A silicone-based flexible tactile sensor based on friction nanogenerator and its preparation method and application

Abstract

The present invention belongs to the technical field of friction nanogenerators and sensors, and specifically relates to a silicone-based flexible tactile sensor based on a friction nanogenerator, and a preparation method and application thereof. The sensor is composed of a silicone / polyvinyl alcohol layer, a silicone / graphite carbon nitride layer, and a MXene cotton-spun electrode, wherein the silicone / polyvinyl alcohol layer is a friction power generation layer, the silicone / graphite carbon nitride layer is an intermediate layer, and the MXene cotton-spun electrode is a bottom electrode. The present invention applies a friction nanogenerator to a tactile sensor, and the silicone-based flexible tactile sensor can not only be used for a remote control switch of a smart home, but also can be combined with a convolutional gated recurrent unit machine learning algorithm to realize real-time recognition of different objects by a robot manipulator, and has important application potential in smart homes, robot tactile perception, intelligent auxiliary sorting systems, human-computer interaction, and the like.

Description

Technical Field

[0001] The present invention belongs to the technical field of friction nanogenerators and sensors, and specifically relates to a silicone-based flexible tactile sensing sensor based on a friction nanogenerator, and a preparation method and application thereof. Background Art

[0002] With the development of technologies such as the Internet of Things, intelligent robots, and human-computer interaction, tactile sensors, as important devices for tactile perception, have also received more and more widespread attention. At present, the mainstream tactile sensors mainly include piezoelectric and piezoresistive types. Piezoresistive sensors generate resistance change signals and consume additional electrical energy; piezoelectric sensors generate low voltages, and the subsequent signal processing is very complicated. In recent years, friction nanogenerators based on the combination of triboelectric effect and electrostatic induction have been favored because of their advantages such as self-powered, strong adaptability, and high voltage generated by triboelectric electricity.

[0003] At present, many flexible tactile sensors with excellent performance have been developed. These sensors use flexible substrates and new nanomaterials and have the characteristics of high sensitivity and rapid response. However, there are still problems such as poor stability, severe drift, complex manufacturing process, and high cost of tactile sensors. This leads to huge challenges in the practical application of tactile sensors.

[0004] Flexible tactile sensors based on friction nanogenerators have broad application potential in wearable devices, human-machine interaction, robot touch, and other fields. For example, flexible tactile sensors are attached to human wrists or clothing, and touch signals are collected and processed through the Internet of Things and wireless transmission technology. In addition, flexible tactile sensors are attached to robot fingers and combined with artificial intelligence algorithms to give robots the ability to sense touch. Summary of the invention

[0005] In view of the shortcomings of the prior art, the present invention provides a silicone-based flexible tactile perception sensor based on a friction nanogenerator and a preparation method and application thereof, which realizes touch detection of the smart home system and also realizes real-time recognition of different objects by combining with a robot manipulator and a machine learning algorithm.

[0006] In order to achieve the above-mentioned objectives, the present invention adopts the following technical scheme: a silicone-based flexible tactile sensing sensor based on a friction nanogenerator, which is composed of a silicone / polyvinyl alcohol layer, a silicone / graphite phase carbon nitride layer and a MXene cotton-spun electrode, with the silicone / polyvinyl alcohol layer as the friction power generation layer, the silicone / graphite phase carbon nitride layer as the middle layer, and the MXene cotton-spun electrode as the bottom electrode.

[0007] Furthermore, the sensing principle of the sensor is:

[0008] The electrical signal generation of the silicone-based flexible tactile sensor is based on the coupling effect of contact electrification and electrostatic induction; when an object contacts the sensor, positive and negative charges are accumulated on the surfaces of the two due to their different electron affinities; when the object separates from the sensor, a potential difference is formed between the two electrodes because opposite charges are separated from each other, thereby generating an electrical signal;

[0009] The presence of the silica gel / g-C3N4 layer weakens the charge repulsion and provides more carriers for the injection of new triboelectric charges. The silica gel / PVA layer generates triboelectric charges, the silica gel / g-C3N4 layer mainly plays the role of charge transfer and weakening charge repulsion, and the MXene cotton electrode maintains the induced charge. The multiple layers work together to obtain the maximum charge density, ultimately improving the electrical output performance.

[0010] A method for preparing a silicone-based flexible tactile sensor based on a friction nanogenerator, the preparation steps comprising:

[0011] (1) Preparation of MXene solution

[0012] Lithium fluoride was added to the hydrochloric acid solution and stirred, and titanium aluminum carbide powder was slowly added to the above solution, and stirred continuously at 35°C for 24 hours to complete the reaction. The resulting solution was washed with deionized water several times under the action of a centrifuge until the pH value of the supernatant was greater than 6, and then the solution was ultrasonically treated for 60 minutes and centrifuged at 5000 rpm. Finally, the dark green supernatant was collected as a MXene solution for standby use;

[0013] (2) Preparation of graphite phase carbon nitride powder

[0014] Melamine and urea were mixed and heated to 450°C at a heating rate of 3°C per minute in a quartz tube furnace under an argon atmosphere. After maintaining a high temperature of 450°C for 4 hours, the synthesized powder was ground and rinsed with deionized water, and then filtered and ground in a mortar to obtain relatively pure g-C3N4. Finally, the product was dried in vacuum to powder for later use.

[0015] (3) Preparation of MXene cotton-spun electrodes

[0016] The cotton yarn was treated with 1 wt% NaOH in a 99°C water bath for 30 minutes and dried, and then the cotton yarn was immersed in a MXene solution for 10 minutes to fully soak it, and then placed in a 35°C vacuum drying oven to dry for 10 hours to obtain a MXene cotton spinning electrode;

[0017] (4) Preparation of silica gel / g-C3N4 layer

[0018] The g-C3N4 was uniformly dispersed in anhydrous ethanol by ultrasonic treatment and added to the silica gel matrix under magnetic stirring, and then heated to completely evaporate the ethanol in the uniform mixture. The silica gel curing agent was added to the prepared mixture and spin-coated on the surface of the flexible MXene cotton-spun electrode, and finally dried in a vacuum at 25 °C for 4 h to obtain the silica gel / g-C3N4 layer.

[0019] (5) Preparation of silica gel / polyvinyl alcohol layer

[0020] The polyvinyl alcohol powder was uniformly dispersed in the silica gel matrix by magnetic stirring and ultrasonic treatment, and the silica gel curing agent was added to the mixture and spin-coated on the surface of the silica gel / g-C3N4 layer, and finally dried in a vacuum at 25°C for 4 hours to obtain the silica gel / PVA layer;

[0021] (6) Preparation of silicone-based flexible tactile sensors

[0022] The silica gel / g-C3N4 layer and the silica gel / PVA layer were superimposed from bottom to top on the MXene cotton spinning electrode by spin coating to produce a silica gel-based flexible tactile sensing sensor.

[0023] Furthermore,

[0024] The mass ratio of lithium fluoride to titanium aluminum carbide in (1) is 1:1, and the concentration of the hydrochloric acid solution is 9M.

[0025] Furthermore,

[0026] The mass ratio of melamine to urea in (2) is 1:1.

[0027] Furthermore,

[0028] The mass ratio of the silica gel matrix to the curing agent in (4) is 1:1, the mass fraction ratio of the g-C3N4 is 1.6wt%, and the thickness of the silica gel / g-C3N4 layer is 200-300μm.

[0029] Furthermore,

[0030] The mass ratio of the silica gel matrix to the curing agent in (5) is 1:1, the mass fraction ratio of the PVA is 10wt%, and the thickness of the silica gel / PVA layer is 200-300μm.

[0031] Application of a silicone-based flexible tactile sensor based on a friction nanogenerator,

[0032] The silicone-based flexible tactile sensor is used in remote control switches of smart homes to control different electronic and electrical devices, and can also be combined with the Internet of Things for touch detection;

[0033] Alternatively, the silicone-based flexible tactile sensing sensor is used in a tactile sensing detection system, and the tactile sensing detection system is used to generate different electrical signals when a manipulator equipped with a sensing sensor touches an object and can classify and identify the object.

[0034] Furthermore, the tactile perception detection system includes a silicone-based flexible tactile perception sensor, a signal processing module, a Bluetooth data transmission module, a single-chip computer data acquisition module, an Android APP, and a CNN-GRU machine learning classification network built according to specific input features.

[0035] Furthermore,

[0036] The remote control switch of the smart home is to connect the voltage of the silicone-based flexible tactile sensor to the signal processing module. After the processing is completed, real-time monitoring can be achieved on the Internet of Things cloud through the WIFI module, and the corresponding relay can be controlled by the single-chip microcomputer to achieve remote control and monitoring of the smart home.

[0037] Alternatively, the tactile perception detection system is to assemble a silicone-based flexible tactile perception sensor on a manipulator, and connect the electrical signals generated when detecting touching different balls to a signal processing module. After the processing is completed, the signal is connected to the ADC data acquisition port of the microcontroller and the signal waveform can be monitored in real time on the Android APP through the Bluetooth module; at the same time, the data is identified and classified through the constructed CNN-GRU machine learning classification model, and finally the four subcategories are identified, with a recognition accuracy of 96.8%.

[0038] The benefits of the present invention's flexible tactile sensor based on a triboelectric nanogenerator and its preparation method and application are as follows:

[0039] (1) The present invention is based on the contact electrification of the friction nanogenerator, which has the advantages of self-powered, low cost and high sensitivity. At the same time, the principle of contact separation is to utilize the tiny contact pressure difference between the friction layer and the object to generate a signal. Therefore, the silicone-based flexible tactile sensor in the present invention has a high perception of external changes, can directly contact with the object to generate a signal, greatly improves the perception ability, and has the advantages of small size, low cost and no need for external power supply equipment.

[0040] (2) The shape and size of the silicone-based flexible tactile sensing sensor in the present invention will not affect its sensing process. Therefore, the shape and size of the silicone-based flexible tactile sensing sensor can be changed, so that multiple sensors can be integrated together and placed in different sensing scenarios.

[0041] (3) The present invention uses a silicone-based flexible tactile sensor to detect the electrical signals when touching different balls, and then identifies the balls. It also uses a machine learning algorithm to predict the detection results, showing its application potential in robot tactile perception, intelligent assisted sorting system, human-computer interaction, etc. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 This is a diagram of the preparation process of a silicone-based flexible tactile sensor according to an example of the present invention;

[0043] Figure 2 This is a photo of a silicone-based flexible tactile sensor according to an example of the present invention;

[0044] Figure 3 a is a SEM image of g-C3N4 of the present invention;

[0045] Figure 3 b is a SEM image of the MXene of the present invention;

[0046] Figure 3 c, d are SEM images of MXene cotton-spun electrodes of the present invention;

[0047] Figure 4 X-ray diffraction (XRD) diagrams of MXene and g-C3N4 of the present invention;

[0048] Figure 5 This is a test diagram of the flexibility of the silicone-based flexible tactile sensor according to the present invention;

[0049] Figure 6 a is a diagram showing the operating mechanism of the silicone-based flexible tactile sensing sensor according to an example of the present invention;

[0050] Figure 6 b is a schematic diagram of the functions of each layer of the silicone-based flexible tactile sensing sensor according to the present invention;

[0051] Figure 7 a The voltage output characteristic diagram of the present invention example when it is separated from the PI contact at 1 Hz, 2 Hz, 4 Hz, and 8 Hz;

[0052] Figure 7 b The current output characteristic diagram of the present invention example generated by contact and separation with PI at frequencies of 1 Hz, 2 Hz, 4 Hz, and 8 Hz;

[0053] Figure 7 c Voltage output characteristic diagram of the present invention example when in contact and separation with PI at pressures of 1 kPa, 2 kPa, 3 kPa, 4 kPa, 5 kPa, and 6 kPa;

[0054] Figure 7d Current output characteristic diagram of the present invention example when in contact and separation with PI at pressures of 1 kPa, 2 kPa, 3 kPa, 4 kPa, 5 kPa, and 6 kPa;

[0055] Figure 7 e The repeatability test diagram of the present invention example under the conditions of 8Hz and 5kPa, running continuously for 2000s;

[0056] Figure 8 This is a circuit diagram of a wireless intelligent switch assembled with a silicone-based flexible tactile sensor, a microcontroller (MCU), and a relay according to an example of the present invention;

[0057] Fig. 9 a is a photo of a wireless smart switch wristband based on a silicone-based flexible tactile sensing sensor according to an example of the present invention;

[0058] Fig. 9 b is a real photo of the simulated micro smart home in the example of the present invention;

[0059] Fig.10 a is a computer interface of an intelligent cloud-based real-time detection platform based on a silicone-based flexible tactile sensing sensor in an example of the present invention;

[0060] Fig.10 b is a mobile phone interface of an intelligent cloud-based real-time detection platform based on a silicone-based flexible tactile sensing sensor according to an example of the present invention;

[0061] Fig.11 a is a simplified schematic diagram of a silicone-based flexible tactile sensor combined with a robot manipulator and a machine learning algorithm for a ball recognition scenario according to an example of the present invention;

[0062] Fig.11 b is a diagram of voltage signals generated when the silicone-based flexible tactile sensing sensor of the present invention contacts four different types of balls (bouncy ball, tennis ball, hockey ball and baseball);

[0063] Fig.12 This is a flowchart of the CNN-GRU algorithm recognition example of the present invention;

[0064] Fig.13 a is a schematic diagram of the process of combining the silicone-based flexible tactile perception sensor and the Android APP to process signals in an example of the present invention;

[0065] Fig.13 b is a flowchart of the compilation and training of the CNN-GRU example of the present invention;

[0066] Fig.14 a is the confusion matrix of the predicted categories and actual categories of the training set of the example of the present invention;

[0067] Fig.14b is a training history diagram of the predicted categories and actual categories of the training set of the example of the present invention;

[0068] Fig.15 This is a real picture of the present invention's example of a silicone-based flexible tactile sensing sensor combined with a robot manipulator and a machine learning algorithm for use in a ball recognition scenario. DETAILED DESCRIPTION

[0069] The specific implementation manner will be further described below in conjunction with the accompanying drawings.

[0070] The reagents and instruments in the following examples are all conventional experimental reagents and instruments.

[0071] Embodiment 1:

[0072] A silicone-based flexible tactile sensor based on a friction nanogenerator is composed of a silicone / polyvinyl alcohol layer, a silicone / graphite carbon nitride layer and a MXene cotton-spun electrode. The silicone / polyvinyl alcohol layer is a friction power generation layer, the silicone / graphite carbon nitride layer is an intermediate layer, and the MXene cotton-spun electrode is a bottom layer. The MXene cotton-spun electrode, the silicone / g-C3N4 layer and the silicone / PVA layer are stacked together from bottom to top by spin coating to obtain a tactile sensor with a flexible silicone-based composite friction layer. The actual picture of the tactile sensor based on the friction nanogenerator is shown in the figure. Figure 2 shown.

[0073] Characterization and analysis of silicone-based flexible tactile sensing sensors based on triboelectric nanogenerators:

[0074] Scanning electron microscopy (SEM, Hitachi S-4800, Japan) was used to characterize the microstructure of g-C3N4, MXene, and MXene cotton-spun electrodes. X-ray diffraction (XRD, Rigaku Miniflex 600) was used to obtain the X-ray diffraction patterns of g-C3N4 and MXene.

[0075] Figure 1 a is the SEM image of g-C3N4, where the wrinkled layer structure with multiple stacked layers can be clearly seen, indicating the successful preparation of g-C3N4. Figure 1 b shows the SEM image of MXene, showing that MXene is a loose, multilayered, accordion-like structure, indicating the successful preparation of MXene. Figure 3 The SEM image in c shows the microscopic picture of the MXene cotton-spun electrode, and the results show that MXene is obviously attached to the surface of the MXene cotton-spun electrode. Figure 3 d SEM further demonstrates that MXene is attached to the surface of a single MXene cotton-spun electrode, characterizing the successful preparation of the MXene cotton-spun electrode. Figure 4The crystal structures of MXene and g-C3N4 were studied by XRD, and the (002) peak of MXene and the (002) peak of g-C3N4 further verified the successful preparation of the two. Figure 5 Intuitively, the silicone-based flexible tactile sensor can be twisted, bent, held and rolled arbitrarily. This shows that the silicone-based flexible tactile sensor has good flexibility.

[0076] Sensor principle of silicone-based flexible tactile sensing sensor:

[0077] The sensing mechanism of silicone-based flexible tactile sensors is as follows Figure 6 As shown in a, it is mainly based on the coupling effect of contact charging and electrostatic induction. When the silicone-based flexible tactile sensor contacts an object (here polyimide film is used as the mechanism to explain the object) ( Figure 6 ai), negative charges accumulate on the surface of the polyimide film (PI), while positive charges accumulate on the surface of the silicone / PVA layer. This is because the former has a stronger electron affinity than the latter. At the same time, the triboelectric charge on the surface will decrease due to the electric field and the concentration gradient of the charge. Then when the silicone-based flexible tactile sensor is separated from the PI ( Figure 6 a-ii), opposite charges are separated from each other, and a potential difference is formed between the two electrodes. Driven by the electric field formed by this potential difference, an electrical signal is generated. When completely separated ( Figure 6 a-iii), the charge of the whole device is balanced and no electrical signal is generated. When they approach each other again ( Figure 6 a-iv), due to the reduction in the interlayer distance, the MXene cotton electrode has a higher electric potential, which causes the electrons to flow back to the bottom cotton electrode to form a reverse electrical signal. Therefore, through periodic contact and separation, the device can generate an alternating electrical output signal in this electrostatic induction process. It should be emphasized here that the presence of the silica gel / g-C3N4 layer weakens the charge repulsion and provides more carriers for the injection of new friction charges. Therefore, the total friction charge increases, thereby improving the power generation performance of the silicone-based flexible tactile sensor. The charge distribution of the silicone-based flexible tactile sensor and the functions of each layer of the silicone-based flexible tactile sensor are as follows: Figure 6 As shown in b. The silica gel / PVA layer generates triboelectric charge; the silica gel / g-C3N4 layer mainly plays the role of charge transfer and weakening charge repulsion; the MXene cotton electrode maintains the induced charge. The multiple layers work together to obtain the maximum charge density, ultimately improving the electrical output performance.

[0078] Electrical output performance of silicone-based flexible tactile sensing sensor:

[0079] In order to test the electrical output performance of the silicone-based flexible tactile sensor, we fabricated a 5×10 cm2 The effect of frequency and pressure on the output electrical performance is studied by using a silicone-based flexible tactile sensor. Figure 7 As shown in a and b, the effect of frequency on the silicone-based flexible tactile sensor was tested by increasing the applied frequency of constant pressure (5kPa) from 1Hz to 8Hz. We found that at all frequencies, the open circuit voltage of the silicone-based flexible tactile sensor was maintained at 310V, and the short circuit current increased from 3μA to 16μA, which indicates that the silicone-based flexible tactile sensor has reliable contact sensing ability and electrical output performance. Figure 7 c, d, when the applied pressure increases from 1kPa to 6kPa at a frequency of 1Hz, the output performance of the silicone-based flexible tactile sensor gradually increases, and the output performance tends to be stable when it exceeds 5kPa. This enables the silicone-based flexible tactile sensor to detect smaller pressures more effectively. In order to determine the functional stability of the silicone-based flexible tactile sensor, we conducted a 2000-second cyclic contact and separation experiment to verify the long-term stability of the silicone-based flexible tactile sensor ( Figure 6 e). The experiment shows that the output voltage of the silicone-based flexible tactile sensor remains stable in nearly 16,000 cycles, and the output voltage does not weaken. Therefore, the silicone-based flexible tactile sensor has good stable output and excellent durability.

[0080] A method for preparing a silicone-based flexible tactile sensor based on a friction nanogenerator, the schematic diagram of the preparation process is shown in Figure 1 As shown, follow the steps below:

[0081] (1) 2 g of lithium fluoride was added to 20 mL of 9 mol / L hydrochloric acid solution and stirred, and then 2 g of titanium aluminum carbide powder was slowly added to the above solution, and stirred continuously at 35 ° C for 24 hours to complete the reaction, and the resulting solution was washed with deionized water for multiple times under the action of a centrifuge until the pH value of the supernatant was greater than 6, and then the solution was ultrasonically treated for 60 minutes and centrifuged at 5000 rpm, and finally the dark green supernatant was collected as a MXene solution for standby use;

[0082] (2) melamine and urea were mixed in a mass ratio of 1:1 and heated to 450°C in an argon atmosphere in a quartz tube furnace at a heating rate of 3°C per minute. After maintaining the high temperature of 450°C for 4 hours, the synthesized powder was ground and rinsed with deionized water, and then filtered and ground in a mortar to obtain relatively pure g-C3N4. Finally, the product was dried in vacuum to form a powder for later use;

[0083] (3) The cotton yarn was treated with 1 wt% NaOH in a 99°C water bath for 30 minutes to remove impurities and improve the adsorption capacity of the cotton yarn. The treated cotton yarn was taken out and placed in a vacuum at 50°C for 5 hours for drying, and the dried cotton yarn was immersed in the prepared MXene solution for 10 minutes to fully soak it, and then placed in a vacuum drying oven at 35°C for 10 hours to obtain a MXene cotton spinning electrode;

[0084] (4) The prepared g-C3N4 with a mass fraction of 1.6 wt% was uniformly dispersed in anhydrous ethanol by ultrasonic treatment, and added to the silica gel matrix under magnetic stirring, and then heated to completely evaporate the ethanol in the uniform mixture. The silica gel curing agent was added to the prepared mixture at a mass ratio of 1:1 and spin-coated on the surface of the flexible MXene cotton spinning electrode, and finally dried in a vacuum at 25°C for 4 hours to obtain a silica gel / g-C3N4 layer with a thickness of 200-300 μm;

[0085] (5) PVA powder with a mass fraction of 10 wt% was uniformly dispersed in a silica gel matrix by magnetic stirring and ultrasonic treatment, a silica gel curing agent was added to the mixture (the mass ratio of silica gel matrix to silica gel curing agent was 1:1) and stirred evenly, and then the mixture was spin-coated on the surface of the silica gel / g-C3N4 layer, and finally dried in a vacuum at 25°C for 4 hours to obtain a 200-300 μm silica gel / PVA layer.

[0086] (6) The silica gel / g-C3N4 layer and the silica gel / PVA layer were stacked from bottom to top on the MXene cotton spinning electrode by spin coating to produce a silica gel-based flexible tactile sensing sensor.

[0087] Embodiment 2:

[0088] An application of a silicone-based flexible tactile sensing sensor based on a friction nanogenerator is used in remote control switches of smart homes, and combined with the cloud Internet of Things to achieve remote control and monitoring of various electronic and electrical equipment.

[0089] Figure 8 The basic circuit diagram of a silicone-based flexible tactile sensing sensor combined with a microcontroller (MCU) and a relay for a remote control switch in a smart home is shown.

[0090] Fig. 9 a shows a real photo of a wireless smart switch wristband based on a silicone-based flexible tactile sensing sensor, which can be worn on a human wrist to accurately control different smart home devices by touching the silicone-based flexible tactile sensing sensor.

[0091] Fig. 9 b shows the construction of a simulated smart home platform for testing the wireless smart switch wristband.

[0092] Fig.10 a and b show the interface diagram of the intelligent speech real-time detection platform that combines silicone-based flexible tactile sensing sensors with the Internet of Things cloud, which can perform real-time information transmission and cloud monitoring on computers and mobile phones.

[0093] Embodiment 3:

[0094] Based on the application of silicone-based flexible tactile perception sensors, the touch electrical signals of four types of balls were tested by combining with a manipulator. The CNN-GRU algorithm was used to successfully identify the type of touched ball with an accuracy rate of up to 96.8%. This shows that the device can be used as a robot tactile perception sensor in the fields of robot tactile perception, intelligent sorting systems, and human-computer interaction.

[0095] Fig.11 a shows the combination of silicone-based flexible tactile sensing sensors with a robot and machine learning algorithms to classify different types of balls. The index finger of the robot is equipped with a silicone-based flexible tactile sensing sensor to identify and classify different types of balls.

[0096] Fig.11 b shows the voltage signals generated by the silicone-based flexible tactile sensor when four types of balls (bouncy ball, tennis ball, hockey ball, and baseball) come into contact with the robotic finger. The four voltage signals are highly similar, but a careful examination of the profile of the signal peaks reveals some repeatable differences, such as the ratio of positive and negative peaks in the voltage signal and the interval between positive and negative peaks. Although the differences are small, repeatable features can be used as the basis for machine learning classification.

[0097] Fig.12The recognition flow chart of CNN-GRU is shown. First, the touch voltage signals of different balls are collected through sensors. In order to standardize the input signal, batch normalization is used on the data before the model starts. The model optimizer is Adam with default parameters, and the loss function is binary cross entropy. Four sets of touch voltage signals of different balls are collected, including bouncy balls, tennis balls, hockey balls, and baseballs. Each set contains 500 signals with a length of 80 samples. A total of 2000 signals are shuffled and then split into a training set (1600) and a validation set (400). In the recognition flow chart, the model receives a signal with a length of 80 samples and finally gives the feature value of the relevant material. The model mainly consists of a convolutional neural network (CNN) and two gated recurrent units (GRU), where the CNN layer extracts local features and the GRU unit better captures the long-term dependencies in the data. The GRU unit consists of a reset gate and an update gate, which enables the GRU unit to remember the information of many earlier timestamps and predict the current state based on the information obtained from the previous state. These gates help the GRU determine when to send past information to future states. The GRU unit helps the model extract long-term temporal dependencies in activity data.

[0098] Fig.13 Figure a shows the schematic diagram of the process of combining the silicone-based flexible tactile sensor with the Android APP to process signals. First, the signals collected by the silicone-based flexible tactile sensor are transmitted to the Android APP via a Bluetooth device. Then, the pre-trained CNN-GRU model deployed in advance on the Android APP is used to identify and predict these collected signals to obtain the type of ball.

[0099] Fig.13 b shows the flow chart of CNN-GRU compilation and training. First, the touch voltage signals of four kinds of balls are received, and all the signals are divided into two categories: training set and validation set. Then, the CNN-GRU model is improved by learning and training the training set signals. Finally, the accuracy of the model is verified by the validation set.

[0100] Fig.14 a shows the confusion matrix of the predicted labels and true labels of the training set. The results show that the system can recognize different types of balls with an average prediction accuracy of up to 96.8%.

[0101] Fig.14 b shows the training history diagram of the predicted categories and actual categories of the training set. The results show that after 30 trainings, the accuracy of the model converges to close to 1.

[0102] Fig.15The actual picture shows the combination of sensors, robot manipulators, and machine learning algorithms for ball recognition scenarios. Based on the voltage signals obtained from 5 touches, the Android APP that deploys the pre-trained model can accurately predict the type of ball that is touched. Then, the Android APP controls the manipulator to accurately send the ball to the corresponding basket.

[0103] The application of silicone-based flexible tactile sensing sensors illustrates the huge application potential of silicone-based flexible tactile sensing sensors assisted by CNN-GRU algorithm in the fields of robot manipulators, robot tactile perception, intelligent sorting systems, and human-computer interaction.

[0104] The above embodiments are only for illustrating the technical concept and features of the present invention, and their purpose is to enable ordinary technicians in the field to understand the content of the present invention and implement it accordingly, and they cannot be used to limit the protection scope of the present invention. Any equivalent changes or modifications made based on the essence of the content of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for preparing a silicone-based flexible tactile sensor based on a friction nanogenerator, wherein the silicone-based flexible tactile sensor is composed of a silicone / polyvinyl alcohol layer, a silicone / graphite carbon nitride layer and a MXene cotton-spun electrode, wherein the silicone / polyvinyl alcohol layer is a friction power generation layer, the silicone / graphite carbon nitride layer is an intermediate layer, and the MXene cotton-spun electrode is a bottom electrode, wherein the silicone / polyvinyl alcohol layer is a friction power generation layer, the silicone / graphite carbon nitride layer is a middle ... and the method is characterized in that: The preparation steps include: (1) Preparation of MXene solution Lithium fluoride was added to the hydrochloric acid solution and stirred, and titanium aluminum carbide powder was slowly added to the above solution, and stirred continuously at 35°C for 24 hours to complete the reaction. The resulting solution was washed with deionized water several times under the action of a centrifuge until the pH value of the supernatant was greater than 6, and then the solution was ultrasonically treated for 60 minutes and centrifuged at 5000 rpm. Finally, the dark green supernatant was collected as a MXene solution for standby use; (2) Preparation of graphite phase carbon nitride powder Melamine and urea were mixed and heated to 450°C at a heating rate of 3°C per minute in a quartz tube furnace under an argon atmosphere. After maintaining a high temperature of 450°C for 4 hours, the synthesized powder was ground and rinsed with deionized water, and then filtered and ground in a mortar to obtain relatively pure g-C3N4. Finally, the product was dried in vacuum to powder for later use. (3) Preparation of MXene cotton-spun electrodes The cotton yarn was treated with 1 wt% NaOH in a 99°C water bath for 30 minutes and dried, and then the cotton yarn was immersed in a MXene solution for 10 minutes to fully soak it, and then placed in a 35°C vacuum drying oven to dry for 10 hours to obtain a MXene cotton spinning electrode; (4) Preparation of silica gel / g-C3N4 layer The g-C3N4 was uniformly dispersed in anhydrous ethanol by ultrasonic treatment and added to the silica gel matrix under magnetic stirring, and then heated to completely evaporate the ethanol in the uniform mixture. The silica gel curing agent was added to the prepared mixture and spin-coated on the surface of the flexible MXene cotton-spun electrode, and finally dried in a vacuum at 25 °C for 4 h to obtain the silica gel / g-C3N4 layer. (5) Preparation of silica gel / polyvinyl alcohol layer The polyvinyl alcohol powder was uniformly dispersed in the silica gel matrix by magnetic stirring and ultrasonic treatment, and the silica gel curing agent was added to the mixture and spin-coated on the surface of the silica gel / g-C3N4 layer, and finally dried in a vacuum at 25°C for 4 hours to obtain the silica gel / PVA layer; (6) Preparation of silicone-based flexible tactile sensors The silica gel / g-C3N4 layer and the silica gel / PVA layer were superimposed from bottom to top on the MXene cotton spinning electrode by spin coating to produce a silica gel-based flexible tactile sensing sensor.

2. The method for preparing a silicone-based flexible tactile sensor based on a friction nanogenerator according to claim 1, characterized in that: The mass ratio of lithium fluoride to titanium aluminum carbide in (1) is 1:1, and the concentration of the hydrochloric acid solution is 9M.

3. The method for preparing a silicone-based flexible tactile sensor based on a friction nanogenerator according to claim 1, characterized in that: The mass ratio of melamine to urea in (2) is 1:

1.

4. The method for preparing a silicone-based flexible tactile sensor based on a friction nanogenerator according to claim 1, characterized in that: The mass ratio of the silica gel matrix to the curing agent in (4) is 1:1, the mass fraction ratio of the g-C3N4 is 1.6wt%, and the thickness of the silica gel / g-C3N4 layer is 200-300μm.

5. The method for preparing a silicone-based flexible tactile sensor based on a friction nanogenerator according to claim 1, characterized in that: The mass ratio of the silica gel matrix to the curing agent in (5) is 1:1, the mass fraction ratio of the PVA is 10wt%, and the thickness of the silica gel / PVA layer is 200-300μm.

6. An application of a silicone-based flexible tactile sensor based on a triboelectric nanogenerator prepared by the method according to any one of claims 1 to 5, characterized in that: The silicone-based flexible tactile sensing sensor is used in remote control switches of smart homes to control different electronic and electrical devices, and is combined with the Internet of Things for touch detection.

7. The use according to claim 6, characterized in that: The silicone-based flexible tactile sensing sensor is used in a tactile sensing detection system, and the tactile sensing detection system is used to generate different electrical signals when a manipulator equipped with a sensing sensor touches an object and can classify and identify the object.

8. The use according to claim 7, characterized in that: The tactile perception detection system includes a silicone-based flexible tactile perception sensor, a signal processing module, a Bluetooth data transmission module, a single-chip computer data acquisition module, an Android APP, and a CNN-GRU machine learning classification network built according to specific input features.

9. The application of the silicone-based flexible tactile sensor based on the friction nanogenerator according to claim 6 is characterized by: The remote control switch of the smart home connects the voltage of the silicone-based flexible tactile sensing sensor to the signal processing module. After the processing is completed, real-time monitoring on the Internet of Things cloud is achieved through the WIFI module and the corresponding relay is controlled by the single-chip microcomputer to realize remote control and monitoring of the smart home.

10. The application of the silicone-based flexible tactile sensor based on the friction nanogenerator according to claim 7 is characterized by: The tactile perception detection system is to assemble a silicone-based flexible tactile perception sensor on a manipulator, and connect the electrical signals generated when detecting touching different balls to a signal processing module. After the processing is completed, the signal is connected to the ADC data acquisition port of the single-chip microcomputer and the signal waveform can be monitored in real time on the Android APP through the Bluetooth module; at the same time, the data is identified and classified through the constructed CNN-GRU machine learning classification model, and finally the four subcategories are identified, with an identification accuracy of 96.8%.

Images