Method for detecting impurities in marinated food

Abstract

This invention belongs to the technical field of food quality testing and relates to a method for detecting impurities in braised foods. The method includes the following steps: First, the braised food is pretreated to enhance its acoustic surface characteristics. Then, dual-frequency acoustic waves are used for synergistic excitation and scanning to form a mixed-spectrum sound field inside and on the surface of the food. The original signal sequence is obtained by collecting the surface vibration response, and the difference-frequency characteristic signal generated solely by impurities is demodulated using lock-in amplification technology, thereby determining the presence or absence of impurities. Finally, the impurity determination result is correlated with the scanning coordinates to construct a three-dimensional impurity distribution map and integrate it with a three-dimensional contour model of the food to generate an intuitive and visual inspection report. This invention solves the problems of low sensitivity and lack of visual localization in the detection of non-metallic impurities in existing technologies.

Description

Technical Field

[0001] This invention belongs to the technical field of food quality testing and relates to a method for detecting impurities in braised food. Background Technology

[0002] The core problem currently facing the production of braised food is the difficulty in controlling the contamination of impurities. During various stages such as raw material handling, processing, packaging, and transportation, impurities in various forms, including hair, plastic fragments, metal shavings, and stones, may be introduced from sources such as tools, packaging materials, and environmental pollution. Once these impurities enter the consumer market, they will threaten consumer food safety and trigger food safety incidents.

[0003] Currently, the commonly used testing methods in the food industry mainly include manual visual inspection, metal detector detection, and X-ray scanning. Manual inspection depends on the inspector's work condition and professional level, which limits the detection efficiency and is prone to missed detection due to visual fatigue; metal detectors can only effectively detect conductive metal impurities and are basically ineffective for insulating or low-conductivity impurities such as plastics and hair; although X-ray scanning can detect various impurities, the equipment cost is high and scanning a single food item takes a long time.

[0004] Existing detection technologies suffer from several drawbacks. Their detection principles fail to fully leverage the inherent physical differences between impurities and the food matrix, such as nonlinear acoustic responses. This makes it difficult to achieve universal detection of impurities. Furthermore, their detection sensitivity and specificity are insufficient, resulting in high rates of missed detections and false alarms. They also lack the ability to capture and visually represent the location of impurities, making it difficult to provide effective location guidance for subsequent manual re-inspection or automated removal. This restricts the overall efficiency of detection and handling, hindering the achievement of efficient, universal, accurate, and intuitive detection of impurities in braised foods. Summary of the Invention

[0005] To address the aforementioned problems, this invention provides a method for detecting impurities in braised foods.

[0006] A method for detecting impurities in braised food includes the following steps:

[0007] S1. Perform surface acoustic property optimization pretreatment on braised food with liquid brine on the surface to generate target food without free liquid and with an acoustically enhanced surface.

[0008] S2. Perform dual-frequency acoustic wave coordinated excitation and scanning on the target food to form a mixed spectrum sound field inside and on the surface of the target food;

[0009] S3. Use non-contact vibration measurement equipment to collect the surface vibration response of the target food under the action of a mixed spectrum sound field, and generate the original vibration signal sequence;

[0010] S4. Perform lock-in amplification on the original vibration signal sequence and demodulate the difference frequency characteristic signal with background suppression caused by impurities.

[0011] S5. Determine whether there are impurities based on the effectiveness of the difference frequency characteristic signal of background suppression, and whether there is a judgment result for the generated impurities;

[0012] S6. Associate the results of the impurity presence / absence judgment with the scanning coordinates to construct a three-dimensional impurity coordinate point set;

[0013] S7. Integrate the three-dimensional impurity coordinate point set with the three-dimensional contour model of the target food to generate a visual impurity distribution report.

[0014] A further aspect of the present invention involves generating a target food product free of free liquid and coated with an acoustically enhanced surface, comprising the following steps:

[0015] A vortex air curtain with a relative humidity lower than the ambient humidity is generated using a vortex air knife device.

[0016] The free liquid on the surface of braised food is removed by vortex air curtain to form a clean interface;

[0017] By utilizing the surface tension changes caused by vortex air curtain, instantaneous stress is applied to impurities attached to the surface of braised food, generating a target food with no free liquid and an acoustically enhanced surface.

[0018] A further aspect of the present invention, step S2, includes the following steps:

[0019] The frequency of the emission to the target food is The pumping sound waves establish a uniform low-frequency vibration field;

[0020] With frequency The high-frequency detection wave is focused into a millimeter-sized spot to perform three-dimensional scanning on the target food.

[0021] By utilizing the nonlinear characteristics of hard impurities, the pump acoustic wave and the high-frequency probe wave undergo heterodyne mixing, generating a frequency of [frequency value missing] in the mixed-spectrum acoustic field. Difference frequency signal.

[0022] A further aspect of the present invention generates the original vibration signal sequence, comprising the following steps:

[0023] A laser Doppler vibrometer is used to scan the measurement point synchronously with the focal point of the high-frequency probe wave;

[0024] The laser measurement point is used to measure and record the vibration displacement or velocity of the target food surface in the vertical direction in real time, and the continuously collected data is combined in time sequence to generate the original vibration signal sequence.

[0025] A further aspect of the present invention generates a background-suppressed difference-frequency characteristic signal, comprising the following steps:

[0026] Calculate the pump acoustic frequency With high frequency detection wave frequency absolute value of the difference And generate a signal of that frequency as a reference frequency input to the lock-in amplifier;

[0027] The original vibration signal sequence is input as the signal to be measured into the lock-in amplifier;

[0028] Through internal mixing and low-pass filtering in the lock-in amplifier, noise unrelated to the reference frequency and noise at frequencies of [missing information] are filtered out. and The fundamental frequency signal is used to generate a difference frequency characteristic signal with background suppression.

[0029] A further aspect of the present invention involves generating a determination result regarding the presence or absence of impurities, comprising the following steps:

[0030] A preset effective signal amplitude threshold is used to distinguish between real signals and electronic noise.

[0031] The amplitude of the difference frequency characteristic signal of background suppression is monitored and compared with the effective signal amplitude threshold in real time. If the threshold is exceeded, the presence of impurities is determined.

[0032] Based on the judgment result, a logical marker sequence of presence or absence synchronized with the scan time is generated to form the judgment result of the presence or absence of impurities.

[0033] A further aspect of the present invention involves constructing a three-dimensional impurity coordinate point set, comprising the following steps:

[0034] Establish a three-dimensional coordinate system corresponding to the target food;

[0035] Record the coordinates of the high-frequency probe wave focal point at the moment when impurities are detected;

[0036] Collect all recorded coordinate points to construct a three-dimensional impurity coordinate point set.

[0037] A further aspect of the present invention generates a visualized impurity distribution report, comprising the following steps:

[0038] Call the 3D contour model of the target food;

[0039] The three-dimensional impurity coordinate point set is rendered onto the three-dimensional contour model as a highlighted mark.

[0040] Output the rendered 3D model image and generate a visual impurity distribution report.

[0041] A further aspect of the present invention, after forming a clean interface, includes the following steps:

[0042] Utilizing optical sensors to measure the surface finish of clean interfaces;

[0043] The surface finish is compared with a preset finish standard to generate an acoustic coupling condition evaluation result.

[0044] A further embodiment of the present invention, after generating the difference frequency signal, further includes:

[0045] Analyze the spectral width of the difference frequency signal components and match it with a preset database of impurity type spectral widths to generate preliminary classification information of impurity materials.

[0046] In summary, the present invention has the following beneficial technical effects:

[0047] 1. By establishing a uniform low-frequency pump acoustic background field and a localized focused scan of high-frequency probe acoustic waves, when the focal point reaches the impurity location, the strong nonlinear characteristics of the impurity will generate a difference frequency signal of a specific frequency under the synergistic effect of the two acoustic waves. Soft tissue, due to its extremely strong linear characteristics, will not generate this signal, thus achieving the distinction between impurities and background. Compared with the single-frequency excitation method, the detection rate of non-metallic impurities is improved through a nonlinear acoustic mechanism.

[0048] 2. Lock-in amplification technology is employed to accurately demodulate the difference frequency characteristic signal generated solely by the nonlinear effect of impurities from the complex original vibration signal. This reduces background interference from the two strong fundamental frequency components of the pump wave and probe wave, effectively suppressing the influence of environmental noise and electronic noise. This background suppression detection method improves detection sensitivity and reliability, enabling the capture of impurities that are small in size or have weak characteristics, thus reducing the false negative rate.

[0049] 3. By associating the logical judgment results of the presence or absence of impurities with the three-dimensional scanning coordinates of high-frequency detection waves, the three-dimensional coordinate position of impurities inside or on the surface of the target food is constructed. This provides clear positioning guidance for subsequent manual re-inspection operations or provides accurate coordinate input for automated rejection, thereby improving the work efficiency of the entire detection and processing process. Attached Figure Description

[0050] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. The drawings are used to provide a further understanding of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0051] Figure 1 A flowchart illustrating an embodiment of this application is disclosed.

[0052] Figure 2Structural schematic diagrams of embodiments of this application are disclosed. Detailed Implementation

[0053] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. All other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0054] The following is in conjunction with the appendix Figure 1 - Figure 2 A preferred description of the present invention is provided below.

[0055] See attached document Figure 1 This invention proposes a method for detecting impurities in braised foods, comprising the following steps:

[0056] S1. Perform surface acoustic property optimization pretreatment on braised food with liquid brine on the surface to generate target food without free liquid and with an acoustically enhanced surface.

[0057] S2. Perform dual-frequency acoustic wave coordinated excitation and scanning on the target food to form a mixed spectrum sound field inside and on the surface of the target food;

[0058] S3. Use non-contact vibration measurement equipment to collect the surface vibration response of the target food under the action of a mixed spectrum sound field, and generate the original vibration signal sequence;

[0059] S4. Perform lock-in amplification on the original vibration signal sequence and demodulate the difference frequency characteristic signal with background suppression caused by impurities.

[0060] S5. Determine whether there are impurities based on the effectiveness of the difference frequency characteristic signal of background suppression, and whether there is a judgment result for the generated impurities;

[0061] S6. Associate the results of the impurity presence / absence judgment with the scanning coordinates to construct a three-dimensional impurity coordinate point set;

[0062] S7. Integrate the three-dimensional impurity coordinate point set with the three-dimensional contour model of the target food to generate a visual impurity distribution report.

[0063] In one embodiment of the present invention, step S1 includes the following steps:

[0064] A vortex air curtain with a relative humidity lower than the ambient humidity is generated using a vortex air knife device.

[0065] The free liquid on the surface of braised food is removed by vortex air curtain to form a clean interface;

[0066] By utilizing the surface tension changes caused by vortex air curtain, instantaneous stress is applied to impurities attached to the surface of braised food, generating a target food with no free liquid and an acoustically enhanced surface.

[0067] Specifically, to obtain the braised food to be tested, the operator places the braised food, such as a whole braised goose with liquid brine adhering to its surface, at the feed end of a conveyor belt. The conveyor belt transports the braised food at a uniform speed to a specific pre-processing station. This station is equipped with a vortex air knife device, an industrial drying device whose core components include a high-pressure blower and a narrow, tapering nozzle designed based on the Venturi principle. When the high-pressure airflow passes through the nozzle, the flow velocity increases sharply, forming a high-speed, low-pressure laminar or micro-vortex air curtain at the outlet to achieve non-contact surface treatment. The rotating vortex characteristics of the air curtain are achieved by an array of guide vanes at the nozzle outlet. These vanes are arranged at a specific angle to convert the linear airflow into a rotating flow.

[0068] First, activate the vortex air knife device. The high-speed centrifugal fan inside generates a powerful airflow, which, after passing through a specially designed narrow nozzle, forms a high-speed rotating vortex air curtain to enhance the peeling efficiency of the surface liquid film. The operator needs to preset and control the relative humidity of this vortex air curtain, ensuring it is consistently lower than the ambient humidity of the testing environment. For example, the air curtain humidity should be controlled at 30%, while the ambient humidity is 50%, meaning the relative humidity difference should be maintained above 20%. The airflow rate can be set to 15m / s-25m / s, and the temperature to 20℃-25℃.

[0069] It should be noted that the relative humidity being lower than the ambient humidity is a key parameter setting for this air curtain. The setting is based on a large number of experiments, such as comparative tests under 200 different humidity conditions. It was found that when the air curtain humidity is more than 10% lower than the ambient humidity, it can most effectively remove moisture while preventing excessive water loss from the food surface. The acoustically enhanced surface is usually free of visible liquid film and has a surface roughness Ra value between 0.2μm and 10μm. This range also allows sound waves to penetrate effectively without damaging the food surface.

[0070] Next, the conveyor belt precisely delivers the braised food into the effective range of this vortex air curtain. When the surface of the braised food comes into contact with the vortex air curtain, the high-speed rotating airflow generates a shear force on the food surface. This shear force, which is relatively parallel to the object's surface, is generated when the vortex air curtain acts on the liquid layer, and is the mechanical factor that causes the liquid layer to deform and shift. When the shear force acts on the liquid layer on the surface of the braised food, it can physically roll away and blow away the free water and oil film attached to the food surface within a short time window, such as within 0.5 seconds. This time threshold is set based on high-speed camera observation and surface humidity sensor data to determine that the liquid film can be effectively removed within this time range without affecting the meat quality. This process removes the main obstacle to sound wave coupling, forming a relatively dry and clean interface on the food surface without liquid coverage. This is a description of the state of the food surface after treatment. After removing the free liquid, the surface can directly and effectively couple with sound waves.

[0071] Meanwhile, due to the drying effect caused by the vortex air curtain, the microenvironment on the surface of the braised food changes, causing changes in local surface tension. This is because the rapid removal of the liquid layer leads to a rapid change in the energy state of the solid-gas interface. This change in physical state applies instantaneous physical stress to flexible impurities such as hair or plastic film that are attached to the food surface. This stress is a brief mechanical stimulus induced by the change in surface tension and acts on the impurities. This stress causes the internal structure of these originally soft impurities to tighten rapidly, which is usually manifested macroscopically as a brief hardening and embrittlement phenomenon. Their mechanical properties undergo measurable changes. The key to this state transition lies in the improvement of acoustic coupling conditions: the impurities originally encased in liquid form a liquid-solid-liquid three-layer medium with the food surface, and sound waves undergo multiple reflections and attenuation at this interface; after treatment, it becomes an air-solid-solid interface, and the impedance mismatch between the impurities and the food matrix increases, thereby enhancing the scattering of high-frequency probe waves and the nonlinear mixing response. Here, hardening does not refer to a change in the material of the impurities, but rather to the change in the mechanical contact between them and the matrix from flexible coupling to rigid coupling.

[0072] After the above continuous physical processes, the original braised food is transformed into a target food with optimized surface properties. Its surface is defined as a target food with an acoustically enhanced surface. It is worth mentioning that this method is not only applicable to block foods such as braised goose and braised meat, but also to other solid braised foods with soft tissue properties and possible liquid adhesion, such as braised dried tofu.

[0073] In one embodiment of the present invention, in order to further quantify the preprocessing effect and ensure the optimal coupling conditions for subsequent acoustic detection, the system may also perform an acoustic coupling condition evaluation.

[0074] Specifically, this includes an optical sensor module, such as a laser scattering sensor based on diffuse reflection or a high-resolution linear CCD sensor, located downstream of the vortex air knife device and before the target food enters the acoustic detection chamber. When the clean interface of the target food passes through this module, the optical sensor rapidly scans its surface and measures its surface finish parameters. These parameters are typically characterized by a surface roughness Ra value or a characteristic value of the scattered light intensity distribution.

[0075] The measured surface finish data is transmitted to the processing unit in real time. The processing unit contains a pre-stored database of the correlation between surface finish and acoustic coupling efficiency, established based on numerous experiments, and sets a standard range for surface finish corresponding to optimal coupling conditions. For example, the range for the scattered light intensity ratio is set to 0.1-0.3, or the acceptable range for the surface roughness Ra value is set to 0.5μm < Ra < 10μm. The system compares and calculates the real-time measured surface finish values ​​with the preset standard ranges.

[0076] The comparison results generate an acoustic coupling condition evaluation result. This result is typically a qualitative or semi-quantitative output, such as: Excellent - surface finish is within the standard range, coupling conditions are ideal; Good - slightly deviates from the standard, but can be tested; Poor - seriously deviates from the standard, recommending reprocessing or alarm. This evaluation result can be displayed on the human-machine interface to prompt operators, can also be archived as quality control data, and provide background reference for subsequent acoustic testing signal processing. For example, when the evaluation result is Good, the system can automatically fine-tune the signal detection threshold to compensate for possible coupling loss.

[0077] Afterward, the target food is conveyed out of the pre-processing and evaluation station by a conveyor belt, preparing it for subsequent steps.

[0078] It should be noted that braised food is the example to be tested, such as a whole braised goose or a large piece of braised meat, the surface of which is usually covered with liquid braising liquid composed of soy sauce, spices and oil.

[0079] Acoustic-enhanced surfaces are defined as the surfaces of target foods after pretreatment. They are characterized by being clean and having their attached impurities temporarily modified, making them more likely to generate nonlinear responses in a sound field.

[0080] For example, when detecting fine hairs with a diameter of 50-100 μm, although the hair itself has a very small mass, the interfacial nonlinear effect it causes will affect the difference frequency. A characteristic signal is generated at this location. This signal is then amplified by a lock-in amplifier at the difference frequency. For reference, coherent detection can be performed, achieving 10 6The improved signal-to-noise ratio enables the identification of difference frequency signals with amplitudes as low as nanovolts. During the detection process, interference elimination mechanisms are typically implemented, such as controlling the ambient temperature at 20±0.5℃ to avoid stress changes caused by thermal expansion of the food; using air-floating vibration isolation on the detection platform to eliminate external vibration interference; scanning with a standard sample before each test to establish a background signal baseline, and then differentially processing the real-time signal against this baseline; repeatedly exciting and acquiring signals at each scanning point, suppressing random noise through time-domain averaging; and using a lock-in amplifier to simultaneously monitor the amplitude and phase of the signal. The signal phase generated by the slow deformation of the food itself is random, while the nonlinear signal generated by impurities has a definite phase relationship, i.e., it is phase-locked with the pump wave.

[0081] It should be noted that the vortex air knife treatment only acts on the free brine on the outermost layer of the food, which is usually less than 0.5 mm thick. The treatment time is usually 0.5-1 second, and the humidity of the air curtain is controlled at a relative humidity of 30-40%, which can reduce the loss of internal moisture in the food and does not affect the main flavor of the food. For specific products with strict requirements on the surface brine, an atomized brine replenishment unit can be added at the end of the testing line. Micron-level atomizing nozzles are used to evenly spray the seasoned brine onto the surface of the tested product, which can restore the appearance and surface flavor without affecting the completed impurity detection results.

[0082] For example, taking a whole braised goose weighing approximately 2 kg as an example, assuming its initial surface is coated with about 5 mL of dark brown braising liquid, with a noticeable oily sheen, it is treated by a vortex air knife device with a relative humidity of 35%. The goose passes through the air curtain zone at a speed of 0.1 m / s for 0.5 seconds. After treatment, the goose's surface appears matte and feels dry to the touch, indicating that free moisture and oil film have been removed, forming a clean interface. If a human hair about 2 cm long is attached to the wing folds, it will be soft and adhered before treatment; after treatment, the hair will typically become straight and slightly raised due to instantaneous physical stress, thus directly exposing it to the sound field. This transition from submersion to exposure, and the change in shape, enhances its interaction with the incident sound waves. For example, the reflection and scattering efficiency makes its nonlinear acoustic response easier to detect. In this case, the braised goose becomes a target food with an acoustically enhanced surface.

[0083] In one embodiment of the present invention, step S2 includes the following steps:

[0084] The frequency of the emission to the target food is The pumping sound waves establish a uniform low-frequency vibration field;

[0085] With frequency The high-frequency detection wave is focused into a millimeter-sized spot to perform three-dimensional scanning on the target food.

[0086] By utilizing the nonlinear characteristics of hard impurities, the pump acoustic wave and the high-frequency probe wave undergo heterodyne mixing, generating a frequency of [frequency value missing] in the mixed-spectrum acoustic field. The difference frequency signal.

[0087] Specifically, the operator or automated conveyor system smoothly transports and positions the target food item with the acoustically enhanced surface to a designated location within the acoustic detection chamber. A low-frequency pump wave transducer assembly, installed on one side of the detection chamber or surrounding the target food item, is then activated. This assembly consists of transducers capable of emitting low-frequency ultrasonic waves, providing a uniform, low-frequency acoustic excitation background field for the entire detection area. The transducer assembly is configured so that, upon power-up, the emission frequency directed towards the target food is pre-set to a specific frequency. The pumping sound waves. This beam of low-frequency sound waves has a wide coverage area and can penetrate most of the volume of the target food, so that its overall structure is in a uniform low-frequency vibration field. It is the mechanical environment established by the pumping sound waves inside the target food, so that all points of the food are subjected to periodic stress of the same frequency, which is macroscopically manifested as the food producing subtle periodic mechanical vibrations.

[0088] Simultaneously, another independently controlled high-frequency sounding transducer array is activated. This array emits and focuses high-frequency ultrasonic waves to generate a movable high-frequency acoustic focus point for fine scanning. This array consists of multiple small transducer units and can preset the emission frequency. The high-frequency probe wave is focused into a concentrated acoustic spot with a diameter on the order of millimeters by using acoustic lenses or phased array focusing technology. The millimeter-sized spot is the concentrated acoustic energy area formed after the high-frequency probe wave is focused. Its diameter is usually about 1 to 3 millimeters. This size is set based on the minimum size of common impurities to ensure detection resolution.

[0089] According to the preset scanning path program, the millimeter-sized light spot is driven to perform precise three-dimensional spatial scanning of the interior and surface of the target food. The scanning path is usually a row-by-row or layer-by-layer rasterized path to ensure coverage of the entire volume of the target food to be inspected. During the scanning process, the millimeter-sized light spot moves at a constant speed. When there are no hard impurities such as plastic, stones, metal fragments, or hair that has changed the coupling state at the scanning location of the light spot, and only the normal tissue structure of the target food itself exists, such as bones and fascia, since soft tissue is an acoustically linear medium, only frequencies of 1000-2000 Hz exist in the sound field. The pump acoustic wave and frequency are The two fundamental frequency components of the high-frequency probe wave;

[0090] In one embodiment of the present invention, normal tissue structures and abnormal impurities can be distinguished through the following mechanism:

[0091] The system incorporates a standard 3D anatomical model of similar braised foods, containing the location, shape, and size data of normal hard structures such as bones, joints, and cartilage. Real-time scan coordinates are spatially matched with this database, and signals located in the expected skeletal region are marked as normal structures.

[0092] Normal bone and muscle tissue form a gradual interface, and its nonlinear response exhibits specific spatial gradient characteristics. In contrast, abnormal impurities, such as plastics and metals, form abrupt impedance interfaces with soft tissue, producing sharp spatial variations in the nonlinear response peak. The system identifies these impurities by analyzing the spatial derivative of the response signal.

[0093] Normal bone exhibits a continuous spectrum characteristic in its nonlinear response at different frequencies, while foreign bodies, especially non-metallic foreign bodies, show a different response in specific frequency bands, such as... Discrete resonance peaks will appear nearby.

[0094] The system can train a classification model using historical detection data and automatically classify impurities and normal structures based on multi-dimensional features such as signal amplitude, phase, spectral width, and spatial distribution.

[0095] When the focal point of a millimeter-scale light spot scans a hard impurity, the situation changes. The presence of this hard impurity induces local acoustic nonlinear effects, and it occurs at a frequency of [frequency missing]. In the periodic stress field established by the pumping sound waves, the internal elastic parameters change nonlinearly with the period of the pumping sound waves. The hard impurities here include, but are not limited to, metal shavings, glass fragments, sand particles, hard plastic sheets, and dried, hardened bone fragments. These materials have nonlinear acoustic constants that differ from those of soft food tissue. At this point, the frequency is... When a high-frequency probe wave acts on an impurity whose elastic parameters change periodically, a heterodyne mixing effect occurs. This is a specific manifestation of acoustic nonlinearity, referring to the process in which two sound waves of different frequencies interact in a nonlinear medium to generate new signals of sum and difference frequencies.

[0096] The result of this physical process is that, originally only and In the sound field, within the tiny local region where the impurities are located, new molecules with a frequency of are generated. minus The difference frequency signal component. Therefore, throughout the scanning process, the sound field in the space where the target food is located is a dynamically changing mixed-spectrum sound field, whose spectral components always include the background. and These two strong fundamental frequency components, and the weak frequency that appears instantaneously only when the scan point is located on a hard impurity, are... The newly generated difference frequency signal component is one of the new signals generated by heterodyne mixing. Its characteristic is that the frequency is strictly equal to the difference between the two input frequencies and is generated only by a nonlinear medium.

[0097] Formula for heterodyne mixing frequency:

[0098] ;

[0099] in, The frequency of the low-frequency pump wave is represented by Hz. This represents the frequency of the high-frequency probe wave, measured in Hz. The frequency of the newborn difference frequency signal is represented by Hz.

[0100] This formula is a scalar subtraction, frequency The frequency is typically set between 20 kHz and 50 kHz. This setting is based on the fact that sound waves in this frequency band can effectively penetrate the soft tissue of braised foods and excite impurities to produce a sufficient nonlinear response, while avoiding mechanical damage to the food. The frequency band is typically set between 0.5 MHz and 2 MHz, based on the fact that this band can achieve millimeter-level acoustic focusing to obtain sufficient spatial resolution, and is compatible with... The difference can generate a difference frequency signal that is easy for subsequent electronic systems to detect and separate. The value is from and The set value is calculated directly, for example, when , hour, .

[0101] Among them, the dual-frequency acoustic wave synergistic excitation scan is to simultaneously apply two different frequencies of acoustic waves and make them work synergistically in space; the mixed spectrum acoustic field is an acoustic state description of the space where the target food is located, which contains acoustic waves with multiple frequency components coexisting.

[0102] For example, a whole braised goose is placed as the target food for acoustic detection. The system activates a low-frequency pump wave transducer array, and the transmission frequency is... The pump sound waves cover the entire goose. Simultaneously, the high-frequency probe wave transducer array is activated, with a transmission frequency of... The high-frequency probe wave is focused into a millimeter-sized spot with a diameter of approximately 2 mm. The spot scans along a pre-defined three-dimensional path, spiraling downwards from the goose's head to its tail. When the spot reaches a pre-defined location inside the goose's chest cavity, if a polypropylene plastic sheet of approximately 3 mm in size is present, this plastic sheet, acting as a hard impurity, is affected by the local acoustic nonlinear effect it induces, and is then focused by a frequency of [frequency value missing]. The pump acoustic wave is modulated and has a frequency of The high-frequency probe wave undergoes heterodyne mixing, thereby instantaneously generating a frequency of at that location. The newly generated difference frequency signal component. The generation of this signal component directly verifies the existence of hard impurities and their acoustic nonlinear characteristics. This newly generated signal is different from the original... and The fundamental frequency signal together constitutes the mixed spectral sound field at this point in the chest cavity of the target food, the goose.

[0103] In one embodiment of the present invention, step S3 includes the following steps:

[0104] A laser Doppler vibrometer is used to scan the measurement point synchronously with the focal point of the high-frequency probe wave;

[0105] The laser measurement point is used to measure and record the vibration displacement or velocity of the target food surface in the vertical direction in real time, and the continuously collected data is combined in time sequence to generate the original vibration signal sequence.

[0106] Specifically, during the continuous operation of the established mixed-spectrum sound field, a non-contact laser Doppler vibrometer is simultaneously activated. This non-contact laser Doppler vibrometer is an optical measurement device that utilizes the laser Doppler effect to accurately measure the vibration velocity or displacement of an object's surface without contact. The operator or control system first performs initial calibration to ensure that the trajectory of the laser measurement point emitted by the vibrometer in three-dimensional space is strictly synchronized with the trajectory of the focal point of the millimeter-scale light spot generated by the high-frequency probe wave transducer array. This synchronization is achieved by connecting both to the same motion control unit and ensuring they follow the same pre-programmed three-dimensional scanning path and time sequence. In other words, synchronized scanning is the operation of keeping the spatial motion of the laser measurement point and the spatial motion of the high-frequency probe wave focal point consistent in path and time. This is set to ensure that the measured surface vibration response is exactly the response directly above the current acoustic excitation point, achieving spatiotemporal alignment of excitation and detection.

[0107] After calibration, the system begins scanning. At each moment during the scan, the millimeter-scale spot of the high-frequency probe wave forms an acoustic excitation point inside or on the surface of the target food, while the laser measurement point of the laser Doppler vibrometer precisely illuminates the corresponding position directly above the target food surface. The core function of the laser Doppler vibrometer is to measure and record in real time the instantaneous vibration state of the target food surface along a direction perpendicular to the surface at the location illuminated by its laser measurement point. This state can be characterized as vibration displacement or vibration velocity. It should be noted that vibration displacement or velocity are physical quantities directly measured by the laser Doppler vibrometer. Displacement refers to the distance the surface moves relative to its equilibrium position, and velocity refers to the instantaneous rate of surface movement. The system is usually set to a speed measurement mode due to its higher sensitivity. Real-time measurement means that the measurement process and the scanning excitation process are synchronized, with extremely low data acquisition delay; the measurement moment can be considered the excitation moment.

[0108] Due to the overall frequency of the target food The pump is excited by acoustic waves, and local points are subjected to frequency The high-frequency probe wave is focused and excited, and at hard impurities, it may also excite a frequency of The resulting acoustic excitations, acting together on the food, cause complex vibrations on its surface containing multiple frequency components. The photodetector of the laser Doppler vibrometer continuously captures the Doppler frequency shift of the reflected laser light caused by the surface vibrations and converts it into an electrical signal proportional to the vibration velocity or displacement.

[0109] The data acquisition system performs analog-to-digital conversion on this continuously changing electrical signal at a constant and sufficiently high sampling rate, generating a series of discrete data points arranged chronologically. Each data point represents the instantaneous vibration value of the target food surface at a specific moment at the location of the laser measurement point. As the scan continues, the system combines all these continuously acquired data points strictly according to their chronological order of generation into a complete one-dimensional time-series data set. Arranging the discrete data points into a one-dimensional array or data stream according to their sampling timestamps is a standard method for constructing time-series signals. This complete data set, recording the vibration response of all points traversed on the target food surface throughout the entire scanning cycle, is defined as the original vibration signal sequence containing multiple frequency components.

[0110] Doppler effect formula:

[0111] ;

[0112] in The instantaneous vibration velocity of the target surface along the laser beam direction, after angle correction; The Doppler frequency shift between the reflected laser and the incident laser is caused by the surface motion velocity. The wavelength of the laser. Cosθ is the angle between the laser incident direction and the normal direction of the target surface. When the laser is incident perpendicularly, cosθ = 1.

[0113] Among them, surface vibration response is the physical motion generated on the surface of the target food under the excitation of a mixed-spectrum acoustic field, which includes the physical motion generated by the surface of the target food under the excitation of a mixed-spectrum acoustic field. , and possible Periodic displacement or velocity changes caused by equal frequency components; the original vibration signal sequence is a discrete digital signal sequence arranged in time order, whose data values ​​correspond to the vibration amount, and whose sequence length is equal to the sampling rate multiplied by the total scanning time.

[0114] The laser measurement point is the spot formed by the laser beam emitted by the laser Doppler vibrometer illuminating the surface of the target food, and its position is controlled by the scanning mechanism of the vibrometer; the continuously acquired data point refers to a series of discrete values ​​obtained by the analog-to-digital converter sampling the continuous analog voltage signal at fixed time intervals.

[0115] For example, continuing the above example, while a millimeter-scale light spot driving a high-frequency probe wave performs a three-dimensional scan of the entire braised goose, the laser measurement point of the laser Doppler vibrometer moves synchronously on the goose's surface along the same trajectory. When the light spot and the laser point scan together to the location in the goose's chest cavity where the polypropylene plastic sheet exists, the surface of the goose at that point contains... 35 kHz For 1.05 MHz and new Complex vibrations were generated under the influence of a mixed-spectrum sound field with a component of 1.015 MHz. A laser Doppler vibrometer was used to measure the vibration velocity at this point in real time, and the velocity was recorded at a rate of 5 × 10⁻⁶. 6 The data is recorded at a sampling rate of 10 times per second. During the brief period of scanning the impurity region, such as 10 ms, 50,000 consecutive data points are collected, exhibiting complex waveforms containing multiple frequencies. Embedding this 10 ms data into the hundreds of millions of data points collected throughout the entire scanning process, and strictly arranging them chronologically, constitutes the complete original vibration signal sequence for this braised goose. Spectral analysis of the data segment corresponding to the moment the impurity appeared in this sequence will verify the presence of... , and The frequency components directly prove the effectiveness of synchronous scanning and vibration acquisition.

[0116] In one embodiment of the present invention, step S4 includes the following steps:

[0117] Calculate the pump acoustic frequency With high frequency detection wave frequency absolute value of the difference And generate a signal of that frequency as a reference frequency input to the lock-in amplifier;

[0118] The original vibration signal sequence is input as the signal to be measured into the lock-in amplifier;

[0119] Through internal mixing and low-pass filtering in the lock-in amplifier, noise unrelated to the reference frequency and noise at frequencies of [missing information] are filtered out. and The fundamental frequency signal is used to generate a difference frequency characteristic signal with background suppression.

[0120] Specifically, the operator or control system uses this raw vibration signal sequence as input data to be processed and transmits it to the signal input port of the lock-in amplifier. Simultaneously, the known pump acoustic wave frequency is utilized... With high frequency detection wave frequency Calculate the difference frequency value The system generates a pure sine wave of that frequency as a reference signal, which is then input to the external reference frequency port of the lock-in amplifier. The core processing unit inside the lock-in amplifier then begins operation. First, its internal mixer performs a multiplication operation on the two input signals; one is an external signal with a frequency of... One path contains a pure reference signal, and the other contains the original vibration signal sequence with complex frequency components. Multiplication is mathematically equivalent to frequency shifting the original vibration signal sequence. After multiplication, the original frequency in the signal is... The components will be converted into DC components and frequencies of . The component, originally with a frequency of The components will be converted to a frequency of and The amount, while the frequency of newly generated impurities is The difference frequency signal component will be converted into a low-frequency DC component with a frequency of The amount.

[0121] Next, the signal passes through a low-pass filter with a very low cutoff frequency set inside the lock-in amplifier. This low-pass filter allows only lower-frequency, near-DC signal components to pass through, while significantly attenuating or filtering out all higher-frequency AC components. After low-pass filtering, the signal contains frequency-dependent components. and The high-frequency AC components generated by the mixing of the fundamental frequency signal are completely removed, and various components related to the reference frequency are eliminated. Irrelevant random noise is also greatly suppressed because it is difficult to generate stable low-frequency components after mixing. Ultimately, the signal that passes through this low-pass filter and is retained is almost entirely composed of signals with frequencies of [frequency range missing]. The stable DC component generated after mixing the nascent difference frequency signal is then processed by a lock-in amplifier, which outputs this processed, amplitude-stable DC voltage signal. This output signal is defined as the final result, i.e., the background-suppressed difference frequency characteristic signal, generated solely by impurity nonlinear effects, with background noise and fundamental frequency interference theoretically eliminated. The amplitude of this signal directly corresponds to the original vibration signal sequence. The intensity of the frequency component, and its presence or absence, directly indicates the presence or absence of impurities at the corresponding scan time. The difference frequency characteristic signal of background suppression is a DC or extremely low frequency voltage signal with stable amplitude, which theoretically does not contain impurities in its background. , Other noise frequency components, only corresponding to The difference frequency information.

[0122] In one embodiment of the present invention, in order to perform preliminary material differentiation on the detected impurities, time-frequency analysis can be further performed on the difference frequency characteristic signal of background suppression.

[0123] Specifically, this includes: after the lock-in amplifier outputs a background-suppressed difference frequency characteristic signal, the background-suppressed difference frequency characteristic signal usually manifests as a DC or narrowband low-frequency signal with increased amplitude. The signal processing unit captures the time-domain waveform of this signal rise segment and extracts its spectral width characteristics, such as -3dB bandwidth, by performing a short-time Fourier transform or wavelet transform on this waveform segment.

[0124] Impurities of different materials, such as metals, hard plastics, hair, and minerals, produce difference frequency signals in heterodyne mixing due to differences in their internal structure, density, elastic modulus, and nonlinear coefficients. These signals are not ideal single frequencies, and their spectral widths are material-specific. For example, metal shavings typically produce difference frequency signals with extremely narrow spectral widths; hard plastics produce signals with moderate spectral widths; while hair or fibrous impurities, due to their non-uniform structure, may produce relatively wide spectral widths.

[0125] The system has a pre-built database of impurity type spectral widths, which stores the correspondence between common impurity materials and typical difference frequency signal spectral widths. This database was established through extensive prior calibration experiments, such as using standard impurity samples of known materials and sizes embedded in standard food for detection. This is a conventional calibration method and will not be elaborated upon here.

[0126] The extracted real-time difference frequency signal spectral width will be quickly matched with the aforementioned database. The matching algorithm can employ nearest neighbor classification or threshold-based judgment. The matching result will generate preliminary classification information for the impurity material. For example: if the spectral width is less than 100 Hz, it is classified as Class A impurity with metallic properties; if it is greater than 100 Hz but less than 500 Hz, it is classified as Class B impurity with hard plastic / glass properties; if it is greater than 500 Hz, it is classified as Class C impurity with fiber / hair properties. These specific numerical limits were obtained through prior calibration experiments using standard samples, such as stainless steel balls, PP plastic granules, and human hair.

[0127] This classification information can be attached to the subsequently generated visual impurity distribution report to provide more targeted guidance for manual re-inspection. For example, it can prompt re-inspectors to pay special attention to possible hair in Category C areas, or provide decision-making reference for automated rejection systems, such as prioritizing the rejection of Category A metallic impurities.

[0128] The core operation of a lock-in amplifier is the multiplication of the input signal and the reference signal:

[0129] The reference signal is: ;

[0130] Input signal including difference frequency: ,in For noise and other frequency components;

[0131] The core operation of a lock-in amplifier is multiplication mixing:

[0132] ;

[0133] Right now: ;

[0134] Using the product-to-sum formula: ;

[0135] For the difference frequency signal term:

[0136] ;

[0137] The lock-in amplifier then passes through a low-pass filter with an extremely low cutoff frequency, which filters out all high-frequency AC components, including harmonic terms. Original fundamental frequency as well as The high-frequency component generated by mixing;

[0138] Therefore, the final output DC or quasi-DC signal is:

[0139] ;

[0140] That is, the difference frequency signal amplitude and phase The relevant DC component;

[0141] in, Difference frequency signal The amplitude is determined by the nonlinear intensity of the impurities; This represents the initial phase of the difference frequency signal; This is the DC voltage signal output by the lock-in amplifier, which is the difference frequency characteristic signal for background suppression.

[0142] Among them, lock-in amplification is a high-precision signal demodulation technique used to extract weak signals synchronized with a specific reference frequency from a background of strong noise; demodulation refers to the process of recovering the original information from a modulated signal, and here it specifically refers to separating the information of impurities carried by the difference frequency component from the original vibration signal sequence;

[0143] The reference frequency is the standard used by the lock-in amplifier for coherent detection; here, it specifically refers to the absolute value of the frequency difference between the pump acoustic wave and the high-frequency probe wave. Its design is based on the following principle: in order to demodulate specific difference frequency components from a mixed signal. It obtains a DC output, and the reference signal must maintain the same frequency as the target signal; the fundamental frequency signal specifically refers to the frequency of the original vibration signal sequence directly caused by the pump wave and the high-frequency probe wave. and The strong amplitude signal components are the main background interference; the nonlinear effect here specifically refers to the acoustic nonlinear behavior exhibited by impurities in the heterodyne mixing process, which is the physical source of the difference frequency signal.

[0144] For example, the system will include 35 kHz For 1.05 MHz and potential The original vibration signal sequence from the detection of braised goose, containing components of 1.015 MHz, is input to the lock-in amplifier. The system calculates the difference frequency to be 1.015 MHz and uses a sine wave with a frequency of 1.015 MHz as the reference frequency input. The lock-in amplifier internally performs mixing and low-pass filtering. When processing the signal segment corresponding to the location of the plastic sheet inside the goose's breast cavity on the scanning time axis, because this signal segment contains... The 1.015 MHz component, after being mixed with the 1.015 MHz reference signal, results in a difference frequency term that returns to zero. This difference frequency term, after passing through a low-pass filter, outputs a raised DC voltage signal, which is the difference frequency characteristic signal for background suppression. However, when scanning other signal segments without impurities, since there is no... The output of the lock-in amplifier is only a baseline noise close to zero. The presence of this DC voltage signal directly verifies that the lock-in amplification process can effectively demodulate and separate the characteristic signal generated only by the nonlinear effect of impurities from the complex original vibration signal sequence, while filtering out the fundamental frequency signals at 35 kHz and 1.05 MHz and other random noise, thus achieving background suppression detection.

[0145] In one embodiment of the present invention, step S5 includes the following steps:

[0146] A preset effective signal amplitude threshold is used to distinguish between real signals and electronic noise.

[0147] The amplitude of the difference frequency characteristic signal of background suppression is monitored and compared with the effective signal amplitude threshold in real time. If the threshold is exceeded, the presence of impurities is determined.

[0148] Based on the judgment result, a logical marker sequence of presence or absence synchronized with the scan time is generated to form the judgment result of the presence or absence of impurities.

[0149] Specifically, the difference frequency characteristic signal for background suppression is a DC or extremely low-frequency voltage signal. The variation in its amplitude over time represents the change in the intensity of the nonlinear effect of impurities. First, an effective signal amplitude threshold is set based on statistical analysis of impurity-free background noise. The method involves collecting background suppression difference frequency characteristic signal data before formal detection begins, or by scanning a known impurity-free target food product. The amplitude distribution of this data sequence is calculated, typically using the root mean square (RMS) or peak value of the signal as the amplitude measure. Based on this distribution, the effective signal amplitude threshold is set to 3 to 5 times the average amplitude of the background noise, or a fixed safety margin above the peak amplitude of the background noise. The purpose of this threshold is to reliably distinguish the effective signal generated by real impurities from the system's inherent electronic noise or random fluctuations. After setting, the system begins continuous and uninterrupted real-time monitoring of the background suppression difference frequency characteristic signal output in real time.

[0150] The core of the monitoring is to continuously read and calculate the instantaneous amplitude value of the signal at the current moment. The signal processing unit compares this instantaneous amplitude value with a preset effective signal amplitude threshold in real time. The signal processing unit can be a comparator circuit or a digital signal processing software module.

[0151] For most of the time along the scanning timeline, the amplitude of the difference frequency characteristic signal for background suppression remains at a very low level, typically below or close to the effective signal amplitude threshold, indicating the absence of impurities at the scan point. However, when the scan point moves to a location containing hard impurities, the amplitude of the difference frequency characteristic signal for background suppression output in the aforementioned steps suddenly increases. Once the system detects that the instantaneous amplitude of this signal exceeds the previously preset effective signal amplitude threshold at a specific moment, the internal logic unit triggers a judgment action, determining that impurities exist at the current scan moment and corresponding spatial location. Based on this judgment result, the system generates a clear, binary logical flag for that specific moment. If the amplitude exceeds the threshold, a logical value of presence or a Boolean value of True is generated; if the amplitude does not exceed the threshold, a logical value of absence or a Boolean value of False is generated. The system records and combines these logical flags generated in chronological order with the corresponding timestamp information to form a sequence or data stream composed of presence and absence flags synchronized with the entire scanning timeline, outputting a judgment result containing the presence or absence of impurities with or without the flag. The results directly and clearly indicate at which moments in the scanning process effective impurity signals were detected.

[0152] Threshold comparison discriminant:

[0153] ;

[0154] in, For a moment The difference frequency characteristic signal amplitude of the background suppression;

[0155] The preset effective signal amplitude threshold is typically set based on statistical values ​​of background noise, for example: ,in, This represents the mean of the background noise. Standard deviation; This is a coefficient, usually taken as 3 to 5. The basis for this setting is to ensure that the probability of false alarms or misjudgments is extremely low when there are no impurities.

[0156] For a moment Whether there are any impurities or not.

[0157] Among them, the result of the impurity presence or absence judgment is a logical marker sequence synchronized with the scanning time axis, with each marker corresponding to the scanning time and position; the effective signal amplitude threshold is a pre-set voltage amplitude threshold, whose functional feature is to serve as a decision boundary to distinguish effective signals from background noise. Its setting is based on the statistical analysis of pure background noise signals, such as the noise amplitude statistics based on 200 sets of impurity-free sample scanning data, taking the mean plus five times the standard deviation as the threshold.

[0158] Logical symbols are symbols that represent the result of a decision. They are usually binary values ​​of yes or no, which facilitate computer processing and subsequent steps.

[0159] For example, before detecting braised goose, the system first scans similar braised foods known to be free of impurities, assuming that the root mean square value of the background noise of the difference frequency characteristic signal for background suppression is 0.1 mV and the peak noise is 0.5 mV. Based on this, the effective signal amplitude threshold is set. The preset value is 1.5 mV. During the scanning of the target braised goose, the system monitors the signal output in step S4 in real time. The amplitude of the difference-frequency characteristic signal for background suppression is measured at the corresponding moment when the scan reaches the area where the plastic sheet inside the goose's chest cavity is located. The voltage spiked sharply to 5 mV. Since 5 mV > 1.5 mV, the system immediately determined that impurities were present at this moment and generated a corresponding logical marker. However, for most of the scanning of other parts, the signal amplitude was less than 0.3 mV, below the 1.5 mV threshold, and the system continuously generated "no" markers. Finally, the system outputs a marker sequence corresponding to the entire scanning timeline, such as [...no, no, present, present, present, no, no, ...]. This represents the result of determining the presence or absence of impurities in the braised goose, directly verifying the feasibility of identifying and determining the validity of impurities based on threshold comparison.

[0160] In one embodiment of the present invention, step S6 includes the following steps:

[0161] Establish a three-dimensional coordinate system corresponding to the target food;

[0162] Record the coordinates of the high-frequency probe wave focal point at the moment when impurities are detected;

[0163] Collect all recorded coordinate points to construct a three-dimensional impurity coordinate point set.

[0164] Specifically, to simultaneously determine the presence or absence of impurities and obtain real-time coordinate information, a three-dimensional coordinate system corresponding to the target food detection space is first established. The origin and axis of this coordinate system are usually fixed at a certain physical reference point in the detection chamber. For example, the center point of the target food when the conveyor belt enters the detection area can be used as the origin, the conveying direction as the X-axis, the horizontal and vertical directions as the Y-axis, and the height direction as the Z-axis. The establishment of this coordinate system allows each spatial position inside and on the surface of the target food to be accurately represented by a unique three-dimensional coordinate (x, y, z).

[0165] The spatial coordinates of the high-frequency sounding wave focal point are generated in real time by its scanning control system according to a preset path and time sequence, and are strictly synchronized with the scanning time. The result of the impurity presence / absence judgment is also a logical marker sequence strictly synchronized with the scanning time. These two time-synchronized data streams are correlated. The system has an internal data correlation unit that continuously and moment-by-moment compares the logical markers in the impurity presence / absence judgment result with the coordinates of the high-frequency sounding wave focal point at the corresponding moment. When the unit reads that the logical marker of a certain moment in the impurity presence / absence judgment result sequence is "sometimes", it indicates that a valid impurity signal exists at that moment. At this time, the system immediately performs a retrieval and recording operation: it retrieves the real-time position coordinates (x, y, z) of the high-frequency sounding wave focal point in the three-dimensional coordinate system recorded by the high-frequency sounding wave scanning control system at the same moment, and stores this coordinate data (x, y, z) along with its corresponding timestamp as an independent data point entry in a dedicated impurity coordinate storage list.

[0166] Throughout the scanning cycle, the aforementioned retrieval and recording operations are automatically executed once whenever a judgment result is obtained. After the scanning process is completed, the system aggregates all recorded coordinate data point entries representing different impurity locations or the same impurity detected at different scanning times from the storage list, removing redundant adjacent points that may be caused by signal jitter, forming the final data set. This set, containing the spatial location information of all impurities determined to be present and consisting of several (x, y, z) coordinate points, is defined as the three-dimensional impurity coordinate point set. This point set accurately represents the three-dimensional spatial distribution of all detected impurities inside and on the surface of the target food in digital form. The three-dimensional impurity coordinate point set is a point set data structure composed of three-dimensional spatial coordinates, such as a matrix or list containing N rows and 3 columns of data, where N is the number of detected impurity locations.

[0167] Among them, characterizing the impurity location refers to digitally describing and representing the distribution of impurities in space using a set of three-dimensional coordinate points; the scanning coordinates specifically refer to the real-time position of the focal point of the high-frequency probe wave in the three-dimensional coordinate system, and the data is provided in real time by the position encoder or motion controller of the scanning control system.

[0168] For example, a three-dimensional coordinate system is established in the detection chamber. During the scanning of the target braised goose, assuming that when the scan reaches the plastic sheet area inside the goose's chest cavity, a logical marker sequence of "Yes, Yes, Yes" is output within several milliseconds. At the same time, the scanning control system records the coordinate sequence of the high-frequency probe wave focal point within these few milliseconds, such as (105.2, 50.1, -15.3), (105.3, 50.1, -15.4), and (105.4, 50.1, -15.5).

[0169] Upon detecting a marked location, the system immediately retrieves and records the corresponding coordinates. After scanning, the system clusters a series of neighboring coordinates representing the location of the plastic piece, taking the center point coordinates (105.3, 50.1, -15.4) as the representative location of the impurity and adding it to the impurity coordinate storage list. If another hair impurity is detected under the goose's wing, its coordinates (80.5, 120.7, 10.2) will also be recorded in the same way. Finally, these two coordinates are combined to form a three-dimensional impurity coordinate set for the braised goose.

[0170] {(105.3, 50.1, -15.4), (80.5, 120.7, 10.2)}.

[0171] The generation of this point set directly verifies the feasibility of associating the logical judgment of the presence or absence of impurities with precise spatial coordinates and constructing digital location data.

[0172] In one embodiment of the present invention, step S7 includes the following steps:

[0173] Call the 3D contour model of the target food;

[0174] The three-dimensional impurity coordinate point set is rendered onto the three-dimensional contour model as a highlighted mark.

[0175] Output the rendered 3D model image and generate a visual impurity distribution report.

[0176] Specifically, the three-dimensional contour model data can come from two sources: one is a general three-dimensional model that is pre-built and stored in the system database after scanning the same type of standard braised food with an optical three-dimensional scanner or structured light scanning device; the other is a specific three-dimensional contour model that is generated in real time by scanning the target food itself with a synchronously deployed optical three-dimensional scanning system before or during the impurity detection of the current target food.

[0177] The system calls and loads this 3D contour model, performs 3D reconstruction and rendering on it within the visualization software interface or graphics processing unit, forming a digital 3D entity that matches the actual shape and size of the target food. Next, the system reads each coordinate data point from the 3D impurity coordinate point set. For each coordinate point in the set, the system performs spatial coordinate matching and mapping operations: based on the same set of spatial references and scales used when establishing the 3D coordinate system, it finds the corresponding spatial position of that coordinate point in the digital 3D entity.

[0178] Then, the graphics processing program overlays and renders at that spatial location using bright and striking visual markers that are clearly distinguishable from the model's own color, such as a red sphere, a flashing cross, or a yellow cube. This rendering process precisely draws the marked graphic elements, according to their three-dimensional coordinates and preset size, into the image buffer of the three-dimensional contour model, making them appear as if they are attached to the model's surface or embedded in a specific location within the model.

[0179] The system iterates through all coordinate points in the 3D impurity coordinate point set, performing the overlay rendering operation of the highlighted marks one by one. Once all impurity locations are marked, the system performs final lighting, shadow, and viewpoint adjustments on the overlay rendered 3D model image to generate a clear, intuitive, and easily observable 3D visualization image. This image not only displays the overall 3D shape of the target food but also clearly indicates the specific distribution of all detected impurities with prominent highlighted marks. Finally, the system outputs the image as a standard format electronic file, such as PNG, JPG, or PDF, or displays it directly on the human-computer interaction interface. This visualization image file or interface display containing impurity spatial distribution information is defined as a visualization impurity distribution report. This report can be directly used to guide subsequent manual re-inspection operations or as an input signal to drive the robotic arm of an automated rejection device to locate the impurity coordinate points for precise removal.

[0180] Among them, the three-dimensional contour model is a digital three-dimensional representation of the shape of the target food. It is a three-dimensional mesh data composed of triangular facets or point clouds, and is obtained based on the principle of optical three-dimensional measurement.

[0181] For example, the system calls the pre-obtained 3D outline model file of the whole braised goose obtained through optical scanning. The system reads the previously obtained 3D impurity coordinate point set: {(105.3, 50.1, -15.4), (80.5, 120.7, 10.2)}. The graphics processing software first loads the 3D model of the braised goose, and then, according to the coordinate transformation parameters calibrated by the system, maps the first coordinate of the point set (105.3, 50.1, -15.4) to the corresponding position inside the chest cavity on the model, and overlays a red sphere with a diameter of 3mm as a highlight mark at that location.

[0182] Next, the second coordinate (80.5, 120.7, 10.2) is mapped to the corresponding surface position below the wing on the model, and the red sphere is rendered in the same way. After all the markers are superimposed, the system generates a 3D rendering with lighting effects viewed from a side-above perspective. In the image, the braised goose model is brown, and the two red spheres are clearly visible. Finally, the system saves this image as a file named: Braised Goose - Impurity Distribution Report.png. This visualized impurity distribution report intuitively shows that the two impurities are located deep in the goose's chest cavity and below the wing, respectively. Operators can use this report to quickly locate and re-inspect, or guide the robotic arm for precise removal. The generation of this report directly verifies the feasibility and intuitiveness of integrating the coordinate point set with the 3D model for visualization output.

[0183] See appendix Figure 2 The present invention also proposes a system for detecting impurities in braised food, comprising the following modules:

[0184] The surface pretreatment module is used to perform surface acoustic property optimization pretreatment on braised food with liquid brine on the surface, to generate target food without free liquid and with an acoustically enhanced surface.

[0185] The multi-frequency acoustic excitation module is used to perform dual-frequency acoustic co-excitation and scanning on the target food, forming a mixed spectrum sound field inside and on the surface of the target food.

[0186] The vibration signal acquisition module is used to acquire the surface vibration response of the target food under the action of a mixed spectrum sound field using non-contact vibration measurement equipment, and generate the original vibration signal sequence.

[0187] The signal demodulation processing module is used to perform lock-in amplification on the original vibration signal sequence and demodulate the difference frequency characteristic signal with background suppression caused by impurities.

[0188] The validity determination module is used to determine whether impurities exist based on the validity of the difference frequency characteristic signal of background suppression, and to generate a judgment result on whether impurities exist.

[0189] The coordinate association and positioning module is used to associate the result of the impurity presence or absence judgment with the scanning coordinates to construct a three-dimensional impurity coordinate point set;

[0190] The 3D visualization module is used to integrate the 3D impurity coordinate point set with the 3D contour model of the target food to generate a visualized impurity distribution report.

[0191] Each of the modules can be implemented in whole or in part through software, hardware, or a combination thereof. It supports hardware embedded in or independent of the processor in the computer device, and also supports software stored in the memory of the computer device, so that the processor can call and execute the operations corresponding to each of the above modules.

[0192] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A method for detecting impurities in a pickled food product, characterized by, Includes the following steps: S1. Perform surface acoustic property optimization pretreatment on braised food with liquid brine on the surface. Use a vortex air knife device to generate a vortex air curtain with a relative humidity lower than the ambient humidity. Remove free liquid from the surface of the braised food through the vortex air curtain to form a clean interface. Use the surface tension change caused by the vortex air curtain to apply instantaneous stress to the impurities attached to the surface of the braised food, and generate target food with no free liquid and an acoustically enhanced surface. S2, the target food is subjected to dual-frequency acoustic wave synergistic excitation and scanning, a mixed frequency spectrum acoustic field is formed in the interior and surface of the target food, including: emitting a pump acoustic wave with a frequency of to the target food, establishing a uniform low-frequency vibration field; focusing a high-frequency detection wave with a frequency of into a millimeter-level light spot, and performing three-dimensional scanning on the target food; using the nonlinear characteristics of hard impurities, the pump acoustic wave and the high-frequency detection wave are heterodyne mixed, and a difference frequency signal with a frequency of is generated in the mixed frequency spectrum acoustic field; S3. Use non-contact vibration measurement equipment to collect the surface vibration response of the target food under the action of a mixed spectrum sound field, and generate the original vibration signal sequence; S4. Perform lock-in amplification on the original vibration signal sequence and demodulate the difference frequency characteristic signal with background suppression caused by impurities, including: calculating the pump acoustic wave frequency. With high frequency detection wave frequency absolute value of the difference The signal at that frequency is generated and used as a reference frequency input to the lock-in amplifier; the original vibration signal sequence is input to the lock-in amplifier as the signal to be tested; through mixing and low-pass filtering within the lock-in amplifier, noise unrelated to the reference frequency and noise at frequencies of... and The fundamental frequency signal; S5. Determine whether there are impurities based on the effectiveness of the difference frequency characteristic signal of background suppression, and whether there is a judgment result for the generated impurities; S6. Associate the results of the impurity presence / absence judgment with the scanning coordinates to construct a three-dimensional impurity coordinate point set; S7. Integrate the three-dimensional impurity coordinate point set with the three-dimensional contour model of the target food to generate a visual impurity distribution report.

2. The method for detecting impurities in braised food according to claim 1, characterized in that, Generating the original vibration signal sequence includes the following steps: A laser Doppler vibrometer is used to scan the measurement point synchronously with the focal point of the high-frequency probe wave; The laser measurement point is used to measure and record the vibration displacement or velocity of the target food surface in the vertical direction in real time, and the continuously collected data is combined in time sequence to generate the original vibration signal sequence.

3. The method for detecting impurities in braised food according to claim 1, characterized in that, The determination of whether impurities have been generated includes the following steps: A preset effective signal amplitude threshold is used to distinguish between real signals and electronic noise. The amplitude of the difference frequency characteristic signal of background suppression is monitored and compared with the effective signal amplitude threshold in real time. If the threshold is exceeded, the presence of impurities is determined. Based on the judgment result, a logical marker sequence of presence or absence synchronized with the scan time is generated to form the judgment result of the presence or absence of impurities.

4. The method for detecting impurities in braised food according to claim 1, characterized in that, Constructing a three-dimensional set of impurity coordinate points includes the following steps: Establish a three-dimensional coordinate system corresponding to the target food; Record the coordinates of the high-frequency probe wave focal point at the moment when impurities are detected; Collect all recorded coordinate points to construct a three-dimensional impurity coordinate point set.

5. The method for detecting impurities in braised food according to claim 1, characterized in that, Generate a visual impurity distribution report, including the following steps: Call the 3D contour model of the target food; The three-dimensional impurity coordinate point set is rendered onto the three-dimensional contour model as a highlighted mark. Output the rendered 3D model image and generate a visual impurity distribution report.

6. The method for detecting impurities in braised food according to claim 1, characterized in that, After a clean interface is formed, the following steps are also included: Utilizing optical sensors to measure the surface finish of clean interfaces; The surface finish is compared with a preset finish standard to generate an acoustic coupling condition evaluation result.

7. The method for detecting impurities in braised food according to claim 1, characterized in that, After generating the difference frequency signal, the process also includes: Analyze the spectral width of the difference frequency signal components and match it with a preset database of impurity type spectral widths to generate preliminary classification information of impurity materials.

Images