A method and system for early warning of cheese microbial contamination

Abstract

The present application relates to the technical field of intelligent detection, and discloses a cheese microbial contamination early warning method and system, the method comprising: gradient heating of a cheese to be tested, collecting volatile organic compounds at different temperature stages to obtain a gas sample of the cheese to be tested; using a multi-parameter sensor array to detect the gas sample to obtain sensor response data of the cheese to be tested; determining a weight coefficient set of the sensor in contamination index prediction according to real-time response intensity and historical data reliability of the sensor in the multi-parameter sensor array; data fusion of the sensor response data based on the weight coefficient set to obtain a microbial contamination index of the cheese to be tested; when the microbial contamination index exceeds a preset contamination threshold, generating a multi-level microbial contamination report of the cheese to be tested based on the microbial contamination index; the present application can improve the efficiency of cheese microbial contamination early warning.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent detection, in particular to a cheese microbial contamination early warning method and system. BACKGROUND

[0002] The prior art has significant deficiencies in the gas sample collection link of cheese microbial contamination early warning. The measured cheese is not collected in stages using a gradient heating method to collect volatile organic compounds, and only a single temperature heating is used to obtain a gas sample, which cannot comprehensively capture the characteristic gas components released at different temperature stages, resulting in the absence or low concentration of key pollution markers in the gas sample, which makes it difficult to support subsequent accurate detection. At the same time, the collected gas samples are not scientifically mixed, the gas component distribution is uneven, which further reduces the representativeness of the sample, and the quality of the basic sample provided for sensor detection is low, which directly affects the accuracy of the subsequent pollution index calculation.

[0003] The prior art has significant defects in the sensor detection, data fusion and early warning report generation links. In terms of sensor detection, the various sensors in the multi-parameter sensor array are not preheated and baseline calibrated, and the collected original sensor signals are not filtered and denoised, resulting in a large amount of noise interference in the sensor response data, which cannot accurately reflect the true characteristics of the gas sample; when determining the sensor weight, the real-time response intensity and historical data reliability are not quantitatively calculated, but only the subjective experience is used to allocate the weight, so that the role of high-reliability sensors in the data fusion process cannot be fully played, and the interference of low-reliability sensors cannot be excluded. In the data fusion and report generation link, a standardized microbial contamination index scale system is not established, and a neighboring grade weighting judgment strategy is not used for critical region data, but only a simple numerical comparison is used to determine the pollution level, resulting in a large deviation in the pollution index calculation; and a multi-level pollution report is not generated according to the pollution index matching the corresponding report template, the early warning result is presented in a single form, which cannot provide detailed and accurate decision-making basis for subsequent processing, and the overall early warning efficiency and accuracy cannot meet the actual demand. SUMMARY

[0004] The present application provides a cheese microbial contamination early warning method and system to solve the problems raised in the background art.

[0005] To achieve the above-mentioned purpose, the present application provides a cheese microbial contamination early warning method, comprising:

[0006] S1, gradient heating the measured cheese to collect volatile organic compounds at different temperature stages to obtain a gas sample of the measured cheese;

[0007] S2, detecting the gas sample using a multi-parameter sensor array to obtain sensor response data of the measured cheese;

[0008] S3, determining a weight coefficient set of the sensors in predicting the pollution index according to real-time response intensity and historical data reliability of the sensors in the multi-parameter sensor array;

[0009] S4, data fusion on the sensor response data based on the weight coefficient set to obtain the microbial pollution index of the cheese to be tested;

[0010] S5, when the microbial pollution index exceeds a preset pollution threshold, generating a multi-level microbial pollution report of the cheese to be tested based on the microbial pollution index.

[0011] In a preferred embodiment, the cheese to be tested is gradient heated, and volatile organic compounds at different temperature stages are collected to obtain a gas sample of the cheese to be tested, including:

[0012] A temperature sequence of gradient heating is configured, the temperature sequence including a low temperature stage, a medium temperature stage and a high temperature stage, and a temperature parameter set of the cheese to be tested is obtained;

[0013] According to the temperature parameter set, the cheese to be tested is heated in turn at the low temperature stage, the medium temperature stage and the high temperature stage;

[0014] During the heating stage, volatile organic compounds released by the cheese to be tested are captured to obtain a sub-stage gas sample of the cheese to be tested;

[0015] The sub-stage gas sample is mixed to obtain a gas sample of the cheese to be tested.

[0016] In a preferred embodiment, the gas sample is detected using a multi-parameter sensor array to obtain sensor response data of the cheese to be tested, including:

[0017] The metal oxide sensor, the electrochemical sensor and the optical sensor are preheated and baseline calibrated to obtain a multi-parameter sensor array;

[0018] The gas sample is uniformly introduced into a detection chamber of the multi-parameter sensor array, and the resistance change signal of the metal oxide sensor, the current response signal of the electrochemical sensor and the light intensity change signal of the optical sensor are synchronously collected to obtain original sensing signals of the multi-parameter sensor array;

[0019] The original sensing signals are filtered and denoised to obtain sensor response data of the cheese to be tested.

[0020] In a preferred embodiment, determining the weighting coefficient set of the sensors in pollution index prediction based on the real-time response intensity and historical data reliability of the sensors in the multi-parameter sensor array includes:

[0021] Obtain the detection accuracy data of the sensors in the multi-parameter sensor array in historical monitoring, and evaluate the reliability of the historical data of the sensors based on the accuracy data;

[0022] The amplitude of the sensor's response signal to the gas sample is analyzed during the current detection. The real-time response intensity is evaluated based on the degree of deviation of the signal amplitude from the baseline signal, and the real-time response intensity evaluation result of the sensor is obtained.

[0023] Combining the historical data reliability and the real-time response strength evaluation results, sensors with both high historical reliability and high real-time response strength are assigned higher weight coefficients, while sensors with low historical reliability or weak real-time response strength are assigned lower weight coefficients, thus obtaining the initial set of weight coefficients for the sensors.

[0024] The initial set of weight coefficients is normalized to obtain the set of weight coefficients for the sensor in pollution index prediction.

[0025] In a preferred embodiment, the formula for calculating the weight coefficients in the initial weight coefficient set is as follows:

[0026] ;

[0027] In the formula, For the first The weighting coefficients of each sensor For the first The historical reliability factor corresponding to the historical reliability of each sensor. For the first The real-time response intensity factor corresponding to the real-time response intensity evaluation result of each sensor. For the first The historical reliability factor corresponding to the historical reliability of each sensor. For the first The real-time response intensity factor corresponding to the real-time response intensity evaluation result of each sensor. The preset real-time response intensity adjustment index, This is the preset historical reliability adjustment index.

[0028] In a preferred embodiment, the step of data fusion based on the weighted coefficient set of the sensor response data to obtain the microbial contamination index of the cheese to be tested includes:

[0029] The weight coefficient set is used to weight and fuse the sensor response data, wherein the response data of a high-weight sensor has a dominant influence in the fusion process, and the response data of a low-weight sensor serves as an auxiliary reference to obtain a preliminary fusion result of the cheese to be tested;

[0030] A consistency test is performed on the preliminary fusion result, and when there is a significant difference between different sensor data, a data correction strategy based on weight confidence is used to adjust the preliminary fusion result to obtain standardized fusion data of the cheese to be tested;

[0031] The standardized fusion data is mapped to a microbial contamination index scale, and a microbial contamination index of the cheese to be tested is obtained through linear conversion.

[0032] In a preferred embodiment, the mapping of the standardized fusion data to the microbial contamination index scale and the linear conversion to obtain the microbial contamination index of the cheese to be tested include:

[0033] A microbial contamination index scale is established, each scale in the microbial contamination index scale corresponds to a specific numerical interval, and a hierarchical scale system of the microbial contamination index scale is obtained;

[0034] The standardized fusion data is matched with the hierarchical scale system, and the pollution grade to which the standardized fusion data belongs is determined by looking up a pre-established mapping relationship table to obtain a preliminary pollution grade determination result of the cheese to be tested;

[0035] When the standardized fusion data is in a critical region of two pollution grades, a neighboring grade weighted determination strategy is used to determine the grade of the cheese to obtain a final pollution grade of the cheese to be tested;

[0036] According to the value in the numerical interval corresponding to the final pollution grade and the preliminary pollution grade determination result, the specific value of the microbial contamination index is determined to obtain the microbial contamination index of the cheese to be tested.

[0037] In a preferred embodiment, when the standardized fusion data is in a critical region of two pollution grades, a neighboring grade weighted determination strategy is used to determine the grade of the cheese to obtain a final pollution grade of the cheese to be tested, which includes:

[0038] The position of the standardized fusion data on the hierarchical scale system is identified to obtain a critical region identification result of the standardized fusion data;

[0039] The historical data distribution characteristics of the two pollution grades adjacent to the critical region are obtained to obtain neighboring grade characteristic parameters of the standardized fusion data;

[0040] determine a grade matching weight of the cheese to be tested according to the matching degree of the standardized fusion data and the adjacent grade characteristic parameters;

[0041] determine a final pollution grade of the cheese to be tested based on the grade matching weight.

[0042] In a preferred embodiment, when the microbial pollution index exceeds the preset pollution threshold, the multi-grade microbial pollution report of the cheese to be tested is generated based on the microbial pollution index, including:

[0043] based on the microbial pollution index, a corresponding report template is selected from a pre-established report template library to obtain a selected template of the cheese to be tested;

[0044] the microbial pollution index and related detection parameters are filled into the selected template to generate the multi-grade microbial pollution report of the cheese to be tested.

[0045] In order to solve the above problems, the present application also provides a cheese microbial pollution early warning system, the system comprising:

[0046] a stage heating module for gradient heating of the cheese to be tested, collecting volatile organic compounds at different temperature stages to obtain a gas sample of the cheese to be tested;

[0047] a data acquisition module for detecting the gas sample using a multi-parameter sensor array to obtain sensor response data of the cheese to be tested;

[0048] a weight acquisition module for determining a weight coefficient set of the sensor in pollution index prediction according to real-time response intensity and historical data reliability of the sensor in the multi-parameter sensor array;

[0049] a data fusion module for data fusion of the sensor response data based on the weight coefficient set to obtain a microbial pollution index of the cheese to be tested;

[0050] a pollution determination module for generating a multi-grade microbial pollution report of the cheese to be tested based on the microbial pollution index when the microbial pollution index exceeds a preset pollution threshold.

[0051] Compared with the prior art, the present application has the following beneficial effects:

[0052] 1.The application configures a gradient heating temperature sequence containing low-temperature, medium-temperature and high-temperature stages by a stage heating module, heats the cheese to be tested in stages and captures the volatile organic compounds released in each stage, and obtains a gas sample with comprehensive ingredients and strong representation through mixing, thereby providing a high-quality basic sample for subsequent detection; a data acquisition module preheats and baseline calibrates metal oxide sensors, electrochemical sensors and optical sensors, constructs a multi-parameter sensor array, uniformly introduces the gas sample into a detection chamber to synchronously collect multiple types of original sensing signals, and then obtains accurate sensor response data through filtering and denoising, thereby greatly improving the reliability of the detection data and laying a foundation of accurate data for pollution index calculation.

[0053] 2.The weight acquisition module of the application evaluates the reliability of historical data in combination with the historical detection accuracy of the sensors, analyzes the real-time response strength according to the amplitude of the current response signal and the deviation degree from the baseline, calculates and normalizes the scientific weight coefficient set through a standardization formula, and ensures that the sensors with high reliability and high response strength play a leading role in data fusion; the data fusion module fuses the sensor response data based on the weight coefficient set, obtains standardized fusion data through consistency test and correction, and then accurately calculates the microbial pollution index through hierarchical scale system matching and critical region adjacent grade weighted judgment; the pollution determination module selects a corresponding template from a template library and fills in parameters to generate a multi-level pollution report when the index exceeds a threshold, thereby comprehensively improving the accuracy and practicality of cheese microbial pollution early warning and providing a clear decision basis for pollution control. BRIEF DESCRIPTION OF DRAWINGS

[0054] Figure 1 A flowchart of a cheese microbial pollution early warning method provided by an embodiment of the application is shown in the figure.

[0055] Figure 2 A functional module diagram of a cheese microbial pollution early warning system provided by an embodiment of the application is shown in the figure.

[0056] The implementation, functional features and advantages of the application will be further described with reference to the embodiments and the accompanying drawings. DETAILED DESCRIPTION

[0057] It should be understood that the specific embodiments described herein are only used to explain the application, and are not used to limit the application.

[0058] The embodiment of the present application provides a cheese microbial contamination early warning method. The execution subject of the cheese microbial contamination early warning method includes but is not limited to at least one of the electronic devices such as a server and a terminal which can be configured to execute the method provided by the embodiment of the present application. In other words, the cheese microbial contamination early warning method can be executed by the software or hardware installed in the terminal device or the server device. The server includes but is not limited to a single server, a server cluster, a cloud server or a cloud server cluster. The server can be a stand-alone server, or a cloud server providing cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content distribution networks (CDN), and basic cloud computing services such as big data and artificial intelligence platforms.

[0059] Referring to Figure 1 Fig. 1 shows a flowchart of a cheese microbial contamination early warning method provided by an embodiment of the present application. In the embodiment, the cheese microbial contamination early warning method includes the following steps.

[0060] S1, gradient heating is performed on a to-be-tested cheese, volatile organic compounds at different temperature stages are collected respectively, and a gas sample of the to-be-tested cheese is obtained;

[0061] In the embodiment of the present application, the gradient heating is performed on the to-be-tested cheese, the volatile organic compounds at different temperature stages are collected respectively, and the gas sample of the to-be-tested cheese is obtained, including the following steps.

[0062] A temperature sequence of gradient heating is configured, the temperature sequence includes a low-temperature stage, a medium-temperature stage and a high-temperature stage, and a temperature parameter set of the to-be-tested cheese is obtained;

[0063] According to the temperature parameter set, the to-be-tested cheese is sequentially heated at the low-temperature stage, the medium-temperature stage and the high-temperature stage;

[0064] During the heating stage, volatile organic compounds released by the to-be-tested cheese are captured, and a staged gas sub-sample of the to-be-tested cheese is obtained;

[0065] The staged gas sub-sample is mixed and processed, and a gas sample of the to-be-tested cheese is obtained.

[0066] Specifically, a temperature sequence of gradient heating is configured, and first, temperature intervals corresponding to the low-temperature stage, the medium-temperature stage and the high-temperature stage are determined, wherein the temperature interval of the low-temperature stage is set to a range capable of preliminarily stimulating the release of low-boiling-point volatile organic compounds in the cheese, the temperature interval of the medium-temperature stage is set to a range capable of promoting the release of medium-boiling-point volatile organic compounds in the cheese, and the temperature interval of the high-temperature stage is set to a range capable of promoting the release of high-boiling-point volatile organic compounds in the cheese. The three temperature intervals are integrated to form a temperature parameter set of the cheese to be tested.

[0067] Further, according to the three temperature intervals determined in the temperature parameter set, first, the cheese to be tested is placed in a heating device with precise temperature control function, the temperature of the device is adjusted to the temperature interval corresponding to the low-temperature stage, and the cheese is heated at this temperature until the release of volatile organic compounds in the cheese tends to be stable in this stage. After the low-temperature stage heating is completed, the temperature of the device is adjusted to the temperature interval corresponding to the medium-temperature stage, and the cheese is continuously heated in the same way until the release of volatile organic compounds in the cheese tends to be stable in this stage. Then, the temperature of the device is adjusted to the temperature interval corresponding to the high-temperature stage, and the cheese is continuously heated until the release of volatile organic compounds in the cheese tends to be stable in this stage, so as to ensure that the cheese to be tested completes the heating in the three temperature stages in turn.

[0068] Further, during the process of each heating stage, a gas collection device with adsorption function is connected to the gas discharge channel of the heating device. When the cheese to be tested is heated in the low-temperature stage to release volatile organic compounds, these compounds flow into the collection device with the gas and are adsorbed and retained by the adsorption material in the device to form a gas sub-sample corresponding to the low-temperature stage. After entering the medium-temperature stage, the volatile organic compounds released by the cheese also enter the collection device through the gas discharge channel and are captured by the adsorption material to form a gas sub-sample corresponding to the medium-temperature stage. After entering the high-temperature stage, the volatile organic compounds released in this stage are collected in the same way to form a gas sub-sample corresponding to the high-temperature stage. The three gas sub-samples from different heating stages together constitute a staged gas sub-sample of the cheese to be tested.

[0069] Further, the gas sub-samples from the low-temperature stage, the medium-temperature stage and the high-temperature stage in the staged gas sub-sample are transferred into a clean and dry sealed container respectively, so as to ensure that the gas sub-sample of each stage can be completely introduced into the container. Then, the container is placed on a shaking device and slowly shaken by the shaking device to fully mix the gas sub-samples of different stages in the container. After the shaking is completed, the mixed gas in the container is the gas sample of the cheese to be tested.

[0070] In general, the gradient temperature parameter set containing low, medium and high temperature stages can adapt to the release characteristics of different volatile organic compounds in cheese, avoid missing key pollution markers at a single temperature, and provide a comprehensive gas composition basis for detection.

[0071] In general, the cheese is heated in sequence according to the temperature parameter set to ensure that the volatile organic compounds are fully volatilized and not decomposed at each stage, which improves the stability and representativeness of the gas composition compared to disordered heating, and provides a reliable sample source for accurate detection.

[0072] In general, the sub-samples of the gas captured in stages can retain the characteristic components at each temperature, avoid interference between gases at different stages, solve the information loss problem caused by mixed collection, and provide independent data support for pollution characteristic analysis.

[0073] In general, the gas sample obtained by mixing the sub-samples in stages can comprehensively reflect the overall pollution status of the cheese and has uniform composition, reduces the detection deviation of the sensor, significantly improves the representativeness of the sample, and lays a high-quality foundation for subsequent detection.

[0074] S2, using a multi-parameter sensor array to detect the gas sample and obtaining sensor response data of the cheese to be tested;

[0075] In the embodiments of the present application, the use of a multi-parameter sensor array to detect the gas sample and obtain sensor response data of the cheese to be tested includes:

[0076] Preheat and baseline calibration are performed on the metal oxide sensor, the electrochemical sensor and the optical sensor to obtain a multi-parameter sensor array;

[0077] The gas sample is uniformly introduced into the detection chamber of the multi-parameter sensor array, and the resistance change signal of the metal oxide sensor, the current response signal of the electrochemical sensor and the light intensity change signal of the optical sensor are synchronously collected to obtain the original sensing signal of the multi-parameter sensor array;

[0078] The original sensing signal is filtered and denoised to obtain the sensor response data of the cheese to be tested.

[0079] Specifically, a stable metal oxide sensor, an electrochemical sensor and an optical sensor are selected, and the three sensors are installed on a detection support according to a preset array layout to form an initial sensor combination. The initial sensor combination is connected to a dedicated preheating device, the preheating device is turned on to make the sensors in a constant preheating environment, and the heating is continued until the output signals of the sensors are stable, so that the internal elements of the sensors reach an optimal working state. After preheating is completed, the sensor array is placed in a clean and interference-free gas environment, and the output signal values of each sensor at this time are recorded as reference signals. Subsequently, the internal parameters of the sensors are adjusted to make the output signals of the sensors in the absence of target gas consistent with the reference signals, baseline calibration is completed, and finally a multi-parameter sensor array is obtained.

[0080] Further, the prepared gas sample to be detected is loaded into a gas storage container with a controllable valve, and the container is connected to the inlet of the detection chamber of the multi-parameter sensor array through a gas guide pipe to ensure that the gas guide pipe and the interface are sealed and leak-free. The valve of the gas storage container is slowly opened, and the gas flow rate control device is adjusted to make the gas sample pass through the gas guide pipe into the detection chamber at a uniform and stable speed, so as to avoid affecting the detection results due to flow rate fluctuations. At the same time of contacting the gas sample with the sensors, a signal acquisition system is started, which establishes data connection with the metal oxide sensor, the electrochemical sensor and the optical sensor respectively, and acquires the resistance change signal of the metal oxide sensor caused by gas adsorption, the current response signal of the electrochemical sensor caused by electrochemical reaction, and the light intensity change signal of the optical sensor caused by gas absorption or scattering of light. The three types of signals collected are integrated and stored to obtain the original sensing signal of the multi-parameter sensor array.

[0081] Further, the original sensing signal of the multi-parameter sensor array is introduced into a signal processing system, and first, trend analysis is performed on the original sensing signal to identify irrelevant fluctuation components in the signal caused by factors such as equipment vibration and slight fluctuations in ambient temperature. For the identified irrelevant fluctuations, a special filtering processing method is used to filter and remove high-frequency noise and burst interference signals in the original sensing signal one by one, while the effective information related to the gas sample in the original sensing signal is always maintained. After filtering processing, the processed signal is subjected to smoothing verification to check whether the signal curve is continuous and smooth, so as to ensure that there is no residual noise affecting the accuracy of the signal, and finally the sensor response data of the cheese to be detected is obtained.

[0082] In summary, the preheating and baseline calibration of the three sensors obtain the multi-parameter sensor array, which can eliminate the instability and deviation of the initial signal, solve the problem of low detection reliability caused by uncalibration, and provide a stable hardware foundation for signal acquisition.

[0083] In general, the uniform introduction of the gas sample and the synchronous acquisition of the three types of signals obtain the original sensing signals, ensure sufficient contact between the sample and the sensor, cover multiple dimensions of signals, improve the comprehensiveness of the signals, and provide rich original data for data processing.

[0084] In general, filtering and denoising of the original signals obtain sensor response data, remove irrelevant noise, avoid signal distortion, solve the problem of large deviation of directly using the original signals, and provide accurate basis for subsequent weight calculation and data fusion.

[0085] S3, according to the real-time response intensity and historical data reliability of the sensors in the multi-parameter sensor array, determining the weight coefficient set of the sensors in the pollution index prediction;

[0086] In the embodiment of the application, according to the real-time response intensity and historical data reliability of the sensors in the multi-parameter sensor array, determining the weight coefficient set of the sensors in the pollution index prediction, comprising:

[0087] Obtaining the detection accuracy data of the sensors in the multi-parameter sensor array in historical monitoring, and evaluating the historical data reliability of the sensors based on the accuracy data;

[0088] Analyzing the response signal amplitude of the sensors to the gas sample in the current detection, evaluating the real-time response intensity according to the deviation degree of the signal amplitude and the baseline signal, and obtaining the real-time response intensity evaluation result of the sensors;

[0089] Combining the historical data reliability and the real-time response intensity evaluation result, assigning a higher weight coefficient to the sensors with high historical reliability and high real-time response intensity, and assigning a lower weight coefficient to the sensors with low historical reliability or weak real-time response intensity, to obtain the initial weight coefficient set of the sensors;

[0090] Normalizing the initial weight coefficient set to obtain the weight coefficient set of the sensors in the pollution index prediction.

[0091] The calculation formula of the weight coefficient in the initial weight coefficient set is as follows:

[0092] ;

[0093] In the formula, is the weight coefficient of the i-th sensor, is the historical reliability factor corresponding to the historical reliability in the i-th sensor, is the real-time response intensity factor corresponding to the real-time response intensity evaluation result in the i-th sensor, ​​​For the first The historical reliability factor corresponding to the historical reliability of each sensor. For the first The real-time response intensity factor corresponding to the real-time response intensity evaluation result of each sensor. The preset real-time response intensity adjustment index, This is the preset historical reliability adjustment index.

[0094] Specifically, the historical monitoring database of the multi-parameter sensor array is used to retrieve recorded data from all past monitoring tasks for each sensor. These records include the actual value of the target substance and the sensor's detected value for each monitoring session. For each sensor, the difference between each detected value and the actual value of the target substance is compared, and the number of times the detected value falls within the allowable error range is counted. This number is then divided by the total number of tests to obtain the detection accuracy data for each sensor. The reliability of the sensor's historical data is evaluated based on the detection accuracy data. If a sensor's detection accuracy data is high, it indicates that its past detection results deviate little from the actual situation, and the sensor's historical data is considered highly reliable. If the detection accuracy data is low, the sensor's historical data is considered unreliable. This process completes the evaluation of the reliability of historical data for all sensors.

[0095] Furthermore, during the current gas sample detection process, the response signal changes of each sensor in the multi-parameter sensor array are recorded in real time, while the baseline signal recorded by each sensor during the baseline calibration phase is retrieved. For each sensor, the amplitude of its current detection response signal is compared with the corresponding baseline signal, and the degree of deviation of the signal amplitude relative to the baseline signal is calculated. If the response signal amplitude of a sensor is much higher or much lower than the baseline signal, it indicates that the sensor is sensitive to the target component in the gas sample, with a large deviation, and its real-time response intensity is evaluated as high; if the response signal amplitude is close to the baseline signal, with a small deviation, its real-time response intensity is evaluated as weak, and the real-time response intensity evaluation result of each sensor is finally obtained.

[0096] Furthermore, a dual evaluation standard of historical data reliability and real-time response intensity is established. Historical data reliability is divided into three levels: high, medium, and low, and real-time response intensity is also divided into three levels: high, medium, and low. For sensors that are simultaneously at the high level of both historical data reliability and real-time response intensity, it indicates that their past performance has been stable and that they are currently sensitive to gas samples, so they are assigned a higher weight coefficient. For sensors with low historical data reliability or low real-time response intensity, regardless of the other evaluation result, they are determined to have low reference value in the current pollution index prediction, so they are assigned a lower weight coefficient. For sensors with medium historical data reliability and real-time response intensity, a medium weight coefficient is assigned based on the combination of the two evaluation results. The weight coefficients of all sensors are compiled and summarized to obtain the initial set of weight coefficients for the sensors.

[0097] Furthermore, the sum of all sensor weight coefficients in the initial weight coefficient set is calculated. For each sensor's initial weight coefficient, it is divided by the sum of all sensor weight coefficients. This calculation method readjusts the weight coefficients of all sensors to values ​​within the same range and sum to 1. During the calculation process, the weight coefficient calculation results for each sensor are checked one by one to ensure that the calculation is accurate and there are no numerical anomalies. After adjustment, the final weight coefficients of all sensors are organized into a standardized set, resulting in the weight coefficient set of sensors for pollution index prediction.

[0098] Specifically, no. The historical reliability factor of a sensor is derived from the analysis of its historical monitoring data. First, records of the detected values ​​and the actual values ​​of the target substance from all past monitoring tasks are retrieved. The number of times the detected values ​​fell within the allowable error range is counted, and the detection accuracy is calculated. Then, the corresponding historical reliability factor is determined based on the accuracy rate; a higher accuracy rate results in a larger historical reliability factor, and vice versa. The real-time response intensity factor of each sensor is derived from the analysis of the sensor's response signal to the gas sample during the current detection. The amplitude of the current response signal is compared with the baseline signal, and the real-time response intensity factor is determined based on the degree of deviation. A larger deviation results in a larger real-time response intensity factor value, and a smaller deviation results in a smaller real-time response intensity factor value. The historical reliability factor and real-time response strength factor of the first sensor are obtained in the same way as the first sensor. The preset historical reliability adjustment index is a fixed value preset according to the importance of the historical performance of the sensor in the actual application scene. If more importance is attached to the past stable performance of the sensor, the adjustment index value is set to be larger. If the importance of the historical performance is lower, the adjustment index value is set to be smaller. The preset real-time response intensity adjustment index is also preset according to the importance of the current response sensitivity of the sensor in the actual application scene. If more attention is paid to the current reaction sensitivity of the sensor to the gas sample, the adjustment index value is set to be larger. If the importance of the current response sensitivity is lower, the adjustment index value is set to be smaller.

[0099] Further, the significance of the formula is that by combining the historical reliability and real-time response intensity of the sensor, the weight coefficient of each sensor in the pollution index prediction is scientifically calculated. First, the historical reliability factor of each sensor is combined with the historical reliability adjustment index, and the real-time response intensity factor is combined with the real-time response intensity adjustment index. Then, the results of the two are multiplied to obtain a comprehensive value that can comprehensively reflect the historical performance and current response ability of the sensor. Subsequently, the sum of the comprehensive values of all sensors is calculated, and the comprehensive value of a single sensor is divided by the sum of the comprehensive values of all sensors to obtain the weight coefficient of the sensor. The weight coefficient can accurately reflect the reference value of the sensor in the pollution index prediction. The sensor with high historical reliability and strong real-time response intensity has a larger weight coefficient and plays a more important role in the prediction.

[0100] Further, the trend of the formula is that when the historical reliability factor of a sensor increases, the comprehensive value of the sensor will increase, and the weight coefficient will also increase if other parameters remain unchanged. Conversely, when the historical reliability factor decreases, the weight coefficient will decrease. When the real-time response intensity factor of the sensor increases, the comprehensive value will increase, and the weight coefficient will also increase if other parameters remain unchanged. When the real-time response intensity factor decreases, the weight coefficient will decrease.

[0101] Further, if the preset historical reliability adjustment index increases, the influence of the historical reliability factor on the comprehensive value will be enhanced. If the historical reliability factor of the sensor is originally large, the weight coefficient will further increase. If the historical reliability factor is originally small, the weight coefficient will further decrease. If the preset real-time response intensity adjustment index increases, the influence of the real-time response intensity factor on the comprehensive value will be enhanced. The weight coefficient of the sensor with a large real-time response intensity factor will be larger, and the weight coefficient of the sensor with a small real-time response intensity factor will be smaller.

[0102] Furthermore, when the combined values ​​of other sensors change, if the combined value of a certain sensor remains unchanged while the combined values ​​of other sensors increase, the proportion of the combined value of that sensor decreases and its weighting coefficient decreases; if the combined values ​​of other sensors decrease, the proportion of the combined value of that sensor increases and its weighting coefficient increases.

[0103] In summary, obtaining historical detection accuracy data from sensors can assess reliability, allowing for the selection of reliable sensors based on past performance, avoiding interference from historical biases, and providing an objective basis for weight allocation.

[0104] In summary, analyzing the deviation of the current response signal from the baseline to assess the real-time intensity can determine the sensor's sensitivity to the current sample, ensure that the weights match the current detection capability, and prevent highly sensitive sensors from failing to perform their intended function.

[0105] In summary, by combining the weights of the two evaluations to obtain the initial set, high-reliability and high-response sensors are given higher weights, low-quality data interference is reduced, and the problem of weight and performance being disconnected is solved.

[0106] In summary, normalization of the initial set yields a weight set, which unifies the weight range and clarifies the proportions, providing a standardized basis for subsequent data weighting and fusion and ensuring the accuracy of the fusion.

[0107] In summary, this weighting coefficient calculation formula can comprehensively calculate the weights based on the sensor's historical reliability and real-time response intensity. By using historical reliability factors and real-time response intensity factors, it correlates the sensor's past stable performance with its current detection sensitivity, avoiding weight bias caused by relying solely on historical or real-time data. It also solves the problem of the single dimension of weight allocation in existing technologies, making the weights more closely aligned with the sensor's overall performance.

[0108] In summary, the historical reliability adjustment index and real-time response intensity adjustment index preset in the formula can be flexibly adjusted according to the importance attached to historical performance and real-time performance in the actual detection scenario. This ensures that the weight allocation can adapt to different detection needs, avoids the problem that fixed weight logic cannot cope with diverse scenarios, and improves the flexibility and applicability of weight calculation.

[0109] In summary, by calculating the overall performance value of a single sensor in the numerator and the sum of the overall performance values ​​of all sensors in the denominator, and then determining the weight of each sensor by the ratio of the two, the weight of each sensor can be positively correlated with its own overall performance. Sensors with high overall performance have higher weights, and sensors with low overall performance have lower weights, thus avoiding the problem of performance differences and weight mismatch and ensuring the fairness and scientific nature of weight allocation.

[0110] In summary, the final weighting coefficients can be directly used for the weighted fusion of subsequent sensor response data without additional complex processing. Furthermore, the total weights are fixed, ensuring that the contribution ratio of each sensor data is clear and quantifiable during the data fusion process. This avoids confusion in fusion calculations caused by inconsistent weighting formats, provides a standardized and reliable weighting basis for accurate data fusion, and improves the accuracy of microbial contamination index calculation.

[0111] S4. Based on the weighted coefficient set, perform data fusion on the sensor response data to obtain the microbial contamination index of the cheese to be tested;

[0112] In this embodiment of the invention, the step of performing data fusion on the sensor response data based on the weighted coefficient set to obtain the microbial contamination index of the cheese to be tested includes:

[0113] The sensor response data are weighted and fused using the set of weight coefficients, wherein the response data of high-weight sensors have a dominant influence in the fusion process, and the response data of low-weight sensors serve as an auxiliary reference, thereby obtaining the preliminary fusion result of the cheese to be tested.

[0114] The initial fusion results are subjected to a consistency test. When there are significant differences between the data from different sensors, the initial fusion results are adjusted based on a data correction strategy with weighted confidence to obtain the standardized fusion data of the cheese to be tested.

[0115] The standardized fusion data is mapped to a microbial contamination index scale, and the microbial contamination index of the cheese to be tested is obtained through linear transformation.

[0116] The process of mapping the standardized fused data to a microbial contamination index scale and obtaining the microbial contamination index of the cheese to be tested through linear transformation includes:

[0117] A microbial contamination index scale is established, wherein each scale in the microbial contamination index scale corresponds to a specific numerical range, thereby obtaining a hierarchical scaling system for the microbial contamination index scale;

[0118] The standardized fusion data is matched with the grading scale system, and the pollution level of the standardized fusion data is determined by looking up the pre-established mapping relationship table to obtain the preliminary pollution level judgment result of the cheese to be tested.

[0119] When the standardized fusion data is in the critical region of two pollution levels, a neighboring level weighted judgment strategy is adopted to determine the level of the cheese and obtain the final pollution level of the cheese to be tested.

[0120] Based on the median value of the numerical range corresponding to the final contamination level and the preliminary contamination level determination results, the specific value of the microbial contamination index is determined, and the microbial contamination index of the cheese to be tested is obtained.

[0121] When the standardized fused data is in a critical region between two contamination levels, a neighboring level weighted judgment strategy is used to determine the contamination level of the cheese, thereby obtaining the final contamination level of the cheese to be tested, including:

[0122] Identify the position of the standardized fused data on the hierarchical scaling system, and obtain the critical region identification result of the standardized fused data;

[0123] By acquiring the historical data distribution characteristics of two pollution levels adjacent to the critical area, the adjacent level characteristic parameters of the standardized fused data are obtained.

[0124] The matching weight of the grade of the cheese to be tested is determined based on the matching degree between the standardized fusion data and the adjacent grade feature parameters.

[0125] Based on the grade matching weights, the final contamination grade of the cheese to be tested is determined.

[0126] Specifically, the weighting coefficients of each sensor are correlated with the corresponding sensor response data. For sensors with high weighting coefficients, their response data is given a greater weight during data fusion. Specifically, the response data of such sensors is included in the fusion calculation at a higher proportion, so that the signal changes detected in real time occupy the main part in the final result. For sensors with low weighting coefficients, their response data is included in the fusion calculation at a lower proportion, serving as auxiliary reference information to supplement details not covered by the data of high-weighted sensors. In this way, the response data of all sensors are integrated together to obtain the preliminary fusion result of the cheese to be tested.

[0127] Furthermore, the preliminary fusion results are compared and analyzed with the response data of each sensor in the multi-parameter sensor array. The degree of difference between the response data of different sensors is checked one by one. If the response data of some sensors deviates significantly from the preliminary fusion results, and this deviation exceeds the normal detection error range, it is determined that there is a significant difference between the data of different sensors. At this time, a data correction strategy based on weighted confidence is initiated to re-verify the response data of high-weight sensors to confirm whether their abnormality is caused by sudden interference. If the data of high-weight sensors is normal, their data is used as a benchmark to appropriately reduce the proportion of low-weight sensor data with large differences in the fusion results. If the data of high-weight sensors is abnormal, the preliminary fusion results are adjusted with reference to the consistent response data of multiple second-highest weighted sensors. After correction, the standardized fusion data of the cheese to be tested is obtained.

[0128] Furthermore, a microbial contamination index scale is pre-established, covering a complete range from no microbial contamination to severe microbial contamination. Each level corresponds to a specific numerical range, and each numerical range corresponds one-to-one with the possible value range of the standardized fusion data. The standardized fusion data is matched with the microbial contamination index scale to find the corresponding numerical range within the scale. Then, by proportionally converting the actual values ​​of the standardized fusion data to the corresponding numerical range, a specific value within the scale is obtained. This value is the microbial contamination index of the cheese to be tested.

[0129] Specifically, we first define the different levels of microbial contamination, dividing them into multiple levels from no contamination to extremely severe contamination. Each level corresponds to a unique contamination level name, such as "no contamination," "slight contamination," "moderate contamination," "severe contamination," and "extremely severe contamination." For each contamination level name, we set a specific numerical range. The lower and upper limits of the numerical range are determined based on a large amount of experimental data on microbial contamination in cheese, ensuring that each range accurately corresponds to the actual degree of microbial contamination. All contamination level names and their corresponding specific numerical ranges are organized into a structured system, which is the hierarchical scaling system of the microbial contamination index scale.

[0130] Furthermore, numerical intervals corresponding to all pollution levels are extracted from the grading scale system, and a mapping table between standardized fused data and numerical intervals is established. This table explicitly records the pollution level name corresponding to each numerical interval. The standardized fused data of the cheese to be tested is then substituted into this mapping table, and each standardized fused data is compared with the lower and upper limits of each numerical interval to determine which interval the data falls into. If the standardized fused data falls between the lower and upper limits of a certain numerical interval, the pollution level name corresponding to that interval is determined as the pollution level to which the standardized fused data belongs. This yields the preliminary pollution level determination result for the cheese to be tested.

[0131] Furthermore, a critical region corresponding to the numerical intervals of two adjacent pollution levels in the grading scale system is pre-determined. The critical region extends from near the upper limit of the numerical interval of the preceding level to near the lower limit of the numerical interval of the following level. This range is verified through multiple experiments to ensure accurate coverage of the transition between the two level numerical intervals. When the value of the standardized fused data falls within this critical region, a neighboring level weighted judgment strategy is activated. First, the two neighboring pollution levels corresponding to the critical region are determined. Then, weights are assigned based on the distance of the standardized fused data from the center of the numerical intervals of the two neighboring levels. The closer the data is to the center of a certain level's numerical interval, the greater its weight. The weights of the standardized fused data and the two neighboring levels are combined for calculation. The level with the larger weight is the final pollution level to which the standardized fused data is assigned, thus obtaining the final pollution level of the cheese to be tested.

[0132] Furthermore, the numerical range corresponding to the preliminary contamination level determination result is located in the grading scale system, and the median of this range is calculated by adding the upper and lower limits of the range and taking the average. This average value is the median of the numerical range corresponding to the preliminary contamination level determination result. If the final contamination level matches the preliminary contamination level determination result, the median of the numerical range corresponding to the preliminary contamination level determination result is directly used as the specific value of the microbial contamination index. If the final contamination level does not match the preliminary contamination level determination result, the numerical range corresponding to the final contamination level is located, and its median is calculated. This median value is used as the specific value of the microbial contamination index of the cheese to be tested.

[0133] Specifically, the critical region range between all adjacent pollution levels is extracted from the grading scaling system. Each critical region range is a specific range extending from the upper limit of the numerical range of the previous pollution level to the lower limit of the numerical range of the next pollution level. This specific range is determined based on the data distribution at the boundary between the two levels in historical monitoring. The standardized fused data is compared with these critical region ranges. If the value of the standardized fused data falls within a certain critical region range, it is determined that the standardized fused data is in that critical region, and the names of the two adjacent pollution levels corresponding to the critical region are recorded, forming the critical region identification result of the standardized fused data.

[0134] Furthermore, all historical monitoring data corresponding to the two adjacent pollution levels in the critical area identification results are retrieved from the historical monitoring database. This data includes standardized fusion data of all cheese samples previously identified as belonging to these two pollution levels. Statistical analysis is performed on the historical monitoring data for each pollution level, calculating the most frequent values ​​in the dataset, the concentration range of the data distribution, and the average amplitude of data fluctuations. These statistical results collectively constitute the historical data distribution characteristics of each pollution level. By integrating the historical data distribution characteristics of two adjacent pollution levels, the adjacent level characteristic parameters of the standardized fusion data are obtained.

[0135] Furthermore, the standardized fused data is compared with the historical data distribution characteristics of two pollution levels in the adjacent level feature parameters. First, it is determined whether the standardized fused data falls within the concentrated range of historical data for a certain pollution level. If it does, the base score for matching that level is higher. If it does not, the difference between the standardized fused data and the most frequently occurring value in the historical data for that level is calculated; the smaller the difference, the higher the base score for matching. Then, the base score is adjusted based on the average fluctuation range of the historical data for that pollution level. If the difference between the standardized fused data and that level is within the average range, the score is appropriately increased. Finally, the matching degree between the standardized fused data and each adjacent pollution level is obtained, and this matching degree is the level matching weight of the cheese to be tested.

[0136] Furthermore, the matching weights of the standardized fused data with two adjacent contamination levels are compared. If the matching weight of one contamination level is greater than that of the other, the contamination level with the greater weight is determined as the final contamination level of the cheese to be tested. If the matching weights of the two contamination levels are equal, the contamination level corresponding to the majority of the final classification results of the data within this critical area in historical testing is referred to as the final contamination level of the cheese to be tested.

[0137] In summary, weighted fusion of sensor response data based on a set of weighted coefficients allows high-weighted sensor data to dominate the fusion process while low-weighted data serves only as an auxiliary component. This approach highlights the effective information from sensors with high reliability and high response intensity, reduces the interference of low-quality data on the fusion results, and addresses the problem of existing technologies failing to differentiate sensor performance during data fusion, which leads to the obscuring of effective information. This provides preliminary fusion results that are more closely aligned with the actual pollution situation for subsequent processing.

[0138] In summary, performing consistency checks on the preliminary fusion results and adjusting the reliability by weighting the data when significant differences exist can promptly identify and correct biases caused by outliers. This involves adjusting low-weight outliers based on high-weight sensor data or correcting high-weight outliers by referring to consistent data from the second-highest weight sensor. This avoids distortion of the fusion results due to individual sensor failures or interference, ensuring that the final standardized fusion data is accurate and reliable.

[0139] In summary, mapping standardized fusion data to a contamination index scale and obtaining a microbial contamination index through linear transformation can transform abstract fusion data into intuitive and quantifiable contamination level indicators. Furthermore, the linear transformation ensures a stable correspondence between the data and the contamination index, solving the problem of ambiguous contamination level determination caused by the lack of standardized scales in existing technologies. This allows testing personnel to clearly understand the microbial contamination status of cheese, providing a clear basis for subsequent contamination early warning and control.

[0140] In summary, establishing microbial contamination index scales with each scale corresponding to a specific numerical range, forming a hierarchical scaling system, can provide clear criteria for standardized data integration, avoid the problem of chaotic contamination level determination caused by the lack of a unified scale in existing technologies, make the classification of contamination levels more standardized and operable, and lay a clear standard foundation for subsequent level matching.

[0141] In summary, by matching standardized and integrated data with a grading scale system and determining the pollution level through a mapping table, a preliminary judgment result can be obtained. This can quickly link data with pollution levels, reduce subjective errors in manual judgment, solve the problems of low efficiency and poor accuracy of existing technologies that rely on experience, and improve the efficiency and accuracy of preliminary pollution level judgment.

[0142] In summary, when data is in a critical region, using a neighboring grade weighted judgment strategy to determine the final pollution level can avoid judgment bias caused by the ambiguity of the critical data's classification. By combining neighboring grade characteristics with data matching degree, the level can be accurately classified, solving the problem that simple numerical comparison in existing technologies cannot handle critical situations, and ensuring the rigor of pollution level determination.

[0143] In summary, determining the specific value of the microbial contamination index based on the median of the numerical range corresponding to the final pollution level and the preliminary judgment result can transform the abstract level judgment into an intuitive and quantifiable index, avoiding the problem of vague description of the degree of pollution. At the same time, it ensures the stability of the correspondence between the index and the level, providing accurate quantitative basis for subsequent pollution early warning and report generation, and improving the practicality of the overall early warning.

[0144] In summary, identifying the position of standardized fusion data in the hierarchical scaling system to obtain critical area identification results can accurately locate whether the data is in the transition zone between two pollution levels, avoiding the problem of critical situations being misjudged as a single level due to unclear data location. This provides an accurate basis for subsequent targeted weighted judgment strategies, ensuring that critical data is not missed or misjudged.

[0145] In summary, obtaining the historical data distribution characteristics of two adjacent pollution levels in a critical area to obtain the characteristic parameters of adjacent levels can extract the typical characteristics of each level based on a large amount of past detection data. This provides an objective reference for determining the level of the current critical data, solves the problem that existing technologies lack historical data support and rely solely on subjective judgment when processing critical data, and improves the scientific nature of level determination.

[0146] In summary, determining the level matching weight based on the matching degree between standardized fused data and adjacent level feature parameters can quantify the weight by the degree of fit between data and level features, allowing levels with higher matching degrees to receive higher weights. This avoids the judgment bias caused by indiscriminate treatment of adjacent levels, solves the problem of arbitrary determination of critical data attribution in existing technologies, ensures that the weight allocation is consistent with the actual characteristics of the data, and provides a reasonable basis for the final level determination.

[0147] In summary, determining the final contamination level based on the matching weights of the levels allows for direct identification of data attribution based on the weights – the level with the higher weight is the final attribution level. If the weights are equal, the majority attribution results of historical critical data are referenced. This avoids the problem of critical data remaining in a “fuzzy zone” for a long time and failing to be accurately classified, ensuring that each critical data point can receive a rigorous and accurate level determination, and further improving the overall accuracy of microbial contamination level determination.

[0148] S5. When the microbial contamination index exceeds the preset contamination threshold, a multi-level microbial contamination report of the cheese to be tested is generated based on the microbial contamination index.

[0149] In this embodiment of the invention, the step of generating a multi-level microbial contamination report for the cheese under test based on the microbial contamination index when the microbial contamination index exceeds a preset contamination threshold includes:

[0150] Based on the microbial contamination index, a corresponding report template is selected from a pre-established report template library to obtain the selected template for the cheese to be tested;

[0151] The microbial contamination index and related detection parameters are filled into the selected template to generate a multi-level microbial contamination report for the cheese to be tested.

[0152] Specifically, the pre-established report template library contains multiple report templates, each corresponding to a different range of microbial contamination indices. The template content is set with different levels of detail based on the degree of contamination. Templates for low-contamination ranges include basic contamination information and simple suggestions, while templates for medium- and high-contamination ranges add risk analysis and specific treatment plans. The microbial contamination index of the cheese to be tested is compared with the index ranges corresponding to each template in the library. The range in which the index falls is then identified, and the corresponding report template is retrieved. This template is the selected template for the cheese to be tested.

[0153] Furthermore, parameters related to the testing of the cheese are collected, including testing time, composition of the multi-parameter sensor array used in the test, number and type of metal oxide sensors, type of electrochemical sensors, model of optical sensors, gas sample preparation process record, stage division of gradient heating, heating environment of each stage, processing method of sensor response data, and instructions for generating standardized fusion data. Following the format requirements of the selected template, the microbial contamination index is filled in the designated location on the template, and the collected relevant detection parameters are then filled in the corresponding parameter columns of the template. It is ensured that the description of each parameter accurately reflects the actual testing situation. After completion, the template content is checked to confirm there are no omissions or errors, and finally, a multi-level microbial contamination report for the cheese is generated.

[0154] In summary, selecting the corresponding report template from a pre-established template library based on the microbial contamination index allows for matching appropriate templates to different levels of contamination—basic information templates for low contamination indices and detailed templates containing risk analysis and treatment recommendations for medium to high contamination indices. This avoids the problems of existing technology-based early warning reports being too simplistic in form and having inappropriate levels of detail, ensuring that the report content accurately matches the severity of contamination and improving the report's relevance.

[0155] In summary, by filling in the microbial contamination index and related detection parameters into the selected template to generate a multi-level report, the core detection results and key process information can be fully presented. This solves the problem that existing technical reports lack process data and only present results. The report contains both clear contamination conclusions and traceable detection evidence, providing comprehensive information support for subsequent pollution control and cause analysis, and improving the practicality and credibility of early warning reports.

[0156] like Figure 2 The diagram shown is a functional block diagram of a cheese microbial contamination early warning system provided in an embodiment of the present invention.

[0157] The cheese microbial contamination early warning system 100 of this invention can be installed in an electronic device. Depending on the functions implemented, the cheese microbial contamination early warning system 100 may include a stage heating module 101, a data acquisition module 102, a weight acquisition module 103, a data fusion module 104, and a contamination determination module 105. The module described in this invention can also be referred to as a unit, which refers to a series of computer program segments that can be executed by the processor of an electronic device and can perform a fixed function, and are stored in the memory of the electronic device.

[0158] In this embodiment, the functions of each module / unit are as follows:

[0159] The stage heating module 101 is used to perform gradient heating on the cheese to be tested, and collect volatile organic compounds at different temperature stages to obtain a gas sample of the cheese to be tested.

[0160] The data acquisition module 102 is used to detect the gas sample using a multi-parameter sensor array to obtain the sensor response data of the cheese to be tested.

[0161] The weight acquisition module 103 is used to determine the weight coefficient set of the sensor in pollution index prediction based on the real-time response intensity and historical data reliability of the sensors in the multi-parameter sensor array.

[0162] The data fusion module 104 is used to perform data fusion on the sensor response data based on the weight coefficient set to obtain the microbial contamination index of the cheese to be tested.

[0163] The contamination determination module 105 is used to generate a multi-level microbial contamination report of the cheese to be tested based on the microbial contamination index when the microbial contamination index exceeds a preset contamination threshold.

[0164] In the several embodiments provided by this invention, it should be understood that the disclosed methods and systems can be implemented in other ways. For example, the system embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and other division methods may be used in actual implementation.

[0165] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0166] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in the form of hardware plus software functional modules.

[0167] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0168] The embodiments of this application can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0169] Finally, it should be noted that 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 preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for early warning of microbial contamination of cheese, characterized in that, The method comprises: S1, gradient heating is performed on the to-be-tested cheese, volatile organic compounds at different temperature stages are collected respectively, and a gas sample of the to-be-tested cheese is obtained; S2, the gas sample is detected using a multi-parameter sensor array, and sensor response data of the to-be-tested cheese is obtained; S3, according to real-time response intensity and historical data reliability of sensors in the multi-parameter sensor array, a weight coefficient set of the sensors in pollution index prediction is determined, comprising: obtaining detection accuracy data of sensors in the multi-parameter sensor array in historical monitoring, and evaluating historical data reliability of the sensors based on the accuracy data; analyzing response signal amplitudes of the sensors to the gas sample in the current detection, evaluating real-time response intensity according to deviation degrees of the signal amplitudes from baseline signals, and obtaining real-time response intensity evaluation results of the sensors; combining the historical data reliability and the real-time response intensity evaluation results, assigning higher weight coefficients to sensors that have both high historical reliability and high real-time response intensity, and assigning lower weight coefficients to sensors that have low historical reliability or weak real-time response intensity, to obtain an initial weight coefficient set of the sensors, wherein a calculation formula of the weight coefficients in the initial weight coefficient set is as follows: ； In the formula, For the first The weighting coefficients of each sensor For the first The historical reliability factor corresponding to the historical reliability of each sensor For the first The real-time response intensity factor corresponding to the real-time response intensity evaluation result of each sensor For the first The historical reliability factor corresponding to the historical reliability of each sensor For the first The real-time response intensity factor corresponding to the real-time response intensity evaluation result of each sensor The preset real-time response intensity adjustment index, The preset historical reliability adjustment index; normalizing the initial weight coefficient set to obtain the weight coefficient set of the sensors in pollution index prediction; S4, data fusion is performed on the sensor response data based on the weight coefficient set, to obtain a microbial pollution index of the to-be-tested cheese, comprising: performing weighted fusion processing on the sensor response data by using the weight coefficient set, wherein response data of high-weight sensors dominates the fusion process, and response data of low-weight sensors serves as auxiliary reference, to obtain a preliminary fusion result of the to-be-tested cheese; performing consistency test on the preliminary fusion result, when there is a significant difference between data of different sensors, adjusting the preliminary fusion result based on a data correction strategy of weight confidence, to obtain standardized fusion data of the to-be-tested cheese; mapping the standardized fusion data to a microbial pollution index scale, and obtaining the microbial pollution index of the to-be-tested cheese through linear conversion; S5, when the microbial pollution index exceeds a preset pollution threshold, generating a multi-level microbial pollution report of the to-be-tested cheese based on the microbial pollution index.

2. A method of early warning of microbial contamination of cheese as claimed in claim 1, wherein, The gradient heating of the to-be-tested cheese to collect volatile organic compounds at different temperature stages and obtain a gas sample of the to-be-tested cheese comprises: configuring a temperature sequence of gradient heating, the temperature sequence comprising a low-temperature stage, a medium-temperature stage and a high-temperature stage, and obtaining a temperature parameter set of the to-be-tested cheese; heating the to-be-tested cheese in the low-temperature stage, the medium-temperature stage and the high-temperature stage in sequence according to the temperature parameter set; during the heating stage, obtaining sub-stage gas samples of the to-be-tested cheese by capturing volatile organic compounds released by the to-be-tested cheese; mixing the sub-stage gas samples to obtain the gas sample of the to-be-tested cheese.

3. A method of detecting early stages of microbial contamination of cheese as claimed in claim 1, wherein, The detecting the gas sample by using the multi-parameter sensor array obtains sensor response data of the cheese to be tested, including: The metal oxide sensor, the electrochemical sensor and the optical sensor are preheated and baseline calibrated to obtain a multi-parameter sensor array; The gas sample is uniformly introduced into the detection chamber of the multi-parameter sensor array, and the resistance change signal of the metal oxide sensor, the current response signal of the electrochemical sensor and the light intensity change signal of the optical sensor are synchronously collected to obtain the original sensing signal of the multi-parameter sensor array; The original sensing signal is filtered and denoised to obtain the sensor response data of the cheese to be tested.

4. A method of detecting contamination of cheese by microorganisms as claimed in claim 1, wherein, The standardized fusion data is mapped to a microbial pollution index scale, and the microbial pollution index of the cheese to be tested is obtained through linear conversion, including: A microbial pollution index scale is established, each scale in the microbial pollution index scale corresponds to a specific numerical interval, and a hierarchical scale system of the microbial pollution index scale is obtained; The standardized fusion data is matched with the hierarchical scale system, the pollution grade to which the standardized fusion data belongs is determined by looking up a pre-established mapping relationship table, and a preliminary pollution grade determination result of the cheese to be tested is obtained; When the standardized fusion data is in a critical region of two pollution grades, a neighboring grade weighted judgment strategy is adopted to determine the grade of the cheese, and a final pollution grade of the cheese to be tested is obtained; According to the value in the numerical interval corresponding to the final pollution grade and the preliminary pollution grade determination result, the specific value of the microbial pollution index is determined, and the microbial pollution index of the cheese to be tested is obtained.

5. A method of early warning of microbial contamination of cheese as claimed in claim 4, wherein, When the standardized fusion data is in a critical region of two pollution grades, a neighboring grade weighted judgment strategy is adopted to determine the grade of the cheese, and a final pollution grade of the cheese to be tested is obtained, including: The position of the standardized fusion data on the hierarchical scale system is identified to obtain a critical region identification result of the standardized fusion data; The historical data distribution characteristics of the two pollution grades adjacent to the critical region are obtained, and the adjacent grade characteristic parameters of the standardized fusion data are obtained; According to the matching degree of the standardized fusion data and the adjacent grade characteristic parameters, the grade matching weight of the cheese to be tested is determined; Based on the grade matching weight, the final pollution grade of the cheese to be tested is determined.

6. A method for early warning of microbial contamination of cheese as claimed in claim 1, wherein, When the microbial pollution index exceeds a preset pollution threshold, a multi-level microbial pollution report of the cheese to be tested is generated based on the microbial pollution index, including: Based on the microbial pollution index, a corresponding report template is selected from a pre-established report template library to obtain a selected template of the cheese to be tested; The microbial pollution index and related detection parameters are filled into the selected template to generate a multi-level microbial pollution report of the cheese to be tested.

7. A cheese microbial pollution early warning system for realizing the cheese microbial pollution early warning method of claim 1, the system comprising: a stage heating module configured to perform gradient heating on the cheese to be tested, and collect volatile organic compounds at different temperature stages to obtain a gas sample of the cheese to be tested; a data acquisition module configured to detect the gas sample using a multi-parameter sensor array to obtain sensor response data of the cheese to be tested; a weight acquisition module configured to determine a weight coefficient set of the sensors in pollution index prediction according to real-time response intensity and historical data reliability of the sensors in the multi-parameter sensor array; a data fusion module configured to perform data fusion on the sensor response data based on the weight coefficient set to obtain a microbial pollution index of the cheese to be tested; a pollution determination module configured to generate a multi-level microbial pollution report of the cheese to be tested based on the microbial pollution index when the microbial pollution index exceeds a preset pollution threshold.

Images