Pulverized coal concentration measuring method based on multiple frequencies and total cross section

Abstract

The invention relates to a multi-frequency and total cross-section-based pulverized coal concentration measurement method, which comprises the following steps of: acquiring multi-frequency and total cross-section microwave attenuation data through a microwave sensor, and constructing a high-dimensional data set; performing global feature extraction on the high-dimensional data set by using a PCA algorithm to obtain a data set after dimension reduction; performing local optimal parameter combination screening on the data set after dimension reduction through an OMP algorithm; and constructing a pulverized coal concentration prediction model, inputting measurement data, and predicting the pulverized coal concentration in real time. The method has the beneficial effects that a multi-frequency and total cross-section measurement method is adopted, a true value is taken as a target, and an optimal subset is gradually selected in an iteration mode, so that a result approaches to an optimal solution, a large amount of invalid calculation is avoided, the calculation amount can be effectively reduced, and the risk of calculation bottleneck is avoided. Moreover, compared with the single-frequency linear measurement, the multi-frequency and total cross-section measurement uses more information in the physical space and the frequency domain.

Description

Technical Field

[0001] The present invention relates to the technical field of pulverized coal concentration detection, and more precisely, it relates to a method for measuring pulverized coal concentration based on multiple frequencies and full cross-section. Background Art

[0002] In coal-fired power plants, the combustion conditions of boilers are closely related to the pulverized coal transportation conditions of each burner. The pulverized coal velocity, concentration, and the air-pulverized coal uniformity of each pulverized coal burner will directly affect the stability of the combustion conditions in the furnace and the combustion efficiency of the boiler. Once problems such as flame deflection and unstable ignition occur in the furnace, it will cause huge economic losses to the power plant equipment and greatly affect the safe operation of the unit. In order to precisely control the air flow rate and pulverized coal flow rate entering each burner to achieve the best combustion in the boiler, one of the key factors is to be able to measure the pulverized coal parameters entering the boiler in real time and accurately.

[0003] The flow of pulverized coal in the pulverized coal conveying pipe of a coal-fired power plant belongs to a dilute-phase gas-solid two-phase flow. Affected by factors such as the geometric shape of the flow channel, the temperature and humidity of the air flow, and the physical properties of pulverized coal (such as particle size), its flow patterns are diverse and randomly distributed, making this process extremely complex. Therefore, the measurement of pulverized coal flow parameters is extremely challenging.

[0004] Existing measurement technologies have obvious limitations: invasive measurement methods are vulnerable to contamination and wear, while non-invasive measurement methods are easily affected by external interference.

[0005] Among many non-invasive technologies, microwave detection technology, as an effective solution, has been successfully applied to various industrial environments and is also used for the measurement of pulverized coal concentration. Common microwave detection technologies include the transmission method, the resonant cavity method, microwave tomography (MWT), and the Doppler method. They usually only use a single frequency for linear measurement. However, single-frequency linear measurement has the following problems: First, linear measurement cannot solve the problem of uneven distribution in pulverized coal transportation, and the measurement results are one-sided; second, signals in different frequency bands have different sensitivities to ambient noise, which often causes changes in the detection effect; third, microwave characteristics such as attenuation and phase shift will obtain different results when measured at different frequencies; fourth, if multi-frequency full cross-section measurement is used, although more information can be obtained, redundant information will also be brought. If effective screening methods are not adopted, a calculation bottleneck may be faced. Summary of the Invention

[0006] The purpose of the present invention is to propose a method for measuring pulverized coal concentration based on multiple frequencies and full cross-section in view of the deficiencies of the existing technology.

[0007] In the first aspect, a method for measuring pulverized coal concentration based on multiple frequencies and full cross-section is provided, including:

[0008] S1. Collect microwave attenuation data of multiple frequencies and the entire cross-section through a microwave sensor, and construct a high-dimensional data set;

[0009] S2. Use the PCA algorithm to extract global features from the high-dimensional data set to obtain a reduced-dimensional data set;

[0010] S3. Screen for the locally optimal parameter combination from the reduced-dimensional data set through the OMP algorithm;

[0011] S4. Construct a pulverized coal concentration prediction model, input measurement data, and predict the pulverized coal concentration in real time.

[0012] Preferably, in S1, multiple signal paths are constructed through a host computer, a microwave signal conditioning system, and a microwave sensor, and a high-dimensional data set is constructed by traversing all frequencies and signal paths.

[0013] Preferably, S2 includes:

[0014] S201. Centralize the data of each signal path in the high-dimensional data set to generate a preprocessing matrix;

[0015] S202. Calculate the covariance matrix of the preprocessing matrix;

[0016] S203. Perform eigenvalue decomposition on the covariance matrix to extract the principal components;

[0017] S204. Select the principal components with a cumulative interpretation rate exceeding a preset threshold, generate a data matrix, and obtain a reduced-dimensional data set.

[0018] Preferably, S3 includes:

[0019] S301. Flatten the data matrix in the frequency dimension and the principal component dimension;

[0020] S302. Given a response vector, initialize the residual, the number of iterations, and an empty set, and perform iterative calculations;

[0021] S303. At the end of the iteration, obtain the optimal subset and the estimated value, that is, the optimal parameter combination and the predicted pulverized coal concentration.

[0022] Preferably, in S302, the iterative calculation includes:

[0023] Select the column with the largest inner product with the residual and add it to the empty set; perform a pseudoinverse calculation on the matrix composed of the currently selected columns to obtain the estimated regression coefficient, and then update the residual with the new estimated value.

[0024] In a second aspect, a pulverized coal concentration measurement system based on multiple frequencies and the entire cross-section is provided for performing any of the methods in the first aspect, including:

[0025] A building block for collecting multi - frequency and full - cross - section microwave attenuation data through a microwave sensor to construct a high - dimensional data set;

[0026] A dimensionality reduction module for globally extracting features from the high - dimensional data set using the PCA algorithm to obtain a dimensionality - reduced data set;

[0027] A screening module for locally screening the optimal parameter combination of the dimensionality - reduced data set using the OMP algorithm;

[0028] A prediction module for constructing a pulverized coal concentration prediction model and inputting measurement data to predict the pulverized coal concentration in real time.

[0029] In a third aspect, a pulverized coal concentration measurement device based on multi - frequency and full - cross - section is provided for performing any of the methods described in the first aspect, including: a host computer, a microwave signal conditioning system, and a microwave sensor;

[0030] Wherein, the host computer and the microwave signal conditioning system are communicatively connected, and the microwave signal conditioning system and the microwave sensor are communicatively connected.

[0031] In a fourth aspect, a computer storage medium is provided, and a computer program is stored in the computer storage medium; when the computer program runs on a computer, the computer is enabled to execute any of the methods described in the first aspect.

[0032] In a fifth aspect, an electronic device is provided, including:

[0033] A memory for storing a computer program;

[0034] A processor for executing the computer program to implement any of the methods described in the first aspect.

[0035] The beneficial effects of the present invention are:

[0036] 1. The present invention adopts a method of multi - frequency and full - cross - section measurement. Taking the true value as the target, it gradually selects the optimal subset through an iterative method, making the result approach the optimal solution, avoiding a large amount of invalid calculations, effectively reducing the computational amount, and avoiding the risk of computational bottlenecks. Moreover, compared with single - frequency linear measurement, multi - frequency and full - cross - section measurement uses more information in both physical space and frequency domain. In addition, since the pulverized coal distribution is random and the flow process is complex, it is difficult to master the distribution of the entire cross - section only by linear measurement. Using full - cross - section measurement can make up for this deficiency; signals in different frequency bands have different sensitivities to ambient noise and the measured object. Multi - frequency measurement reduces the influence of ambient noise and improves the sensitivity to the measured object.

[0037] 2. Excessive frequencies and signal paths can bring redundant information. The present invention performs double dimensionality reduction on the high-dimensional pulverized coal concentration dataset, which can scientifically and effectively screen frequencies and signal paths, and reduce the amount of data required for prediction on the premise of ensuring accuracy.

[0038] 3. The present invention is designed for a specific hardware system, but can be applied to different parameters of such hardware systems, with high versatility. Brief Description of the Drawings

[0039] Figure 1 It is a schematic structural diagram of the pulverized coal concentration measurement device based on multiple frequencies and full cross-section provided by this application;

[0040] Figure 2 It is a schematic diagram of the pulverized coal conveying process provided by this application;

[0041] Figure 3 It is a flowchart of the pulverized coal concentration measurement method based on multiple frequencies and full cross-section provided by this application;

[0042] Figure 4a It is a schematic structural diagram of a common microwave sensor;

[0043] Figure 4b It is a schematic structural diagram of the microwave sensor provided by this application;

[0044] Figure 5 It is a schematic diagram of the result distribution of linear measurement and full cross-section measurement provided by this application;

[0045] Explanation of the reference numerals: host computer 1, microwave signal conditioning system 2, microwave sensor 3, boiler 4, pulverized coal conveying pipe 5, coal mill 6. Detailed Description of the Embodiments

[0046] The following further describes the present invention in conjunction with the embodiments. The description of the following embodiments is only used to help understand the present invention. It should be noted that for those of ordinary skill in the art, without departing from the principle of the present invention, several modifications can still be made to the present invention, and these improvements and modifications also fall within the protection scope of the claims of the present invention.

[0047] Embodiment 1:

[0048] In view of the obvious limitations of the technical solutions in the current field of pulverized coal concentration measurement, the present invention provides a pulverized coal concentration measurement method based on multiple frequencies and the full cross-section, innovatively combining the PCA (Principal Component Analysis) and OMP (Orthogonal Matching Pursuit) algorithms to achieve double dimensionality reduction of complex high-dimensional microwave pulverized coal concentration data sets, and using the optimal parameter combination to fit a regression model to predict the real-time pulverized coal concentration. The label of each sample in the microwave pulverized coal concentration data set consists of three parts, namely the acquisition time, frequency, and channel number. Therefore, this data set can be expressed as matrix. The whole measurement method mainly includes two steps: full cross-section data dimensionality reduction, and selection of the optimal frequency and signal path.

[0049] Specifically, as Figure 3 shown, the pulverized coal concentration measurement method based on multiple frequencies and the full cross-section includes:

[0050] S1. Collect microwave attenuation data of multiple frequencies and the full cross-section through a microwave sensor to construct a high-dimensional data set.

[0051] S2. Use the PCA algorithm to perform global feature extraction on the high-dimensional data set to obtain a dimensionality-reduced data set.

[0052] Full cross-section data dimensionality reduction is performed in S2. Full cross-section data dimensionality reduction is mainly achieved through the PCA algorithm. The PCA algorithm aims to identify k new variables that can capture the main features of the data set, effectively compress the original data matrix, and retain the minimum dimension representing the most important features. During the PCA calculation process, the covariance matrix is calculated and its eigenvalues and corresponding eigenvectors are deduced. Each eigenvector corresponds to a principal component, representing the projection of the original data in the feature space. For the microwave pulverized coal concentration data set, the role of PCA is to extract the features on the signal path dimension at each frequency, so as to represent each frequency with fewer principal components to achieve the purpose of dimensionality reduction.

[0053] S3. Use the OMP algorithm to screen the optimal local parameter combination for the dimensionality-reduced data set.

[0054] S3 performs optimal parameter combination screening. The optimal parameter combination screening mainly relies on the OMP algorithm to achieve. For the large number of frequencies and their principal components used in the measurement process, the OMP method can effectively avoid the computational bottleneck caused by iterative fitting. OMP is a sparse signal reconstruction algorithm based on a greedy strategy. This algorithm constructs a sparse representation of the signal by iteratively selecting the dictionary atoms most relevant to the residual. In each iteration, OMP updates the residual and coefficients through orthogonal projection to ensure the orthogonality and stability among the selected atoms. For the microwave pulverized coal concentration dataset, the OMP algorithm will be used to screen each frequency and its principal component after the application of the PCA algorithm, and finally obtain a combination of the frequency and its principal component that best matches the characteristics of the current microwave pulverized coal concentration dataset to construct a sparse representation of the pulverized coal concentration.

[0055] The method proposed in the present invention extracts features from a pre-collected dataset. Through two steps of full cross-section data dimensionality reduction and screening of the optimal parameter combination, it overcomes the limitations of using any single algorithm alone and realizes double dimensionality reduction processing for complex high-dimensional datasets. On the premise of maximizing the approximation of the result to the true value, a pulverized coal concentration prediction model is constructed. The PCA algorithm ensures the retention of important features, while the OMP algorithm further optimizes feature selection through optimal parameter combination screening. The collaborative work of the two realizes a more efficient data processing flow: compared with using the OMP algorithm alone, this solution significantly reduces the amount of data to be processed through PCA preprocessing, resulting in a significant improvement in the computational efficiency of the OMP algorithm. Through experimental verification, if the PCA algorithm reduces 240 signal paths composed of a total of 16 electrodes to 15 principal components, the data processing volume can be reduced by approximately 30%. This double dimensionality reduction strategy not only makes the finally constructed regression model have higher prediction accuracy (the mean absolute error is reduced to less than 1.5%), but also significantly improves the overall performance of the system. At the same time, the PCA algorithm effectively filters out the noise components in the data, while the OMP algorithm further improves the anti-interference ability of the model through sparse representation, making the measurement results more stable and reliable. Through the combination of PCA preprocessing and OMP optimization processing, this solution significantly improves the data processing speed while ensuring accuracy, enabling it to meet the requirements of real-time measurement, and the data processing volume is reduced by more than 96% compared with the traditional traversal method.

[0056] It should be noted that the parameters of the PCA and OMP algorithms in this method are not fixed and can be specified according to the needs of the user. For example, the PCA algorithm can specify the amount of information of the original dataset retained after dimensionality reduction, and the OMP algorithm can specify the number of subsets in the final prediction model.

[0057] S4. Construct a pulverized coal concentration prediction model, input the measurement data, and predict the pulverized coal concentration in real time.

[0058] During actual measurement, the microwave sensing system is adjusted to obtain the input parameters required by the pulverized coal concentration prediction model. Through model operation, the function of measuring the pulverized coal concentration in the pulverized coal conveying pipe of the coal combustion system is realized.

[0059] In addition, this method is not only applicable to the pulverized coal concentration measurement scenario, but can also be used in similar gas-solid two-phase flow scenarios.

[0060] Embodiment 2:

[0061] Based on Embodiment 1, Embodiment 2 of the present application provides a more specific method for measuring pulverized coal concentration based on multiple frequencies and the entire cross-section, including:

[0062] S1. Collect microwave attenuation data of multiple frequencies and the entire cross-section through a microwave sensor, and construct a high-dimensional data set.

[0063] In S1, as Figure 1 shown, multiple signal paths are constructed through the host computer 1, the microwave signal conditioning system 2, and the microwave sensor 3. By traversing all frequencies and signal paths, a high-dimensional data set is constructed.

[0064] Specifically, the host computer 1 controls the microwave signal conditioning system 2 to send a microwave signal of a specified frequency. After being amplified, attenuated, filtered, etc., it is connected to any two electrodes in the microwave sensor 3 to form a signal path, and then the signal is transmitted back to the microwave signal conditioning system 2 for detection processing. Finally, the microwave signal reaches the host computer 1 in the form of a digital signal. Each time of measurement, the working frequency of the microwave signal conditioning system 2 and the signal path of the microwave sensor 3 can be adjusted according to the host computer 1. By traversing all frequencies and signal paths, samples at each moment can be obtained and constructed into a microwave pulverized coal concentration data set, the size of which is .

[0065] It should be noted that the size of the high-dimensional data set depends on the frequency range of the microwave signal conditioning system 2 and the number of electrodes of the microwave sensor 3. For example, is determined by the output frequency range and the step size of the voltage-controlled oscillator in the microwave signal conditioning system 2, while is determined by the number of electrodes of the microwave sensor 3. When the number of electrodes is N, , where A represents the permutation formula, represents taking 2 from N elements for permutation. Therefore, the size of the high-dimensional data set is not fixed, and this method is applicable to all sizes.

[0066] S2. Use the PCA algorithm to perform global feature extraction on the high-dimensional data set to obtain a reduced-dimensional data set.

[0067] S2 includes:

[0068] S201. Centralize each signal pathway data in the high-dimensional dataset to generate a preprocessing matrix.

[0069] Exemplarily, group the microwave pulverized coal concentration dataset by frequency to obtain a matrix of size . Centralize the data for each column corresponding to the pathway number, that is, subtract the mean value of each column of data to move the center of the data to the origin. The centralized matrix is denoted as :

[0070]

[0071] where Y is the matrix obtained by grouping the microwave pulverized coal concentration dataset by frequency, is the row number, representing the data acquisition time, is the column number, representing the pathway number of the data, is the mean value of the

[0072] S202. Calculate the covariance matrix of the preprocessing matrix.

[0073] where the calculation formula for the covariance matrix C is:

[0074]

[0075] where is the transpose of

[0076] S203. Perform eigenvalue decomposition on the covariance matrix to extract the principal components.

[0077] Specifically, perform eigenvalue decomposition on the covariance matrix to find a set of orthogonal basis vectors such that the projection of the data on these basis vectors has the maximum variance. The eigenvalues and eigenvectors of the covariance matrix are:

[0078]

[0079] Arrange the eigenvalues in descending order, and select the eigenvectors corresponding to the first k largest eigenvalues as the principal components (Principal Components, abbreviated as PC).

[0080] S204. Select the principal components whose cumulative explained variance rate exceeds the preset threshold, generate a data matrix, and obtain the dimensionality-reduced dataset.

[0081] The variance explained rate of each principal component It is the ratio of this eigenvalue to the sum of all eigenvalues.

[0082]

[0083] Exemplarily, ensure that the eigenvectors of each frequency retain more than 80% of the information in the original matrix, that is, the cumulative variance explanation rate exceeds 80%, and form a new data matrix with dimensions . Then, the data volume of each frequency can be minimized without distortion.

[0084] S3. Screen for the locally optimal parameter combination from the dimension-reduced data set through the OMP algorithm.

[0085] After completing the dimension reduction in the path number dimension, start the optimal parameter combination screening step. Specifically, S3 includes:

[0086] S301. Flatten the data matrix according to the frequency dimension and the principal component dimension.

[0087] Exemplarily, for the frequency dimension and the principal component dimension in the data matrix to be flattened, then a matrix with dimensions can be obtained.

[0088] S302. Given a response vector, initialize the residual, the number of iterations, and an empty set, and perform iterative calculations.

[0089] Exemplarily, given a response vector , initialize the residual , the number of iterations , and an empty set .

[0090] In S302, the iterative calculation includes:

[0091] Select the column with the largest inner product with the residual, and add it to the empty set; perform a pseudo-inverse calculation on the matrix composed of the currently selected columns to obtain the estimated regression coefficients, and then update the residual using the new estimated values.

[0092] Specifically, the following operations will be performed in each iteration:

[0093] Select the column with the largest inner product with the residual , and add it to the set :

[0094]

[0095]

[0096] Among them, represents the j-th column of the matrix ; perform pseudo-inverse calculation on the matrix formed by the currently selected columns to obtain the estimated regression coefficients:

[0097]

[0098] Then update the residuals using the new estimated values:

[0099]

[0100] Through these steps, the pseudo-inverse calculation assists in minimizing the residuals on the currently selected subset in each iteration, thereby gradually approaching the optimal solution. Finally, when the iteration ends, the optimal subset and the estimated values are obtained, that is, the optimal parameter combination and the predicted pulverized coal concentration:

[0101]

[0102] S303. When the iteration ends, the optimal subset and the estimated values are obtained, that is, the optimal parameter combination and the predicted pulverized coal concentration.

[0103] S4. Construct a pulverized coal concentration prediction model and input the measurement data to predict the pulverized coal concentration in real time.

[0104] Use the parameters in S3 as the preset model parameters. During actual measurement, only the full-section microwave attenuation data at the frequencies included in the optimal subset Λ need to be collected to achieve the detection of the pulverized coal concentration.

[0105] In summary, the present invention innovatively combines the PCA and OMP dual dimensionality reduction algorithms and works in cooperation with a microwave sensing system with adjustable working frequency and signal path to achieve efficient feature extraction of high-dimensional complex pulverized coal concentration data and screening of the optimal parameter combination, providing a high-precision and high-efficiency real-time non-invasive technical solution for pulverized coal concentration measurement. Compared with the prior art, this method has the following innovative points:

[0106] 1. Establish a scientific frequency and signal path screening mechanism, take the lead in adopting a dual dimensionality reduction strategy, perform global feature extraction through the PCA algorithm, and combine the OMP algorithm to achieve local optimal parameter screening, ensuring that the extracted features contain both comprehensive information and the highest contribution rate, and accurately identifying the parameter combination most valuable for pulverized coal concentration prediction from the massive data;

[0107] 2. The computational efficiency is significantly optimized, avoiding the computational bottleneck problem caused by iterative fitting. Taking the microwave sensor of the present invention as an example, assuming it can operate at 100 frequency points, if the traditional traversal method is used for optimal strategy selection, it will face the dilemma of explosive computational complexity (as shown in Table 1). However, this method greatly reduces the computational complexity through a dual dimensionality reduction strategy and an optimal parameter screening mechanism, achieving efficient data processing.

[0108] Table 1 The explosive growth of the traversal quantity with the number of frequencies used

[0109]

[0110] 3. Compared with the traditional single-frequency and pure linear measurement schemes, the multi-frequency and full-section measurement technology adopted by this method significantly improves the reliability and accuracy of the measurement results. As shown in Table 2, the classification experiment data shows that: compared with the linear measurement with 2 electrodes, the accuracy rate of the full-section measurement with 16 electrodes is increased by 33%.

[0111] Table 2 The influence of the number of electrodes on the prediction accuracy

[0112]

[0113] Meanwhile, as Figure 5 shown, the regression experiment data shows that the average absolute error of the multi-frequency measurement is reduced by 3.2% compared with the single-frequency measurement. These quantitative analyses fully verify the significant advantages of this method in improving the measurement accuracy.

[0114] It should be noted that the parts that are the same or similar to those in Embodiment 1 in this embodiment can be referred to each other and will not be elaborated in this application.

[0115] Embodiment 3:

[0116] Based on Embodiment 2, Embodiment 3 of the present application provides a pulverized coal concentration measurement system based on multi-frequency and full-section, including:

[0117] A construction module, configured to collect multi-frequency and full-section microwave attenuation data through a microwave sensor and construct a high-dimensional data set;

[0118] A dimensionality reduction module, configured to perform global feature extraction on the high-dimensional data set by using the PCA algorithm to obtain a dimensionality-reduced data set;

[0119] A screening module, configured to perform local optimal parameter combination screening on the dimensionality-reduced data set by using the OMP algorithm;

[0120] A prediction module, configured to construct a pulverized coal concentration prediction model and input measurement data to predict the pulverized coal concentration in real time

[0121] It should be noted that the system provided in this embodiment is the system corresponding to the method provided in Embodiment 2. Therefore, for the parts that are the same or similar in this embodiment and Embodiment 2, reference can be made to each other and will not be elaborated in this application.

[0122] Embodiment 4:

[0123] Based on Embodiment 2, Embodiment 4 of this application provides a pulverized coal concentration measurement system based on multiple frequencies and the full cross-section, including: a host computer 1, a microwave signal conditioning system 2, and a microwave sensor 3;

[0124] Among them, the host computer 1 and the microwave signal conditioning system 2 are communicatively connected, and the microwave signal conditioning system 2 and the microwave sensor 3 are communicatively connected. As Figure 1 shown, the arrow direction is the direction of the main signal or information, which is divided into a DC signal and a microwave signal. The components of the multi-frequency pulverized coal concentration measurement system communicate bidirectionally to complete data acquisition and instruction issuance.

[0125] Figure 2 It is a schematic structural diagram of the pulverized coal conveying process, including a boiler 4, a pulverized coal conveying pipe 5, and a coal mill 6. The pulverized coal flow direction is from bottom to top. After being diverted from the coal mill 6 into multiple pulverized coal conveying pipes 5, it is fed into the boiler 4 for combustion from different angles respectively.

[0126] It should be noted that the device provided in this embodiment is the device corresponding to the method provided in Embodiment 2. Therefore, for the parts that are the same or similar in this embodiment and Embodiment 2, reference can be made to each other and will not be elaborated in this application.

Claims

1. A method for measuring pulverized coal concentration based on multi - frequencies and full cross - section, characterized in that, Including: S1. Collect microwave attenuation data of multiple frequencies and full cross-section through a microwave sensor, and construct a high-dimensional data set; S2. Use the PCA algorithm to extract global features from the high-dimensional data set to obtain a reduced-dimensional data set; S3. Screen local optimal parameter combinations for the reduced-dimensional data set through the OMP algorithm; S4. Construct a pulverized coal concentration prediction model, input measurement data, and predict the pulverized coal concentration in real time.

2. The pulverized coal concentration measurement method based on multiple frequencies and full cross-section according to claim 1, characterized in that, In S1, multiple signal paths are constructed through a host computer, a microwave signal conditioning system, and a microwave sensor. By traversing all frequencies and signal paths, a high-dimensional data set is constructed.

3. The method for measuring pulverized coal concentration based on multiple frequencies and full cross-section according to claim 2, characterized in that S2 Including: S201. Centralize the data of each signal path in the high-dimensional data set to generate a preprocessing matrix; The preprocessing matrix is denoted as , and the calculation formula is: ； Wherein, Y is a matrix obtained by grouping the microwave pulverized coal concentration data set according to frequency, is the row number, representing the data acquisition time, is the column number, representing the data path number, is the mean value of the S202. Calculate the covariance matrix of the preprocessing matrix; The calculation formula of the covariance matrix C is: ； Among them, represents the acquisition time of the data, is the transpose of the preprocessing matrix ; S203. Perform eigenvalue decomposition on the covariance matrix to extract the principal components; S204. Select the principal components whose cumulative interpretation rate exceeds a preset threshold, generate a data matrix, and obtain a reduced-dimensional data set.

4. The method for measuring pulverized coal concentration based on multiple frequencies and full cross-section according to claim 3, wherein S3 Including: S301. Flatten the data matrix in the frequency dimension and the principal component dimension; S302. Given a response vector, initialize the residual, the number of iterations, and an empty set, and perform iterative calculations; S303. At the end of the iteration, obtain the optimal subset and the estimated value, that is, the optimal parameter combination and the predicted pulverized coal concentration.

5. The pulverized coal concentration measurement method based on multiple frequencies and full cross-section according to claim 4, characterized in that, In S302, the iterative calculation includes: Select the column with the largest inner product with the residual and add it to the empty set; perform pseudo-inverse calculation on the matrix composed of the currently selected columns to obtain the estimated regression coefficient, and then update the residual with the new estimated value.

6. A pulverized coal concentration measurement system based on multi - frequency and full cross - section, characterized in that, For implementing the method according to any one of claims 1 to 5, including: A construction module for collecting microwave attenuation data of multiple frequencies and full cross-section through a microwave sensor and constructing a high-dimensional data set; A dimensionality reduction module for extracting global features from the high-dimensional data set by using the PCA algorithm to obtain a reduced-dimensional data set; A screening module for screening local optimal parameter combinations for the reduced-dimensional data set through the OMP algorithm; A prediction module for constructing a pulverized coal concentration prediction model, inputting measurement data, and predicting the pulverized coal concentration in real time.

7. A pulverized coal concentration measuring device based on multiple frequencies and full cross-section, characterized in that, For implementing the method according to any one of claims 1 to 5, including: a host computer, a microwave signal conditioning system, and a microwave sensor; Wherein, the host computer and the microwave signal conditioning system are communicatively connected, and the microwave signal conditioning system and the microwave sensor are communicatively connected.

8. A computer storage medium, characterized in that, The computer storage medium stores a computer program; when the computer program runs on a computer, the computer is enabled to execute the method according to any one of claims 1 to 5.

9. An electronic device, characterized in that, Including: A memory for storing a computer program; A processor for executing the computer program to implement the method according to any one of claims 1 to 5.

Images