity conditions, focusing on the impact of humidity on the device's operating voltage and turn-on time; and second, the study of humidity field programming. The specific contents are as follows:

[0050] (1) The device performance was jointly controlled by using a semiconductor parameter analyzer and a homemade humidity control device. First, the resistance change behavior of the device was studied at a relative humidity of 30%, 50%, 70%, and 90%, with the voltage retracement behavior from 0 volts to 1.5 volts and a scanning speed of 0.01 volts / second. The results show that under low humidity (such as 30%), the humidity-sensitive memristor cannot achieve resistance change. Under high humidity (such as 50%, 70%, and 90%), the humidity-sensitive memristor can achieve resistance change, and the threshold voltage decreases with increasing humidity, while the high-resistance current increases with increasing humidity, as shown in Figure 2. Figure 2 shown.

[0051] (2) A semiconductor parameter analyzer was used to apply a step voltage signal to the humidity-sensitive memristor. The signal jumped at 10ms. The voltage before the jump was 0.1V. The voltage after the jump was divided into four groups, ranging from 0.7V to 1V at intervals of 0.1V. Under different humidity conditions (30%, 50%, 70%, 90%), the resistance change of the humidity-sensitive memristor after the voltage jump was observed. The time difference between the voltage jump time and the resistance change time of the humidity-sensitive memristor was the device's turn-on time. The research results show that the increase in humidity and voltage can speed up the turn-on of the humidity-sensitive memristor and shorten the turn-on time. Figure 3 shown.

[0052] (3) Humidity field programming: Humidity field programming is to increase or decrease the humidity of the environment in which the humidity-sensitive memristor is located through a humidity control device at a voltage of 0.3 volts, thereby causing the humidity-sensitive memristor to undergo a resistance state change. The results show that under high humidity conditions, the voltage of 0.3 volts is greater than the threshold voltage for turning on the device, and the humidity-sensitive memristor switches from a high-resistance state to a low-resistance state; and when the ambient humidity decreases, the voltage of 0.3 volts is less than the threshold voltage for turning on the device, and the humidity-sensitive memristor returns to a high-resistance state; therefore, humidity field programming of the humidity-sensitive memristor can be achieved by changing the ambient humidity, such as Figure 4 shown.

[0053] Wind direction measurement is an important research topic in meteorology, industry, and other fields. It often requires detection instruments to have characteristics such as long-term stability, fast response, low power consumption, and high precision. In previous wind direction monitoring research, there have been studies based on traditional mechanical rotating wind vanes, as well as studies relying on various sensors such as ultrasonic and humidity sensors. However, research using neuromorphic devices such as memristors is still blank. Due to the integrated storage and computing characteristics of humidity-sensitive memristors, they can effectively solve the rate limitation problem brought by the von Neumann architecture in traditional circuits. Therefore, the present embodiment has considerable feasibility in building a LIF neuron system for wind direction monitoring using humidity-sensitive memristors. In order to realize this system, this embodiment starts from the fact that humidity-sensitive memristors can simultaneously sense and process changes in humidity in the wind, and proposes a LIF neuron system. The LIF neuron system includes a humidity-sensitive memristor and an RC circuit (a circuit composed of a resistor (Resistor, R) and a capacitor (Capacitor, C)). The humidity signal is converted into an asynchronous spike pulse sequence using the above-mentioned humidity-sensitive memristor and RC circuit. Specifically, as Figure 5 As shown, (200k ) is the current limiting resistor to prevent the loop current from being too large, and the load resistor ( , 20k ) is connected in series with the memristor and the capacitor ( , 0.33 ) in parallel. At a stable input voltage ( , 1.5 volts), detected by oscilloscope Voltage output on ( ). Apply back, Start charging when When the voltage on the humidity sensitive memristor reaches the critical threshold of the humidity sensitive memristor, the humidity sensitive memristor will switch to a low resistance state, and the current will be shunted from the humidity sensitive memristor, resulting in The voltage on the humidity sensitive memristor drops below the critical threshold, causing the humidity sensitive memristor to switch back to a high resistance state. In the process of the humidity sensitive memristor switching from high resistance to low resistance and then back to high resistance, the current will cause The voltage changes on the humidity sensitive resistor generate spike pulses. The change of relative humidity will cause the switching threshold and the resistance value of the humidity sensitive memristor to change, making the resistance switching behavior of the humidity sensitive memristor slower or more frequent, and the current flowing through the humidity sensitive memristor to decrease or increase, resulting in The frequency and amplitude of the output spike pulse on the sensor change. This circuit can effectively convert the humidity signal into a pulse spike signal of the sensor, realizing the fusion of perception and calculation.

[0054] In order to apply the LIF neuron system to wind direction detection, the wind direction identification device is designed as follows: Figure 1 As shown, there are 9 humidity-sensitive memristors in the middle of the device. The 9 humidity-sensitive memristors are arranged in a 3×3 pattern and surrounded by a circle of pure water. When wind enters the device, the water allows the wind to carry more humidity information, thereby improving the device's perception ability. The 9 humidity-sensitive memristors are connected to the LIF circuit (also 9) at the bottom of the device via wires. The output results of each LIF circuit are respectively output to the pulse neural network in the FPGA after passing through 3 comparators for wind direction recognition. The wind direction recognition results are finally output through the display. Figure 6 As shown, this embodiment uses a blowing device to simulate the wind input device. Figure 6 The simulation in Figure 2 simulates northwest wind, and shows the output of the LIF circuits of the nine humidity-sensitive memristors in the detection device. As the wind blows, C1 first receives humidity information from the wind and begins emitting spike pulses, generating the output. As the wind continues, C5 and C9 also receive humidity information. However, as the humidity decays as the wind moves, the humidity information received by C5 and C9 also weakens, causing the frequency and amplitude of the spike pulses they emit to decrease, resulting in an asynchronous spike pulse train. Based on these output results, three comparators are set to extract wind direction information from different spike pulses. The comparator thresholds are set to 3.5 volts, 2.5 volts, and 1.5 volts, respectively. Only when the spike pulse amplitude exceeds the corresponding comparator threshold will it be fed into the spiking neural network for wind direction identification. Therefore, the output of each LIF circuit, after passing through the comparators, is ultimately input into the spiking neural network for wind direction identification. In this embodiment, comparators with different thresholds (3.5 volts, 2.5 volts, and 1.5 volts) are used to screen and distinguish spike pulses output by each LIF circuit. Amplitudes above the corresponding thresholds are selected and converted into specific outputs. This converts continuously varying spike pulses into discrete logic signals recognizable by a spiking neural network, extracting key voltage features. This provides clear, distinguishable input signals for the subsequent spiking neural network to accurately identify wind direction, facilitating the identification of different wind directions and wind-related humidity information.

[0055] The pulse neural network of this embodiment is a pre-trained network model, which can determine the target direction based on the humidity sensing result, thereby realizing wind direction recognition. Figure 7As shown, the spiking neural network consists of 27 input nodes in the input layer, two hidden layers (with 54 and 30 nodes, respectively), and an output layer with 8 nodes. The input to the spiking neural network is a physical quantity related to humidity (e.g., humidity information). The online training process of the spiking neural network uses a backpropagation algorithm. This algorithm is based on experimentally measured electrical properties of memristors (e.g., the relationship between resistance change and humidity-pulse behavior) and the dependence of the humidity-sensing neural network's spike frequency and amplitude on input humidity (e.g., the dependence of humidity on spike frequency / amplitude). During training, the error between the network's predicted output and the actual wind direction label is compared, and the inter-layer connection weights are adjusted inversely, allowing the spiking neural network to learn to determine wind direction based on humidity signals. After training the spiking neural network using the backpropagation algorithm, the network output can be classified into eight wind direction categories (east, south, west, north, northeast, northwest, southeast, and southwest), thus enabling wind direction recognition. After training, the spiking neural network accurately associates the input humidity-related physical quantity with the wind direction category. In subsequent actual recognition, it can output reliable wind direction recognition results based on the forward propagated signal. Figure 8 Shows the evolution of accuracy during training of the spiking neural network, which reaches 84.32% accuracy after 50 training epochs.

[0056] Specifically, the wind from the blower carries humidity information and acts on humidity-sensitive memristors. Nine humidity-sensitive memristors correspond to nine LIF circuits. The output of each LIF circuit is processed by three comparators with different thresholds, generating a total of 9 × 3 = 27 humidity-related electrical signals, such as pulse frequency and amplitude-converted voltage characteristics. These signals are fed into the 27 input nodes of the spiking neural network's input layer, transforming the raw humidity information into a recognizable signal for the network. Next, the 27 input nodes pass the signals to the first hidden layer (54 nodes). The nodes are connected via weights, and the input signals are multiplied by the weights, accumulated, and passed through the neuron activation function to generate the outputs of the first hidden layer nodes. The outputs of the 54 nodes of the first hidden layer are then passed to the second hidden layer (30 nodes) in a similar manner for further signal feature extraction and transformation. The outputs of the 30 nodes of the second hidden layer are then passed to the output layer (8 nodes). Ultimately, each node in the output layer corresponds to the probability of a wind direction category (east, south, west, north, northeast, northwest, southeast, southwest). The 8 nodes in the output layer correspond to 8 types of wind directions. After the network forward propagation, the numerical value output by the node represents the probability of the corresponding wind direction. The wind direction corresponding to the node with the highest probability value is the final wind direction recognition result, realizing the recognition of wind direction from humidity perception.

[0057] Based on the above embodiments, the present invention also provides a wind direction recognition method based on humidity-sensitive memristive neurons, such as Figure 8The method of this embodiment includes the following steps:

[0058] Step S100: preparing a humidity-sensitive memristor, and building a LIF neuron system for wind direction monitoring based on the humidity-sensitive memristor;

[0059] Step S200: After using an air blowing device to simulate wind input, the output results of the LIF circuit on the humidity-sensitive memristor are detected, and the output results of each LIF circuit are respectively transmitted to the pulse neural network in the FPGA after passing through three comparators for wind direction recognition to obtain the wind direction recognition result;

[0060] Step S300: output the wind direction recognition result via a display.

[0061] In this example, to prepare a humidity-sensitive memristor, 1 ml of a 5% Nafion solution was diluted in 3.6 ml of anhydrous ethanol and shaken for 5 minutes to thoroughly mix, yielding a Nafion mixed solution. A glass substrate with a 185-nanometer-thick ITO (Indium Tin Oxide) electrode was ultrasonically cleaned with deionized water, then blown dry with nitrogen and dried at 120°C for 30 minutes. 200 μL of the Nafion mixed solution was spin-coated onto the cleaned ITO glass substrate, yielding a 30-nanometer-thick Nafion polymer film. The spin-coating speed was 500 rpm for 5 seconds, followed by 1500 rpm for 40 seconds, and then annealed at 100°C for 45 minutes. Electrodes were formed using a pre-patterned metal mask to produce the humidity-sensitive memristor. Specifically, in this embodiment, a mask plate is covered on a spin-coated glass substrate, and 50 nanometers of silver are evaporated on the glass substrate by thermal evaporation. The evaporation rate is controlled at 0.08-0.12 angstroms / second, specifically 0.1 angstroms / second. The mask plate is then removed to obtain a humidity-sensitive memristor.

[0062] In other implementations, this embodiment may also select a suitable humidity-sensitive material as the memristor dielectric layer, such as PEDOT:PSS, polystyrene sulfonate (PSS), aluminum oxide ( )、Molybdenum disulfide( When designing the structure of the memristor, this embodiment can also change the device structure of the memristor, such as a three-terminal memristor transistor. In addition, the electrode material of the memristor can also be copper, platinum, nickel, aluminum, etc.

[0063] Combine Figure 6 As shown, this embodiment uses a blowing device to simulate the wind input device. Figure 6The simulation in Figure 2 simulates northwest wind, and shows the output of the LIF circuits of the nine humidity-sensitive memristors in the detection device. As the wind blows, C1 first receives humidity information from the wind and begins emitting spike pulses, generating the output. As the wind continues, C5 and C9 also receive humidity information. However, as the humidity decays as the wind moves, the humidity information received by C5 and C9 also weakens, causing the frequency and amplitude of the spike pulses they emit to decrease, resulting in an asynchronous spike pulse train. Based on these output results, three comparators are set to extract wind direction information from different spike pulses. The comparator thresholds are set to 3.5 volts, 2.5 volts, and 1.5 volts, respectively. Only when the spike pulse amplitude exceeds the corresponding comparator threshold will it be fed into the spiking neural network for wind direction identification. Therefore, the output of each LIF circuit, after passing through the comparators, is ultimately input into the spiking neural network for wind direction identification. In this embodiment, comparators with different thresholds (3.5 volts, 2.5 volts, and 1.5 volts) are used to screen and distinguish spike pulses output by each LIF circuit. Amplitudes above the corresponding thresholds are selected and converted into specific outputs. This converts continuously varying spike pulses into discrete logic signals recognizable by a spiking neural network, extracting key voltage features. This provides clear, distinguishable input signals for the subsequent spiking neural network to accurately identify wind direction, facilitating the identification of different wind directions and wind-related humidity information.

[0064] The pulse neural network of this embodiment is a pre-trained network model, which can determine the target direction based on the humidity sensing result, thereby realizing wind direction recognition. Figure 7 As shown, the spiking neural network consists of 27 input nodes in the input layer, two hidden layers (with 54 and 30 nodes, respectively), and an output layer with 8 nodes. The input to the spiking neural network is a physical quantity related to humidity (e.g., humidity information). The online training process of the spiking neural network uses a backpropagation algorithm. This algorithm is based on experimentally measured electrical properties of memristors (e.g., the relationship between resistance change and humidity-pulse behavior) and the dependence of the humidity-sensing neural network's spike frequency and amplitude on input humidity (e.g., the dependence of humidity on spike frequency / amplitude). During training, the error between the network's predicted output and the actual wind direction label is compared, and the inter-layer connection weights are adjusted inversely, allowing the spiking neural network to learn to determine wind direction based on humidity signals. After training the spiking neural network using the backpropagation algorithm, the network output can be classified into eight wind direction categories (east, south, west, north, northeast, northwest, southeast, and southwest), thus enabling wind direction recognition. After training, the spiking neural network accurately associates the input humidity-related physical quantity with the wind direction category. In subsequent actual recognition, it can output reliable wind direction recognition results based on the forward propagated signal. Figure 8Shows the evolution of accuracy during training of the spiking neural network, which reaches 84.32% accuracy after 50 training epochs.

[0065] Specifically, the wind from the blower carries humidity information and acts on humidity-sensitive memristors. Nine humidity-sensitive memristors correspond to nine LIF circuits. The output of each LIF circuit is processed by three comparators with different thresholds, generating a total of 9 × 3 = 27 humidity-related electrical signals, such as pulse frequency and amplitude-converted voltage characteristics. These signals are fed into the 27 input nodes of the spiking neural network's input layer, transforming the raw humidity information into a recognizable signal for the network. Next, the 27 input nodes pass the signals to the first hidden layer (54 nodes). The nodes are connected via weights, and the input signals are multiplied by the weights, accumulated, and passed through the neuron activation function to generate the outputs of the first hidden layer nodes. The outputs of the 54 nodes of the first hidden layer are then passed to the second hidden layer (30 nodes) in a similar manner for further signal feature extraction and transformation. The outputs of the 30 nodes of the second hidden layer are then passed to the output layer (8 nodes). Ultimately, each node in the output layer corresponds to the probability of a wind direction category (east, south, west, north, northeast, northwest, southeast, southwest). The eight nodes in the output layer correspond to eight types of wind directions. After forward propagation of the network, the numerical value output by the node represents the probability of the corresponding wind direction. The wind direction corresponding to the node with the highest probability value is the final wind direction recognition result, realizing the recognition of wind direction from humidity perception. In other implementations, this embodiment can also design different neural networks, such as artificial neural networks, binary neural networks, feedforward neural networks, recurrent neural networks, etc.

[0066] The wind direction recognition method based on humidity-sensitive memristive neurons in this embodiment is similar in principle to the various modules in the above-mentioned device embodiment, and will not be described in detail here.

[0067] Based on the above embodiment, the present invention further provides a terminal, the principle block diagram of the terminal can be as follows: Figure 9 The terminal may include one or more processors 100 ( Figure 9 Only one is shown in the figure), memory 101, and computer program 102 stored in memory 101 and executable on one or more processors 100. For example, a wind direction identification program based on humidity-sensitive memristive neurons. When one or more processors 100 execute computer program 102, each step of an embodiment of a method for wind direction identification based on humidity-sensitive memristive neurons can be implemented. Alternatively, when one or more processors 100 execute computer program 102, the functions of each module / unit in an embodiment of a wind direction identification system based on humidity-sensitive memristive neurons can be implemented, without limitation herein.

[0068] In one embodiment, the processor 100 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor.

[0069] In one embodiment, memory 101 may be an internal storage unit of an electronic device, such as a hard drive or memory. Memory 101 may also be an external storage device of the electronic device, such as a plug-in hard drive, smart media card (SMC), secure digital (SD) card, flash memory card, etc. Furthermore, memory 101 may include both an internal storage unit of the electronic device and an external storage device. Memory 101 is used to store computer programs and other programs and data required by the terminal. Memory 101 may also be used to temporarily store data that has been output or is about to be output.

[0070] Those skilled in the art will understand that Figure 9 The principle block diagram shown in the figure is only a block diagram of a partial structure related to the solution of the present invention, and does not constitute a limitation on the terminal to which the solution of the present invention is applied. The specific terminal may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0071] Those skilled in the art will understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, operation database or other media used in the embodiments provided by the present invention may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in many forms such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), Synchronous Link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct RAM bus dynamic RAM (DRDRAM), and RAM bus dynamic RAM (RDRAM), etc.

[0072] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A wind direction recognition device based on humidity-sensitive memristive neurons, characterized in that: The device comprises: A humidity-sensitive memristor, the humidity-sensitive memristor being disposed in the middle of the device and surrounded by a circle of pure water; A LIF circuit, wherein the LIF circuit is disposed at the bottom of the device and connected to the humidity-sensitive memristor via a wire; FPGA, wherein the FPGA is connected to each LIF circuit, wherein each LIF circuit is respectively connected to a comparator, and the output result of each LIF circuit is respectively transmitted to the pulse neural network in the FPGA through the comparator for wind direction recognition; A display, the display being used to output a wind direction identification result; The number of each of the LIF circuit and the humidity-sensitive memristor is 9; The spiking neural network consists of 27 input nodes in the input layer, 2 hidden layers, and 8 nodes in the output layer; The number of comparators is three, and the thresholds are set to 3.5 volts, 2.5 volts, and 1.5 volts respectively.

2. A wind direction recognition method based on humidity-sensitive memristive neurons, characterized in that: The method is applied to the wind direction recognition device based on humidity-sensitive memristive neurons according to claim 1, and the method comprises: Prepare humidity-sensitive memristors and build a LIF neuron system for wind direction monitoring based on humidity-sensitive memristors; After using an air blowing device to simulate northwest wind input, the output results of the LIF circuit on the humidity-sensitive memristor are detected. The output results of each LIF circuit are transmitted to the pulse neural network in the FPGA after passing through three comparators for wind direction recognition. The wind direction identification result is outputted via a display.

3. The method for wind direction recognition based on humidity-sensitive memristive neurons according to claim 2, characterized in that: The method for preparing a humidity-sensitive memristor comprises: Take 1 ml of 5% Nafion solution and dilute it in 3.6 ml of anhydrous ethanol, shake for 5 minutes to mix thoroughly, and obtain a Nafion mixed solution; A glass substrate with a 185 nm thick ITO electrode was ultrasonically cleaned with deionized water, blown dry with nitrogen, and dried at 120 °C for 30 min. 200 μL of the nafion mixed solution was taken and spin-coated on a cleaned ITO glass substrate to obtain a 30 nm thick nafion polymer film, which was then annealed at 100 degrees Celsius for 45 minutes. Electrodes are prepared using a metal mask with a pre-fabricated pattern to obtain a humidity-sensitive memristor.

4. The method for wind direction recognition based on humidity-sensitive memristive neurons according to claim 3, characterized in that: The spin coating speed conditions were 500 rpm for 5 seconds and then 1500 rpm for 40 seconds.

5. The method for wind direction recognition based on humidity-sensitive memristive neurons according to claim 3, characterized in that: The method of preparing electrodes by using a metal mask having a prefabricated pattern to obtain a humidity-sensitive memristor comprises: The mask plate is covered on the spin-coated glass substrate, and 50 nanometers of silver is evaporated on the glass substrate by thermal evaporation. After removing the mask plate, a humidity-sensitive memristor is obtained.

6. The method for wind direction recognition based on humidity-sensitive memristive neurons according to claim 5, characterized in that: The evaporation rate is controlled at 0.08-0.12 angstroms / second.

7. The method for wind direction recognition based on humidity-sensitive memristive neurons according to claim 3, characterized in that: The online training process of the spiking neural network is performed using a back-propagation algorithm based on experimentally measured electrical properties of humidity-sensitive memristors and the dependence of the spike frequency and amplitude of the humidity-sensing neural network on input humidity.

Images