Portable near-infrared nondestructive testing device and method for kiwi fruit quality

Abstract

The invention discloses a portable near-infrared nondestructive testing device and a portable near-infrared nondestructive testing method for kiwi fruit quality. The device comprises an optical detection unit, a signal processing and control unit, a man-machine interaction unit and the like, wherein the optical detection unit adopts an MEMS Fourier transform spectrometer and a special probe with an annular light shield. The method comprises the following steps: establishing a kiwi fruit quality quantitative model based on partial least square regression and characteristic wavelength screening in a laboratory; the device is used for collecting spectrums on site, and a built-in model is called to predict indexes such as soluble solids, dry matters and acidity in real time. According to the invention, lossless, rapid, multi-index and on-site detection of the internal quality of kiwi fruits is realized, and the system has the advantages of high precision, good portability, strong anti-interference capability and the like, and is suitable for various scenes such as orchards, packaging lines, supermarkets and the like.

Description

Technical Field

[0001] This invention belongs to the field of non-destructive testing technology and intelligent sensing for agricultural products. Specifically, it relates to a portable device and method that combines near-infrared spectroscopy analysis, embedded systems, chemometrics and Internet of Things technologies. It is particularly suitable for rapid, non-destructive, on-site testing of the internal quality of kiwifruit (including soluble solids content (SSC), dry matter content (DM), acidity, etc.). Background Technology

[0002] The internal quality of kiwifruit is a key factor determining its commercial value. Currently, traditional quality testing methods mainly rely on destructive sampling, such as refractometer measurement of sugar content and oven drying method for dry matter determination. While these methods offer high accuracy, they have significant drawbacks:

[0003] 1. Damaging the fruit, causing economic losses;

[0004] 2. The testing process is cumbersome and inefficient, making it difficult to achieve large-scale sorting;

[0005] 3. The sampling test is not representative enough and cannot fully assess the quality of the batch.

[0006] In existing technologies, large-scale near-infrared spectrometers at the laboratory level offer high precision, but they are expensive, bulky, and complex to operate, making them unsuitable for use in orchards, packaging lines, and other field settings. Meanwhile, commercially available portable fruit refractometers often use finite wavelengths for estimation, resulting in low model accuracy, poor anti-interference capabilities, and limited functionality. Furthermore, they typically use general-purpose models that are insufficiently adapted to the specific characteristics of kiwifruit, leading to limited reliability of the test results.

[0007] Therefore, there is an urgent need to develop a near-infrared non-destructive testing device and method specifically for kiwifruit that combines high precision, multiple indicators, true portability, and ease of operation. Summary of the Invention

[0008] Purpose of the invention

[0009] The primary objective of this invention is to provide a portable near-infrared non-destructive testing device for kiwifruit quality. This device can enable real-time, non-destructive, and multi-index testing in scenarios such as orchards, packing plants, and supermarkets.

[0010] Another objective of this invention is to provide a detection method that works in conjunction with the device. This method covers the entire technical process from laboratory modeling to on-site prediction, ensuring that the detection results are accurate, reliable, and repeatable.

[0011] Technical solution

[0012] To achieve the above objectives, the present invention adopts the following technical solution:

[0013] I. Portable near-infrared non-destructive testing device

[0014] The device includes:

[0015] 1. Optical detection unit:

[0016] Light source: A small halogen tungsten lamp (5W) with a spectral range of 400-2500nm is used. It is equipped with constant current drive and temperature control circuit to ensure that the output light intensity stability error is less than 1%.

[0017] Spectrometer: Employs a MEMS miniature Fourier transform infrared spectrometer with a working wavelength range of 900-1700nm, featuring high light throughput, high wavelength repeatability, and vibration resistance.

[0018] Probe: A reflective contact probe with a hard anodized aluminum alloy housing. A receiving optical fiber is located at the center of the end face, surrounded by multiple light-emitting optical fibers arranged in a ring, forming a 45° illumination / 0° reception optical path layout. The end face window is made of 8mm diameter sapphire glass. A retractable annular light shield is located at the front of the probe, which presses firmly against the fruit surface during measurement to form a light-blocking cavity.

[0019] 2. Signal processing and control unit:

[0020] Embedded motherboard: Equipped with an ARM Cortex-A series quad-core processor, 4GB LPDDR4 RAM and 32GB eMMC storage for real-time spectral processing and model calculations.

[0021] Power management module: Built-in 5000mAh lithium polymer battery, supports fast charging, and provides more than 8 hours of battery life on a single charge.

[0022] 3. Human-computer interaction and display unit:

[0023] Touchscreen: 4.3-inch IPS capacitive screen with a resolution of 800×480 and a maximum brightness of 500 nits.

[0024] Wireless communication module: integrates dual-band Wi-Fi and Bluetooth 5.2, supporting data upload and model updates.

[0025] 4. External interface unit: including a USB-C interface for charging and data transfer.

[0026] II. Detection Methods

[0027] The method comprises two stages:

[0028] Phase 1: Model Building (Offline)

[0029] 1. Construct a representative kiwifruit sample set (≥500 samples) covering major varieties, production areas, and maturity levels.

[0030] 2. Use the device to collect the near-infrared diffuse reflectance spectrum of the sample.

[0031] 3. Use standard methods to determine the reference values ​​of SSC, DM, acidity, etc. of the sample.

[0032] 4. Preprocess the original spectrum, including multivariate scattering correction and Savitzky-Golay derivative processing.

[0033] 5. A competitive adaptive reweighted sampling method is used to screen characteristic wavelengths from the full spectrum.

[0034] 6. Using partial least squares regression algorithm, a quantitative correction model for spectral and quality parameters is established based on characteristic wavelengths.

[0035] 7. Validate the model using an independent validation set. Performance requirements: R²p ≥ 0.90, RMSEP ≤ 0.6°Brix.

[0036] 8. Solidify the verified model parameters into an encrypted file and deploy it to each detection device.

[0037] Phase Two: On-site Prediction (Online)

[0038] 1. The device is powered on and performs self-test, light source preheating, and background calibration.

[0039] 2. Place the probe close to the surface of the kiwi fruit to be tested and trigger spectral acquisition.

[0040] 3. Perform real-time preprocessing on the acquired spectra and extract characteristic wavelength data.

[0041] 4. Call the device's built-in PLS model to calculate the predicted values ​​of indicators such as SSC, DM, and acidity.

[0042] 5. Display the test results and quality level on the touch screen, and upload the data to the cloud.

[0043] Beneficial effects

[0044] 1. Non-destructive and rapid: The test does not damage the fruit, and the test time for a single sample is less than 3 seconds, enabling full inspection.

[0045] 2. High-precision multi-indicator: Based on full spectrum and dedicated chemometric model, it can simultaneously and accurately predict multiple indicators such as SSC, DM, and acidity.

[0046] 3. Portable and robust: The device is lightweight and has a long battery life. The dedicated probe and MEMS spectrometer design ensure measurement stability in complex field environments.

[0047] 4. Intelligent interconnection: Supports data cloud synchronization and remote model updates, enabling continuous optimization of the detection system. Attached Figure Description

[0048] Figure 1 This is a hardware system architecture diagram of the device of the present invention.

[0049] Figure 2 A flowchart for establishing the chemometric model of this invention.

[0050] Figure 3 This is a flowchart of the rapid on-site prediction process of the present invention. Detailed Implementation

[0051] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0052] Example:

[0053] Take the SSC test of "Hayward" kiwifruit as an example.

[0054] 1. Model Establishment:

[0055] 800 samples were collected, with SSC ranging from 9.5 to 19.0°Brix.

[0056] according to Figure 2 The process involved spectral preprocessing and CARS feature screening to obtain 32 key wavelengths.

[0057] A PLSR model (8 principal factors) was established, with validation set R²p=0.93 and RMSEP=0.48°Brix.

[0058] Deploy the model files to the device.

[0059] 2. On-site prediction:

[0060] On the packaging line, workers use a device to press a probe against the fruit surface and take measurements.

[0061] After about 2 seconds, the screen displays: "Sweetness: 16.1°Brix, Dry matter: 18.5%, Grade: Extra".

[0062] The results can be graded automatically or manually.

[0063] The apparatus and method of this invention are also applicable to the quality testing of other kiwifruit varieties and other fruits; only the corresponding model needs to be re-established.

Claims

1. A portable near-infrared non-destructive testing device for kiwifruit quality, characterized in that, include: The optical detection unit includes a light source, a MEMS miniature Fourier transform infrared spectrometer, and an integrated reflective contact probe, which is equipped with a ring-shaped light shield and a sapphire glass window. Signal processing and control unit, including an embedded motherboard and a power management module, is used for spectral data processing and model calculation; Human-computer interaction and display unit, including touch screen and wireless communication module; The light source is a halogen tungsten lamp with a spectral range of 400-2500nm; the spectrometer operates in the 900-1700nm band; and the light source fiber and the receiving fiber of the probe are arranged in a 45° illumination / 0° receiving configuration.

2. The apparatus according to claim 1, characterized in that, The embedded motherboard uses an ARM Cortex-A series processor and is equipped with no less than 4GB of RAM and 32GB of storage.

3. The apparatus according to claim 1, characterized in that, The probe housing is made of hard anodized aluminum alloy, the sapphire window has a diameter of 8mm, and the light shield is a retractable ring structure.

4. A non-destructive testing method for kiwifruit quality using the apparatus described in any one of claims 1-3, characterized in that, Includes the following steps: (1) Model building stage: a. Collect near-infrared spectra of representative kiwifruit samples; b. Determine the reference values ​​for the quality parameters of the sample; c. Perform spectral preprocessing and characteristic wavelength screening; d. Establish a quantitative calibration model between spectral data and quality parameters; e. Validate and deploy the model; (2) On-site prediction stage: a. Use the device to collect the spectrum of the kiwifruit to be tested; b. Preprocess the spectrum and extract characteristic wavelength data; c. Call the built-in quantitative correction model to calculate the predicted values ​​of quality parameters; d. Output the detection results.

5. The method according to claim 4, characterized in that, In step (1)c, the spectral preprocessing includes multivariate scattering correction and Savitzky-Golay derivative processing; The characteristic wavelength selection adopts a competitive adaptive reweighted sampling method.

6. The method according to claim 4, characterized in that, In step (1)d, the quantitative correction model is established using the partial least squares regression algorithm.

7. The method according to claim 4, characterized in that, The quality parameters include at least one of soluble solids content, dry matter content, and acidity.

8. The method according to claim 4, characterized in that, In step (2)d, the detection results are displayed on the touch screen and automatically uploaded to the cloud server.

Images