Method and device for constructing discrimination model based on solid waste fingerprint features

Abstract

The invention discloses a method and device for constructing a discrimination model based on solid waste fingerprint features, and belongs to the field of solid waste discrimination. The method comprises the following steps: acquiring a plurality of groups of solid waste samples of a predetermined category in a predetermined area, and selecting a K alpha spectral line to analyze the oxide content of the sample as various fingerprint factors to obtain a plurality of groups of preliminarily screened fingerprint factors; according to the statistical magnitude of the multiple groups of preliminary screening fingerprint factors, removing the fingerprint factors which do not meet the screening conditions to obtain secondary screening fingerprint factors; on the basis of the fingerprint factors screened for the second time, new fingerprint factors are introduced one by one according to a preset principle, and an optimal fingerprint factor combination is screened out; and constructing a discrimination model based on the optimal fingerprint factor combination, inputting the solid waste sample of the to-be-discriminated category into the discrimination model to obtain a discrimination score, and obtaining the category of the solid waste sample of the to-be-discriminated category according to a discrimination rule of a predetermined category and the discrimination score. According to the invention, unknown solid wastes can be accurately and rapidly identified.

Description

Technical Field

[0001] The present invention belongs to the field of solid waste identification, and in particular relates to a method and a device for constructing an identification model based on solid waste fingerprint characteristics. Background Art

[0002] The non-compliant dumping of solid waste has resulted in a certain amount of unknown solid waste being stored in the wild. Without knowing the solid waste, we cannot understand the harm it causes to the environment. Therefore, it is crucial to identify unknown solid waste.

[0003] Experimental testing and analysis of unknown solid waste can reveal its category, but the method is complicated. How to quickly and accurately identify unknown solid waste is a problem that needs to be solved urgently. Summary of the invention

[0004] In order to solve the technical problems existing in the prior art, the present invention provides a method for constructing a discrimination model based on solid waste fingerprint characteristics, comprising: Step S1, obtaining multiple groups of solid waste samples of predetermined categories in a predetermined area, selecting Kα spectral line analysis sample oxide content as various fingerprint factors of each of the solid waste samples, eliminating fingerprint factors with no statistical significance, and obtaining multiple groups of preliminary screened fingerprint factors; Step S2, according to the statistics of multiple groups of preliminary screening fingerprint factors, the fingerprint factors that do not meet the screening conditions are eliminated to obtain secondary screening fingerprint factors; Step S3, based on the secondary screening of fingerprint factors, new fingerprint factors are introduced one by one according to preset principles to screen out the best fingerprint factor combination; Step S4, constructing a discriminant model based on the best fingerprint factor combination, inputting the solid waste sample of the to-be-discriminated category into the discriminant model to obtain a discriminant score, and obtaining the category of the solid waste sample of the to-be-discriminated category according to the discriminant rules and the discriminant score of the predetermined category.

[0005] Preferably, the secondary screening fingerprint factor is screened by the following method: The oxide content data of all groups of solid waste samples of a single fingerprint factor are mixed to obtain the average value, and the average value is used as the theoretical prediction value. The number of data with actual observation values ​​greater than the average value and less than the average value in each group of sample data is counted. The number of data is the true frequency, and the theoretical prediction value is the predicted frequency, which is summarized into a contingency table; among them, The degree of deviation between the actual observed value and the theoretical predicted value of the contingency table is calculated to determine whether the difference between the fingerprint factors in different groups of samples is greater than the preset value, and the fingerprint factors with a difference less than the preset value are eliminated to obtain the secondary screening fingerprint factors.

[0006] Preferably, the following formula is used to calculate the degree of deviation between the actual observed value and the theoretical predicted value of the contingency table: (1) In the formula, Indicates the degree of deviation between the actual observed value of the contingency table of the single fingerprint factor and the theoretical predicted value, Indicates The true frequency of the interval, Indicates The predicted frequency of the interval, Indicates the number of intervals, Indicates the interval number.

[0007] Preferably, in step S3, new fingerprint factors are gradually introduced according to the principle of minimum reference statistics.

[0008] Preferably, the reference statistic calculation formula is as follows:

[0009] In the formula, represents the reference statistic, represents the sum of squares of within-group deviations, represents the sum of squares of the between-group deviations.

[0010] Preferably, a discriminant function is used in the discriminant model to calculate the discriminant score, and the discriminant function formula is as follows:

[0011] In the formula represents the discrimination score of a single solid waste sample, is the oxide content data of a single solid waste sample, Indicates the number of individual solid waste samples. The fingerprint factor value of the exponential factor, represents the number of fingerprint factors, Indicates the number of individual solid waste samples. The discriminant coefficient of the exponential factor.

[0012] Preferably, the classification of solid waste samples includes classification of two categories and classification of multiple categories, that is, the predetermined category includes two categories of solid waste and the predetermined category includes multiple categories of solid waste. For two categories, the category judgment is performed according to a preset discrimination score threshold, and for multiple categories, the category judgment is performed according to the degree of proximity between the discrimination score of the solid waste to be judged and each category of solid waste group.

[0013] Preferably, for two categories, the sample data of solid wastes of the same category are input into a single discriminant function, and the discriminant score threshold for distinguishing the two categories of solid wastes is calculated; for the solid waste to be discriminated, its fingerprint factor value is substituted into the discriminant function to obtain the discriminant score, and the discriminant score is compared with the discriminant score threshold to obtain the corresponding solid waste category. The calculation formula of the discriminant score threshold is as follows:

[0014] In the formula, represents the discrimination score threshold, and Respectively represent the mean discrimination scores of the two categories of solid waste.

[0015] Preferably, for multiple categories: the same category of solid waste sample data is input into multiple discriminant functions to calculate multiple discriminant scores respectively, and the average score in each function is taken as the group centroid. For the solid waste to be discriminated, its fingerprint factor value is brought into multiple discriminant functions to obtain the discriminant score, and the comparison parameter is calculated using the following formula:

[0016] The solid waste category to which the comparison parameter is smaller belongs is the category of the solid waste to be identified; In the formula, For comparison parameters, is a vector consisting of multiple discriminant scores, is the centroid vector of the solid waste group, is the covariance matrix and T represents the transpose.

[0017] Preferably, the following method is used to test the accuracy of the discriminant function classification: assuming that the total number of solid waste samples is N, N-1 samples are used to establish the discriminant function each time, and the remaining 1 sample is used for category discrimination. The cycle is repeated until all samples have completed category discrimination, and the accuracy of the discriminant function classification is recorded.

[0018] The present invention also provides a device for constructing a discrimination model based on solid waste fingerprint characteristics, comprising a processor for executing the steps of the above-mentioned method for constructing a discrimination model based on solid waste fingerprint characteristics.

[0019] Compared with the prior art, the present invention has the following beneficial effects: 1. The present invention eliminates fingerprint factors with insignificant differences, obtains secondary screening fingerprint factors, and simplifies the subsequent solid waste identification process.

[0020] 2. For the discrimination of unknown solid wastes in multiple categories, the present invention obtains the average score of each function as the group centroid. For the solid waste to be discriminated, its fingerprint factor value is substituted into multiple discriminant functions to obtain the discrimination score, and the comparison parameter is calculated. The solid waste category with the smaller comparison parameter is the category of the solid waste to be discriminated, so as to accurately identify the unknown solid waste. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 The result of classifying 7 types of solid wastes by using discriminant functions 1 and 2 according to an embodiment of the present invention; Figure 2 The present invention is a flowchart of a method for constructing a discrimination model based on solid waste fingerprint characteristics according to an embodiment of the present invention. DETAILED DESCRIPTION

[0022] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. The described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0023] like Figure 2 As shown, the present invention provides a method for constructing a discrimination model based on solid waste fingerprint characteristics, comprising: Step S1, obtaining multiple groups of solid waste samples of predetermined categories in a predetermined area, selecting Kα spectral line analysis sample oxide content as various fingerprint factors of each of the solid waste samples, eliminating fingerprint factors with no statistical significance, and obtaining multiple groups of preliminary screened fingerprint factors; Step S2, according to the statistics of multiple groups of preliminary screening fingerprint factors, the fingerprint factors that do not meet the screening conditions are eliminated to obtain secondary screening fingerprint factors; Step S3, based on the secondary screening of fingerprint factors, new fingerprint factors are introduced one by one according to preset principles to screen out the best fingerprint factor combination; Step S4, constructing a discriminant model based on the best fingerprint factor combination, inputting the solid waste sample of the to-be-discriminated category into the discriminant model to obtain a discriminant score, and obtaining the category of the solid waste sample of the to-be-discriminated category according to the discriminant rules and the discriminant score of the predetermined category.

[0024] According to a specific embodiment of the present invention, the secondary screening fingerprint factor is screened and obtained by the following method: The oxide content data of all groups of solid waste samples of a single fingerprint factor are mixed to obtain the average value, and the average value is used as the theoretical prediction value. The number of data with actual observation values ​​greater than the average value and less than the average value in each group of sample data is counted. The number of data is the true frequency, and the theoretical prediction value is the predicted frequency, which is summarized into a contingency table; among them, The degree of deviation between the actual observed value and the theoretical predicted value of the contingency table is calculated to determine whether the difference between the fingerprint factors in different groups of samples is greater than the preset value, and the fingerprint factors with a difference less than the preset value are eliminated to obtain the secondary screening fingerprint factors.

[0025] According to a specific embodiment of the present invention, the following formula is used to calculate the degree of deviation between the actual observed value and the theoretical predicted value of the contingency table: (1) In the formula, Indicates the degree of deviation between the actual observed value of the contingency table of the single fingerprint factor and the theoretical predicted value, Indicates The true frequency of the interval, Indicates The predicted frequency of the interval, Indicates the number of intervals, Indicates the interval number.

[0026] According to a specific implementation of the present invention, in step S3, new fingerprint factors are gradually introduced according to the principle of minimum reference statistics.

[0027] According to a specific embodiment of the present invention, the reference statistic calculation formula is as follows:

[0028] In the formula, represents the reference statistic, represents the sum of squares of within-group deviations, represents the sum of squares of the between-group deviations.

[0029] According to a specific embodiment of the present invention, a discriminant function is used in the discriminant model to calculate the discriminant score, and the discriminant function formula is as follows:

[0030] In the formula, represents the discrimination score of a single solid waste sample, is the oxide content data of a single solid waste sample, Indicates the number of individual solid waste samples. The fingerprint factor value of the exponential factor, represents the number of fingerprint factors, Indicates the number of individual solid waste samples. The discriminant coefficient of the exponential factor.

[0031] According to a specific embodiment of the present invention, the classification of solid waste samples includes classification of two categories and multiple categories, that is, the predetermined category includes two categories of solid waste and the predetermined category includes multiple categories of solid waste. For two categories, the category judgment is performed according to a preset discrimination score threshold, and for multiple categories, the category judgment is performed according to the degree of proximity between the discrimination score of the solid waste to be judged and each category of solid waste group.

[0032] According to a specific embodiment of the present invention, for two categories, the solid waste sample data of the same category is input into a single discriminant function, and the discrimination score threshold for distinguishing the two categories of solid waste is calculated; for the solid waste to be discriminated, its fingerprint factor value is brought into the discriminant function to obtain the discrimination score, and the discrimination score is compared with the discrimination score threshold to obtain the corresponding solid waste category. The discrimination score threshold calculation formula is as follows:

[0033] In the formula, represents the discrimination score threshold, and Respectively represent the mean discrimination scores of the two categories of solid waste.

[0034] According to a specific embodiment of the present invention, for multiple categories: the same category of solid waste sample data is input into multiple discriminant functions to calculate multiple discriminant scores, and the average score of each function is taken as the group centroid. For the solid waste to be discriminated, its fingerprint factor value is brought into multiple discriminant functions to obtain the discriminant score, and the comparison parameter is calculated using the following formula:

[0035] The solid waste category to which the comparison parameter is smaller belongs is the category of the solid waste to be identified; In the formula, For comparison parameters, is a vector consisting of multiple discriminant scores, is the centroid vector of the solid waste group, is the covariance matrix and T represents the transpose.

[0036] According to a specific embodiment of the present invention, the following method is used to test the accuracy of discriminant function classification: assuming that the total number of solid waste samples is N, N-1 samples are used to establish the discriminant function each time, and the remaining 1 sample is used for category discrimination. The cycle is repeated until all samples have completed category discrimination, and the accuracy of the discriminant function classification is recorded.

[0037] The present invention also provides a device for constructing a discrimination model based on solid waste fingerprint characteristics, comprising a processor for executing the steps of the above-mentioned method for constructing a discrimination model based on solid waste fingerprint characteristics.

[0038] Example 1 like Figure 2 The present invention provides a method for constructing a discrimination model based on solid waste fingerprint characteristics, comprising: Step S1, obtaining multiple groups of solid waste samples of predetermined categories in a predetermined area, selecting Kα spectral line analysis sample oxide content as various fingerprint factors of each of the solid waste samples, eliminating fingerprint factors with no statistical significance, and obtaining multiple groups of preliminary screened fingerprint factors; Step S2, according to the statistics of multiple groups of preliminary screening fingerprint factors, the fingerprint factors that do not meet the screening conditions are eliminated to obtain secondary screening fingerprint factors; Step S3, based on the secondary screening of fingerprint factors, new fingerprint factors are introduced one by one according to preset principles to screen out the best fingerprint factor combination; Step S4, constructing a discriminant model based on the best fingerprint factor combination, inputting the solid waste sample of the to-be-discriminated category into the discriminant model to obtain a discriminant score, and obtaining the category of the solid waste sample of the to-be-discriminated category according to the discriminant rules and the discriminant score of the predetermined category.

[0039] Example 2 like Figure 2 The present invention provides a method for constructing a discrimination model based on solid waste fingerprint characteristics, comprising: Step S1, obtaining multiple groups of solid waste samples of predetermined categories in a predetermined area, selecting Kα spectral line analysis sample oxide content as various fingerprint factors of each of the solid waste samples, eliminating fingerprint factors with no statistical significance, and obtaining multiple groups of preliminary screened fingerprint factors; Step S2, according to the statistics of multiple groups of preliminary screening fingerprint factors, the fingerprint factors that do not meet the screening conditions are eliminated to obtain secondary screening fingerprint factors; Step S3, based on the secondary screening of fingerprint factors, new fingerprint factors are introduced one by one according to preset principles to screen out the best fingerprint factor combination; Step S4, constructing a discriminant model based on the best fingerprint factor combination, inputting the solid waste sample of the to-be-discriminated category into the discriminant model to obtain a discriminant score, and obtaining the category of the solid waste sample of the to-be-discriminated category according to the discriminant rules and the discriminant score of the predetermined category.

[0040] Furthermore, the secondary screening fingerprint factor is obtained by screening using the following method: The oxide content data of all groups of solid waste samples of a single fingerprint factor are mixed to obtain the average value, and the average value is used as the theoretical prediction value. The number of data with actual observation values ​​greater than the average value and less than the average value in each group of sample data is counted. The number of data is the true frequency, and the theoretical prediction value is the predicted frequency, which is summarized into a contingency table; among them, The degree of deviation between the actual observed value and the theoretical predicted value of the contingency table is calculated to determine whether the difference between the fingerprint factors in different groups of samples is greater than the preset value, and the fingerprint factors with a difference less than the preset value are eliminated to obtain the secondary screening fingerprint factors.

[0041] Furthermore, the following formula is used to calculate the degree of deviation between the actual observed value and the theoretical predicted value of the contingency table: (1) In the formula, Indicates the degree of deviation between the actual observed value of the contingency table of the single fingerprint factor and the theoretical predicted value, Indicates The true frequency of the interval, Indicates The predicted frequency of the interval, Indicates the number of intervals, Indicates the interval number.

[0042] Furthermore, in step S3, new fingerprint factors are gradually introduced according to the principle of minimum reference statistics.

[0043] Furthermore, the reference statistic calculation formula is as follows:

[0044] In the formula, represents the reference statistic, represents the sum of squares of within-group deviations, represents the sum of squares of the between-group deviations.

[0045] Furthermore, the discriminant model uses a discriminant function to calculate the discriminant score, and the discriminant function formula is as follows:

[0046] In the formula, represents the discrimination score of a single solid waste sample, is the oxide content data of a single solid waste sample, Indicates the number of individual solid waste samples. The fingerprint factor value of the exponential factor, represents the number of fingerprint factors, Indicates the number of individual solid waste samples. The discriminant coefficient of the exponential factor.

[0047] Furthermore, the classification of solid waste samples includes classification of two categories and classification of multiple categories, that is, the predetermined category includes two categories of solid waste and the predetermined category includes multiple categories of solid waste. For two categories, the category judgment is performed according to a preset discrimination score threshold, and for multiple categories, the category judgment is performed according to the degree of proximity between the discrimination score of the solid waste to be judged and each category of solid waste group.

[0048] Furthermore, for two categories, the sample data of solid waste of the same category are input into a single discriminant function to calculate the discriminant score threshold for distinguishing the two categories of solid waste; for the solid waste to be discriminated, its fingerprint factor value is substituted into the discriminant function to obtain the discriminant score, and the discriminant score is compared with the discriminant score threshold to obtain the corresponding solid waste category. The calculation formula of the discriminant score threshold is as follows:

[0049] In the formula, represents the discrimination score threshold, and Respectively represent the mean discrimination scores of the two categories of solid waste.

[0050] Furthermore, for multiple categories: the same category of solid waste sample data is input into multiple discriminant functions to calculate multiple discriminant scores, and the average score in each function is taken as the group centroid. For the solid waste to be discriminated, its fingerprint factor value is brought into multiple discriminant functions to obtain the discriminant score, and the comparison parameter is calculated using the following formula:

[0051] The solid waste category to which the comparison parameter is smaller belongs is the category of the solid waste to be identified; In the formula, For comparison parameters, is a vector consisting of multiple discriminant scores, is the centroid vector of the solid waste group, is the covariance matrix and T represents the transpose.

[0052] Furthermore, the following method is used to test the accuracy of the discriminant function classification: assuming that the total number of solid waste samples is N, N-1 samples are used to establish the discriminant function each time, and the remaining 1 sample is used for category discrimination. The cycle is repeated until all samples have completed category discrimination, and the accuracy of the discriminant function classification is recorded.

[0053] Example 3 The following is a detailed description of the method for constructing a discrimination model based on the fingerprint characteristics of solid waste according to a specific discrimination process of unknown solid waste. The specific calculation process that is not detailed is carried out with reference to Example 2.

[0054] Assume that there are seven types of solid waste in a certain area, namely fly ash, coal gangue, desulfurized gypsum, red mud, lithium slag, slag and steel slag. After determining the above predetermined categories, obtain relevant solid waste samples and use a portable X-ray fluorescence spectrometer (XRF) to measure them, and use the Kα spectrum to analyze the oxide content of the sample as the fingerprint factor. More than ten fingerprint factors such as Na2O, MgO, Al2O3, SiO2, P2O5, etc. are obtained through this test method.

[0055] The fingerprint factors measured by the instrument were preliminarily screened, and six factors with no statistical significance, such as CuO and Cd2O3, were removed. In all samples of the above seven types of solid waste, the six fingerprint factors did not reach the instrument detection limit, and the measured content was zero. The 13 fingerprint factors that passed the preliminary screening include Al2O3, MgO, K2O, Na2O, Fe2O3, SiO2, CaO, SO3, TiO2, MnO, As2O3, ZnO, and P2O5.

[0056] Remove redundant fingerprint factors. The null hypothesis of this test is that the content of a certain type of substance in different solid wastes is the same. The following method is used to screen and obtain secondary screening fingerprint factors: The oxide content data of all groups of solid waste samples of a single fingerprint factor are mixed to obtain the average value, and the average value is used as the theoretical prediction value. The number of data with actual observation values ​​greater than the average value and less than the average value in each group of sample data is counted. The number of data is the true frequency, and the theoretical prediction value is the predicted frequency, which is summarized into a contingency table; among them, The degree of deviation between the actual observed value and the theoretical predicted value of the contingency table is calculated to determine whether the difference between the fingerprint factors in different groups of samples is greater than the preset value, and the fingerprint factors with a difference less than the preset value are eliminated to obtain the secondary screening fingerprint factors.

[0057] Furthermore, the following formula is used to calculate the degree of deviation between the actual observed value and the theoretical predicted value of the contingency table: (1) In the formula, Indicates the degree of deviation between the actual observed value of the contingency table of the single fingerprint factor and the theoretical predicted value, Indicates The true frequency of the interval, Indicates The predicted frequency of the interval, Indicates the number of intervals, Indicates the interval number.

[0058] The following is an example of the contingency table using the fingerprint factor Al2O3 content data of 8 fly ash samples and 10 coal gangue samples, as shown in Table 1. The average value is calculated from the Al2O3 content of all 18 samples. First, assuming that there is no difference in the Al2O3 factor between the two solid wastes, then the number of samples with a content greater than the average Al2O3 content and less than the average Al2O3 content in a type of solid waste should be equal. For example, in the fly ash samples, the number of samples greater than the average and the number of samples less than the average should both be 4. This 4 is the theoretical prediction value, and the actual number of samples greater than and less than the average (6 and 4) is the actual observation value. The degree of deviation is calculated by formula (1) to determine whether to reject the hypothesis.

[0059] Table 1. Example of a contingency table

[0060] The calculated deviation value between the actual observed value and the theoretical predicted value is compared with the critical value. If the deviation value between the actual observed value and the theoretical predicted value is large, the null hypothesis is rejected, and it is considered that the inter-group difference of the fingerprint factor is large, and the fingerprint factor can be used to distinguish different types of solid waste.

[0061] The specific fingerprint factor test statistics results are shown in Table 2 below.

[0062] Table 2. Fingerprint factor test statistics results table

[0063] In Table 1, " indicates that the deviation degree value is significant compared with the critical value at the 0.05 level.

[0064] In this embodiment, the significance represents the probability value corresponding to the fingerprint factor, which can be calculated by relevant software, but is not the focus of the present invention and will not be described in detail here.

[0065] The results show that among the 13 fingerprint factors, only three fingerprint factors, namely MnO, As2O3 and P2O5, have no significant differences among the groups. The remaining 10 fingerprint factors, namely Al2O3, MgO, K2O, Na2O, Fe2O3, SiO2, CaO, SO3, TiO2 and ZnO, have significant differences among the seven types of solid waste and can be used as fingerprint indicator factors.

[0066] New fingerprint factors are gradually introduced based on the principle of minimum reference statistics. The reference statistics calculation formula is as follows:

[0067] In the formula, represents the reference statistic, represents the sum of squares of within-group deviations, represents the sum of squares of the between-group deviations.

[0068] The 10 fingerprint factors were analyzed according to the principle of minimum reference statistic. The reference statistic is the ratio of the sum of squares of the intra-group deviations to the sum of squares of the total deviations. It can be used to describe whether there are significant differences in the means between the groups. When the means of all groups are equal, the reference statistic is 1; when the total variation is greater than the intra-group variation, the reference statistic is close to 0. Therefore, the smaller the reference statistic of the composite fingerprint factor group, the greater the difference between the groups.

[0069] Each time a fingerprint factor is introduced to construct the best composite fingerprint factor group, the reference statistics are calculated to ensure that each introduced fingerprint factor minimizes the change in the reference statistics of the overall fingerprint factor group. Taking the first introduction of the fingerprint factor as an example (Table 3 below), among all the candidate factors, ZnO has the smallest change in the reference statistics when it enters the fingerprint factor group, so it is used as the first fingerprint factor to form the best composite fingerprint factor group.

[0070] Table 3. Reference statistics calculated based on the candidate fingerprint factors

[0071] In Table 4, when the composite fingerprint factor group is MgO, K2O, Na2O, Fe2O3, SiO2, CaO, SO3, TiO2 and ZnO, the reference statistic of the composite fingerprint factor group is the smallest, and Al2O3 with a large change in reference statistic is eliminated.

[0072] Table 4. Reference statistics when the composite fingerprint factor group is MgO, K2O, Na2O, Fe2O3, SiO2, CaO, SO3, TiO2 and ZnO

[0073] The discriminant function was constructed using the best composite fingerprint factor group, and the discriminant coefficient of each fingerprint factor was obtained by substituting the covariance matrix into the solution. In this embodiment, a total of 6 discriminant functions were constructed, as shown in Table 5. The use of 6 discriminant functions to identify 7 types of solid waste can explain 100% of the variation.

[0074] Table 5. Results of using six discriminant functions to identify seven types of solid waste

[0075] The resulting discriminant function is shown below:

[0076]

[0077]

[0078]

[0079]

[0080]

[0081] Taking the classification of lithium slag and furnace slag by function 1 as an example, the two types of solid waste samples are substituted into function 1 to calculate the average discrimination scores of 17.80 and -2.89, and the discrimination score threshold is 7.455. Assuming that the discrimination score of an unknown solid waste sample substituted into function 1 is greater than the discrimination score threshold, it is classified as lithium slag, otherwise it is classified as furnace slag.

[0082] Figure 1 To discriminate the results of the classification of 7 types of solid waste by functions 1 and 2, the discrimination of multiple types of solid waste takes the classification of desulfurized gypsum, lithium slag and slag by functions 1 and 2 as an example. The three types of solid waste samples are substituted into functions 1 and 2 to calculate the group centroid composed of the average discrimination score and the discrimination score of an unknown sample. At the same time, the following formula is used to calculate the comparison parameters:

[0083] The solid waste category to which the comparison parameter is smaller belongs is the category of the solid waste to be identified; In the formula, For comparison parameters, is a vector consisting of multiple discriminant scores, is the centroid vector of the solid waste group, is the covariance matrix, T represents the transpose; The obtained comparison parameters are shown in Table 6. The comparison parameters of the unknown sample are closest to the centroid of the lithium slag group and are identified as lithium slag by the discrimination model.

[0084] Table 6. Comparison parameters

[0085] The following method is used to test the accuracy of the discriminant function classification: in this embodiment, the total number of solid waste samples is 84, 83 samples are used each time to establish the discriminant function, and the remaining 1 sample is used for type discrimination, and the cycle is repeated until all samples have completed type discrimination. The training sample used to establish the discriminant function is only 1 less than the initial data set, which makes the trained model basically the same as the expected model. This verification method is closer to the actual results and is especially suitable for cases with small data sets.

[0086] Table 7. Verification results of 84 samples in this embodiment

[0087] The verification results are shown in Table 7. The final discriminant function classification prediction accuracy is 92.9%, and the accuracy rate exceeds 70%, which is a relatively accurate model.

[0088] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention are included in the protection scope of the present invention.

Claims

1. A method for constructing a discrimination model based on solid waste fingerprint characteristics, characterized in that: include: Step S1, obtaining multiple groups of solid waste samples of predetermined categories in a predetermined area, selecting Kα spectral line analysis sample oxide content as various fingerprint factors of each of the solid waste samples, eliminating fingerprint factors with no statistical significance, and obtaining multiple groups of preliminary screened fingerprint factors; Step S2, according to the statistics of multiple groups of preliminary screening fingerprint factors, the fingerprint factors that do not meet the screening conditions are eliminated to obtain secondary screening fingerprint factors; Step S3, based on the secondary screening of fingerprint factors, new fingerprint factors are introduced one by one according to the preset principle to screen out the best fingerprint factor combination; Step S4, constructing a discriminant model based on the best fingerprint factor combination, inputting the solid waste sample of the to-be-discriminated category into the discriminant model to obtain a discriminant score, and obtaining the category of the solid waste sample of the to-be-discriminated category according to the discriminant rules and the discriminant score of the predetermined category.

2. The method for constructing a discrimination model based on solid waste fingerprint characteristics according to claim 1, characterized in that: In step S2, the secondary screening fingerprint factor is screened by the following method: The oxide content data of all groups of solid waste samples of a single fingerprint factor are mixed to obtain the average value, and the average value is used as the theoretical prediction value. The number of data with actual observation values ​​greater than the average value and less than the average value in each group of sample data is counted. The number of data is the true frequency, and the theoretical prediction value is the predicted frequency, which is summarized into a contingency table; among them, The degree of deviation between the actual observed value and the theoretical predicted value of the contingency table is calculated to determine whether the difference between the fingerprint factors in different groups of samples is greater than the preset value, and the fingerprint factors with a difference less than the preset value are eliminated to obtain the secondary screening fingerprint factors.

3. The method for constructing a discrimination model based on solid waste fingerprint characteristics according to claim 2, characterized in that: The following formula is used to calculate the degree of deviation between the actual observed value and the theoretical predicted value of the contingency table: （1） In the formula, Indicates the degree of deviation between the actual observed value of the contingency table of the single fingerprint factor and the theoretical predicted value, Indicates The true frequency of the interval, Indicates The predicted frequency of the interval, Indicates the number of intervals, Indicates the interval number.

4. The method for constructing a discrimination model based on solid waste fingerprint characteristics according to claim 1, characterized in that: In step S3, new fingerprint factors are gradually introduced according to the principle of minimum reference statistics.

5. The method for constructing a discrimination model based on solid waste fingerprint characteristics according to claim 4, characterized in that: The reference statistic calculation formula is as follows: In the formula, represents the reference statistic, represents the sum of squares of within-group deviations, represents the sum of squares of the between-group differences.

6. The method for constructing a discrimination model based on solid waste fingerprint characteristics according to claim 1, characterized in that: The discriminant model uses a discriminant function to calculate the discriminant score, and the discriminant function formula is as follows: In the formula, represents the discrimination score of a single solid waste sample, is the oxide content data of a single solid waste sample, Indicates the number of individual solid waste samples. The fingerprint factor value of the exponential factor, represents the number of fingerprint factors, Indicates the number of individual solid waste samples. The discriminant coefficient of the exponential factor.

7. The method for constructing a discrimination model based on solid waste fingerprint characteristics according to claim 6, characterized in that: The classification of solid waste samples includes two categories and multiple categories, that is, the predetermined category includes two categories of solid waste and the predetermined category includes multiple categories of solid waste. For two categories, the category judgment is performed according to a preset discrimination score threshold, and for multiple categories, the category judgment is performed according to the degree of proximity between the discrimination score of the solid waste to be judged and each category of solid waste groups.

8. The method for constructing a discrimination model based on solid waste fingerprint characteristics according to claim 7, characterized in that: For two categories, the sample data of solid waste of the same category are input into a single discriminant function, and the discriminant score threshold for distinguishing the two categories of solid waste is calculated; for the solid waste to be discriminated, its fingerprint factor value is brought into the discriminant function to obtain the discriminant score, and the discriminant score is compared with the discriminant score threshold to obtain its corresponding solid waste category. The calculation formula of the discriminant score threshold is as follows: In the formula, represents the discrimination score threshold, and Respectively represent the mean discrimination scores of the two categories of solid waste.

9. The method for constructing a discrimination model based on solid waste fingerprint characteristics according to claim 7, characterized in that: For multiple categories: input the same category of solid waste sample data into multiple discriminant functions to calculate multiple discriminant scores, and take the average score of each function as the group centroid. For the solid waste to be discriminated, bring its fingerprint factor value into multiple discriminant functions to obtain the discriminant score, and use the following formula to calculate the comparison parameter: The solid waste category to which the comparison parameter is smaller belongs is the category of the solid waste to be identified; In the formula, For comparison parameters, is a vector consisting of multiple discriminant scores, is the centroid vector of the solid waste group, is the covariance matrix and T represents the transpose.

10. A device for constructing a discrimination model based on solid waste fingerprint characteristics, characterized in that: It includes a processor for executing the steps of the method for constructing a discrimination model based on solid waste fingerprint characteristics as described in any one of claims 1 to 9.

Images