Milk powder production quality management early warning method

Abstract

The invention relates to the technical field of milk powder production management, in particular to a milk powder production quality management early warning method. The method comprises the following steps: acquiring a plurality of infrared spectrum curves of a plurality of milk powder samples of a to-be-detected batch, and determining characteristic values of the infrared spectrum curves; taking the characteristic values of the infrared spectrum curves as weights, and respectively carrying out weighted fusion on the absorbance corresponding to the same wavelength in the plurality of infrared spectrum curves to obtain a target infrared spectrum curve; clustering a plurality of characteristic values corresponding to the plurality of milk powder samples under the condition that the fact that the to-be-detected batch of protein is not abnormal is determined according to the target infrared spectrum curve, so as to determine the similarity value of the to-be-detected batch of milk powder; and when the similarity value is smaller than a preset similarity threshold value, outputting first prompt information. By means of the technical scheme, quality early warning of the milk powder production process can be better achieved.

Description

Technical Field

[0001] The present application relates to the technical field of milk powder production management, and in particular to a milk powder production quality management early warning method. Background Art

[0002] As one of the important sources of nutrition for infants in their early years, the safety and quality of infant formula powder are receiving increasing attention. The control of process parameters during the production of milk powder is crucial to product quality. Slight fluctuations in parameters such as processing temperature, processing humidity and processing time during the production process may lead to unstable quality of milk powder products, thereby causing changes in the composition of milk powder. Therefore, it is necessary to test the quality of the produced milk powder to prevent milk powder products with abnormal quality from entering the market.

[0003] In order to detect the quality of milk powder, a Chinese patent application document with publication number CN107655852A provides a near infrared spectroscopy rapid detection method for essential nutrients in infant formula milk powder, including: using a milk powder analyzer to collect sample spectra, wherein the ambient temperature is controlled at 20°C, the relative humidity is 45%, and the scanning range is 10000-4000 , scan times 32, deduct air background each time, deduct strong absorption of water and carbon dioxide in the air in real time; Spectral preprocessing: perform various preprocessing on sample spectra: first-order derivative, smoothing, standard normal variable transformation, multivariate scattering correction; establish PLS model: the preprocessed spectrum is calculated using partial least squares PLS method to establish PLS model; the PLS model can be used to predict the content of essential nutrients in the sample to be tested.

[0004] In the related art, partial least squares is used to calculate the acquired sample spectrum, so as to realize the quality detection of milk powder samples. However, in the process of milk powder production, the related technology is difficult to realize the early warning of the production quality of milk powder more accurately. Summary of the invention

[0005] In order to overcome the problems existing in the related art, the present application provides a milk powder production quality management early warning method, comprising: obtaining multiple infrared spectrum curves of multiple milk powder samples of a batch to be tested, and determining characteristic values ​​of the infrared spectrum curves; the characteristic values ​​are used to characterize the peak values ​​of the absorbance in multiple wavelength ranges where the protein is located in the infrared spectrum curve, and the degree of difference between the absorbance threshold value corresponding to the wavelength range; using the characteristic values ​​of the infrared spectrum curve as weights, respectively performing weighted fusion on the absorbances corresponding to the same wavelength in the multiple infrared spectrum curves to obtain a target infrared spectrum curve; when it is determined that there is no abnormality in the protein of the batch of milk powder to be tested according to the target infrared spectrum curve, clustering the multiple characteristic values ​​of the multiple milk powder samples to determine the similarity value of the batch of milk powder to be tested. , exp is an exponential function with a natural constant as the base, D is the number of clusters, is the average distance from the eigenvalue in the mth cluster to the cluster center of the cluster, The maximum distance between the cluster center of the mth cluster and the cluster centers of other clusters; is the similarity of the eigenvalues ​​in the mth cluster, It is determined according to the DTW distance between the infrared spectrum curves corresponding to the characteristic values ​​in the mth cluster; when the similarity value is less than the preset similarity threshold, the first prompt information is output; the first prompt information is used to prompt that the quality of the tested batch of milk powder is abnormal.

[0006] In this way, multiple infrared spectral curves of multiple milk powder samples of the batch to be tested are obtained, and the absorbances corresponding to the same wavelength in the multiple infrared spectral curves are weightedly fused to obtain the target infrared spectral curve. When it is determined that there is no abnormality in the protein of the batch of milk powder to be tested according to the target infrared spectral curve, the similarity value of the batch of milk powder to be tested can be further determined. When the similarity value is less than a preset similarity threshold, the first prompt information for prompting that there is an abnormality in the quality of the batch of milk powder to be tested can be output. When it is preliminarily determined that there is no abnormality in the protein of the batch of milk powder to be tested, further detection of the quality of the batch of milk powder to be tested can be achieved, so as to better realize quality early warning of the production process of milk powder.

[0007] Since the target infrared spectral curve used to determine whether there is an abnormality in the protein of the tested batch of milk powder is obtained by weighted fusion of the absorbance corresponding to the same wavelength in multiple infrared spectral curves, it can avoid the data uniformity and sampling error when determining a single sample. Therefore, it can more accurately determine whether there is an abnormality in the protein of the tested batch of milk powder, so as to further determine whether there is an abnormality in other components other than protein in the tested batch of milk powder.

[0008] Optionally, the characteristic value of the infrared spectrum curve is determined by: , C is the characteristic value of the infrared spectrum curve, n is the number of wavelength ranges where the protein is located, is the peak value of the absorbance of the infrared spectrum curve in the i-th wavelength range where the protein is located, is the absorbance threshold corresponding to the i-th wavelength range where the protein is located, is the wavelength corresponding to the peak value of the absorbance of the infrared spectrum curve in the i-th wavelength range where the protein is located, is the wavelength threshold corresponding to the i-th wavelength range where the protein is located.

[0009] In this way, the peak value of the absorbance of the infrared spectrum curve in the wavelength range where the protein is located is compared with the absorbance threshold corresponding to the wavelength range where the protein is located, and the wavelength corresponding to the peak value of the absorbance of the infrared spectrum curve in the wavelength range where the protein is located is compared with the wavelength threshold corresponding to the wavelength range where the protein is located. The obtained characteristic value can better characterize the probability of abnormality in the protein of the milk powder sample corresponding to the infrared spectrum curve.

[0010] Optionally, the absorbance thresholds corresponding to the multiple wavelength ranges where the protein is located are determined in the following manner: for a target wavelength range among the multiple wavelength ranges where the protein is located, the average value of the absorbances corresponding to the target wavelength range of multiple infrared spectral curves of multiple milk powder samples in the batch to be tested is taken as the absorbance threshold corresponding to the target wavelength range where the protein is located.

[0011] Optionally, the similarity of feature values ​​in clusters is determined by: , is the similarity of the eigenvalues ​​in the mth cluster, exp is the indicator function with a natural constant as the base, is the maximum of the DTW distances between the two infrared spectral curves corresponding to any two eigenvalues ​​in the mth cluster. is the range of the eigenvalues ​​in the mth cluster.

[0012] In this way, the maximum of the DTW distances between the two infrared spectral curves corresponding to any two eigenvalues ​​in the same cluster and the range of the eigenvalues ​​in the same cluster can be used to better obtain the similarity used to characterize the similarity of the eigenvalues ​​in the same cluster, so that the similarity of different milk powder samples in the same cluster in terms of other components except protein can be characterized through the similarity of the eigenvalues ​​in the same cluster.

[0013] Optionally, the characteristic value of the infrared spectrum curve is used as a weight, and the absorbances corresponding to the same wavelength in multiple infrared spectrum curves are weightedly fused to obtain a target infrared spectrum curve, including: for a target wavelength among all wavelengths in the infrared spectrum curve, the characteristic value of the infrared spectrum curve is used as a weight, and the absorbances of multiple infrared spectrum curves at the target wavelength are weighted summed to obtain a weighted absorbance at the target wavelength after weighted fusion; and the target infrared spectrum curve is obtained according to different weighted absorbances corresponding to different target wavelengths.

[0014] In this way, the absorbances of multiple infrared spectral curves at the target wavelength are weighted and summed to obtain the weighted absorbance at the target wavelength after weighted fusion, so as to obtain the target infrared spectral curve, so as to more accurately determine whether the protein of the batch of milk powder to be tested is abnormal.

[0015] Optionally, the target infrared spectral curve determines that there is no abnormality in the protein of the batch of milk powder to be tested in the following manner: from the absorbance of the target infrared spectral curve, determine a plurality of target peaks respectively located in the plurality of wavelength ranges where the protein is located; compare the target peak with the corresponding preset peaks in the plurality of wavelength ranges where the protein is located, and when the target peak is consistent with the preset peak, determine that there is no abnormality in the target infrared spectral curve at the target peak; when it is determined that there is no abnormality in all target peaks of the target infrared spectral curve, determine that there is no abnormality in the protein of the batch to be tested.

[0016] In this way, by determining multiple target peaks in the absorbance of the target infrared spectrum curve that are respectively located in multiple wavelength ranges where the protein is located, and comparing them with the corresponding preset peaks, it is possible to better determine whether there is any abnormality in the protein of the batch of milk powder to be tested.

[0017] Optionally, the average distance of the feature values ​​in the cluster to the cluster center of the cluster is determined by: ,in, is the average distance from the eigenvalue in the mth cluster to the cluster center of the cluster, is an exponential function with a natural constant as base, is the similarity of the eigenvalues ​​in the mth cluster, is the initial average distance from the feature value in the mth cluster to the cluster center of the cluster.

[0018] Optionally, the method further includes: when it is determined that the protein is abnormal according to the target infrared spectrum curve, outputting second prompt information, wherein the second prompt information is used to prompt that the protein of the batch of milk powder to be tested is abnormal.

[0019] The technical solution provided by the embodiments of the present application may include the following beneficial effects: obtaining multiple infrared spectral curves of multiple milk powder samples of a batch to be tested, and performing weighted fusion on the absorbances corresponding to the same wavelength in the multiple infrared spectral curves to obtain a target infrared spectral curve; when it is determined according to the target infrared spectral curve that there is no abnormality in the protein of the batch of milk powder to be tested, the similarity value of the batch of milk powder to be tested can be further determined; when the similarity value is less than a preset similarity threshold, a first prompt information for prompting that there is an abnormality in the quality of the batch of milk powder to be tested can be output; when it is preliminarily determined that there is no abnormality in the protein of the batch of milk powder to be tested, further detection of the quality of the batch of milk powder to be tested can be achieved, so as to better realize quality warning of the production process of milk powder.

[0020] Since the target infrared spectral curve used to determine whether there is an abnormality in the protein of the tested batch of milk powder is obtained by weighted fusion of the absorbance corresponding to the same wavelength in multiple infrared spectral curves, it can avoid the data uniformity and sampling error when determining a single sample. Therefore, it can more accurately determine whether there is an abnormality in the protein of the tested batch of milk powder, so as to further determine whether there is an abnormality in other components other than protein in the tested batch of milk powder.

[0021] It should be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0023] Figure 1 The present invention is a flowchart of a method for early warning of milk powder production quality management according to an exemplary embodiment. DETAILED DESCRIPTION

[0024] Here, exemplary embodiments are described in detail, and examples thereof are shown in the accompanying drawings. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The implementations described in the following exemplary embodiments do not represent all implementations consistent with the present application.

[0025] First, the application scenario of the embodiment of the present application is briefly introduced. In the application scenario of the present application, the data of influencing factors in the production process of milk powder can be collected, and the influencing factor data can be used to predict whether there is any abnormality in the quality of the milk powder. However, the influencing factor data of the production process is difficult to directly reflect the actual situation of the milk powder, making it difficult to accurately realize the early warning of the production quality of milk powder.

[0026] In view of the above technical problems, the embodiment of the present application provides a milk powder production quality management early warning method, which can be applied to a device for early warning of milk powder production quality. Figure 1 is a flow chart of a method for early warning of milk powder production quality management according to an exemplary embodiment, which can be Figure 1 As shown, the method includes the following steps.

[0027] In step S101, a plurality of infrared spectrum curves of a plurality of milk powder samples of a batch to be tested are obtained, and characteristic values ​​of the infrared spectrum curves are determined.

[0028] Multiple milk powder samples can be extracted for the batch to be tested, and corresponding infrared spectrum curves can be obtained for different milk powder samples. The infrared spectrum curve includes the absorbance of the same milk powder sample to infrared light of different wavelengths; for example, the horizontal axis of the infrared spectrum curve is the wavelength of infrared light, and the vertical axis of the infrared spectrum curve is the absorbance of the milk powder sample to infrared light.

[0029] The milk powder sample can be prepared by tabletting, attenuated total reflection or coating method, and the infrared spectrum curve of the milk powder sample can be obtained by Fourier transform infrared spectrometer or other types of infrared spectrometers, so as to analyze the quality of the milk powder sample.

[0030] The characteristic value is used to characterize the degree of difference between the peak value of the absorbance in the multiple wavelength ranges where the protein is located in the infrared spectrum curve and the absorbance threshold value corresponding to the wavelength range; the larger the characteristic value of the infrared spectrum curve, the higher the degree of difference between the peak value of the absorbance in the multiple wavelength ranges where the protein is located in the infrared spectrum curve and the absorbance threshold value corresponding to the wavelength range; on the contrary, the smaller the characteristic value of the infrared spectrum curve, the higher the degree of difference between the peak value of the absorbance in the multiple wavelength ranges where the protein is located in the infrared spectrum curve and the absorbance threshold value corresponding to the wavelength range.

[0031] The absorption peaks at different wavelengths in the infrared spectrum curve can reflect the absorption of infrared light by different substances in the milk powder sample. For example, protein is composed of amino acids, which contain a variety of functional groups, such as amide I band, amide II band and amide III band. Among them, the absorption peak of amide I band is approximately at 1650 to 1600 In this wavelength range, the absorption peak of the amide II band is approximately at 1550 Up to 1500 In this wavelength range, the absorption peak of the amide III band is approximately at 1250 to 1200 within this wavelength range.

[0032] By observing the infrared spectrum curve in the three wavelength ranges corresponding to the amide I band, the amide II band and the amide III band, the protein situation of the milk powder to be tested can be determined; for example, it can be determined whether the infrared spectrum curve of the milk powder to be tested has an absorption peak at the corresponding position, and whether the peak shape of the absorption peak changes, and the type and structural characteristics of the protein in the milk powder to be tested can be inferred, which will not be elaborated here.

[0033] By determining whether there is an absorption peak at the corresponding position of the infrared spectrum curve of the tested milk powder and whether the peak shape of the absorption peak has changed, the type and structural characteristics of the protein in the tested milk powder can be inferred, and the characteristic value is used to characterize the peak value of the absorbance in multiple wavelength ranges where the protein is located in the infrared spectrum curve, and the degree of difference between the absorbance threshold value corresponding to the wavelength range. Therefore, the characteristic value can reflect the probability of abnormality of the protein in the milk powder sample. By determining the characteristic value of the infrared spectrum curve, it is helpful to determine whether the protein in the milk powder sample is abnormal.

[0034] In one embodiment, the characteristic value of the infrared spectrum curve is determined by: , C is the characteristic value of the infrared spectrum curve, n is the number of wavelength ranges where the protein is located, is the peak value of the absorbance of the infrared spectrum curve in the i-th wavelength range where the protein is located, is the absorbance threshold corresponding to the i-th wavelength range where the protein is located, is the wavelength corresponding to the peak value of the absorbance of the infrared spectrum curve in the i-th wavelength range where the protein is located, is the wavelength threshold corresponding to the i-th wavelength range where the protein is located.

[0035] In the case where there is an abnormality in the protein of milk powder, the absorption peak of the corresponding protein in the infrared spectrum curve of the milk powder may be shifted; or there is no absorption peak in the wavelength range of the corresponding protein in the infrared spectrum curve of the milk powder, but there is an absorption peak in the infrared spectrum curve of the milk powder in other ranges except the wavelength range of the corresponding protein; therefore, by comparing the peak values ​​of the infrared spectrum curves of the tested batches of milk powder within the wavelength range of the corresponding protein, the infrared spectrum curve of the milk powder sample can be better utilized to determine the protein information of the milk powder sample.

[0036] In this way, the peak value of the absorbance of the infrared spectrum curve in the wavelength range where the protein is located is compared with the absorbance threshold corresponding to the wavelength range where the protein is located, and the wavelength corresponding to the peak value of the absorbance of the infrared spectrum curve in the wavelength range where the protein is located is compared with the wavelength threshold corresponding to the wavelength range where the protein is located. The obtained characteristic value can better characterize the probability of abnormality in the protein of the milk powder sample corresponding to the infrared spectrum curve.

[0037] In one embodiment, the absorbance thresholds corresponding to the multiple wavelength ranges where the protein is located are determined in the following manner: for the target wavelength range among the multiple wavelength ranges where the protein is located, the average value of the absorbances corresponding to the target wavelength range of multiple infrared spectral curves of multiple milk powder samples in the batch to be tested is taken as the absorbance threshold corresponding to the target wavelength range where the protein is located.

[0038] For different milk powder samples from the same batch, when there is no abnormality in the production process of the batch to be tested, the production conditions and production raw materials of the milk powder samples have a high consistency, so that among multiple milk powder samples in the batch to be tested, the absorbance performance of the infrared spectral curves of different milk powder samples at the peak value within the wavelength range corresponding to the protein is usually also highly consistent.

[0039] In this way, the average value of the absorbance corresponding to multiple infrared spectral curves of multiple milk powder samples in the test batch in the target wavelength range is used as the absorbance threshold corresponding to the target wavelength range where the protein is located, which can better determine whether there is any abnormality in the absorbance of the milk powder samples in the test batch in the target wavelength range where the protein is located.

[0040] In one embodiment, the wavelength thresholds corresponding to the multiple wavelength ranges where the protein is located are determined in the following manner: for the target wavelength range among the multiple wavelength ranges where the protein is located, the average value of the wavelengths of multiple infrared spectral curves of multiple milk powder samples in the batch to be tested at the peak values ​​within the target wavelength range is taken as the wavelength threshold corresponding to the target wavelength range where the protein is located.

[0041] For different milk powder samples from the same batch, when there is no abnormality in the production process of the batch to be tested, the production conditions and production raw materials of the milk powder samples have a high degree of consistency, so that among multiple milk powder samples in the batch to be tested, the wavelengths at the peaks of the infrared spectral curves of different milk powder samples within the wavelength range corresponding to protein usually also have a high degree of consistency.

[0042] In this way, by taking the average value of the wavelengths at the peaks of multiple infrared spectrum curves of multiple milk powder samples in the test batch within the target wavelength range as the wavelength threshold corresponding to the target wavelength range where the protein is located, it can be better determined whether there is an abnormality in the wavelength corresponding to the peak of the milk powder samples in the test batch within the target wavelength range where the protein is located.

[0043] The absorbance threshold corresponding to the target wavelength range where the protein is located can also be determined based on the infrared spectral curves of multiple reference batches of milk powder samples that have passed the inspection; for example, the infrared spectral curves of multiple reference batches of milk powder samples that have passed the inspection can be obtained; for the target wavelength range among the multiple wavelength ranges where the protein is located, the average value of the absorbance corresponding to the target wavelength range of the infrared spectral curves of multiple reference batches of milk powder samples that have passed the inspection is used as the absorbance threshold corresponding to the target wavelength range where the protein is located.

[0044] The reference batch of milk powder and the batch of milk powder to be tested are the same milk powder; for example, the reference batch of milk powder and the batch of milk powder to be tested have the same expected ingredients. Due to slight differences in production conditions or slight differences in production raw materials during the production of milk powder, there may be certain differences between the actual ingredients of the batch to be tested and the expected ingredients, and such differences may be sufficient to make the batch of milk powder to be tested fail to meet quality requirements. The reference batch of milk powder samples can provide a better reference for the detection of the batch of milk powder samples to be tested, thereby more accurately realizing early warning of the production quality of milk powder.

[0045] In step S102, the characteristic values ​​of the infrared spectrum curves are used as weights to perform weighted fusion on the absorbances corresponding to the same wavelength in multiple infrared spectrum curves to obtain a target infrared spectrum curve.

[0046] In order to facilitate the quality inspection of the batch of milk powder to be tested, multiple infrared spectrum curves of multiple milk powder samples of the batch to be tested are fused to avoid the monotony of the infrared spectrum curve of a single milk powder sample. Since the weighted fusion of the absorbance corresponding to the same wavelength in the infrared spectrum curve is performed based on the eigenvalue of the infrared spectrum curve, and the eigenvalue of the infrared spectrum curve can characterize the probability of abnormality in the protein of the milk powder sample, the infrared spectrum curve with a larger eigenvalue has a greater contribution in the weighted summation process. The target infrared spectrum curve obtained after weighting can more easily reflect the possible abnormalities in the protein of the batch to be tested.

[0047] In one embodiment, the characteristic value of the infrared spectrum curve is used as a weight, and the absorbances corresponding to the same wavelength in multiple infrared spectrum curves are weightedly fused to obtain a target infrared spectrum curve, including: for a target wavelength among all wavelengths in the infrared spectrum curve, the characteristic value of the infrared spectrum curve is used as a weight, and the absorbances of multiple infrared spectrum curves at the target wavelength are weighted summed to obtain a weighted absorbance at the target wavelength after weighted fusion; and the target infrared spectrum curve is obtained according to different weighted absorbances corresponding to different target wavelengths.

[0048] The target wavelength can be any wavelength among all the wavelengths in the infrared spectrum curve. The absorbances of multiple infrared spectrum curves at the target wavelength are weighted and summed to obtain the weighted absorbance at the target wavelength after weighted fusion. The weighted absorbance at the target wavelength combines the absorbances of multiple milk powder samples at the target wavelength, so that the absorbances at different wavelengths in the obtained target infrared spectrum curve are changed.

[0049] Since the absorbance at different wavelengths in the obtained target infrared spectrum curve has changed, the peak position in the target infrared spectrum curve has changed, and the infrared spectrum curve with a larger eigenvalue has a greater weight. Compared with the infrared spectrum curve of a single milk powder sample, the target infrared spectrum curve can not only avoid the singleness of the data, but also better reflect the abnormalities of the protein in the batch to be tested.

[0050] For example, in the target infrared spectrum curve, the The absorbance at this wavelength is calculated by taking the characteristic value of the infrared spectrum curve as the weight and calculating the absorbance at 1650 The target infrared spectrum curve is obtained by weighted summation of multiple absorbances at this wavelength, so that the target infrared spectrum curve can better reflect the situation of the amide I band of the protein in the batch of milk powder to be tested.

[0051] In this way, the absorbances of multiple infrared spectral curves at the target wavelength are weighted and summed to obtain the weighted absorbance at the target wavelength after weighted fusion, so as to obtain the target infrared spectral curve, so as to more accurately determine whether the protein of the batch of milk powder to be tested is abnormal.

[0052] In step S103, when it is determined that there is no abnormality in the protein of the tested batch of milk powder according to the target infrared spectrum curve, multiple feature values ​​of multiple milk powder samples are clustered to determine the similarity value of the tested batch of milk powder.

[0053] When there is no abnormality in the protein of the tested batch of milk powder, the infrared spectrum curve of the milk powder sample has a high consistency with the preset infrared spectrum curve within the wavelength range where the protein is located. The preset infrared spectrum curve may be the infrared spectrum curve corresponding to the milk powder sample that has passed the inspection. The milk powder corresponding to the preset infrared spectrum curve has the same production raw materials and production conditions as the tested batch of milk powder.

[0054] Since the target infrared spectral curve is obtained by weighted fusion of infrared spectral curves of multiple milk powder samples, the target infrared spectral curve can better represent the component information of the multiple milk powder samples, and by comparing the target infrared spectral curve with the preset infrared spectral curve within the wavelength range where the protein is located, it can be determined whether there is an abnormality in the protein of the tested batch; wherein, determining whether there is an abnormality in the protein of milk powder based on the infrared spectral image of the milk powder is common knowledge of those skilled in the art, and the embodiments of the present application will not be elaborated herein.

[0055] When it is determined that the protein content of the tested batch of milk powder is abnormal, it means that the protein content of at least one milk powder sample among the multiple milk powder samples of the tested batch is abnormal. Even if the multiple milk powder samples of the tested batch have high consistency, the quality test result of the milk powder samples of the tested batch will be unqualified, and the step of determining the similarity value of the tested batch of milk powder may no longer be performed. In this way, unnecessary calculation processes can be reduced, and the production conditions or production raw materials of the tested batch can be adjusted as soon as possible to obtain milk powder of qualified quality.

[0056] When it is determined that the protein of the test batch of milk powder is abnormal according to the target infrared spectrum curve, it at least means that there is no abnormality in the protein of the test batch of milk powder. In order to determine whether there is abnormality in other components of the test batch of milk powder except protein, a step of determining the similarity value of the test batch of milk powder can be performed.

[0057] Since protein is a more critical component in milk powder, and protein is more easily affected by production conditions and raw materials than other components in milk powder, when there is no abnormality in the protein of the tested batch of milk powder, it means that there is a high probability that there is no abnormality in the production conditions and raw materials of the tested batch of milk powder, and there is a high probability that there is no abnormality in other components except protein in the tested batch of milk powder produced under the same production conditions and raw materials. By determining the consistency of different milk powder samples in the tested batch, it is possible to simply and effectively determine whether there is any abnormality in the quality of the tested batch of milk powder.

[0058] Multiple feature values ​​of multiple milk powder samples are clustered to determine the similarity value of the milk powder batch to be tested. The obtained similarity value can be used to characterize the consistency of multiple milk powder samples in the test batch. If there is no abnormality in the protein of the milk powder sample and the multiple milk powder samples in the test batch have a high consistency, it means that the milk powder samples in the test batch are more likely to meet the requirements in other costs except protein, which means that the quality of the milk powder in the test batch meets the requirements.

[0059] Similarity value of the tested batch of milk powder , exp is an exponential function with a natural constant as the base, D is the number of clusters, is the average distance from the eigenvalue in the mth cluster to the cluster center of the cluster, The maximum distance between the cluster center of the mth cluster and the cluster centers of other clusters; is the similarity of the eigenvalues ​​in the mth cluster, It is determined based on the DTW distance between the infrared spectrum curves corresponding to the eigenvalues ​​in the mth cluster.

[0060] When clustering of multiple eigenvalues ​​corresponding to multiple milk powder samples of the test batches is completed, the more clusters obtained, the higher the degree of dispersion between different clusters obtained after clustering; on the contrary, the fewer the number of clusters obtained, the higher the degree of aggregation between different clusters obtained after clustering.

[0061] is the average distance from the eigenvalue in the mth cluster to the cluster center of the cluster, corresponding to the mth cluster The larger the value, the higher the dispersion of the eigenvalues ​​included in the mth cluster, and the lower the consistency of different milk powder samples in the tested batch; on the contrary, the corresponding mth cluster The smaller it is, the higher the degree of clustering between the eigenvalues ​​included in the mth cluster, and the higher the consistency of different milk powder samples in the tested batches.

[0062] The maximum distance between the cluster center of the mth cluster and the cluster centers of other clusters, corresponding to the mth cluster The larger the value, the higher the difference between the mth cluster and other clusters, and the lower the consistency of different milk powder samples in the tested batches; on the contrary, the value corresponding to the mth cluster is The smaller it is, the lower the difference between the mth cluster and other clusters, and the higher the consistency of different milk powder samples in the tested batches.

[0063] It is used to characterize the similarity between the infrared spectral curves corresponding to the eigenvalues ​​in the mth cluster. The larger the value is, the higher the similarity between the infrared spectrum curves corresponding to the eigenvalues ​​in the mth cluster is. On the contrary, The smaller it is, the lower the similarity between the infrared spectral curves corresponding to the eigenvalues ​​in the mth cluster.

[0064] The similarity value is processed using an exponential function with a natural constant as the base, so that the similarity value H is between 0 and 1, which is convenient for comparing and analyzing different batches of milk powder to be tested.

[0065] In this way, by clustering the distances between different clusters after eigenvalues ​​are clustered, as well as the distances between eigenvalues ​​within the same cluster, the clusters after clustering can be better compared, and the DTW distance between the infrared spectral curves corresponding to the eigenvalues ​​in the same cluster is taken into account, so that the consistency of the milk powder batch to be tested can be better characterized by the similarity value.

[0066] In one embodiment, the average distance from the feature values ​​in the cluster to the cluster center of the cluster is determined by: ,in, is the average distance from the eigenvalue in the mth cluster to the cluster center of the cluster, is an exponential function with a natural constant as base, is the similarity of the eigenvalues ​​in the mth cluster, is the initial average distance from the feature value in the mth cluster to the cluster center of the cluster.

[0067] In this way, for clusters with smaller similarity of included eigenvalues, the denominator in the calculation formula of the average distance from the eigenvalues ​​in the cluster to the cluster center of the cluster is smaller, so that the average distance from the eigenvalues ​​in the cluster to the cluster center of the cluster is larger, thereby improving the contrast between the average distances corresponding to clusters with different similarities of included eigenvalues.

[0068] In one embodiment, the target infrared spectrum curve determines that there is no abnormality in the protein of the batch of milk powder to be tested by the following method: from the absorbance of the target infrared spectrum curve, determine a plurality of target peaks respectively located in the plurality of wavelength ranges where the protein is located; compare the target peak with the corresponding preset peaks in the plurality of wavelength ranges where the protein is located, and when the target peak is consistent with the preset peak, determine that there is no abnormality in the target infrared spectrum curve at the target peak; when it is determined that there is no abnormality in all target peaks of the target infrared spectrum curve, determine that there is no abnormality in the protein of the batch to be tested.

[0069] The preset peak value can be determined based on the infrared spectrum curve of a reference batch of milk powder that has passed the inspection. The reference batch of milk powder and the batch of milk powder to be tested are milk powders with the same formula; different chemical bonds and functional groups in protein will absorb infrared light within a specific wavelength range, thereby forming characteristic peaks in the infrared spectrum curve of milk powder. By determining multiple target peaks located in multiple wavelength ranges where the protein is located on the infrared spectrum curve, the protein component in the milk powder can be determined.

[0070] The multiple wavelength ranges where the protein is located correspond to multiple preset peaks respectively. If the target peak is consistent with the corresponding preset peak in the multiple wavelength ranges where the protein is located, it means that the protein information of the tested batch of milk powder is consistent with the expected performance, and it can be determined that there is no abnormality in the functional group corresponding to the target peak of the tested batch of milk powder.

[0071] In this way, by determining multiple target peaks in the absorbance of the target infrared spectrum curve that are respectively located in multiple wavelength ranges where the protein is located, and comparing them with the corresponding preset peaks, it is possible to better determine whether there is any abnormality in the protein of the batch of milk powder to be tested.

[0072] When at least one target peak value of the target infrared spectrum curve is inconsistent with the corresponding preset peak value, it can be determined that the protein of the tested batch of milk powder is abnormal.

[0073] In one embodiment, the similarity of feature values ​​in the clusters is determined by: , is the similarity of the eigenvalues ​​in the mth cluster, exp is the indicator function with a natural constant as the base, is the maximum of the DTW distances between the two infrared spectral curves corresponding to any two eigenvalues ​​in the mth cluster. is the range of the eigenvalues ​​in the mth cluster.

[0074] DTW (Dynamic Time Warping) is an algorithm used to compare different sequences. DTW can perform nonlinear deformation on data on a certain axis to find the best alignment between two sequences. It calculates the distance matrix between the two sequences, in which each element represents the distance between the corresponding points in the two sequences. It uses a dynamic programming algorithm to find a path from the upper left corner to the lower right corner of the matrix so that the cumulative distance on the path is minimized. This cumulative distance is the DTW distance. The DTW distance between two sequences can be used to characterize the similarity between the two sequences.

[0075] For multiple eigenvalues ​​in the same cluster, the DTW distance between two infrared spectrum curves corresponding to any two eigenvalues ​​can be determined respectively. Different two eigenvalues ​​correspond to different DTW distances, so different DTW distances can be obtained respectively to determine the largest one from multiple DTW distances.

[0076] Since different eigenvalues ​​in the same cluster correspond to different milk powder samples, different milk powder samples corresponding to different eigenvalues ​​in the same cluster have high consistency in protein performance, and the largest DTW distance between two infrared spectrum curves corresponding to any two eigenvalues ​​in the cluster can reflect the maximum difference between infrared spectrum curves corresponding to different eigenvalues ​​in the same cluster, thereby reflecting the similarity of different milk powder samples corresponding to different eigenvalues ​​in the same cluster in other components except protein. Among them, other components except protein can be, for example, fat, vitamins and minerals.

[0077] is the range of the eigenvalues ​​in the mth cluster. The range of the eigenvalues ​​in the mth cluster is equal to the maximum eigenvalue in the mth cluster minus the minimum eigenvalue in the mth cluster. The range of the eigenvalues ​​in the mth cluster can reflect the maximum amplitude of the difference in the eigenvalues ​​in the mth cluster.

[0078] In this way, the maximum of the DTW distances between the two infrared spectral curves corresponding to any two eigenvalues ​​in the same cluster and the range of the eigenvalues ​​in the same cluster can be used to better obtain the similarity used to characterize the similarity of the eigenvalues ​​in the same cluster, so that the similarity of different milk powder samples in the same cluster in terms of other components except protein can be characterized through the similarity of the eigenvalues ​​in the same cluster.

[0079] In one embodiment, when it is determined that there is an abnormality in the protein according to the target infrared spectrum curve, a second prompt information is output, wherein the second prompt information is used to prompt that there is an abnormality in the protein of the batch of milk powder to be tested.

[0080] When it is determined that there is an abnormality in the protein based on the target infrared spectrum curve, it is no longer necessary to compare the consistency of other components of the test batch except protein, and output a second prompt message; the output second prompt message can be at least one of audio prompt message, light prompt message, text prompt message and vibration prompt message.

[0081] For example, a device used to provide early warning of the production process of milk powder can output a text prompt message or an audio prompt message such as "There is an abnormality in this batch of milk powder, please check relevant information"; or, the device can send a prompt message to the terminal device bound to the user, so that the terminal device outputs a prompt message to indicate that there is an abnormality in the protein of the batch of milk powder to be tested.

[0082] In step S104, when the similarity value is less than the preset similarity threshold, the first prompt information is output.

[0083] Since the similarity value of the milk powder of the test batch can characterize the consistency of the milk powder samples of the test batch, when the similarity value is less than the preset similarity threshold, it means that the consistency of different milk powder samples in the same test batch is low. For example, the consistency of different milk powder samples in the test batch may be low due to uneven mixing of production raw materials. By outputting the first prompt information for prompting that the quality of the milk powder of the test batch is abnormal, the user can check the production situation of the test batch to determine the reason for the low consistency of the milk powder of the test batch, thereby obtaining milk powder with higher consistency.

[0084] The first prompt information outputted may be at least one of an audio prompt information, a light prompt information, a text prompt information and a vibration prompt information. For example, a device used to provide an early warning for the production process of milk powder may output a text prompt information or an audio prompt information of “The quality of this batch of milk powder is abnormal, please check the relevant information”; alternatively, the device may send a prompt message to a terminal device bound to the user, so that the terminal device outputs a prompt information for prompting that the quality of the batch of milk powder to be tested is abnormal.

[0085] By using the milk powder production quality management and early warning method provided in the embodiment of the present application, multiple infrared spectral curves of multiple milk powder samples of the tested batch are obtained, and the absorbances corresponding to the same wavelength in the multiple infrared spectral curves are weightedly fused to obtain the target infrared spectral curve. When it is determined that there is no abnormality in the protein of the tested batch of milk powder according to the target infrared spectral curve, the similarity value of the tested batch of milk powder can be further determined. When the similarity value is less than a preset similarity threshold, the first prompt information for prompting that there is an abnormality in the quality of the tested batch of milk powder can be output. When it is preliminarily determined that there is no abnormality in the protein of the tested batch of milk powder, further detection of the quality of the tested batch of milk powder can be achieved, so as to better achieve quality early warning of the milk powder production process.

[0086] Since the target infrared spectral curve used to determine whether there is an abnormality in the protein of the tested batch of milk powder is obtained by weighted fusion of the absorbance corresponding to the same wavelength in multiple infrared spectral curves, it can avoid the data uniformity and sampling error when determining a single sample. Therefore, it can more accurately determine whether there is an abnormality in the protein of the tested batch of milk powder, so as to further determine whether there is an abnormality in other components other than protein in the tested batch of milk powder.

[0087] It should be understood that the features of some embodiments of the various applications described herein may be combined with each other unless specifically stated otherwise.

[0088] Although terms such as "first", "second", and "third" may be used herein to describe various components, parts, regions, layers, or sections, these components, parts, regions, layers, or sections are not limited to these terms. Instead, these terms are only used to distinguish one component, part, region, layer, or section from another component, part, region, layer, or section. Therefore, without departing from the teachings of the examples described herein, the first component, part, region, layer, or section mentioned in the examples may also be referred to as the second component, part, region, layer, or section.

[0089] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In the description herein, the meaning of "plurality" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined.

[0090] In addition, the word "exemplary" is used herein to mean serving as an example, instance, or diagram. Any aspect or design described herein as "exemplary" is not necessarily to be understood as being advantageous over other aspects or designs. Instead, the use of the word exemplary is intended to present concepts in a concrete manner. As used herein, the term "or" is intended to mean an inclusive "or" rather than an exclusive "or".

[0091] Likewise, although the present application has been shown and described with respect to one or more implementations, equivalent variations and modifications will occur to those skilled in the art upon reading and understanding this specification and the accompanying drawings. With particular regard to the various functions performed by the components (e.g., elements, resources, etc.) described above, unless otherwise noted, the terms used to describe such components are intended to correspond to any component (functionally equivalent) that performs the specific functions of the described components, even if not structurally equivalent to the disclosed structures.

[0092] Additionally, while particular features of the present application may have been disclosed with respect to only one of several implementations, such features may be combined with one or more other features of other implementations as may be desired and advantageous for any given or particular application.

[0093] Those skilled in the art will readily appreciate other embodiments of the present application after considering the specification and practicing the invention disclosed herein. The present application is intended to cover any modification, use or adaptation of the present application, which follows the general principles of the present application and includes common knowledge or customary technical means in the art that are not disclosed in the present application, and the specification and embodiments are only considered as exemplary.

[0094] It should be understood that the present application is not limited to the exact construction that has been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof.

Claims

1. A milk powder production quality management early warning method, characterized in that: include: Obtaining multiple infrared spectrum curves of multiple milk powder samples of a batch to be tested, and determining characteristic values ​​of the infrared spectrum curves; The characteristic value is used to characterize the difference between the peak value of the absorbance in the multiple wavelength ranges where the protein is located in the infrared spectrum curve and the absorbance threshold value corresponding to the wavelength range; The characteristic value of the infrared spectrum curve is used as a weight, and the absorbance corresponding to the same wavelength in multiple infrared spectrum curves is weightedly fused to obtain the target infrared spectrum curve; When it is determined that the protein of the tested batch of milk powder does not have abnormality according to the target infrared spectrum curve, multiple feature values ​​of multiple milk powder samples are clustered to determine the similarity value of the tested batch of milk powder , exp is an exponential function with a natural constant as the base, D is the number of clusters, is the average distance from the eigenvalue in the mth cluster to the cluster center of the cluster, The maximum distance between the cluster center of the mth cluster and the cluster centers of other clusters; is the similarity of the eigenvalues ​​in the mth cluster, It is determined based on the DTW distance between the infrared spectrum curves corresponding to the eigenvalues ​​in the mth cluster; When the similarity value is less than a preset similarity threshold, outputting a first prompt message; The first prompt information is used to prompt that the quality of the tested batch of milk powder is abnormal.

2. The milk powder production quality management early warning method according to claim 1, characterized in that: The characteristic values ​​of the infrared spectrum curve are determined in the following way: , C is the characteristic value of the infrared spectrum curve, n is the number of wavelength ranges where the protein is located, is the peak value of the absorbance of the infrared spectrum curve in the i-th wavelength range where the protein is located, is the absorbance threshold corresponding to the i-th wavelength range where the protein is located, is the wavelength corresponding to the peak value of the absorbance of the infrared spectrum curve in the i-th wavelength range where the protein is located, is the wavelength threshold corresponding to the i-th wavelength range where the protein is located.

3. The milk powder production quality management early warning method according to claim 1, characterized in that: The absorbance thresholds corresponding to the multiple wavelength ranges where the protein is located are determined in the following way: For a target wavelength range in the multiple wavelength ranges where the protein is located, the average value of the absorbances of multiple infrared spectral curves of multiple milk powder samples of the tested batches in the target wavelength range is used as the absorbance threshold corresponding to the target wavelength range where the protein is located.

4. The milk powder production quality management early warning method according to claim 1, characterized in that: The similarity of feature values ​​in clusters is determined in the following way: , is the similarity of the eigenvalues ​​in the mth cluster, exp is the indicator function with a natural constant as the base, is the maximum of the DTW distances between the two infrared spectral curves corresponding to any two eigenvalues ​​in the mth cluster. is the range of the eigenvalues ​​in the mth cluster.

5. The milk powder production quality management early warning method according to claim 1, characterized in that: The characteristic value of the infrared spectrum curve is used as the weight, and the absorbance corresponding to the same wavelength in multiple infrared spectrum curves is weighted and fused to obtain the target infrared spectrum curve, including: For a target wavelength among all wavelengths in the infrared spectrum curve, the characteristic value of the infrared spectrum curve is used as a weight, and the absorbances of multiple infrared spectrum curves at the target wavelength are weighted summed to obtain the weighted absorbance at the target wavelength after weighted fusion; According to different weighted absorbances corresponding to different target wavelengths, the target infrared spectrum curve is obtained.

6. The milk powder production quality management early warning method according to claim 1, characterized in that: The target infrared spectrum curve determines that there is no abnormality in the protein of the tested batch of milk powder by the following method: From the absorbance of the target infrared spectrum curve, a plurality of target peaks respectively located in the plurality of wavelength ranges where the protein is located are determined; Comparing the target peak with corresponding preset peaks in the multiple wavelength ranges where the protein is located, and determining that there is no abnormality in the target infrared spectrum curve at the target peak when the target peak is consistent with the preset peak; When it is determined that there is no abnormality in all target peaks of the target infrared spectrum curve, it is determined that there is no abnormality in the protein batch to be tested.

7. The milk powder production quality management early warning method according to claim 1, characterized in that: The average distance of the feature values ​​in the cluster to the cluster center of the cluster is determined as follows: ,in, is the average distance from the eigenvalue in the mth cluster to the cluster center of the cluster, is an exponential function with a natural constant as base, is the similarity of the eigenvalues ​​in the mth cluster, is the initial average distance from the feature value in the mth cluster to the cluster center of the cluster.

8. The milk powder production quality management early warning method according to claim 1, characterized in that: The method further comprises: When it is determined that the protein is abnormal according to the target infrared spectrum curve, a second prompt information is output, wherein the second prompt information is used to prompt that the protein of the batch of milk powder to be tested is abnormal.

Images