A coal slag sample classification detection method and system

Abstract

The present invention belongs to the field of spectral analysis technology, and specifically relates to a method and system for classifying and detecting coal slag samples. The method comprises: obtaining a noise-reduced spectrum of each slag sample; obtaining the noise index of each wavelength of the spectrum of each slag sample based on the difference between the spectrum of the slag sample and the corresponding noise-reduced spectrum and the fluctuation frequency of the spectrum of the slag sample; obtaining the structural performance of each wavelength based on the difference in the noise index of each wavelength in all different slag samples of the same type; obtaining the component performance of each wavelength based on the difference in absorbance at the same wavelength of different slags and the difference in absorbance at the same wavelength of slag samples of the same type with different degrees of crushing; obtaining characteristic wavelengths by combining the structural performance and component performance of the wavelengths, and establishing a classification detection model for the slag samples. The present invention constructs a classification detection model for slag samples, thereby improving the accuracy and efficiency of classification detection of slag samples.

Description

Technical Field

[0001] The present invention relates to the field of spectral analysis technology, and more specifically, to a method and system for classifying and detecting coal slag samples. Background Art

[0002] Coal is a complex mixture consisting primarily of moisture, ash, and fixed carbon. Different compositions of coal have different industrial uses. For example, anthracite is used to make coal gas or directly as fuel. Different types of coal produce different types of slag, which are categorized into eight types: powdery, weakly bonded, non-expanding, and strongly impact-fused. Furthermore, the combustion properties of coal also influence the composition and morphology of slag. Therefore, analyzing the composition and type of slag can provide insights into coal combustion properties, quality, and type.

[0003] To classify and detect coal slag, near-infrared spectroscopy is currently commonly used to analyze its composition. Characteristic groups in the slag, such as CH, OH, and SH, absorb near-infrared light, generating characteristic peaks in the spectrum, enabling component detection. A neural network is then trained to identify the composition of different types of slag to complete slag classification. For example, Chinese patent application CN106990066B discloses a method and apparatus for identifying coal types. This method utilizes a coal classification model based on visible-near-infrared hyperspectral data to obtain coal type information, and employs a multi-layer perceptron classification model to improve classification accuracy, enabling non-destructive identification of coal samples.

[0004] However, there are certain differences between coal samples and slag samples. Slag can be powdery or lumpy, containing bubbles. These structures affect near-infrared light reflection, weakening absorption and increasing spectral noise, affecting component detection and slag classification. Grinding all samples loses the morphological characteristics of the slag, requiring complex pre-processing for slag classification and reducing classification efficiency. Summary of the Invention

[0005] In order to solve the above-mentioned technical problems of insufficient accuracy and low efficiency in classification and detection of coke residue samples, the present invention provides solutions in the following aspects.

[0006] In a first aspect, the present invention provides a method for classifying and detecting coal slag samples, comprising:

[0007] Several types of coke slag are obtained, each type of coke slag is crushed and sampled to obtain coke slag samples of several crushing degrees of various coke slags, and a near-infrared spectrometer is used to obtain the spectrum of each coke slag sample, wherein the spectrum is an absorbance sequence corresponding to several wavelengths; based on a filtering algorithm, a noise reduction spectrum of each coke slag sample is obtained; according to the difference between the spectrum of the coke slag sample and the corresponding noise reduction spectrum and the fluctuation frequency of the spectrum of the coke slag sample, the noise index of each wavelength of the spectrum of each coke slag sample is obtained; according to the law of the difference in the noise index of each wavelength in all coke slag samples of the same type, the structural performance of each wavelength is obtained; according to the difference in absorbance at the same wavelength of different coke slags and the difference in absorbance at the same wavelength of coke slag samples of the same type but with different crushing degrees, the component performance of each wavelength is obtained; the characteristic wavelength is obtained by combining the structural performance and component performance of the wavelength; based on the characteristic wavelength and the neural network model, a classification and detection model for coke slag samples is established.

[0008] This method analyzes the spectra of coke slag samples with varying degrees of pulverization to determine the effect of pulverization on the slag sample spectra. This analysis then identifies the structural representation of the spectral wavelengths and uses spectral noise as a feature for classification and detection of the slag samples. This method establishes a classification and detection model for slag samples, enabling automated classification and detection of coal slag samples without complex pre-processing, thereby improving the efficiency of slag sample classification and detection.

[0009] Preferably, obtaining the noise index of each wavelength of the spectrum of each coke residue sample includes:

[0010] The spectrum of any coke slag sample is recorded as a target spectrum, and any wavelength of the target spectrum is recorded as a target wavelength; the neighborhood of the target wavelength is obtained; the error sequence of the target spectrum and the error sign difference sequence of the target spectrum are obtained;

[0011] ;

[0012] Where, Indicates the noise figure at the target wavelength; The number of wavelengths representing the neighborhood of the target wavelength; The i-th value of the neighborhood corresponding to the target wavelength in the error sequence of the target spectrum; The cth value of the neighborhood corresponding to the target wavelength in the error sign difference sequence of the target spectrum; represents the absolute value function.

[0013] The present invention obtains the noise index of each wavelength of the spectrum of the coke slag sample, can quantify the noise generated by the coke slag sample due to factors such as the porous structure, and improves the accuracy of classification detection of the coke slag sample.

[0014] Preferably, the step of obtaining the error sequence of the target spectrum and the error symbol difference sequence of the target spectrum includes:

[0015] The absorbance sequence of the target spectrum is subtracted from the absorbance sequence of the corresponding denoised spectrum to obtain the error sequence of the target spectrum; the error sequence of the target spectrum is symbolized using a sign function to obtain the error symbol sequence of the target spectrum; the error symbol sequence of the target spectrum is forward differentiated to obtain the error symbol difference sequence of the target spectrum.

[0016] Preferably, the obtaining of the structural performance of each wavelength includes: taking any coke slag as the target coke slag, obtaining the noise index of the ath wavelength of the spectrum of all coke slag samples of the target coke slag, using the least squares method to perform straight line fitting to obtain the noise index fitting line of the ath wavelength of the target coke slag; based on the fitting effect of the noise index fitting line of the ath wavelength of the target coke slag and the distribution law of the noise index of the ath wavelength of the spectrum of all coke slag samples of the target coke slag according to the degree of crushing, obtaining the structural characteristics of the ath wavelength relative to the target coke slag; and taking the average of the structural characteristics of the ath wavelength relative to all types of coke slag as the structural performance of the ath wavelength.

[0017] By fitting the noise index of the same wavelength of all coke slag samples of the same type of coke slag in the present invention, it is possible to obtain the regular changes of the same wavelength of the same type of coke slag as the degree of crushing of the coke slag samples changes, thereby obtaining the structural performance of the wavelength, making the characteristic dimensions of the classification detection of the coke slag samples more, and improving the accuracy of the classification detection of the coke slag samples.

[0018] Preferably, the ath wavelength satisfies the expression relative to the structural characteristics of the target coke residue:

[0019] ;

[0020] Where, Indicates the structural characteristics of the ath wavelength relative to the target coke residue; The number of slag samples representing the target slag; It represents the distance between the hth coke residue sample of the target coke residue and the noise index fitting line of the ath wavelength of the target coke residue; Indicates the type of crushing degree; 、 、 The average noise index of the ath wavelength of the spectrum of the coke slag sample representing the q+1th, qth, and q-1th degree of crushing of the target coke slag; represents an exponential function with a natural constant as its base; Represents the normalization function.

[0021] Preferably, obtaining the component representation of each wavelength includes:

[0022] Obtain the absorbance difference of each wavelength relative to all coke residues; obtain the regularity of the absorbance change of each wavelength relative to all coke residues;

[0023] The components of any wavelength satisfy the expression:

[0024] ;

[0025] Where, Indicates the component performance of the ath wavelength; It represents the absorbance difference of the ath wavelength relative to all coke residues; Indicates the regularity of the absorbance variation of the ath wavelength relative to all coke residues; Indicates the number of coke slag types; represents the absorbance variance of the ath wavelength of all coke residue samples of the uth type; Represents an exponential function with a natural constant as its base.

[0026] The present invention obtains the component performance of the wavelength through the difference in absorbance of the same wavelength of different types of coke slag, the change law of absorbance of the same wavelength of different types of coke slag, and the discrete situation of absorbance of the same wavelength of coke slag samples of the same type of coke slag, and obtains the wavelength that can characterize the difference in component content of different types of coke slag, thereby ensuring the accuracy of classification detection of coke slag samples.

[0027] Preferably, the absorbance difference of each wavelength relative to all coke residues satisfies the expression:

[0028] ;

[0029] Where, It represents the absorbance difference of the ath wavelength relative to all coke residues; Indicates the number of coke slag types; 、 It represents the average absorbance of the ath wavelength of the uth and vth types of coke residues; represents the absolute value function; Represents the normalization function.

[0030] Preferably, obtaining the regularity of absorbance variation of each wavelength relative to all coke residues includes:

[0031] Obtain the average absorbance sequence of the ath wavelength of all types of coke residues in ascending order of type number, perform forward difference on the average absorbance sequence of the ath wavelength of all types of coke residues, and obtain the difference sequence of the average absorbance sequence of the ath wavelength of all types of coke residues;

[0032] The number of values ​​greater than 0 in the difference sequence of the average absorbance sequence of the ath wavelength of all types of coke residues is compared with the number of values ​​less than 0, and recorded as the regularity of the absorbance change of the ath wavelength relative to all coke residues.

[0033] Preferably, the method of obtaining the characteristic wavelength includes: comparing the structural performance of the a-th wavelength with the composition performance as the first ratio of the a-th wavelength; comparing the composition performance of the a-th wavelength with the structural performance as the second ratio of the a-th wavelength; taking the maximum value of the first ratio and the second ratio of the a-th wavelength as the characteristic goodness of the a-th wavelength; and obtaining the Z wavelengths with the largest characteristic goodness among all wavelengths as the characteristic wavelengths.

[0034] The present invention integrates the structural performance and composition performance of the wavelength to obtain several characteristic wavelengths with the strongest characteristic performance, thereby further improving the accuracy of classification detection of coke slag samples.

[0035] In a second aspect, the present invention provides a coal slag sample classification and detection system, comprising a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the above-mentioned coal slag sample classification and detection method is implemented.

[0036] By adopting the above technical solution, the above-mentioned coal slag sample classification detection method is generated into a computer program and stored in a memory to be loaded and executed by a processor, so that a terminal device is made based on the memory and the processor for easy use.

[0037] The beneficial effects of the present invention are:

[0038] (1) The present invention extracts several effective wavelengths from the wavelength range of near-infrared light as characteristic wavelengths, thereby improving the classification accuracy of the classification detection model for coke slag samples affected by invalid wavelengths;

[0039] (2) The present invention uses spectral noise as a feature to classify and detect coke slag samples, thereby increasing the feature richness of coke slag sample classification detection;

[0040] (3) The present invention establishes a classification detection model for coal slag samples by extracting several characteristic wavelengths that characterize the types of coal slag, so that the slag samples can be automatically classified and detected more accurately and efficiently. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 is a flow chart schematically illustrating a method for classifying and detecting coal slag samples in the present invention;

[0042] Figure 2 Schematic diagram schematically showing eight types of coke residues. DETAILED DESCRIPTION

[0043] The embodiment of the present invention discloses a method for classifying and detecting coal slag samples, referring to Figure 1 , including steps S1 to S4:

[0044] S1: obtaining several kinds of coke slag, crushing and sampling each kind of coke slag to obtain coke slag samples of several crushing degrees of each kind of coke slag, and using a near-infrared spectrometer to obtain a spectrum of each coke slag sample.

[0045] It should be noted that slag is divided into eight types according to its physical properties such as viscosity, melting degree and expansion height, such as Figure 2 This is a schematic diagram of eight types of coke residue. As the type number increases, the viscosity, degree of melting, and expansion height of the coke residue gradually increase. These physical properties will affect the intensity transmission of near-infrared light, and thus affect the spectral results and component analysis. Therefore, the present invention analyzes the impact of the morphology of coke residue on the spectrum, eliminates the complex pre-processing before coke residue spectral detection, and improves the accuracy of classification detection of coke residue samples.

[0046] Specifically, several types of coke slag are obtained, wherein the coke slag includes all types of coke slag; a crushing device is used to crush and sample the various coke slags in several batches to obtain coke slag samples with several crushing degrees of the various coke slags; a near-infrared spectrometer is used to obtain a spectrum of each coke slag sample, wherein the spectrum is an absorbance sequence corresponding to several wavelengths.

[0047] The pulverization device is capable of crushing the slag into a specified size, such as a small pulverizer. The degree of pulverization includes several finenesses, such as 60 mesh and 120 mesh, as well as a pulverization degree of 0, which indicates the original slag fineness without pulverization. The volume of the slag sample is uniform and can be set by the implementer based on actual implementation. For example, the volume can be set to 10 mL. The wavelength range of the spectrum is 780-2526 nm.

[0048] It should be noted that in order to ensure the robustness of the detection, the number of slag samples with the same slag and the same degree of crushing is greater than 1.

[0049] So far, spectra of coke slag samples of several types and at several degrees of pulverization of coke slag have been obtained.

[0050] S2: Based on the filtering algorithm, the noise reduction spectrum of each coke slag sample is obtained; according to the difference between the spectrum of the coke slag sample and the corresponding noise reduction spectrum and the fluctuation frequency of the spectrum of the coke slag sample, the noise index of each wavelength of the spectrum of each coke slag sample is obtained; according to the difference in the noise index of each wavelength in all different coke slag samples of the same type of coke slag, the structural performance of each wavelength is obtained.

[0051] It should be noted that the coke slag with a higher expansion height has a porous structure, which will affect the reflection of light and lead to a decrease in the spectral signal-to-noise ratio. Therefore, there are certain noise differences in the spectra of coke slag samples with different degrees of crushing of the coke slag with a higher expansion height. This difference reflects the structural characteristics of the coke slag. Therefore, the present invention obtains the structural characteristics of the coke slag at each wavelength based on the spectral noise differences of different coke slag samples of the same coke slag.

[0052] It should be noted that filtering algorithms are commonly used to denoise spectral data. For example, SG filtering smoothes and denoises the spectrum by performing a k-order polynomial least squares fit on the data within a sliding window. Working backwards, the denoising effect of the filtering algorithm can be used to determine the noise profile of the spectrum. Therefore, the present invention filters and denoises the spectra of the coke residue samples. Based on the difference in absorbance at each wavelength between the denoised spectrum and the original spectrum, the noise index at each wavelength of the spectrum of each coke residue sample is obtained.

[0053] Specifically, the spectra of all coke residue samples are filtered using the SG filtering algorithm to obtain a de-noised spectrum of the spectra of each coke residue sample. It should be noted that the parameters of the SG filtering algorithm include the polynomial order and window length. The parameters are set by the implementer based on the actual implementation situation. For example, the polynomial order can be set to 3, and the window length can be set to 5.

[0054] It should be noted that for any coke slag sample spectrum and its corresponding noise-reduced spectrum, the greater the difference in absorbance between wavelengths within the same wavelength neighborhood, the greater the degree of noise reduction and the higher the wavelength noise index. Furthermore, noise fluctuates; the coke slag sample spectrum fluctuates around the noise-reduced spectrum. The higher the frequency of fluctuation in the coke slag sample spectrum, the higher the wavelength noise index.

[0055] Preferably, the noise index of each wavelength of the spectrum of each coke slag sample is obtained according to the difference between the spectrum of the coke slag sample and the corresponding noise reduction spectrum and the fluctuation frequency of the spectrum of the coke slag sample:

[0056] The spectrum of any coke residue sample is recorded as the target spectrum, any wavelength of the target spectrum is recorded as the target wavelength, and the target wavelength and its A wavelengths to the left and A wavelengths to the right are recorded as the target wavelength neighborhood. It should be noted that A is set by the implementer based on actual implementation conditions; for example, A can be set to 5.

[0057] The target spectrum's absorbance sequence is subtracted from the corresponding denoised spectrum's absorbance sequence to obtain the target spectrum's error sequence. The target spectrum's error sequence is symbolized using a sign function to obtain the target spectrum's error symbol sequence. The target spectrum's error symbol sequence is forward-differentiated to obtain the target spectrum's error symbol difference sequence. It should be noted that the target spectrum's error symbol sequence only contains -1, 0, and 1, removing the physical meaning of the specific numerical values ​​and representing the error direction of each wavelength in the target spectrum. The target spectrum's error symbol difference sequence only contains -1, 0, and 1. The number of -1s and 1s reflects the target spectrum's fluctuation frequency. The more 0s in the target spectrum's error symbol difference sequence, the more stable the sequence and the lower the fluctuation frequency. The more -1s and 1s in the target spectrum's error symbol difference sequence, the higher the frequency of the sequence's fluctuations.

[0058] The noise figure at the target wavelength satisfies the expression:

[0059] ;

[0060] Where, Indicates the noise figure at the target wavelength; The number of wavelengths representing the neighborhood of the target wavelength; The i-th value of the neighborhood corresponding to the target wavelength in the error sequence of the target spectrum; The cth value of the neighborhood corresponding to the target wavelength in the error sign difference sequence of the target spectrum; represents the absolute value function.

[0061] Where, Indicates that the values ​​of the corresponding positions of the target wavelength and the neighboring wavelengths in the error sequence of the target spectrum are accumulated to represent the total amount of error in the neighborhood of the target wavelength. The larger the value, the greater the noise index of the target wavelength. It means that the values ​​of the corresponding positions of the target wavelength and the neighboring wavelengths in the error symbol difference sequence of the target spectrum are accumulated, which represents the fluctuation frequency in the neighborhood of the target wavelength. The larger the value, the higher the fluctuation frequency in the neighborhood of the target wavelength, which means that the noise index of the target wavelength is larger.

[0062] At this point, the noise index of each wavelength of the spectrum of each slag sample has been obtained.

[0063] It should be noted that for coke slag samples of the same type with different degrees of crushing, if the wavelength noise index decreases as the degree of crushing increases, it means that the wavelength can strongly reflect the structural characteristics of the coke slag; if the wavelength noise index can reflect the structural characteristics of the coke slag in all coke slags, it means that the degree of crushing affects the absorbance data of the target wavelength. The absorbance of the wavelength can reversely characterize the degree of crushing of the coke slag. Because the degree of crushing of the coke slag is relatively intuitive and can be easily measured, the present invention extracts the wavelength that reflects the structural characteristics of the coke slag. On the one hand, it can be used as a feature to characterize the type of coke slag, and on the other hand, it can be used as a feature to characterize the volume of the coke slag sample, thereby improving the accuracy of the classification and detection of coke slag samples.

[0064] Preferably, the structural performance of each wavelength is obtained according to the rule of the difference in noise index of each wavelength in all samples of the same coke residue and different coke residues:

[0065] Taking any coke residue as the target coke residue, the noise index of the ath wavelength of the spectrum of all coke residue samples of the target coke residue is obtained, and the least squares method is used for straight line fitting to obtain the noise index fitting line of the ath wavelength of the target coke residue.

[0066] It should be noted that the better the fitting effect of the noise index fitting line of the a-th wavelength of the target coke residue, and the more regularly it is distributed according to the degree of crushing, the stronger the structural characteristics of the a-th wavelength relative to the target coke residue.

[0067] The structural characteristics of each wavelength relative to the target coke residue satisfy the expression:

[0068] ;

[0069] Where, Indicates the structural characteristics of the ath wavelength relative to the target coke residue; The number of slag samples representing the target slag; It represents the distance between the hth coke residue sample of the target coke residue and the noise index fitting line of the ath wavelength of the target coke residue; Indicates the type of crushing degree; 、 、 The average noise index of the ath wavelength of the spectrum of the coke slag sample representing the q+1th, qth, and q-1th degree of crushing of the target coke slag; represents an exponential function with a natural constant as its base; Represents the normalization function.

[0070] Where, and Compared with the average distance between all coke residue samples representing the target coke residue and the noise index fitting line of the a-th wavelength of the target coke residue, this value reflects the fitting effect of the noise index fitting line of the a-th wavelength of the target coke residue. The larger the value, the worse the fitting effect of the noise index fitting line of the a-th wavelength of the target coke residue, and the worse the structural characteristics of the a-th wavelength relative to the target coke residue; It indicates the regularity of the change of the noise index of the coke slag samples with adjacent crushing degrees of the target coke slag. The larger the value, the more obvious the change of the noise index of the coke slag sample of the target coke slag with the crushing degree, thereby reflecting that the structural characteristics of the ath wavelength relative to the target coke slag are stronger.

[0071] It should be noted that if the noise index of the ath wavelength can strongly reflect the structural characteristics of the coke slag in all coke slags, the structural performance of the ath wavelength is higher.

[0072] The average value of the structural characteristics of the ath wavelength relative to all types of coke residue is taken as the structural representation of the ath wavelength.

[0073] So far, the structural representation of arbitrary wavelength has been obtained.

[0074] S3: According to the consistency of the absorbance difference of different coke residues at the same wavelength and the difference in the types of coke residues, the component characteristics of each wavelength relative to any type of coke residue are obtained.

[0075] It should be noted that the essential reason for the morphological differences among different types of coke residues is the difference in composition. With the decrease of ash and volatile matter and the increase of colloid content, the coke residue gradually shows adhesiveness and expansibility. Therefore, among all wavelengths, the wavelength whose absorbance varies with the type of coke residue and does not vary with the degree of coke residue crushing can better be used as a wavelength that reflects the composition of the coke residue.

[0076] Specifically, the absorbance difference of any wavelength relative to all coke residues satisfies the expression:

[0077] ;

[0078] Where, It represents the absorbance difference of the ath wavelength relative to all coke residues; Indicates the number of coke slag types; 、 It represents the average absorbance of the ath wavelength of the uth and vth types of coke residues; represents the absolute value function; Represents the normalization function.

[0079] Where, It represents the average absorbance difference of the ath wavelength of all different types of coke residues. The larger the value, the greater the difference in the absorbance of the ath wavelength of different types of coke residues, and the greater the absorbance difference of the ath wavelength relative to all coke residues.

[0080] It should be noted that, the greater the difference in absorbance of the a-th wavelength relative to all coke residues, and the closer the absorbance of the a-th wavelength of the same coke residues are, the more suitable the a-th wavelength is for component detection; in addition, the absorbance of the a-th wavelength changes regularly with the increase of the type number, which means that the reason for the difference in absorbance of the a-th wavelength due to the type of coke residue is the gradual change in composition, thus indicating that the a-th wavelength can be used for component detection.

[0081] Preferably, the average absorbance sequence of the ath wavelength of all types of coke slag is obtained in ascending order of type number, and the average absorbance sequence of the ath wavelength of all types of coke slag is forward differentiated to obtain the differential sequence of the average absorbance sequence of the ath wavelength of all types of coke slag; the number of values ​​greater than 0 in the differential sequence of the average absorbance sequence of the ath wavelength of all types of coke slag is compared with the number of values ​​less than 0, and the result is recorded as the regularity of absorbance change of the ath wavelength relative to all coke slags.

[0082] Preferably, the component performance at any wavelength satisfies the expression:

[0083] ;

[0084] Where, Indicates the component performance of the ath wavelength; It represents the absorbance difference of the ath wavelength relative to all coke residues; Indicates the regularity of the absorbance variation of the ath wavelength relative to all coke residues; Indicates the number of coke slag types; represents the absorbance variance of the ath wavelength of all coke residue samples of the uth type; Represents an exponential function with a natural constant as its base.

[0085] In the formula, the greater the regularity of the absorbance change of the ath wavelength relative to all coke residues, the more consistent the absorbance change of the ath wavelength is with the change of composition; It represents the sum of the absorbance variances of all coke residue samples at the a-th wavelength. The larger the value, the more discrete the absorbance of all coke residues at the a-th wavelength. Therefore, the absorbance variation of the a-th wavelength is less stable in coke residue samples of the same coke residue, making it difficult to use it as a characteristic detection component.

[0086] Thus, the component expression of arbitrary wavelengths is obtained.

[0087] S4: Combine the structural and compositional characteristics of the wavelength to obtain the characteristic wavelength; based on the characteristic wavelength and the neural network model, establish a classification and detection model for the coke slag sample.

[0088] It should be noted that, since the wavelength with greater structural expression has greater absorbance difference in the coke slag samples with different crushing degrees of the same coke slag, and the wavelength with greater composition expression has smaller absorbance difference in the coke slag samples with different crushing degrees of the same coke slag, it is impossible for a wavelength to have both high structural expression and composition expression at the same time. In order to make the characteristic significance of the wavelength stronger, it is necessary to screen the wavelength so that the wavelength only has strong structural expression significance or composition expression significance.

[0089] It should be noted that since the slag type and degree of pulverization corresponding to the slag samples are known, a supervised machine learning model can be used to learn the absorbance at the characteristic wavelengths of all slag samples, thereby enabling automated sample detection of subsequent coal slag samples. Random forests have strong generalization capabilities and can handle a large number of features, which is consistent with the characteristics of spectral data. Therefore, this paper uses random forests to construct a classification detection model for slag samples.

[0090] Specifically, the structural performance and the compositional performance at the ath wavelength are compared as the first ratio of the ath wavelength; the compositional performance and the structural performance at the ath wavelength are compared as the second ratio of the ath wavelength; the maximum of the first and second ratios of the ath wavelength is taken as the characteristic goodness of the ath wavelength; and the Z wavelengths with the largest characteristic goodness among all wavelengths are obtained as the characteristic wavelengths. It should be noted that the Z value is set by the implementer based on actual implementation conditions; for example, the Z value can be set to 20.

[0091] Construct a classification and detection model for slag samples: The absorbance at the characteristic wavelengths of all slag samples is used as a training set. Bootstrap sampling is performed on the training set to generate several sub-datasets. Randomly select features from each sub-dataset to construct a decision tree. All decision trees form a random forest, which serves as the classification and detection model for slag samples. The input is the absorbance at the characteristic wavelength of the slag sample, and the output is the type of slag sample. The number of decision trees is set by the implementer based on actual implementation circumstances; for example, the number of decision trees can be set to 100.

[0092] The absorbance of the characteristic wavelength of the coal slag sample to be tested is input into the classification detection model of the slag sample to obtain the type of the coal slag sample to be tested.

[0093] The classification and testing of coal slag samples have been completed.

[0094] An embodiment of the present invention further discloses a coal slag sample classification detection system, comprising a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, a coal slag sample classification detection method according to the present invention is implemented.

[0095] The above system also includes other components well known to those skilled in the art, such as a communication bus and a communication interface. The configuration and functions of these components are known in the art and will not be described in detail here.

[0096] Although this specification has shown and described several embodiments of the present invention, it is obvious to those skilled in the art that such embodiments are provided by way of example only. Those skilled in the art will conceive of many modifications, changes and substitutions without departing from the idea and spirit of the present invention.

Claims

1. A method for classifying and detecting coal slag samples, characterized in that: include: Obtaining several types of coke residues, crushing and sampling each type of coke residue to obtain coke residue samples of various crushing degrees of each type of coke residue, and using a near-infrared spectrometer to obtain a spectrum of each coke residue sample, wherein the spectrum is an absorbance sequence corresponding to several wavelengths; Based on the filtering algorithm, the noise-reduced spectrum of each coke residue sample is obtained; According to the difference between the spectrum of the coke slag sample and the corresponding noise reduction spectrum and the fluctuation frequency of the spectrum of the coke slag sample, the noise index of each wavelength of the spectrum of each coke slag sample is obtained, including: recording the spectrum of any coke slag sample as the target spectrum and recording any wavelength of the target spectrum as the target wavelength; obtaining the neighborhood of the target wavelength; obtaining the error sequence of the target spectrum and the error sign difference sequence of the target spectrum; , represents the noise figure at the target wavelength, represents the number of wavelengths in the neighborhood of the target wavelength, Represents the i-th value of the neighborhood corresponding to the target wavelength in the error sequence of the target spectrum, The cth value of the neighborhood corresponding to the target wavelength in the error symbol difference sequence of the target spectrum, represents the absolute value function; According to the law of the difference in noise index of each wavelength in all different coke slag samples of the same type, the structural performance of each wavelength is obtained, including: taking any coke slag as the target coke slag, obtaining the noise index of the ath wavelength of the spectrum of all coke slag samples of the target coke slag, using the least squares method to perform straight line fitting to obtain the noise index fitting line of the ath wavelength of the target coke slag; based on the fitting effect of the noise index fitting line of the ath wavelength of the target coke slag and the distribution law of the noise index of the ath wavelength of the spectrum of all coke slag samples of the target coke slag according to the degree of crushing, obtaining the structural characteristics of the ath wavelength relative to the target coke slag; and taking the average of the structural characteristics of the ath wavelength relative to all types of coke slag as the structural performance of the ath wavelength; Based on the absorbance difference of the same wavelength of different coke residues and the absorbance difference of the same wavelength of coke residue samples with different crushing degrees, the component performance of each wavelength is obtained, including: obtaining the absorbance difference of each wavelength relative to all coke residues; obtaining the regularity of the absorbance change of each wavelength relative to all coke residues; the component performance of any wavelength satisfies the expression: , Indicates the component performance of the ath wavelength, It represents the absorbance difference of the ath wavelength relative to all coke residues, Indicates the regularity of the absorbance variation of the ath wavelength relative to all coke residues, Indicates the number of coke slag types, represents the absorbance variance of the ath wavelength of all coke residue samples of the uth type of coke residue, represents an exponential function with a natural constant as its base; Combining the structural performance and the compositional performance of the wavelength to obtain a characteristic wavelength includes: comparing the structural performance and the compositional performance of the a-th wavelength as a first ratio of the a-th wavelength; comparing the compositional performance and the structural performance of the a-th wavelength as a second ratio of the a-th wavelength; taking the maximum value of the first ratio and the second ratio of the a-th wavelength as the characteristic goodness of the a-th wavelength; and obtaining the Z wavelength with the largest characteristic goodness among all wavelengths as the characteristic wavelength; Based on the characteristic wavelength and neural network model, a classification and detection model for coke slag samples is established.

2. A coal slag sample classification detection method according to claim 1, characterized in that: The step of obtaining the error sequence of the target spectrum and the error symbol difference sequence of the target spectrum includes: The absorbance sequence of the target spectrum is subtracted from the absorbance sequence of the corresponding denoised spectrum to obtain the error sequence of the target spectrum; the error sequence of the target spectrum is symbolized using a sign function to obtain the error symbol sequence of the target spectrum; the error symbol sequence of the target spectrum is forward differentiated to obtain the error symbol difference sequence of the target spectrum.

3. A coal slag sample classification detection method according to claim 1, characterized in that: The ath wavelength satisfies the expression relative to the structural characteristics of the target coke residue: ； Where, Indicates the structural characteristics of the ath wavelength relative to the target coke residue; The number of slag samples representing the target slag; It represents the distance between the hth coke residue sample of the target coke residue and the noise index fitting line of the ath wavelength of the target coke residue; Indicates the type of crushing degree; 、 、 The average noise index of the ath wavelength of the spectrum of the coke slag sample representing the q+1th, qth, and q-1th degree of crushing of the target coke slag; represents an exponential function with a natural constant as its base; Represents the normalization function.

4. A coal slag sample classification detection method according to claim 1, characterized in that: The absorbance difference of each wavelength relative to all coke residues satisfies the expression: ； Where, It represents the absorbance difference of the ath wavelength relative to all coke residues; Indicates the number of coke slag types; 、 It represents the average absorbance of the ath wavelength of the uth and vth types of coke residues; represents the absolute value function; Represents the normalization function.

5. The method for classifying and detecting coal slag samples according to claim 1, characterized in that: The obtaining of the regularity of absorbance variation of each wavelength relative to all coke residues includes: Obtain the average absorbance sequence of the ath wavelength of all types of coke residues in ascending order of type number, perform forward difference on the average absorbance sequence of the ath wavelength of all types of coke residues, and obtain the difference sequence of the average absorbance sequence of the ath wavelength of all types of coke residues; The number of values ​​greater than 0 in the difference sequence of the average absorbance sequence of the ath wavelength of all types of coke residues is compared with the number of values ​​less than 0, and recorded as the regularity of the absorbance change of the ath wavelength relative to all coke residues.

6. A coal slag sample classification detection system, characterized in that: include: A processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, a coal slag sample classification and detection method according to any one of claims 1 to 5 is implemented.

Images