Kiwi storage days prediction system based on perception gloves
By designing a portable sensing glove and combining it with a CNN network prediction model, the issues of portability, cost, and intelligence of kiwi fruit storage day prediction equipment were solved, enabling convenient prediction of kiwi fruit storage days and improving the operational efficiency and customer satisfaction of supermarkets and fruit stores.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- ZHENGZHOU UNIV
- Filing Date
- 2022-08-16
- Publication Date
- 2026-05-22
AI Technical Summary
Existing kiwifruit storage days prediction devices are poorly portable, expensive, have limited functionality, and low intelligence, making them difficult to popularize in supermarkets and fruit shops, and they cannot accurately predict the storage days of kiwifruit.
Design a kiwi fruit storage days prediction system based on a sensing glove, including modules for acquiring information on hardness, volume, weight, temperature and humidity. Combine wireless Bluetooth communication and microcontroller processing, a storage days prediction model is built by training a CNN network, and the results are displayed through a human-computer interaction interface.
It enables portable, multi-functional kiwi fruit information measurement, reducing equipment costs and improving measurement efficiency. Furthermore, it accurately predicts the storage days of kiwi fruit through predictive algorithms, thereby enhancing customer satisfaction and reducing waste for businesses.
Smart Images
Figure CN115345368B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of food testing technology, and in particular to a kiwifruit storage days prediction system based on sensing gloves. Background Technology
[0002] With societal development, market demand for kiwifruit is increasing, and quality requirements are gradually rising. Quality grading plays an indispensable role in the kiwifruit supply chain. Identifying the shelf life of kiwifruit is crucial for supermarkets and fruit shops to rationally store and sell it. Currently, there are many research findings on kiwifruit characteristics both domestically and internationally. Research on kiwifruit characteristics mainly uses CCD cameras, employing non-destructive testing and computer analysis to analyze and judge each fruit individually. Fruit quality based on mechanical, acoustic, vibration, and optical properties is widely studied. However, the development of methods for predicting the shelf life of kiwifruit has been slow, and the implementation of real-time systems still faces many challenges. Currently, there is no inexpensive and efficient device on the market that can predict the shelf life of kiwifruit.
[0003] Currently, kiwi fruit measurement systems mainly suffer from the following problems: 1. Poor portability and high price. Current kiwi fruit data measurement systems primarily rely on sound waves and spectroscopy, which are cumbersome, have high requirements for the operating environment, and are expensive. This hinders the widespread adoption of kiwi fruit measurement equipment. 2. Limited functionality. Current fruit measurement equipment generally only provides one function, such as measuring firmness, weight, and volume, and the implementation method is fixed, without expansion into other characteristics. This is a major obstacle limiting the system's use. 3. Low level of intelligence. Current measuring devices are limited to measuring fruit parameters and do not predict fruit quality or storage days based on these parameters. This low level of intelligence results in poor practicality. Due to these issues, there are still no kiwi fruit storage day prediction devices available for supermarkets and fruit shops.
[0004] Existing kiwifruit firmness testing devices can be divided into two categories: destructive and non-destructive testing devices. Destructive testing devices can provide accurate measurements of fruit firmness, but they cause irreversible damage to the fruit. Among non-destructive testing sensors, the Shore hardness tester and infrared spectrometer are representative examples. The Shore hardness tester requires manual pressure for measurement, while the infrared spectrometer requires analysis of the fruit's internal characteristics to indirectly measure firmness, making them expensive. Glove-style sensing gloves can only measure finger pressure and cannot measure firmness. Therefore, a cheap, non-destructive sensing glove capable of collecting information on firmness, weight, volume, and ambient temperature and humidity is not yet widely available on the market.
[0005] In daily life, fruit shops and supermarkets often display kiwis of varying ripeness together for customers to choose from. However, most customers lack experience in selecting kiwis, resulting in a large amount of kiwis rotting for the vendors, while customers don't get to eat them at the expected time. Therefore, accurately predicting the shelf life of kiwis is crucial for improving customer satisfaction and reducing waste for vendors. Summary of the Invention
[0006] To address the shortcomings of existing technologies, the present invention aims to provide a kiwi fruit storage days prediction system based on sensing gloves.
[0007] To achieve the objectives of this invention, the technical solution adopted is as follows:
[0008] A sensing glove includes a glove, an information acquisition module, an information display module, and a control module; the information acquisition module and the information display module are installed on the glove.
[0009] The information acquisition module includes a hardness information acquisition module, a volume information acquisition module, a weight information acquisition module, a humidity acquisition module, and a temperature acquisition module; the hardness information acquisition module includes three hardness sensors, distributed at the tips of the thumb, index finger, and middle finger; the volume information acquisition module includes three curvature sensors, distributed at the backs of the thumb, index finger, and middle finger.
[0010] The information acquisition module collects information about kiwifruit and sends it to the control module. After being processed by the microcontroller in the control module, the information is displayed through the information display module. The information display module displays average hardness, average volume, weight, and temperature and humidity information.
[0011] The control module sends data from the microcontroller to the host computer via a wireless Bluetooth module. The host computer then uses a kiwi fruit storage day prediction algorithm to predict the number of days the fruit can be stored, and sends the predicted number of days to the control module's microcontroller for display via an information display module.
[0012] Furthermore, the hardness device includes a top, a housing, a spring, a chassis, and a circular thin-film pressure sensor;
[0013] The top of the housing has a hole for sliding connection with the top head, and the bottom has a groove for fixed connection with the circular diaphragm pressure sensor. The bottom of the chassis is fixed above the circular diaphragm pressure sensor, and the top of the chassis has a groove for fixed connection with the lower end of the spring.
[0014] The top is a combination of a semi-circle and a cylinder. The top is a semi-circle, and the diameter of the lower cylinder is larger than the diameter of the hole at the top of the outer shell, which is used for limiting the position. The bottom of the cylinder has a groove that is fixedly connected to the upper end of the spring.
[0015] The hardness tester stops the measurement when the top of the outer casing is tangent to the object being tested. The measured value is the hardness value.
[0016] Furthermore, the weight information acquisition module includes a block-shaped thin-film pressure sensor, a flexible block material, and a stitching fabric;
[0017] The block-shaped thin-film pressure sensor is fixed in the palm of the glove, the flexible block material is fixed above the block-shaped thin-film pressure sensor, and the sewn cloth is fixed above the flexible block material to cover the flexible block material and the block-shaped thin-film pressure sensor, and further fix it to the glove.
[0018] Furthermore, the humidity acquisition module and temperature acquisition module are fixed to the back of the glove, and the information display module is an OLED screen, which is fixed to the front of the glove at the wrist.
[0019] Furthermore, the control module includes a control shell, PCB, wireless Bluetooth module, microcontroller, voltage regulator, battery, fixing strap, and touch switch; the control shell has a rectangular shell structure, the PCB, wireless module, microcontroller, voltage regulator, and battery are all placed inside the control shell, the touch switch is fixed to the outside of the control shell, and the fixing strap is placed at the bottom of the control shell for carrying.
[0020] The battery output is connected to the voltage regulator input, and the voltage regulator output is connected to the PCB. The PCB controls the voltage conversion to power the various sensors and the microcontroller.
[0021] The touch switch is connected to the microcontroller, which communicates with the host computer via a wireless Bluetooth module. When the touch switch is pressed, the microcontroller sends the collected current time data to the host computer. The host computer uses a kiwi fruit storage day prediction algorithm to predict the number of days that can be stored and sends the number of days that can be stored back to the microcontroller.
[0022] A kiwifruit storage days prediction system based on a sensing glove includes a sensing glove, a human-computer interaction interface, and an operating system that implements the prediction process based on a kiwifruit storage days prediction model.
[0023] Furthermore, the human-computer interaction interface includes glove information, storage days information, kiwi parameter information, and a reset button; the glove information display is located on the left side of the human-computer interaction interface, with a glove icon, thumb, middle finger, and index finger, and from top to bottom, there are real-time display windows for hardness information and bending information, a real-time display window for weight information in the palm position, and below it from left to right, there are real-time display windows for temperature and humidity information;
[0024] The right side displays the kiwi fruit parameter information interface. From top to bottom, the display windows show the storage days, average volume, average hardness, weight, temperature, and humidity information.
[0025] Furthermore, data on kiwifruit are collected to construct a dataset, which is then trained to obtain a kiwifruit storage days prediction model. The model can be constructed based on 1D CNN, SVM, LSTM, or CNN-LSTM.
[0026] Furthermore, the process of constructing the dataset is as follows: Select verified edible kiwifruit, measure them with equipment to determine the hardness range of edible kiwifruit; randomly select a number of overly hard and inedible kiwifruit, collect the weight, hardness, volume of the kiwifruit and the temperature and humidity information of the environment with sensing gloves at room temperature every day, check whether they have reached the edible range with equipment every day, and record the number of days of collection for each sample.
[0027] When a kiwi fruit sample reaches the edible hardness threshold, the current number of days is used as label 0, and the predicted number of days it can be stored is 0. Then, the dataset labels for this sample are summed in reverse order of the number of days of the previously sampled samples to obtain the predicted number of days it can be stored, and the number of labels is added.
[0028] Furthermore, the prediction process implemented by the operating system control includes: wearing a sensing glove, randomly selecting a kiwi, placing the kiwi in the palm of the hand, and pressing a touch switch to start the device; collecting temperature, humidity, and weight information, displaying it on the OLED screen, and sending it to the host computer via Bluetooth; the host computer enters a loop, waiting for the collection of hardness and curvature information, and displaying the temperature, humidity, and weight information on the human-machine interface; using the sensing glove to hold the kiwi until the top surface of the hardness device is tangent to the kiwi, pressing the touch switch, collecting hardness and curvature information, calculating the average value, displaying it on the OLED screen, and sending it to the host computer via Bluetooth; the host computer receives the collected hardness and curvature information and displays it on the human-machine interface.
[0029] The host computer inputs the collected information into the kiwi fruit storage days prediction model and waits for the results to be output. After the calculation is completed, the host computer exits the loop, the kiwi fruit storage days are displayed on the human-computer interaction interface, and sent to the microcontroller via Bluetooth. The microcontroller displays the kiwi fruit storage days on the OLED screen. Other values are reset, and the system waits to detect the next kiwi fruit.
[0030] The beneficial effects of this invention are that, compared with the prior art, this invention addresses the pain point of unpredictable storage days for kiwifruit by designing a kiwifruit storage days prediction system based on sensing gloves. The main advantages are as follows:
[0031] 1. Portable Multifunctional Sensing Glove. This invention addresses the problems of poor portability, high cost, and limited functionality in existing measuring devices by designing a portable sensing glove capable of measuring the volume, weight, firmness, and ambient temperature and humidity of kiwifruit. The fruit measuring device is further simplified by being designed as a glove, allowing the wearer to collect fruit information with just one hand, simplifying the operation of the measuring instrument and improving measurement efficiency.
[0032] 2. Kiwi Fruit Storage Days Prediction Algorithm. This invention addresses the social pain point of difficulty in predicting the storage days of kiwi fruit and the low level of intelligence in existing inexpensive measurement equipment. It constructs a kiwi fruit storage days prediction model by training the collected information using a CNN network, thereby enabling the prediction of kiwi fruit storage days, ensuring customer satisfaction and reducing kiwi fruit waste for merchants.
[0033] 3. A Kiwi Fruit Storage Days Prediction System Based on Sensing Gloves. This invention aims to better apply the kiwi fruit storage days prediction algorithm and simplify operation. It combines a portable sensing glove with the kiwi fruit storage days prediction algorithm to build a kiwi fruit storage days prediction system. Furthermore, a human-computer interaction interface was designed to visualize system information and ensure more convenient human-computer interaction. Attached Figure Description
[0034] Figure 1 This is a schematic diagram of the front and back of the sensory glove;
[0035] Figure 2 This is a schematic diagram of the inside of the hardness tester;
[0036] Figure 3 This is a schematic diagram of the control module;
[0037] Figure 4 This is a schematic diagram illustrating the principle of a kiwifruit storage days prediction system.
[0038] Figure 5 This is a schematic diagram of a human-computer interaction interface;
[0039] Figure 6 This is a schematic diagram of the process for predicting the storage days of kiwifruit;
[0040] Figure 7 It is a microcontroller and PCB circuit design diagram. Detailed Implementation
[0041] The technical solution of the present invention will be further described below with reference to the accompanying drawings and embodiments. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and should not be used to limit the scope of protection of this application.
[0042] This invention addresses current pain points in daily life and existing problems in kiwifruit testing systems. With the aim of improving the accuracy, convenience, and practicality of quality testing while reducing system costs, it designs a portable sensing glove to measure the volume, weight, firmness, and ambient temperature and humidity of kiwifruit. Furthermore, it trains the dataset collected by the sensing glove using a CNN network to construct a kiwifruit storage days prediction model, enabling the prediction of the storage days required for kiwifruit.
[0043] This invention combines a portable sensing glove with a kiwifruit storage days prediction model to form a kiwifruit storage days prediction system. Furthermore, a human-computer interaction interface is designed to visualize system information and ensure more convenient human-computer interaction.
[0044] The portable kiwifruit storage days prediction system designed in this invention is inexpensive and user-friendly, making it easy for individual merchants to use to independently predict storage days. Individual fruit supermarkets are fully capable of deploying this product in their stores; after purchasing kiwifruit, they can use the system to perform a simple prediction to obtain the predicted storage days for the kiwifruit, facilitating reasonable storage and sales for supermarkets and fruit stores.
[0045] The system of this invention predicts the storage days of kiwifruit based on the hardness, weight, volume, and ambient temperature and humidity information of the kiwifruit. This information is collected through a sensing glove.
[0046] like Figure 1 As shown, the sensing glove consists of a glove 1, an information acquisition module, an information display module 7, and a control module. Glove 1 is made of elastic material, with rubber dots on the front to enhance friction and ensure stability when gripping objects. The information acquisition module collects information about the kiwi fruit and sends it to the control module. After processing by the microcontroller in the control module, the information is displayed through the information display module.
[0047] The information acquisition module includes a hardness information acquisition module 2, a volume information acquisition module 3, a weight information acquisition module 4, a humidity acquisition module 5, and a temperature acquisition module 6. The hardness information acquisition module 2 consists of three hardness sensors, distributed at the tips of the thumb, index finger, and middle finger, ensuring the acquisition of fruit hardness information. The volume information acquisition module 3 includes three curvature sensors, distributed on the backs of the thumb, index finger, and middle finger, thereby measuring the external volume of the fruit by measuring the curvature of the fingers.
[0048] like Figure 2As shown, the hardness device includes a top head 21, a housing 22, a spring 23, a base 24, and a circular thin-film pressure sensor 25. The housing 22 has a hole at its top for sliding connection with the top head 21, and a groove at its bottom for fixed connection with the circular thin-film pressure sensor 25. The bottom of the base 24 is fixed above the circular thin-film pressure sensor 25, and a groove at its top is fixedly connected to the lower end of the spring 23. The top head 21 is a combination of a semi-circle and a cylinder; the top is semi-circular, and the diameter of the lower cylinder is larger than the diameter of the hole at the top of the housing 22, used for limiting movement. The bottom of the cylinder has a groove for fixed connection to the upper end of the spring 23. The hardness device stops the measurement when the top of the housing 22 is tangent to the object being measured; the measured value is the hardness value.
[0049] The weight information acquisition module 4 consists of a 20mm×20mm block-shaped thin-film pressure sensor, a flexible block material, and a sewn fabric. The block-shaped thin-film pressure sensor is fixed in the palm of the glove 1. The flexible block material, 2mm high, is fixed above the block-shaped thin-film pressure sensor to ensure even force distribution. The sewn fabric is fixed above the flexible block material to shield both the flexible block material and the block-shaped thin-film pressure sensor, and further secures it to the glove. Through this structural design, the weight information acquisition module 4 can acquire the weight information of a kiwi fruit when it is placed in the palm of the hand.
[0050] Humidity acquisition module 5 and temperature acquisition module 6 are fixed to the back of glove 1 and use temperature and humidity sensors to measure the temperature and humidity information of the environment.
[0051] The information display module 7 is an OLED screen, fixed to the wrist on the front of the glove. It displays average hardness, average volume, weight, temperature and humidity information, as well as the number of days the glove can be stored, as predicted by the algorithm. The number of days the glove can be stored is transmitted from the host computer to the microcontroller via Bluetooth, while the average hardness, average volume, and weight information are obtained through processing by the microcontroller.
[0052] like Figure 4 As shown, the hardness information acquisition module 2, volume information acquisition module 3, weight information acquisition module 4, humidity acquisition module 5, and temperature acquisition module 6 are all connected to the control module, sending the acquired information to the microcontroller for processing. The control module is also connected to the information display module. The control module integrates the curvature sensor, hardness device, pressure sensor, temperature sensor, humidity sensor, and OLED through the microcontroller. The information acquired by the curvature sensor, hardness device, pressure sensor, temperature sensor, and humidity sensor is all sent to the microcontroller for processing, and then the processed information is displayed on the OLED.
[0053] like Figure 3As shown, the control module includes a control housing 8, a PCB 9, a wireless Bluetooth module 10, a microcontroller 11, a voltage regulator 12, a battery 13, a mounting strap 14, and a touch switch 15. The control housing 8 is a rectangular shell structure. The PCB 9, wireless module 10, microcontroller 11, voltage regulator 12, and battery 13 are all housed inside the control housing 8. The touch switch 15 is fixed to the outside of the control housing 8. The mounting strap 14 is located at the bottom of the control housing 8 for carrying. The output of battery 13 is connected to the input of voltage regulator 12, and the output of voltage regulator 12 is connected to PCB 9. The PCB controls the voltage conversion to power the various sensors and microcontroller 11, providing different power supply voltages. The microcontroller 11 is an STM32F407VGT6. Different voltage power supply modules and a linear voltage conversion module are installed on the PCB.
[0054] The curvature sensor and hardness device are connected to a linear voltage conversion module, directly outputting the detection information to the module to convert the resistance signal into a voltage signal. The temperature sensor and humidity sensor are connected to channels 1-8 of the ADC3 of the microcontroller 11 to convert voltage to analog signals. Hardness, volume, temperature, and humidity are collected according to different conversion formulas for different sensors. All data acquisition takes 10ms.
[0055] The touch switch 15 is connected to the general I / O port of the microcontroller 11. The microcontroller 11 communicates with the host computer via the wireless Bluetooth module 10. When the touch switch 15 is pressed, the microcontroller 11 sends the corresponding data collected at the current moment to the host computer. The host computer uses a kiwi fruit storage day prediction algorithm to predict the number of days the fruit can be stored and sends this prediction number back to the microcontroller. The microcontroller 11 transmits the data to the OLED screen via the SCL and SDA ports to display the number of days the fruit can be stored, average hardness, average volume, and weight.
[0056] The microcontroller is connected to the PCB board to ensure circuit simplicity. PCB9 connects to microcontroller 11 and has 42 I / O ports, including 9 ADC interfaces, 4 USART serial ports, 18 TIM ports, and 3 general-purpose I / O ports. A switch button is added to PCB9 to provide separate power supplies for the microcontroller and various sensors, further ensuring hardware safety during operation. The microcontroller and PCB circuit design diagram is shown below. Figure 7 As shown.
[0057] A 1D CNN model was used to construct a kiwifruit storage days prediction model, and the model was trained to obtain a kiwifruit storage days prediction model based on signal values. Data set construction is required before training the kiwifruit storage days prediction model.
[0058] One hundred edible kiwifruit samples, verified by experts, were selected, and their firmness range was determined using equipment. Thirty hundred overly firm and inedible kiwifruit were randomly selected and placed at room temperature. Daily data on the weight, firmness, volume of the kiwifruit, and the temperature and humidity of the environment were collected using sensing gloves for model training. The number of days for each sample collection was recorded, and the samples were checked daily to determine if they had reached an edible range.
[0059] When a kiwi fruit sample reaches the edible firmness threshold, the current day is assigned as label 0, predicting a shelf life of 0 days. The dataset label for this sample is then incremented in reverse order of the number of days since the last sample was collected; for example, if 5 days have passed, the label is set to 5, predicting a shelf life of 5 days. A total of 2000 samples were collected.
[0060] This invention inputs the hardness, curvature, weight, and ambient temperature and humidity values from the collected dataset into a one-dimensional 1D CNN model for training, thereby obtaining a kiwi fruit storage days prediction model based on signal values.
[0061] The 1D CNN model structure consists of a convolutional part and a fully connected part. The convolutional part comprises two sub-blocks. The first sub-block includes convolutional layers, batch normalization (BN) layers, and a ReLU activation function. The convolutional layers perform one-dimensional convolutions on the data, and the BN layers accelerate training and prevent gradient vanishing or exploding. The second sub-block includes convolutional layers, batch normalization (BN) layers, a ReLU activation function, and pooling layers. The pooling layers reduce the dimensionality of the feature maps, thus reducing the number of parameters. The fully connected part consists of fully connected layers, dropout layers, and a softmax function. The dropout layers reduce the risk of overfitting, and the fully connected layers extract and classify the features from the convolutions, feeding them back to the softmax function. The softmax function classifies and predicts signals based on the normalized probability values of the nodes.
[0062] The 1D CNN network has 6 classes, a batch size of 64, uses ADAM as the optimizer, and has an initial learning rate of 2×10⁻⁶. -5 To avoid gradient explosion or vanishing, initial weights are generated using Kaiming. The other network parameters are set to default values.
[0063] In the training process of a 1D CNN network, 60% of the data is randomly selected as the training set, 10% as the validation set, and 30% as the test set. The training set is used for model training, the validation set is used for testing and correcting the model during training, and the test set is used to verify the model's performance.
[0064] The method for constructing the algorithm model of this invention is not limited to 1D CNN. Any network structure that can process one-dimensional data can be used for data processing, including SVM, LSTM, CNN-LSTM, etc.
[0065] This invention combines a sensing glove with a kiwifruit storage days prediction model to construct a kiwifruit storage days prediction system. The system includes a human-computer interaction interface and an operating system, with the operating system used to implement the prediction process.
[0066] like Figure 5 As shown, the human-computer interaction interface is built based on the Tkinter module of Python, with the aim of ensuring the system's visibility, ease of operation, and good interactivity.
[0067] The human-computer interface includes information on the sensing glove, storage days, kiwi fruit parameters, and a reset button. The sensing glove information display is located on the left side of the interface, featuring a glove icon, and showing the thumb, middle finger, and index finger. From top to bottom, there are real-time display windows for hardness and curvature information. A real-time weight information window is located in the palm area, and below it, from left to right, are real-time temperature and humidity information windows. The right side displays the kiwi fruit parameter information, including, from top to bottom, a window showing the storage days, average volume, average hardness, weight, temperature, and humidity information.
[0068] like Figure 6 The diagram illustrates the prediction process controlled by the operating system. The process includes: wearing a sensing glove, randomly selecting a kiwi, placing it in the palm of the hand, and pressing a touch switch to activate the device; collecting temperature, humidity, and weight information, displaying it on the OLED screen, and sending it to the host computer via Bluetooth; the host computer entering a loop, waiting for hardness and curvature information to be collected, and displaying the temperature, humidity, and weight information on the human-machine interface; using the sensing glove to hold the kiwi until the top surface of the hardness device is tangent to the kiwi; when tangent, pressing the touch switch to collect hardness and curvature information, calculating the average value, displaying it on the OLED screen, and sending it to the host computer via Bluetooth; and the host computer receiving the collected hardness and curvature information and displaying it on the human-machine interface.
[0069] The host computer inputs the collected information into the kiwi fruit storage days prediction model, waits for the results to be output, and after the calculation is completed, the host computer exits the loop, displays the kiwi fruit storage days on the human-computer interaction interface, and sends it to the microcontroller via Bluetooth. The microcontroller displays the kiwi fruit storage days on the OLED screen, resets other values, and waits to detect the next kiwi fruit.
[0070] The applicant of this invention has provided a detailed description of the embodiments of the invention in conjunction with the accompanying drawings. However, those skilled in the art should understand that the above embodiments are merely preferred embodiments of the invention. The detailed description is only intended to help readers better understand the spirit of the invention and is not intended to limit the scope of protection of the invention. On the contrary, any improvements or modifications made based on the inventive spirit of the invention should fall within the scope of protection of the invention.
Claims
1. A sensing glove, characterized in that, It includes a glove (1), an information acquisition module, an information display module (7), and a control module; the glove is equipped with an information acquisition module and an information display module (7); The information acquisition module includes a hardness information acquisition module (2), a volume information acquisition module (3), a weight information acquisition module (4), a humidity acquisition module (5), and a temperature acquisition module (6); the hardness information acquisition module (2) includes three hardness devices distributed on the tips of the thumb, index finger, and middle finger; the volume information acquisition module (3) includes three curvature sensors distributed on the backs of the thumb, index finger, and middle finger; The information acquisition module collects kiwi fruit information and sends it to the control module. After being processed by the microcontroller (11) of the control module, it is displayed through the information display module (7). The information display module (7) displays the average hardness, average volume, weight, temperature and humidity information. The control module sends the data information of the microcontroller (11) to the host computer through the wireless Bluetooth module (10). The host computer uses the kiwi fruit storage day prediction algorithm to predict the number of days that can be stored and sends the predicted number of days that can be stored to the microcontroller of the control module, which then displays the data through the information display module (7). The hardness device includes a top (21), a housing (22), a spring (23), a base (24), and a circular thin-film pressure sensor (25). The top of the housing (22) has a hole that is slidably connected to the top (21), and the bottom has a groove that is fixedly connected to the circular thin-film pressure sensor (25). The bottom of the base (24) is fixed above the circular thin-film pressure sensor (25), and the top of the base (24) has a groove that is fixedly connected to the lower end of the spring (23). The top (21) is a combination of a semi-circle and a cylinder. The top is a semi-circle, and the diameter of the lower cylinder is larger than the diameter of the hole at the top of the housing (22) for limiting. The bottom of the cylinder has a groove that is fixedly connected to the upper end of the spring (23). The hardness device is considered to end the measurement when the top of the housing (22) is tangent to the object being measured. The measured value is the hardness value.
2. The sensing glove according to claim 1, characterized in that, The weight information acquisition module (4) includes a block-shaped thin-film pressure sensor, a flexible block material, and a stitching cloth; Among them, the block-shaped thin film pressure sensor is fixed in the palm of the glove (1), the flexible block material is fixed above the block-shaped thin film pressure sensor, and the sewn cloth is fixed above the flexible block material to cover the flexible block material and the block-shaped thin film pressure sensor, and further fix it to the glove.
3. The sensing glove according to claim 1, characterized in that, The humidity acquisition module (5) and temperature acquisition module (6) are fixed to the back of the glove (1), and the information display module (7) is an OLED screen, which is fixed to the wrist on the front of the glove.
4. The sensing glove according to claim 1, characterized in that, The control module includes a control shell (8), a PCB (9), a wireless Bluetooth module (10), a microcontroller (11), a voltage regulator (12), a battery (13), a fixing strap (14), and a touch switch (15). The control shell (8) has a rectangular shell structure. The PCB (9), the wireless Bluetooth module (10), the microcontroller (11), the voltage regulator (12), and the battery (13) are all placed inside the control shell (8). The touch switch (15) is fixed to the outside of the control shell (8). The fixing strap (14) is placed at the bottom of the control shell (8) for carrying. The output of the battery (13) is connected to the input of the voltage regulator (12), and the output of the voltage regulator (12) is connected to the PCB (9). The voltage is converted through the PCB to power each sensor and the microcontroller (11). The touch switch (15) is connected to the microcontroller (11). The microcontroller (11) communicates with the host computer through the wireless Bluetooth module (10). When the touch switch (15) is pressed, the microcontroller (11) sends the collected current time data to the host computer. The host computer uses the kiwi fruit storage day prediction algorithm to predict the number of days that can be stored and sends the number of days that can be stored to the microcontroller.
5. A kiwifruit storage days prediction system based on sensing gloves, characterized in that, Includes the sensing gloves, human-computer interaction interface, and operating system for implementing the prediction process based on the kiwi fruit storage days prediction model as described in any one of claims 1-4.
6. The kiwifruit storage days prediction system based on sensing gloves according to claim 5, characterized in that, The human-computer interaction interface includes glove information, storage days information, kiwi parameter information, and a reset button. The glove information display is located on the left side of the human-computer interaction interface, with a glove icon, thumb, middle finger, and index finger. From top to bottom, there are real-time display windows for hardness information and bending information. There is a real-time display window for weight information in the palm position, and below it, from left to right, there are real-time display windows for temperature and humidity information. The right side displays the kiwi fruit parameter information interface. From top to bottom, the display windows show the storage days, average volume, average hardness, weight, temperature, and humidity information.
7. The kiwifruit storage days prediction system based on sensing gloves according to claim 5, characterized in that, Data sets were constructed by collecting information on kiwifruit and training them to obtain a model for predicting the storage days of kiwifruit. The model was constructed based on 1D CNN, SVM, LSTM, and CNN-LSTM.
8. The kiwifruit storage days prediction system based on sensing gloves according to claim 7, characterized in that, The process of building the dataset is as follows: Select verified edible kiwifruit and determine the hardness range of edible kiwifruit by measuring with equipment; randomly select a number of kiwifruit that are too hard to eat and collect the weight, hardness, volume and environmental temperature and humidity information of the kiwifruit daily at room temperature using sensing gloves; check whether they have reached the edible range every day by using equipment and record the number of days for each sample collection. When a kiwi fruit sample reaches the edible hardness threshold, the current number of days is used as label 0, and the predicted number of days it can be stored is 0. Then, the dataset labels for this sample are summed in reverse order of the number of days of the previously sampled samples to obtain the predicted number of days it can be stored, and the number of labels is added.
9. The kiwifruit storage days prediction system based on sensing gloves according to claim 5, characterized in that, The predictive process controlled by the operating system includes: wearing a sensing glove, randomly selecting a kiwifruit, placing the kiwifruit in the palm of the hand, and pressing a touch switch to start the device; collecting temperature, humidity, and weight information, displaying it on an OLED screen, and sending it to a host computer via Bluetooth; the host computer enters a loop, waiting for hardness and curvature information to be collected, and displaying the temperature, humidity, and weight information on the human-machine interface; using the sensing glove to hold the kiwifruit until the top surface of the hardness device is tangent to the kiwifruit, pressing the touch switch, collecting hardness and curvature information, calculating the average value, displaying it on the OLED screen, and sending it to the host computer via Bluetooth; the host computer receives the collected hardness and curvature information and displays it on the human-machine interface. The host computer inputs the collected information into the kiwi fruit storage days prediction model and waits for the results to be output. After the calculation is completed, the host computer exits the loop, the kiwi fruit storage days are displayed on the human-computer interaction interface, and sent to the microcontroller via Bluetooth. The microcontroller displays the kiwi fruit storage days on the OLED screen. Other values are reset, and the system waits to detect the next kiwi fruit.