Industrial drying process solid particle moisture content automatic modeling system and method based on near infrared technology

Abstract

The invention discloses an industrial drying process solid particle moisture content automatic modeling system and method based on a near infrared technology, and relates to the technical field of industrial drying process solid particle moisture content quantitative analysis. The system comprises a near-infrared emission device, a transmission device, a detection device, a heat flow drying device, a weighing device and a signal processing device. The near-infrared emission device generates broadband near-infrared light and forms high-energy density light spots; the transmission device couples the light spot to the surface of the sample and collects a reflection signal; the detection device analyzes the optical signal to obtain a near infrared spectrum; the heat flow drying device is used for controllably drying the sample; the weighing device monitors the weight change in real time; and the signal processing device calculates a moisture content label based on the spectral data and the weight data, and establishes a prediction model. According to the method, rapid and lossless automatic modeling analysis of the moisture content of the solid particles is realized, and the moisture content detection efficiency and precision in the industrial drying process are remarkably improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of quantitative analysis of water content of solid particles in industrial drying process, and particularly relates to an automatic modeling system and method for water content of solid particles in industrial drying process based on near-infrared technology. BACKGROUND

[0002] Accurate detection of water content of solid particles is of great significance in industrial production process. In the fields of chemical industry, pharmaceutical industry, food industry and agricultural product processing, the water content of solid particles is one of the key indicators for measuring product quality and process stability. Too high water content may cause product caking, mold growth or shortened storage period, while too low water content may cause excessive drying, resulting in energy waste or changes in product physical and chemical properties.

[0003] At present, the commonly used water content detection methods mainly include oven drying method, Karl Fischer method, near-infrared spectroscopy technology, etc. As a traditional method, the oven drying method has accurate determination results, but the detection period is as long as several hours, and manual operation is required, which is low in efficiency and difficult to meet the demand for rapid detection in modern industry. The Karl Fischer method is more suitable for the determination of trace water, but the operation is complex and the equipment cost is high. Although the near-infrared spectroscopy technology can realize rapid detection, its penetration ability is limited, mainly reflecting the water content information of the surface of the material, and the measurement accuracy of solid particles with uneven internal water distribution is limited, and a complex model usually needs to be established and frequently corrected. In addition, most of the existing detection methods are offline detection, which cannot realize real-time and online monitoring and feedback of the water content of materials in the production process. This leads to the inability to timely adjust the drying process parameters in the production process, which not only affects the product consistency, but also causes energy waste.

[0004] Therefore, it is necessary to develop a solid particle water content detection system and method which can realize automatic modeling, rapid and accurate detection, solve the problems of complex sample preparation, tedious water content calibration and difficult online monitoring, etc. SUMMARY

[0005] The present application aims to provide an automatic modeling system and method for water content of solid particles in industrial drying process based on near-infrared technology, which solves the problems of complex sample preparation, tedious water content calibration and difficult online monitoring in the prior art, and can realize rapid and non-destructive automatic modeling analysis of water content of solid particles, significantly improving the detection efficiency and accuracy of water content.

[0006] To achieve the above-mentioned purpose, the technical solutions adopted by the present application are as follows: An automatic modeling system for water content of solid particles in industrial drying process based on near-infrared technology, comprising: The near-infrared emission device comprises a near-infrared light source and a focusing lens, the near-infrared light source generates wide-band near-infrared light and converges the light into a high-energy density light spot through the focusing lens, and the light spot is optically coupled to the near-infrared transmission device. The near-infrared transmission device comprises a first optical fiber coupler, a transmission optical fiber bundle, a second optical fiber coupler and a reflection probe connected in sequence, the first optical fiber coupler receives the high-energy density light spot from the near-infrared emission device, the light spot is transmitted through the transmission optical fiber bundle and the second optical fiber coupler, and the reflection probe projects the light spot onto the surface of the solid particle sample and simultaneously collects the sample signal reflected by the solid particle and transmits the signal to the near-infrared detection device. The near-infrared detection device comprises a detector, which receives the near-infrared light signal carrying the moisture content information from the near-infrared transmission device, analyzes the signal into digital near-infrared spectrum data and outputs the data to the signal processing device. The hot air drying device is arranged above or beside the sample and is used to generate hot air with a certain temperature and speed to dry the solid particle sample on the weighing device. The weighing device is used to monitor and obtain the weight change data of the solid particle sample in the drying process in real time and output the weight data to the signal processing device through serial communication. The signal processing device receives the spectrum data from the near-infrared detection device and the weight data from the weighing device, calculates the moisture content label corresponding to each spectrum according to the weight data, and performs machine learning training on the spectrum data and the moisture content label to obtain a solid particle moisture content prediction model.

[0007] Further, the wavelength range of the near-infrared light source of the near-infrared emission device is 800-2500 nm, and the high-energy density light spot formed after converging through the focusing lens has an area of less than 4 mm².

[0008] Further, the first optical fiber coupler and the second optical fiber coupler adopt SMA905 standard connectors, and the transmission optical fiber is a Y-shaped low-hydroxyl quartz optical fiber bundle.

[0009] Further, the detector adopts a CCD array type near-infrared detector, and the spectrum acquisition range is 800-2500 nm.

[0010] Further, the judgment condition for stopping the drying of the hot air drying device is that the difference between the adjacent two weighing masses of the solid particle sample is less than a preset threshold value within a fixed collection interval.

[0011] Another object of the present application is to provide an industrial drying process solid particle moisture content automatic modeling method based on near-infrared technology, which uses the industrial drying process solid particle moisture content automatic modeling system based on near-infrared technology when executed, comprising: S1: initialization and first data acquisition; S2: periodic drying and data acquisition cycle according to preset time interval; S3: calculating moisture content label based on collected data; S4: preprocessing collected spectral data; S5: selecting characteristic wavelength by principal component analysis (PCA); S6: constructing moisture content quantitative analysis model based on support vector machine (SVM) method.

[0012] Further, the moisture content label in S3 is calculated by the following formula:

[0013] In the formula, is the moisture content, , and are the mass of the moisture-containing solid particle sample, the mass of the moisture, and the mass of the dried solid particle sample, respectively, and i is a time sequence i = 1, 2, 3,....

[0014] Further, S4 includes: performing smoothing processing on the spectrum by using the SG method; performing multivariate scatter correction (MSC) on the smoothed spectrum; performing baseline correction by using the airPLS method to eliminate baseline drift.

[0015] Another object of the present application is to provide a computer readable storage medium storing a computer program, which, when executed by a processor, implements the industrial drying process solid particle moisture content automatic modeling method based on near-infrared technology.

[0016] Another object of the present application is to provide an electronic device including a memory and a processor, the memory being configured to store a computer program, and the processor being configured to execute the computer program to cause the electronic device to perform the industrial drying process solid particle moisture content automatic modeling method based on near-infrared technology.

[0017] In summary, the present application has at least one of the following beneficial technical effects: The present application realizes full-process automation from sample drying, spectrum acquisition, label calculation to model construction, solves the problems of complex sample preparation, tedious calibration, and difficulty in online monitoring in the traditional method, and can realize rapid, non-destructive automatic modeling analysis of solid particle moisture content, thereby significantly improving the moisture content detection efficiency and accuracy of the industrial drying process. BRIEF DESCRIPTION OF DRAWINGS

[0018] Figure 1This is a system composition diagram of the present invention; Figure 2 This is a schematic diagram of the optical fiber arrangement of the reflective probe of the present invention; Figure 3 This is the initial spectrum of the AP solid particles of the present invention; Figure 4 This is the SG smoothed spectrum of the present invention; Figure 5 This is the MSC multivariate scattering correction spectrum of the present invention; Figure 6 This is the baseline-corrected spectrum of the airPLS of the present invention; Figure 7 This is a schematic diagram of the SVM moisture content analysis results of the present invention; Figure 8 This is a schematic diagram of the PSO-SVM moisture content analysis results of the present invention; In the picture: 1-Near-infrared light source, 2-Focusing lens, 3-First fiber optic coupler, 4-Transmission fiber bundle, 5-Second fiber optic coupler, 6-Reflection probe, 7-Detector, 8-PC end, 9-Solid particle sample, 10-Weighing device, 11-Hot flow drying device. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. Furthermore, the technical features involved in the various embodiments of this invention described below can be combined with each other as long as they do not conflict with each other.

[0020] This embodiment provides an automatic modeling system for the moisture content of solid particles in an industrial drying process based on near-infrared technology, such as... Figure 1 As shown, the system includes a near-infrared emitting device, a near-infrared transmitting device, a near-infrared detection device, a hot air drying device 11, a weighing device 10, and a signal processing device. The following detailed explanation uses the detection of moisture content in ammonium perchlorate solid particles as an example.

[0021] Ammonium perchlorate (AP), as an important oxidant, is widely used in energetic material formulations in aerospace propulsion and special materials fields. Accurate detection of its moisture content is crucial for ensuring material stability, combustion performance, and safe production. This system, based on near-infrared spectroscopy analysis technology combined with intelligent drying control and machine learning algorithms, achieves rapid, non-destructive, and automated detection of the moisture content of ammonium perchlorate. The system works by utilizing the characteristic absorption properties of water molecules in the near-infrared region (800–2500 nm). When near-infrared light irradiates an ammonium perchlorate sample, the water in the sample selectively absorbs light energy of specific wavelengths, causing corresponding changes in the reflectance spectrum. By monitoring the correlation between this spectral change and the weight reduction of the sample before and after drying, the system can automatically establish an accurate moisture content prediction model. The system specifically includes the following components.

[0022] 1. A near-infrared emitting device, comprising a near-infrared light source 1 and a focusing lens 2, wherein the near-infrared light source 1 generates broadband near-infrared light and is focused by the focusing lens to form a high-energy-density light spot, which is optically coupled to a near-infrared transmission device.

[0023] Near-infrared light source 1 is a broadband light source using an LED array. Its spectral output range perfectly covers the 800-2500nm wavelength band, providing ample light intensity, especially near the 970nm, 1200nm, and 1450nm wavelengths where the characteristic absorption peaks of moisture are obvious. The light source integrates a constant current drive circuit and a temperature control module to ensure the stability of the output light intensity (fluctuation less than 0.5%) and guarantee the repeatability of long-term measurements.

[0024] The focusing lens 2 system employs an achromatic composite lens design, consisting of a convex lens and a concave lens, effectively correcting chromatic aberration and spherical aberration. The lens surface is coated with an anti-reflective film, achieving an average transmittance exceeding 97% in the 800-2500nm range. The optical system is precisely adjusted to efficiently converge the light emitted from the near-infrared light source 1 into a high-energy-density spot with an area less than 4mm², achieving an energy density of 300-500mW / mm², providing sufficient luminous flux to the first fiber optic coupler 3 of the subsequent near-infrared transmission device.

[0025] The device's casing is made of aluminum alloy, and it features internal heat dissipation fins and a miniature fan to ensure thermal stability during prolonged operation. The power module provides overvoltage and overcurrent protection to guarantee safe operation of the equipment.

[0026] 2. A near-infrared transmission device, comprising a first fiber coupler 3, a transmission fiber bundle 4, a second fiber coupler 5, and a reflection probe 6 connected in sequence. The first fiber coupler 3 receives a high-energy-density light spot from a near-infrared emitting device. The light spot is then transmitted through the transmission fiber bundle 4 and the second fiber coupler 5, and projected onto the surface of a solid particle sample 9 by the reflection probe 6. The reflection probe 6 also simultaneously collects the sample signal reflected by the solid particles and transmits it to the near-infrared detection device.

[0027] The near-infrared transmission device is responsible for transmitting light energy from the emitting device to the sample surface and collecting the reflected signal to return to the detection device.

[0028] The fiber optic coupler uses the SMA905 standard connector, which features low insertion and removal loss (<0.2dB) and good stability for repeated connections. The coupler contains a precision positioning mechanism and employs a six-axis adjustment bracket to achieve micron-level precision adjustment, ensuring optimal coupling efficiency between the light source and the optical fiber. Measured coupling efficiency reaches over 85%.

[0029] The transmission fiber bundle 4 uses Y-type low-hydroxyl silica fiber bundles, a material with extremely low absorption loss (<0.2dB / m) and good thermal stability in the near-infrared region. The numerical aperture of the fiber NA = 0.22 ensures good optical transmission efficiency. The fiber bundle structure is specially designed: the number of fibers N0 = 8 at the near-infrared transmitter end to maximize the reception of light source energy; the number of fibers N1 = 9 at the reflector 6 end, using an 8+1 ring distribution design (e.g., ...). Figure 2 As shown, there are 8 transmitting optical fibers on the periphery and 1 receiving optical fiber in the center; the number of optical fibers at the near-infrared detection device end is N2=1, which efficiently transmits the collected signal light to the detector 7.

[0030] The reflective probe 6 features a stainless steel housing with a sapphire protective window at the front end, offering high hardness (Mohs hardness 9), high temperature resistance (maximum withstand temperature 300℃), and good chemical stability. The internal optical fibers are arranged at specific angles (typically 15°-30°), a design that effectively reduces the influence of specular reflection and improves the collection efficiency of diffuse reflection signals, making it particularly suitable for detecting highly reflective particulate materials such as ammonium perchlorate.

[0031] 3. A near-infrared detection device, including a detector 7, receives near-infrared light signals carrying moisture content information from a near-infrared transmission device, analyzes them into near-infrared spectral data in digital form, and outputs them to a signal processing device.

[0032] Detector 7 employs a CCD array near-infrared detector, a high-sensitivity optical sensor based on the semiconductor photoelectric effect, specifically designed to detect light signals in the near-infrared band (780-2500nm). It converts incident photons into electrons through a photodiode array, then sequentially reads these electron signals via charge coupling, ultimately reconstructing complete spectral information. The CCD uses a linear array sensor, capable of accurately capturing subtle changes in the absorption peaks characteristic of moisture. The optical system adapts to detection requirements with varying signal intensities, exhibiting high diffraction efficiency (>70%) in the 800-2500nm range. The entire optical system is sealed in a nitrogen-filled environment to prevent moisture and oxidation of optical components, ensuring long-term stability. The signal processing circuitry outputs high-quality spectral data. The CCD incorporates a thermoelectric cooling module, stabilizing the sensor temperature at -10°C ± 0.1°C, enhancing detection sensitivity.

[0033] 4. A hot flow drying device 11 is set above or to the side of the sample to generate a hot flow at a certain temperature and rate to dry the solid particle sample 9 placed on the weighing device.

[0034] The hot air drying device 11 consists of an adjustable-speed fan and a temperature-controllable heating module, which generates a heat flow at a specific temperature and rate to dry the ammonium perchlorate solid particle sample 9. The heating module employs a zoned temperature control design, containing three independent nickel-chromium alloy resistance wire heating elements, each with a power of 500W, for a total power of 1500W. The heating elements are surrounded by high-purity alumina ceramic insulation material, ensuring both insulation safety and improved thermal efficiency. The fan is driven by a brushless DC motor, providing a uniform and stable laminar airflow. The airflow adjustment range is 0-80 m³ / h, controlled by a PWM signal to adjust the motor speed and thus the airflow. The temperature control system uses an advanced PID algorithm and is equipped with multi-point temperature sensors, with a temperature control range of 40-200°C.

[0035] Based on the characteristics of ammonium perchlorate, the condition for determining complete drying is set as follows: within a fixed sampling interval Δt (e.g., 2 min), the difference Δm between two consecutive weighings of solid particle sample 9 is less than a preset threshold. (e.g., 50mg). This threshold setting ensures thorough drying while avoiding energy waste caused by over-drying.

[0036] 5. Weighing device 10, used to monitor and acquire the weight change data of solid particle sample 9 during the drying process in real time, and output the weight data to the signal processing device through serial communication.

[0037] The weighing device 10 employs a specially designed high-temperature resistant weighing module based on the principle of a high-precision strain gauge sensor. The sensor elastomer is made of nickel-chromium-molybdenum alloy steel, which undergoes special heat treatment and aging treatment, exhibiting excellent temperature stability and creep performance. The communication interface uses serial communication (RS485 isolated interface), supports the Modbus RTU protocol, and has a maximum baud rate of 115200bps, transmitting weight data to the signal processing device in real time.

[0038] 6. A signal processing unit receives spectral data from a near-infrared detection device and weight data from a weighing device. Based on the weight data, it calculates the moisture content label corresponding to each spectrum and performs machine learning training on the spectral data and moisture content labels to obtain a prediction model for the moisture content of solid particles. The signal processing unit includes a PC (8-bit), built on an industrial-grade computer platform, equipped with a multi-core processor, memory, and a solid-state drive.

[0039] This embodiment also provides an automatic modeling method for the moisture content of solid particles in an industrial drying process based on near-infrared technology, using a signal processing device. The modeling is achieved by executing the following process.

[0040] S1: Initialize and perform the first data acquisition.

[0041] When the operator clicks "Start Modeling" on the PC, the system automatically performs synchronous data acquisition. Near-infrared spectroscopy acquisition: Detector 7 immediately acquires the initial spectral data of solid particle sample 9. The mass benchmark measurement and weighing module simultaneously acquire the initial mass of the sample. The initial acquisition time was recorded as a time reference for the drying process. The initial spectrum of AP (ammonium perchlorate) solid particles acquired in this embodiment is as follows: Figure 3 As shown.

[0042] S2: Perform periodic drying and data collection cycles according to preset time intervals.

[0043] The drying parameters are set, and the timer is set to the drying time interval Δt (usually 2-5 minutes). This interval can be intelligently adjusted according to the characteristics of the sample.

[0044] Alternating execution process: The hot air drying device 11 is started, and the temperature and wind speed controlled by the PID algorithm are used to dry the sample efficiently; after one drying cycle, the system collects the current spectrum. and sample quality All collected data are stored with timestamps to establish a complete data sequence; this process is repeated cyclically, with i incrementing from 1 (i=1,2,3,...) to form a time series dataset.

[0045] The system determines whether drying should stop based on drying termination conditions. Within a fixed sampling interval Δt (e.g., 2 min), the mass difference Δm between two consecutive weighings of solid particle sample 9 is less than a preset threshold. (e.g., 50mg). Considering the cumulative drying time, a maximum cumulative drying time protection value (e.g., <30min) is set. When the termination condition is met, the system automatically stops the drying process and enters the next stage.

[0046] S3: Calculate the moisture content label based on the collected data.

[0047] The system automatically calculates the moisture content at each time point based on the principle of mass conservation. (Based on sample mass...) The final stable mass is taken as the dry solid mass. The baseline is used to calculate the moisture content label value corresponding to each spectrum at each time point, using the following expression:

[0048] In the formula, Moisture content, , and , i represents the mass of the water-containing solid particle sample 9, the mass of the water content, and the mass of the dry solid particle sample 9, respectively, and i is the time series i=1,2,3,...

[0049] Spectral data collected at each time point Compared with the calculated moisture content label value Precise pairings are used to form the dataset needed for modeling.

[0050] S4: Preprocess the collected spectral data.

[0051] First, the SG method is used to smooth the spectrum, reducing noise interference in the spectral data, improving the quality of the spectral data, and enhancing the clarity of spectral features. Figure 4 Then, multivariate scattering correction (MSC) is performed on the smoothed spectrum to eliminate the influence of particle state and scattering intensity changes on the spectrum. Figure 5 Finally, the airPLS method was used for baseline correction to eliminate baseline drift caused by changes in the detection distance during spectral acquisition. Figure 6 ).

[0052] The basic process of SG smoothing is as follows: For each data point x in the spectral vector i Centered on this point, a sub-interval with a width of m=13 is selected. The data points within this sub-interval form a vector. Where k = (m-1) / 2. Within this window, a third-order polynomial is used to fit the data. The smoothed output value X smooth_iIt is calculated using the following convolution formula:

[0053] In the formula, j ranges from -k to k; C j The convolution coefficients are fixed constants determined by the order of the selected polynomial and the window size m, calculated using the least squares method. N is a normalization coefficient, which is the sum of the convolution coefficients.

[0054] The basic process of multivariate scattering correction (MSC) is as follows: ① Calculate the average spectrum: Average the smoothed spectra of all samples to obtain the average spectral vector X̄.

[0055] ② Linear regression: For each spectrum X to be corrected, the least squares method is used. smooth Perform a univariate linear regression with the average spectrum X̄ at each wavelength point, and solve for the regression coefficients:

[0056] In the formula, a i Let b be the slope. i This is the intercept.

[0057] ③ Correction calculation: Using the obtained regression coefficients, the original smoothed spectrum is corrected to obtain the MSC-corrected spectrum X. msc :

[0058] The basic process of airPLS baseline correction is as follows: ① Initialization. Input spectrum X msc denoted as z 0 Initialize the weight vector w 0 =1.

[0059] ② Iterative fitting. In the k-th iteration, the baseline zk is fitted by solving the following weighted least squares problem:

[0060] In the formula: λ is the smoothness parameter, which controls the smoothness of the baseline; D is a second-order difference matrix used to calculate the roughness of the baseline.

[0061] ③ Update weights. Compare the current spectrum X. msc With the fitted baseline z k For spectral points above the current baseline, their weight is reduced; for points below or equal to the baseline, their weight is maintained at a high level. The weight update rules are as follows:

[0062] In the formula, t is an adjustment coefficient, which is taken as t=-1.

[0063] ④ Convergence judgment. Repeat steps ② and ③ until the number of points with changed weights among all spectral points is lower than the preset threshold, or the maximum number of iterations is reached, to obtain the fitted baseline z. final .

[0064] ⑤ Baseline subtraction: Subtract the fitted baseline from the original MSC spectrum to obtain the final corrected spectrum:

[0065] S5: Principal component analysis (PCA) is used for characteristic wavelength selection.

[0066] Principal component analysis (PCA) was used to select features from the near-infrared spectral data of the ammonium perchlorate solid particle drying process. The procedure is as follows: ① Data preprocessing: Standardize the original data so that the mean of each feature is 0 and the variance is 1.

[0067]

[0068] In the formula, μ j σ is the average value. j z is the standard deviation. ij It is the value of the i-th sample in the standardized matrix Z on the j-th feature.

[0069] ② Calculate the covariance matrix: For the standardized dataset, calculate the covariance matrix.

[0070]

[0071] In the formula, Z T It is the transpose of the normalized matrix Z.

[0072] ③ Eigenvalue decomposition: Perform eigenvalue decomposition on the covariance matrix to obtain eigenvalues ​​and corresponding eigenvectors.

[0073]

[0074] In the formula, λ i It is the i-th eigenvalue, v i It corresponds to λ i eigenvectors.

[0075] ④ Select principal components: Based on the size of the eigenvalues, select the top k=10 eigenvectors with the largest variance as principal components, project the original data onto the selected principal components, and obtain the dimensionality-reduced dataset.

[0076]

[0077] In the formula, V k Z is the matrix consisting of the first k=10 columns (i.e., the first 10 eigenvectors) of the eigenvector matrix V; Z is the normalization matrix; P is the calculated n×10 principal component score matrix, which is the coordinates of each sample in the 10 new principal component dimensions.

[0078] S6: Construct a quantitative analysis model for moisture content based on the Support Vector Machine (SVM) method.

[0079] A quantitative analysis model for the moisture content of ammonium perchlorate solid particles was constructed using the support vector machine (SVM) method. Figure 7 In this invention, the SVM method is specifically implemented using Support Vector Regression (SVR), which is suitable for quantitative analysis of continuous variables such as water content. SVR is a regression extension of SVM, which maps data through a kernel function to minimize prediction error. During the modeling process, Particle Swarm Optimization (PSO) is used to seek the optimal combination of parameters such as the SVM kernel function parameter γ and the penalty term c, avoiding overfitting or underfitting problems, enabling the SVM model to achieve optimal performance, and improving the model's prediction accuracy and stability. Figure 8 This is a schematic diagram of the moisture content analysis results using PSO-SVM. The PSO-SVM modeling process is as follows: ①SVR quantitative analysis model: SVR maps the input data to a high-dimensional feature space using a kernel function, and then performs linear regression in that space. The prediction model is represented as follows:

[0080] In the formula, f(x) is the predicted water content of the input sample x; n_sv is the number of support vectors; α i With α i * represents the Lagrange multiplier, obtained during model training by combining it with the optimal penalty factor c; x i Let K(x) be the i-th support vector; i Let x be the kernel function, and choose the radial kernel function. b is the bias term (or intercept) of the model, used to adjust the overall offset of the prediction function.

[0081] ②PSO parameter optimization: The first step is initialization. In (2) -5 <c<2 15 ,2 -15 <γ<2 3 In a two-dimensional search space, a group of particles is randomly initialized, and the position parameter of particle i is denoted as P. i =(C i ,γ i The speed parameter is S. i The optimal position of an individual is P.best_i The optimal position among all particles is denoted as the global optimal position G. best .

[0082] The second step is iterative optimization. The position parameters (C) of particle i are then... i ,γ i Train an SVR model on the training set and use 5-fold cross-validation to calculate the fitness value of the parameter combination.

[0083] In the formula, RMSE_CV represents the root mean square error between the predicted and actual values. The individual and global optima are updated to maximize the fitness value by comparing the current fitness of each particle with its historical best value, P. best_i The fitness of P is compared, and if it is better, then P is updated. best_i Similarly, update the global best G. best Further update particle state:

[0084] In the formula, w is the inertia weight; c1 and c2 are learning factors; and r1 and r2 are random numbers between [0,1] used to increase the randomness of the search.

[0085] The third step is to terminate the search. When the maximum number of iterations is reached or the fitness requirement is met, the iteration stops, and the globally optimal position G is output. best Parameter combination (C) opt ,γ opt This allows us to obtain the SVR model with optimal performance.

[0086] This invention automates the entire process from sample drying, spectral acquisition, label calculation to model construction, solving problems such as complex sample preparation, cumbersome calibration, and difficulty in online monitoring in traditional methods. It enables rapid, non-destructive, and automatic modeling and analysis of the moisture content of solid particles, significantly improving the efficiency and accuracy of moisture content detection in industrial drying processes.

[0087] This embodiment provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the aforementioned method for automatically modeling the moisture content of solid particles in an industrial drying process based on near-infrared technology.

[0088] This embodiment provides an electronic device, including a memory and a processor. The memory stores a computer program, and the processor runs the computer program to enable the electronic device to execute the described automatic modeling method for the moisture content of solid particles in an industrial drying process based on near-infrared technology.

[0089] Embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0090] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0091] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0092] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0093] Contents not described in detail in this specification are prior art known to those skilled in the art. It is hereby indicated that the above description is intended to help those skilled in the art understand this invention, but does not limit the scope of protection of this invention. Any equivalent substitutions, modifications, improvements, or simplifications of the above descriptions that do not depart from the essential content of this invention fall within the scope of protection of this invention.

Claims

1. An automatic modeling system for the moisture content of solid particles in an industrial drying process based on near-infrared technology, characterized in that, include: The near-infrared emitting device includes a near-infrared light source (1) and a focusing lens (2). The near-infrared light source (1) generates a wide-band near-infrared light and is focused by the focusing lens to form a high-energy-density light spot, which is optically coupled to the near-infrared transmission device. The near-infrared transmission device includes a first fiber coupler (3), a transmission fiber bundle (4), a second fiber coupler (5), and a reflection probe (6) connected in sequence. The first fiber coupler (3) receives a high-energy-density light spot from the near-infrared emitting device. The light spot is then projected onto the surface of the solid particle sample (9) by the reflection probe (6) via the transmission fiber bundle (4) and the second fiber coupler (5). The reflection probe (6) also collects the sample signal reflected by the solid particles and transmits it to the near-infrared detection device. The near-infrared detection device includes a detector (7) that receives near-infrared light signals carrying water content information from a near-infrared transmission device, analyzes them into near-infrared spectral data in digital form, and outputs them to a signal processing device. A hot flow drying device (11) is set above or to the side of the sample to generate a hot flow at a certain temperature and rate to dry the solid particle sample (9) placed on the weighing device (10). Weighing device (10) is used to monitor and acquire the weight change data of solid particle sample (9) during the drying process in real time, and output the weight data to the signal processing device through serial communication. The signal processing device receives spectral data from the near-infrared detection device and weight data from the weighing device (10), calculates the moisture content label corresponding to each spectrum based on the weight data, and performs machine learning training on the spectral data and moisture content label to obtain a prediction model for the moisture content of solid particles.

2. The automatic modeling system for the moisture content of solid particles in an industrial drying process based on near-infrared technology according to claim 1, characterized in that: The near-infrared light source (1) of the near-infrared emitting device has a wavelength range of 800-2500nm, and the high energy density light spot area formed after being focused by the focusing lens (2) is less than 4mm².

3. The automatic modeling system for the moisture content of solid particles in an industrial drying process based on near-infrared technology according to claim 1, characterized in that: The first fiber coupler (3) and the second fiber coupler (5) adopt SMA905 standard connectors, and the transmission fiber is a Y-type low hydroxyl silica fiber bundle.

4. The automatic modeling system for the moisture content of solid particles in an industrial drying process based on near-infrared technology according to claim 1, characterized in that: The detector (7) is a CCD array near-infrared detector with a spectral acquisition range of 800-2500nm.

5. The automatic modeling system for the moisture content of solid particles in an industrial drying process based on near-infrared technology according to claim 1, characterized in that: The condition for stopping drying in the hot flow drying device (11) is: within a fixed collection interval, the difference in mass between two adjacent weighings of the solid particle sample (9) is less than a preset threshold.

6. A method for automatically modeling the moisture content of solid particles in an industrial drying process based on near-infrared technology, characterized in that, This method, when executed, uses the automatic modeling system for the moisture content of solid particles in an industrial drying process based on near-infrared technology, as described in claims 1-7, and includes: S1: Initialize and perform initial data acquisition; S2: Perform periodic drying and data collection cycles according to preset time intervals; S3: Calculate the moisture content label based on the collected data; S4: Preprocess the acquired spectral data; S5: Principal component analysis (PCA) is used for characteristic wavelength selection; S6: Construct a quantitative analysis model for moisture content based on the Support Vector Machine (SVM) method.

7. The method for automatically modeling the moisture content of solid particles in an industrial drying process based on near-infrared technology according to claim 6, characterized in that, The moisture content label mentioned in S3 is calculated using the following formula: ； In the formula, Moisture content, , and The mass of the water-containing solid particle sample (9), the mass of the water content, and the mass of the dry solid particle sample (9) are respectively, and i is the time series i=1,2,3,...

8. The automatic modeling method for the moisture content of solid particles in an industrial drying process based on near-infrared technology according to claim 6, characterized in that, S4 include: The SG method was used to smooth the spectrum; Multivariate scattering correction (MSC) is performed on the smoothed spectrum; Baseline correction is performed using the airPLS method to eliminate baseline drift.

9. A computer-readable storage medium storing a computer program that, when executed by a processor, implements the automatic modeling method for the moisture content of solid particles in an industrial drying process based on near-infrared technology, as described in any one of claims 6-8.

10. An electronic device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor runs the computer program to cause the electronic device to perform the automatic modeling method for the moisture content of solid particles in an industrial drying process based on near-infrared technology as described in any one of claims 6-8.

Images