A fiber optic fluorescence sensing system for food pathogens

Abstract

The present invention relates to a fiber optic fluorescence sensing detection system for food pathogens, comprising a sensing module and a detection module. The sensing module includes a fiber optic fluorescence sensing module (optical components and a fluorescence reaction module) and an environmental sensing module. The optical components include an optical fiber array, a filter, a collimating lens, and an image sensor. The fluorescence reaction module includes an excitation light source and a multi-channel fluorescence reaction tank. The excitation light source irradiates the solution in the centrifuge tube in the multi-channel fluorescence reaction tank to stimulate a fluorescence signal. The fiber optic array transmits the fluorescence signal. The filter selects the fluorescence signal. The collimating lens amplifies and focuses the fluorescence signal. The image sensor is used to image the fluorescence signal. The environmental sensing module includes several sensors that collect environmental data. The detection module obtains a fluorescence intensity image obtained by the image sensor and environmental data collected by the environmental sensing module, and performs pathogen detection on the solution of the food sample. The present invention can effectively detect food pathogens.

Description

Technical Field

[0001] The present invention relates to the technical field of food safety detection, and in particular to an optical fiber fluorescence sensing detection system for food pathogens. Background Art

[0002] As a cornerstone for safeguarding human health and social stability, food safety has always been a focus of scientific research and practical applications. However, traditional food safety testing methods, due to complex operations and high costs, are unable to meet the needs of real-time on-site monitoring.

[0003] A variety of advanced detection methods have been developed for numerous food safety risk factors, including viruses, pathogens, toxins, and fungi. These include electrochemical detection, colorimetric analysis, mass spectrometry, and fluorescence detection. Fluorescence detection has garnered widespread attention and application due to its outstanding performance in high sensitivity and selectivity. By utilizing the fluorescence properties of substances under excitation light, fluorescence methods can rapidly and sensitively detect trace amounts of harmful substances in food, offering broad application prospects.

[0004] When it comes to high-precision testing for pathogenic microorganisms and viruses, environmental factors such as food temperature and humidity also significantly impact detection accuracy. Furthermore, just like food processing and production, food transportation is a key factor in food contamination. Real-time monitoring of food at different locations during transportation can provide valuable insights into the evolution of pathogenic microorganism concentrations in food and improve detection accuracy.

[0005] Machine learning provides a powerful guarantee for the accurate identification of pathogenic microorganisms in food safety testing. Changes in pathogenic microorganism concentrations can cause changes in signals such as current, fluorescence, or images. By extracting relevant data features and training models using machine learning algorithms, rapid pathogen classification and concentration identification can be achieved.

[0006] In summary, environmental parameters such as temperature, humidity, and geographic location are key factors in the accurate detection of food safety and the analysis of regular changes. However, the current detection methods based on fluorescence technology have not effectively integrated these environmental parameter information, and the detection accuracy is not high enough.

[0007] Therefore, there is an urgent need for a high-performance, low-cost and integrated new portable food safety testing system to achieve comprehensive integration and intelligent analysis of test data, and provide more advanced and convenient technical means for on-site food safety monitoring. Summary of the Invention

[0008] To this end, the technical problem to be solved by the present invention is to overcome the problem that the food safety detection method based on fluorescence technology in the existing technology has not effectively integrated environmental parameters such as temperature, humidity, and geographical location, and the detection accuracy is not high enough.

[0009] In order to solve the above technical problems, the present invention provides a fiber optic fluorescence sensing detection system for food pathogens, including a sensing module and a detection module. The sensing module includes a fiber optic fluorescence sensing module and an environmental sensing module, wherein:

[0010] The fiber optic fluorescence sensing module includes an optical component and a fluorescence reaction module, wherein the optical component includes an optical fiber array, a filter, a collimating lens, and an image sensor arranged in sequence; the fluorescence reaction module includes an excitation light source and a multi-channel fluorescence reaction tank; the excitation light source is used to irradiate the solution in the centrifuge tube in the multi-channel fluorescence reaction tank to excite the fluorescence signal, and the solution in the centrifuge tube is the solution of the food sample to be detected; the fiber optic array is connected to the multi-channel fluorescence reaction tank for transmitting the fluorescence signal; the filter is used to select the fluorescence signal in the fiber optic array; the collimating lens is used to amplify and focus the fluorescence signal selected by the filter; and the image sensor is used to image the amplified and focused fluorescence signal to obtain a fluorescence intensity image;

[0011] The environmental sensing module includes a geographic location sensor, a temperature sensor, and a humidity sensor, which are used to collect geographic location data, temperature data, and humidity data respectively;

[0012] The detection module is used to obtain the fluorescence intensity image obtained by the image sensor, and the geographical location, temperature data and humidity data collected by the geographical location sensor, temperature sensor and humidity sensor respectively, and perform pathogen concentration detection on the solution of the food sample based on the obtained data.

[0013] In one embodiment of the present invention, in order to eliminate the influence of background light, before collecting the fluorescence intensity image, the detection module first collects the background image obtained by the image sensor when the excitation light source is not excited, and performs differential processing on the fluorescence intensity image collected by the detection module after the excitation light source is excited and the background image to remove the interference of ambient background light.

[0014] In one embodiment of the present invention, it is assumed that there is a linear relationship between the fluorescence intensity I corresponding to the fluorescence intensity image and the sample concentration c, which is expressed as:

[0015] I=k·c+I0

[0016] Where k is the linear conversion coefficient from c to I; I0 is the initial fluorescence intensity when the sample concentration is 0.

[0017] In one embodiment of the present invention, in order to eliminate the influence of noise on the output of the image sensor, the real-time output value of the image sensor is subjected to mean filtering, and the formula is:

[0018]

[0019] Among them, I = [I1, I2, …, I n , n ∈ [1, N], N is the number of fiber array channels, and ΔT is the sampling period.

[0020] In an embodiment of the present invention, the detection module uses the collected fluorescence image, temperature data, humidity data, and geographical location data as the input of the THPIRBFNN model;

[0021] The activation function of the hidden layer of the THPIRBFNN model is a Gaussian radial basis function with respect to the environmental parameter matrix. The environmental parameters include temperature data, humidity data, and geographical location data, and the expression is as follows:

[0022]

[0023] Among them, Y THPI is a Gaussian radial basis function with respect to the environmental parameter matrix, represents the centroid of the j-th neuron, σ represents the width of the Gaussian kernel; the input data is represented by the environmental parameter matrix [T, H, P, I1, I2, I3 … I N , T is the environmental temperature, H is the environmental humidity, P is the geographical location information, and I1, I2, I3 … I N represent the fluorescence intensities detected by multiple channels.

[0024] In an embodiment of the present invention, the detection module further includes an optimization unit. Among them, the optimization unit is used to optimize the linear weights of the output layer neurons and the center point parameters of the hidden layer neurons of the THPIRBFNN model;

[0025] The linear weights of the output layer neurons of the THPIRBFNN model are updated according to the following formula:

[0026] |G k+1 | = |G k | + k G × |η·ΔG|

[0027]

[0028] Among them, |G k+1 | and |G k | are the weights before and after update respectively, k G is the influence coefficient of the environmental parameter matrix on weight optimization, η is the learning rate, and E is the minimum error function; the output state represents different concentration levels or classification results, and m is the total number of these possible output states; Indicates the predicted output value for a certain input sample when the model output state is i, Z i Indicates the actual output value or true concentration corresponding to the sample;

[0029] The center points of the hidden layer neurons of the THPIRBFNN model are updated according to the following formula:

[0030] |c k+1 |=|c k |+k c ×|η·Δc|

[0031]

[0032] Among them, k c is the influence coefficient of the environmental parameter matrix on the optimization of the hidden layer neuron center point, and the width of the Gaussian kernel is updated according to the following formula:

[0033] |σ k+1 |=|σ k |+k σ ×|η·Δσ|

[0034]

[0035] Among them, k σ is the influence coefficient of the environmental parameter matrix on the Gaussian kernel width optimization.

[0036] In one embodiment of the present invention, the THPIRBFNN model uses a preset loss function LOSS during training, which is expressed as follows:

[0037]

[0038] Where N is the total number of samples in the test set, is the concentration prediction state, and Z is the actual concentration of the sample.

[0039] In one embodiment of the present invention, the detection module further includes an evaluation unit, which is used to perform an accuracy rate ACC evaluation on the identification result of the sample concentration, and the formula is:

[0040]

[0041] Among them, TP represents the number of true positive examples, FN represents the number of false negative examples, TN represents the number of true negative examples, and FP is the number of false positive examples.

[0042] In one embodiment of the present invention, the detection module further includes a temperature control module, which is in contact with the multi-channel fluorescence reaction tank to achieve temperature control of the solution in the centrifuge tube in the multi-channel fluorescence reaction tank.

[0043] The above technical solution of the present invention has the following advantages over the prior art:

[0044] The optical fiber fluorescence sensing detection system for food pathogens described in the present invention can be effectively applied in food safety detection. By integrating the sensing module and the detection module, rapid and accurate detection of target substances in food samples is achieved.

[0045] This invention improves the sensitivity and stability of the system by optimizing image sensor (camera) parameters and calibrating multiple channels. At the same time, it combines machine learning models to achieve accurate identification of sample concentrations under different environments.

[0046] The present invention has the advantages of small size, light weight, and easy operation, is suitable for the needs of rapid on-site detection, and provides an efficient and reliable solution for food safety detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to make the contents of the present invention more clearly understood, the present invention is further described in detail below based on specific embodiments of the present invention in conjunction with the accompanying drawings.

[0048] Figure 1 Schematic diagram of the structure of the optical fiber fluorescence sensing module in an embodiment of the present invention;

[0049] Figure 2 Schematic diagram of the structure of the detection module in an embodiment of the present invention;

[0050] Figure 3 Schematic diagram of the human-computer interaction interface of the optical fiber fluorescence sensing detection system according to an embodiment of the present invention;

[0051] Figure 4 This is a temperature control performance test and analysis diagram of the optical fiber fluorescence sensing detection system in an embodiment of the present invention;

[0052] Figure 5 2. It is a schematic diagram of multi-channel fluorescence detection signal calibration of the optical fiber fluorescence sensing detection system according to an embodiment of the present invention;

[0053] Figure 6 This is a comparison chart of the fluorescence performance test of the optical fiber fluorescence sensing detection system in the embodiment of the present invention and a commercial microplate reader;

[0054] Figure 7 This is a stability test diagram of the optical fiber fluorescence sensing detection system in an embodiment of the present invention;

[0055] Figure 8 This is a diagram showing the dynamic monitoring results of African swine fever virus under different environments in an embodiment of the present invention;

[0056] Figure 9This is a graph showing the linearity detection and concentration identification results of African swine fever virus in an embodiment of the present invention;

[0057] Figure 10 This is a diagram showing the dynamic monitoring results of Salmonella in an embodiment of the present invention;

[0058] Figure 11 This is a graph showing the linearity detection and concentration identification results of Salmonella in an embodiment of the present invention. DETAILED DESCRIPTION

[0059] The present invention will be further described below with reference to the accompanying drawings and specific embodiments so that those skilled in the art can better understand the present invention and implement it. However, the embodiments are not intended to limit the present invention.

[0060] Reference Figure 1 As shown, the present invention relates to an optical fiber fluorescence sensing detection system for food pathogens, comprising: a sensing module and a detection module, wherein the sensing module comprises an optical fiber fluorescence sensing module and an environmental sensing module, wherein:

[0061] The fiber optic fluorescence sensing module includes an optical component and a fluorescence reaction module, wherein the optical component includes an optical fiber array, a filter, a collimating lens, and an image sensor arranged in sequence; the fluorescence reaction module includes an excitation light source and a multi-channel fluorescence reaction tank; the excitation light source is used to irradiate the solution in the centrifuge tube in the multi-channel fluorescence reaction tank to excite the fluorescence signal, and the solution in the centrifuge tube is the solution of the food sample to be detected; the fiber optic array is connected to the multi-channel fluorescence reaction tank for transmitting the fluorescence signal; the filter is used to select the fluorescence signal in the fiber optic array; the collimating lens is used to amplify and focus the fluorescence signal selected by the filter; and the image sensor is used to image the amplified and focused fluorescence signal to obtain a fluorescence intensity image;

[0062] The environmental sensing module includes a geographic location sensor, a temperature sensor, and a humidity sensor, which are used to collect geographic location data, temperature data, and humidity data respectively;

[0063] The detection module is used to obtain the fluorescence intensity image obtained by the image sensor, and the geographical location, temperature data and humidity data collected by the geographical location sensor, temperature sensor and humidity sensor respectively, and perform pathogen concentration detection on the solution of the food sample based on the obtained data.

[0064] Furthermore, the image sensor includes but is not limited to a CMOS micro camera, a CCD camera, a photodetector, etc., and is used for detecting multi-channel fluorescence signals.

[0065] Furthermore, the end of the optical fiber array, the filter, and the collimating lens are arranged in the optical component slot.

[0066] Further, the detection module of the present embodiment also includes a touch screen, a signal processing system, and a power supply module. The signal processing system includes a wireless signal transmission module (such as a WIFI signal transmission module) and a main controller (such as a raspberry development board, an STM32 series development board). The signal processing system is respectively connected to an image sensor and an environmental sensing module for collecting the fluorescence intensity image and environmental data of the optical fiber, and the main controller can transmit the data in real time to a remote terminal or a mobile device through the wireless signal transmission module. The power supply module is used to provide electric energy for the detection module.

[0067] Furthermore, the main controller of this embodiment also has a built-in machine learning model for performing pathogen detection on the solution of food samples.

[0068] Furthermore, the detection module of this embodiment also includes a temperature control module, which includes a PID controller and a heating refrigerator. The heating refrigerator is in contact with the multi-channel fluorescence reaction tank. The temperature control module controls the temperature of the heating refrigerator through the PID controller, thereby controlling the temperature of the multi-channel fluorescence reaction tank.

[0069] The following is a detailed introduction to this embodiment:

[0070] like Figure 1 The figure shows a portable fiber optic fluorescence sensing detection system for food pathogens in this embodiment. The fiber optic fluorescence sensing detection system designed in this example integrates multiple functional modules to achieve rapid and accurate detection of foodborne pathogens. When testing a sample, a centrifuge tube containing the sample solution is placed in a multi-channel fluorescence reaction tank for detection. The fluorescence signal is ultimately captured accurately in the form of an image through the fiber optic array and transmitted to the main controller for in-depth analysis. The system uses image segmentation and RGB three-channel pixel intensity extraction to achieve fully automatic detection and analysis of multiple channels and multiple targets. The intensity of the fluorescence signal is closely related to the concentration of the target. In addition, the system also integrates global positioning system GPS positioning and temperature and humidity sensing functions, which can record and display relevant information about the detection environment in real time. With the help of a wireless signal transmission module, temperature, humidity, location information and test results can be transmitted to a mobile phone or personal server in real time, providing users with fast and portable reading and remote control functions for test data. In order to achieve rapid and intelligent analysis of fluorescence detection signals, the system combines digital signal processing technology and advanced machine learning models.

[0071] In the actual product, a multi-channel fluorescence reaction tank is placed at the core of the fiber-optic fluorescence sensing detection system (possibly in the center of the system), with a heating and cooling unit mounted on its back. This heating and cooling unit is closely connected to a proportional-integral-derivative (PID) controller, which precisely controls the heating and cooling unit (TEC) to achieve bidirectional temperature regulation of the multi-channel fluorescence reaction tank. An LED light source (i.e., excitation light source) is cleverly embedded in the multi-channel fluorescence reaction tank and precisely regulated by the main controller to stimulate the generation of a fluorescence signal. A fiber optic array connected to the bottom of the multi-channel fluorescence reaction tank transmits the excited fluorescence signal through filters and collimating lenses to an image sensor (microcamera). The image data captured by the image sensor is then transmitted to the main controller for further data processing and analysis. The main controller is also connected to a geolocation sensor (GPS) and temperature and humidity sensors for real-time environmental data collection. Test results and monitoring data are intuitively displayed on a touchscreen display. This design makes the fiber-optic fluorescence sensing detection system a powerful technical support for rapid food safety testing.

[0072] In one possible implementation, to eliminate the influence of background light, the fiber optic fluorescence sensing detection system first captures an image of the fiber array in the unexcited state of the LED light source (i.e., a background image) before capturing the fluorescence excitation image. The fluorescence image captured after the LED light source is excited is then differentiated from the background image to remove ambient background light interference. The image captured by the micro camera contains multi-channel fluorescence data, and the fluorescence intensity corresponding to each channel can be obtained by the position of different optical fibers in the fluorescence image. The RGB pixel values ​​within the circular area corresponding to the optical fiber end face in the fluorescence image are extracted to obtain the fluorescence intensity of each optical fiber end face, thereby determining the concentration of the detected object in different channels.

[0073] This embodiment assumes that there is a linear relationship between the fluorescence intensity I corresponding to the fluorescence intensity image and the sample concentration x, which can be expressed as:

[0074] I=k·c+I0(1)

[0075] Where k is the linear conversion coefficient from c to I. The initial fluorescence intensity when the sample concentration is 0 is I0.

[0076] At the same time, in order to eliminate the influence of hardware circuit or environmental noise on the image sensor output, the real-time output of the image sensor is processed by mean filtering. The formula is as follows:

[0077]

[0078] Where, I=[I1,I2,…,I n, where \(n\in[1,N]\), \(N\) is the number of optical fiber channels, and \(\Delta T\) is the sampling period.

[0079] The detection module uses the collected fluorescence images, temperature data, humidity data, and geographical location data as the training inputs of the THPIRBFNN model, which is a radial basis function neural network affected by the environmental parameter matrix.

[0080] The activation function of the hidden layer of the THPIRBFNN model is the Gaussian radial basis function with respect to the environmental parameter matrix, and its expression is as follows:

[0081]

[0082] where \(Y\) THPI is the Gaussian radial basis function with respect to the environmental parameter matrix, represents the centroid of the \(j\)-th neuron, \(\sigma\) represents the width of the Gaussian kernel; the input data is represented by \([T, H, P, I_1, I_2, I_3 \cdots I N \), where \(T\) is the environmental temperature, \(H\) is the environmental humidity, \(P\) is the geographical location information, and \(I_1, I_2, I_3 \cdots I N represent the fluorescence intensities detected by multiple channels. By adjusting the center and width of the basis function, the THPIRBFNN model can learn and approximate complex non-linear mapping relationships, so as to be used for accurate detection and recognition of output states.

[0083] The master controller in the detection module of this embodiment further includes an optimization unit. In order to optimize the parameters of the THPIRBFNN model, the linear weights of the output layer neurons of the THPIRBFNN model are updated according to Equations (6) and (7):

[0084] \(\vert G k+1 \vert=\vert G k \vert + k G \times\vert\eta\cdot\Delta G\vert\ (6)

[0085]

[0086] where \(\vert G k+1 \vert\) and \(\vert G k \vert\) are the weights before and after update respectively, and \(k Gis the coefficient of influence of the environmental parameter matrix on weight optimization, η is the learning rate, and E is the minimization error function. By continuously adjusting the weights, the model can better fit the actual data, thereby improving the accuracy of pathogen concentration prediction. By adjusting the learning rate η, faster convergence and higher prediction accuracy can be achieved while ensuring model stability. By minimizing the error function, the model parameters can be continuously optimized, which helps to more accurately identify and predict pathogen concentrations. In this embodiment, the output state represents different concentration levels or classification results, and m is the total number of these possible output states. It represents the predicted output value for a certain input sample when the model output state is i, and Z i It represents the actual output value or true concentration corresponding to the sample. In this embodiment, the goal of the model is to predict the concentration of pathogenic bacteria based on the input fluorescence image and environmental data. Therefore, and Z i The gap between the two reflects the accuracy of the model prediction. By continuously optimizing the model parameters and structure, this gap can be reduced and the prediction accuracy of the model can be improved.

[0087] Similarly, the center points of the hidden layer neurons of the THPIRBFNN model are updated according to equations (8) and (9):

[0088] |c k+1 |=|c k |+k c ×|η·Δc|(8)

[0089]

[0090] Among them, k c is the influence coefficient of the environmental parameter matrix on the optimization of the hidden layer neuron center point, and the width of the Gaussian kernel is updated according to formula (10) and formula (11):

[0091] |σ k+1 |=|σ k |+k σ ×|η·Δσ| (10)

[0092]

[0093] k σ is the influence coefficient of the environmental parameter matrix on the Gaussian kernel width optimization.

[0094] The THPIRBFNN model of this embodiment adopts a preset loss function LOSS during training, and its expression is as follows:

[0095]

[0096] Where N is the total number of samples in the test set, is the concentration prediction state, and Z is the actual concentration of the sample.

[0097] The main controller in the detection module of this embodiment further includes an evaluation unit, which is used to evaluate the recognition result of the sample concentration using the accuracy (ACC), which is calculated as follows:

[0098]

[0099] Here, TP represents the number of true positives, FN represents the number of false negatives, TN represents the number of true negatives, and FP represents the number of false positives. In this example, TP represents the number of samples where the pathogen is actually present and the model correctly predicts the presence of the pathogen; FN represents the number of samples where the model incorrectly predicts the presence of the pathogen; TN represents the number of samples where the pathogen is actually absent and the model correctly predicts the presence of the pathogen; and FP represents the number of samples where the pathogen is actually absent but the model incorrectly predicts the presence of the pathogen. By calculating these metrics, we can comprehensively evaluate the classification performance of the model.

[0100] In this embodiment, if Figure 3 As shown in the figure, in order to realize the control and data visualization of the portable fiber optic fluorescence sensing detection system, a data acquisition program with a human-computer interaction interface is designed using Python. After the program starts running and the system is initialized, a human-computer interaction interface with data display function is created, and the event loop is entered to wait for user operation. The designed human-computer interaction interface is shown in the figure. Figure 3 As shown, the touchscreen displays multiple information simultaneously in digital and graphical form, including the real-time fluorescence signal from the sample, ambient temperature and humidity, GPS coordinates, battery level, and date and time. This data can be transmitted to a mobile phone via a wireless signal transmission module. The touchscreen features "Start," "Pause," "End," "Save," and "Refresh" buttons, as well as a "Category" drop-down menu. Clicking the "Start" button activates the fluorometer's real-time detection function, performing fluorescence detection every 15 seconds and refreshing the data display. Clicking the "Save" button saves the current and collected data. Clicking the "Pause" button pauses data collection and data refreshing. Clicking the "Stop" button ends data collection and data display. Clicking the "Refresh" button retrieves the current GPS data (geographic location data). Select the target type by clicking the "Category" drop-down menu. The collected fluorescence intensity is automatically converted to the concentration of the target object based on the trained model.

[0101] In this embodiment, if Figure 4As shown in the figure, in order to achieve precise control of the temperature of the multi-channel fluorescence reaction tank of the portable fiber optic fluorescence sensing detection system, the PID parameters of the temperature control module are adjusted. The temperature T is set to 39°C, P is 0.056, I is 103.7, D ​​is 25.92, the sampling period is 259ms, and position PID control is adopted. In the simulation experiment, a centrifuge tube containing 50ul sample solution was added to the multi-channel fluorescence reaction tank. After 4 minutes of reaction, it was taken out and centrifuged and shaken, and then placed back into the multi-channel fluorescence reaction tank. This process was repeated three times for a total of 100 minutes. Figure 4 As shown in (a), the experimental results show that the actual temperature is maintained between 38.9-39.1℃ during standby. When adding or taking samples, the temperature fluctuation does not exceed 0.2℃ and can quickly return to stability. Error analysis is shown in Figure 4 As shown in Figure (b), during the measurement process, the error was within the ±0.05°C range for 94.4% of the time, within the ±0.1°C range for 98.3% of the time, and within the ±0.15°C range for 99.3% of the time. This demonstrates that the temperature control module exhibits excellent stability and accuracy after PID parameter tuning. The device's battery life was also tested. The results showed that with temperature control enabled, the battery life was 4.7 hours; without it, the battery life reached 7 hours.

[0102] In this embodiment, if Figure 5 As shown in the figure, in order to eliminate process errors and ensure the accuracy and consistency of multi-channel fluorescence detection signals, a set of 13 FITC solutions with different fluorescence intensities were selected for calibration. Reagents numbered 1-13 were tested in different channels in sequence, with the concentration of FITC increasing gradually from 1 to 13. The acquired fluorescence images are shown in the figure. Figure 5 As shown in (a). It can be seen that when the FITC concentration is low, the excitation light is stronger than the emission light, and the fiber image is darker (actually dark blue). As the FITC concentration gradually increases, the emission light intensity gradually becomes stronger than the excitation light intensity, and the fiber image is brighter (actually bright green). The fluorescence intensity value corresponding to the fiber fluorescence image is as follows Figure 5 As shown in (b), channel 8 was selected as the standard output channel. The remaining seven channels show a positive correlation with the standard output channel. Therefore, a linear fit calibration was performed using the g-channel intensity of channels 1-7 as the independent variable and the g-value of channel 8 as the dependent variable. The results show a good linear correlation and a good fit.

[0103] In this embodiment, if Figure 6As shown in Figure 2, in order to evaluate the sensitivity of the portable fluorescence detector, FITC was used as a standard fluorophore and tested at different concentrations (0-10 μM) and compared with a commercial device microplate reader (Synergy H1, Xinling Bio). The fluorescence photos taken by the fiber optic fluorescence sensing detection system proposed in this embodiment are shown in Figure 2. Figure 6 As shown in (a), as the FITC concentration decreases, the fluorescence image gradually changes from a brighter color (green in practice) to a darker color (blue in practice), and the fluorescence signal gradually weakens. Figure 6 As shown in (b), the horizontal axis is the concentration of FITC solution, and the vertical axis is the fluorescence intensity. Curve FL2 represents the test results of the fiber optic fluorescence sensing detection system, and curve FL1 is the test results of the commercial microplate reader. It can be seen from the figure that the trends of the test results of the two are consistent. In the low concentration range of FITC of 0-1μM, both devices have good linearity. 2 Indicates the degree of linear relationship between the dependent variable and the independent variable, R 2 The value range of is [0,1], and the closer to 1, the better the model fit. The sensitivity of the optical fiber fluorescence sensing detection system proposed in this embodiment is 19459μM -1 、R 2 =0.9902, the sensitivity of the microplate reader is 1774.8μM -1 、R 2 =0.9626. This indicates that in the low fluorometer intensity range, the sensitivity of the portable fiber-optic fluorescence sensing detection system is 11 times higher than that of the commercial microplate reader, and the linearity is better, reflecting the excellent performance of the fiber-optic fluorescence sensing detection system.

[0104] In this embodiment, if Figure 7 As shown in the figure, the stability of the portable fiber optic fluorescence sensing detection system was tested. Two FITC solutions with different concentrations (high concentration of 10μM and low concentration of 1nM) were tested continuously in real time at intervals of 15s for 1 hour. The fiber optic fluorescence images collected during the test are shown in the figure. Figure 7 (a) is shown. The test results are shown in Figure 7 (b) shows good stability.

[0105] In one possible implementation, Figure 8 As shown in the figure, the portable optical fiber fluorescence sensing detection system designed in this example can detect 10 1 -10 4 The African swine fever virus at 100 copies / μl was dynamically monitored and tested for linearity. Considering the interference of environmental factors during actual use, the experiment was conducted at two locations with different temperatures and humidity. Figure 8 .like Figure 8As shown in (a), different concentrations of African swine fever virus were monitored in real time at point A, with a temperature of 19°C and a humidity of 37.5%. The fluorescence intensity gradually increased with the increase of reaction time and reached a stable state within 20 minutes. Similarly, Figure 8 As shown in (b), real-time monitoring of African swine fever at different concentrations was performed at point B, with a temperature of 21.8°C and a humidity of 31.2%. The reaction rates at different concentrations were different, demonstrating strong distinguishability.

[0106] In one possible implementation, see Figure 9 ,like Figure 9 As shown in (a), nucleic acid fluorescence detection was performed on African swine fever virus at 5 different concentrations under two different environments, and 30 data were collected. The horizontal axis represents the nucleic acid concentration of African swine fever virus, and the vertical axis represents the fluorescence intensity. The test results show good linearity, R 2 The value is 0.9815. In addition, Figure 9 As shown in Figure (b), the proposed machine learning algorithm was trained on test data containing different concentrations and environments, achieving 100% concentration recognition accuracy. Experimental results show that the system can effectively identify African swine fever virus under different conditions and successfully avoid interference from environmental factors.

[0107] In one possible implementation, Figure 10 As shown in the figure, the portable optical fiber fluorescence sensing detection system designed in this example can detect the concentration of 10 1 -10 5 Dynamic monitoring and linearity testing of Salmonella CFU / ml were performed. CFU / ml indicates the number of single cells per milliliter of bacterial solution. Real-time monitoring of different concentrations of Salmonella was performed to verify the dynamic testing function of the designed detection system. The fluorescence intensity gradually increased with the increase of reaction time, and the reaction rates between different concentrations were different, with strong distinguishability. 30 tests were conducted on 6 different concentrations of Salmonella. For details, please see Figure 11 , data such as Figure 11 As shown in (a), the horizontal axis is the concentration of Salmonella nucleic acid, and the vertical axis is the fluorescence intensity. The fluorescence intensity images of different concentrations of Salmonella are clearly different, and the test results have good linearity, R 2 =0.9808. Figure 11 As shown in (b), the proposed machine learning algorithm was trained on Salmonella data of different concentrations, achieving concentration recognition with an accuracy of 100%.

[0108] Although the preferred embodiments of the present application have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present application.

[0109] Obviously, the above embodiments are merely examples for clarity of explanation and are not intended to limit the implementation methods. Those skilled in the art will appreciate that other variations or modifications can be made based on the above description. It is not necessary and impossible to enumerate all implementation methods here. Obvious variations or modifications arising therefrom remain within the scope of protection of the present invention.

Claims

1. A fiber optic fluorescence sensing system for food pathogen detection, characterized by: It includes a sensing module and a detection module. The sensing module includes an optical fiber fluorescence sensing module and an environmental sensing module. The fiber optic fluorescence sensing module includes an optical component and a fluorescence reaction module, wherein the optical component includes an optical fiber array, a filter, a collimating lens, and an image sensor arranged in sequence; the fluorescence reaction module includes an excitation light source and a multi-channel fluorescence reaction tank; the excitation light source is used to irradiate the solution in the centrifuge tube in the multi-channel fluorescence reaction tank to excite the fluorescence signal, and the solution in the centrifuge tube is the solution of the food sample to be detected; the fiber optic array is connected to the multi-channel fluorescence reaction tank for transmitting the fluorescence signal; the filter is used to select the fluorescence signal in the fiber optic array; the collimating lens is used to amplify and focus the fluorescence signal selected by the filter; and the image sensor is used to image the amplified and focused fluorescence signal to obtain a fluorescence intensity image; The environmental sensing module includes a geographic location sensor, a temperature sensor, and a humidity sensor, which are used to collect geographic location data, temperature data, and humidity data respectively; The detection module is used to obtain the fluorescence intensity image obtained by the image sensor, as well as the geographical location, temperature data and humidity data collected by the geographical location sensor, temperature sensor and humidity sensor respectively, and perform pathogen concentration detection on the solution of the food sample based on the obtained data; The detection module uses the collected fluorescence image and temperature data, humidity data, and geographic location data as inputs to the THPIRBFNN model; The activation function of the hidden layer of the THPIRBFNN model is a Gaussian radial basis function of the environmental parameter matrix, which includes temperature data, humidity data, and geographic location data. The expression is as follows: Among them, Y THPI is the Gaussian radial basis function of the environment parameter matrix, represents the mass center of the jth neuron, σ represents the width of the Gaussian kernel; the input data is represented by the environment parameter matrix [T,H,P,I1,I2,I3…I N ] indicates that T is the ambient temperature, H is the ambient humidity, P is the geographical location information, I1, I2, I3…I N represents the fluorescence intensity of multi-channel detection; The detection module further includes an optimization unit, wherein the optimization unit is used to optimize the linear weights of the output layer neurons and the center point parameters of the hidden layer neurons of the THPIRBFNN model; The linear weights of the output layer neurons of the THPIRBFNN model are updated according to the following formula: |G k+1 |=|G k |+k G ×|η·ΔG| where, |G k+1 | and |G k | are the weights before and after update respectively, k G is the influence coefficient of the environmental parameter matrix on weight optimization, η is the learning rate, and E is the error function to be minimized; the output states represent different concentration levels or classification results, and m is the total number of these possible output states; represents the predicted output value of a certain input sample when the model output state is i, and Z i represents the actual output value or true concentration corresponding to this sample; The center points of the hidden layer neurons of the THPIRBFNN model are updated according to the following formula: |c k+1 |=|c k |+k c ×|η·Δc| Among them, k c is the influence coefficient of the environmental parameter matrix on the optimization of the hidden layer neuron center point, and the width of the Gaussian kernel is updated according to the following formula: |s k+1 |=|s k |+k σ ×|η·Δσ| Among them, k σ is the influence coefficient of the environmental parameter matrix on the Gaussian kernel width optimization.

2. The optical fiber fluorescence sensing detection system for food pathogens according to claim 1, characterized in that: To eliminate the influence of background light, before collecting the fluorescence intensity image, the detection module first collects the background image obtained by the image sensor when the excitation light source is not excited, and then performs differential processing on the fluorescence intensity image collected by the detection module after the excitation light source is excited and the background image to remove the interference of ambient background light.

3. The optical fiber fluorescence sensing detection system for food pathogens according to claim 1, characterized in that: Assume that there is a linear relationship between the fluorescence intensity I corresponding to the fluorescence intensity image and the sample concentration c, which can be expressed as: I=k·c+I0 Where k is the linear conversion coefficient from c to I; I0 is the initial fluorescence intensity when the sample concentration is 0.

4. The optical fiber fluorescence sensing detection system for food pathogens according to claim 1, characterized in that: In order to eliminate the influence of noise on the output of the image sensor, the real-time output value of the image sensor is subjected to mean filtering, and the formula is: Where, I=[I1,I2,…,I n ], n∈[1,N], N is the number of fiber array channels, and ΔT is the sampling period.

5. The optical fiber fluorescence sensing detection system for food pathogens according to claim 1, characterized in that: The THPIRBFNN model uses a preset loss function LOSS during training, which is expressed as follows: Where N is the total number of samples in the test set, is the concentration prediction state, and Z is the actual concentration of the sample.

6. The optical fiber fluorescence sensing detection system for food pathogens according to claim 1, characterized in that: The detection module also includes an evaluation unit, which is used to evaluate the accuracy rate ACC of the identification result of the sample concentration. The formula is: Among them, TP represents the number of true positive examples, FN represents the number of false negative examples, TN represents the number of true negative examples, and FP is the number of false positive examples.

7. The optical fiber fluorescence sensing detection system for food pathogens according to claim 1, characterized in that: The detection module further includes a temperature control module, which is in contact with the multi-channel fluorescence reaction tank to achieve temperature control of the solution in the centrifuge tube in the multi-channel fluorescence reaction tank.

Images