Flexible wearable optical wireless sensing system for fruit quality monitoring-EEWORLD

Collect

★ Research background

Conventional rigid optoelectronic systems need to be improved to adapt to current food quality monitoring. This paper aims to propose and develop a flexible wearable optical wireless sensing system (FWOS) using laser engraved polyethylene terephthalate/polyimide/copper (PET/PI/Cu) film as a substrate. The FWOS can be composed of different sensor arrays to adapt to different fruit sizes. The FWOS can acquire the spectral data of the fruit and transmit it to a smartphone or personal computer (PC) via Bluetooth (BT). Simulation is used to determine the most appropriate engraving power and speed, and a microscope is used to measure the surface properties of the flexible Cu circuit. The FWOS has the characteristics of simple process, simple operation, low cost, low power consumption, real-time monitoring, and small physical size. The soluble solids content (SSC) of apples, grapes, peaches, and cherries was evaluated using FWOS. The results show that the FWOS has potential application prospects in various fruit quality monitoring. It can reduce the fruit waste rate and ensure food quality and safety.

★ Innovation

Xiao Xinqing's research group at China Agricultural University proposed a near-infrared flexible wearable optical wireless sensing method for fruit quality monitoring. FWOS can be directly worn on most fruits and transmit data wirelessly. The accuracy of FWOS in monitoring fruit SSC was evaluated using MLR and support vector machine. In general, very interesting results were obtained on grapes and cherries. Compared with conventional sensor monitoring previously performed in our laboratory, FWOS successfully overcomes the problems of sensor fixation and the influence of manual grasping. The sensor array also solves the problem of only being able to measure one point. The application causes little damage to the fruit, achieving non-destructive testing. The new FOSM can be used as a connection to build the Internet of Things. When applied to fruit growth, cold chain transportation, storage, and sales, remote monitoring can be achieved without opening the cold storage or fruit plastic wrap. The total cost of the system is only US$40, and a large number of wearable sensors can be installed in various processes of the fruit to collect more data and obtain more accurate maturity predictions.

★ Article analysis

Figure 1 illustrates the design of the FWOS. The flexible optical sensor array consists of PU tape, PET film, PI film, Cu circuit, PDMS film, and electronic components (Figure 1a). The PET film and PI film serve as substrates for the flexible Cu circuit, the electronic components are soldered on the flexible Cu circuit, and the PDMS film is waterproof and prevents the flexible Cu circuit from oxidizing. The PU biotape is non-toxic, waterproof, and non-corrosive to the circuit. The structure of the optical measurement system is shown in Figure 1b. The structures of the flexible sensor array and the flexible single sensor are shown in Figure 1c-d. The AS7263 is the core device used to measure the light reflected by the fruit. The AS7263 features ultra-low power consumption, small size (4.5x4.7x2.5 mm), factory calibration, wide operating temperature range (-40 to +85°C), and integrated interference filters deposited directly onto complementary metal oxide semiconductor (CMOS) silicon. The AS7263 has 2 programmable LED drivers and 2 optional interfaces (UART and I²C). The 2700kLED is able to emit light suitable for the monitoring range of the AS7263. The Flash stores the program to control the AS7263. Bluno receives code instructions and processes sensor data, and then transmits this data to a smartphone or PC via BT4.0 for further processing. Since the same optical sensors have the same address, an I2C multiplexer must be used to receive data simultaneously. The optical sensing array has good bendability (Figure 1e-g). Devices with Bluetooth capabilities can wirelessly obtain changes in fruit quality in real time (Figure 1f).

Figure 1: Overall diagram of FWOS

Figure 2 shows the results of COMSOL simulation. After testing, a single ablation of 10μm will lead to high local temperature and no steady-state solution, so a single ablation of 5μm is used, and ablation is performed twice. The linear fit of the laser ablation depth at different powers and temperatures is higher than 0.97 (Figure 2a). Figure 2b shows the temperature change curve of the 20μm PI film over time. Figures 2c-e show the simulated morphology and temperature changes of the ablation power and speed just above 5μm.

Figure 2: COMSOL laser simulation results

Figure 3 illustrates the flexible circuit preparation process. The UV laser marking machine includes a PC, a UV nanosecond laser, an industrial chiller, and a galvanometer (Figure 3a). The industrial chiller directly acts on the UV nanosecond laser to maintain a low temperature throughout the process. The laser wavelength is 355nm, and the pulse width of the laser beam is about 15nm. The film that has not been ablated after cleaning and drying is fixed on the Al substrate. The designed circuit is imported into the laser ablation software of the PC for operation. When the laser acts on the Cu film, Cu is quickly peeled off and sublimated. A flexible Cu circuit is obtained by ablating the PET/PI/Cu film twice (Figure 3b). The flexible circuit can be obtained after cleaning and drying (Figures c and d). After two ablations, the Cu surface loses its conductivity and has good flexibility (Figures 3e-h).

Figure 3: Flex circuit preparation

Figure 4 shows the principle of measuring fruit quality with a flexible sensor array. The flexible sensor array is wrapped around the fruit, and the light emitted by the LED is irradiated on the surface of the fruit and enters the small holes of the AS7263 through diffuse reflection (Figure 4a, d, e). NIR is between visible light (Vis) (380-780nm) and mid-infrared (MIR) (2526nm-25μm) and far infrared (FIR) (25-1000μm). It has been proven that almost all organic structures and functional groups can be measured in the NIR spectrum with fairly stable spectra. The principle of NIR spectroscopy measurement is to use the anharmonicity of atoms or molecules to change their absorption (Figure 4b and c). When molecules or atoms vibrate, atoms or molecules are stretched, sheared, oscillated and bent, and the state of energy transition from the ground state to the excited state is called the anharmonicity of molecules or atoms. Combination absorption and overtone absorption are two types of unique absorption in the spectrum. The characteristic absorption produced by various vibration forms of chemical bonds at wavenumbers and fundamental frequencies is called combination absorption. The unique absorption produced by multiples of the fundamental frequency of chemical bonds is called overtone. The NIR spectral region expresses the absorption information of key chemical bonds such as CO, C=O, OH and H2O, which can reflect the internal quality of the fruit. AS7263 has 6 NIR sensing channels (610nm, 680nm, 730nm, 760nm, 810nm and 860nm, each channel has a half-peak full width of 20nm, as shown in Figure f). The average value is taken as the final spectral value of the berry. 630 and 690nm correspond to the chlorophyll peak, 730nm corresponds to the third overtone of OH stretching, and 810 and 860nm correspond to the combined bands of the OH groups of sugars.

Figure 4: Principle of near infrared monitoring of fruit quality

In the SSC parameter, the SD is about 20% of the difference between the maximum and minimum values of the parameter, indicating that the calibration is meaningful. The data for each fruit covers a wide and different range, which shows the variability between different fruits. The measurement process and results of different fruits are shown in Figure 5. Figure 5a and b are measured by a single sensor (grapes and cherries). Figure 5c and d are measured by a sensor array (apples and peaches). The sensor array measurement method saves time, but requires sufficient space on the fruit to place these sensors, so both apples and peaches are measured using this method. When the flexible optical sensor array is attached to the surface of the fruit, the impact on the surface of the fruit is small, and the data will be stable after about two seconds. It is worth noting that appropriate preprocessing methods can improve the prediction accuracy of the model to a certain extent, but sometimes improper application will reduce the accuracy. Grapes have the best prediction and peaches have the worst results, which may be related to their absorbance (Figure 5e). Figure 5f-i shows the effect of the optimal SSC prediction model on the four fruits. The RPD values of the peach, apple and cherry prediction models are about 1.5, indicating that FWOS is able to distinguish between high and low values of their SSC. The RPD value of the grape prediction model is around 2, indicating that the single-sensor optical measurement system can roughly distinguish the SSC of grapes. In the classification model, the success rate of grapes is 73.1% and the success rate of cherries is 80%.