High-precision petroleum coke moisture content analysis system

Abstract

The invention relates to the technical field of industrial detection, and discloses a high-precision petroleum coke moisture content analysis system, which comprises a petroleum coke multispectral characteristic processing module, a moisture analysis algorithm matching module and a moisture content execution module, petroleum coke is finely divided into three independent analysis levels of fine powder, medium particles and coarse particles according to particle sizes, detection weights are dynamically distributed based on mass proportions of all the particle sizes, when the proportion of the fine powder exceeds a critical threshold value, the system automatically improves the detection weights of the fine powder, the high water absorption characteristic of the fine powder caused by sharp increase of the specific surface area is captured, and the detection accuracy is improved. Aiming at the capillary moisture adsorption phenomenon caused by fine powder enrichment, the system generates a compensation coefficient positively correlated with the fine powder concentration in real time, the structural deviation caused by neglecting the particle size-moisture relevance in a traditional method is eliminated, the detection precision of a high-fine powder sample reaches a new height, a three-dimensional moisture distribution diagram is constructed through a distributed sensor network, and the detection precision of the high-fine powder sample is improved. The inherent defect that materials are regarded as a static homogeneous system in a traditional method is overcome.

Description

Technical Field

[0001] This invention relates to the field of industrial testing technology, specifically a high-precision petroleum coke moisture content analysis system. Background Technology

[0002] Petroleum coke is a product obtained by separating light and heavy oils from crude oil through distillation, followed by thermal cracking of the heavy oil. It is also a black solid coke produced by cracking and coking vacuum residue of petroleum at 500°C in a coking unit. Generally considered an amorphous carbon body, it is a highly aromatic polymeric carbide containing needle-like or granular structures of tiny graphite crystals. As a residue left over from petroleum refining, petroleum coke can be used as fuel in power plants, improving recovery rates and reducing costs. However, petroleum coke requires the addition of water for processing. Excessive water content can cause the coke to lack adhesion during combustion, hindering complete chemical reactions. Therefore, appropriate moisture testing is necessary.

[0003] The main shortcomings of existing high-precision petroleum coke moisture content analysis systems include: 1. Poor particle size adaptability: After crushing and screening, petroleum coke is divided into multiple particle sizes. Different particle sizes have different water absorption rates. Traditional methods have not established a particle size-moisture correlation model, resulting in a detection deviation of up to 30% for samples with high fine powder content.

[0004] 2. Rigid Algorithm Selection: A single detection algorithm is difficult to adapt to complex working conditions. Near-infrared method fails due to sulfur-hydrogen bond interference when the sulfur content of petroleum coke is greater than 3%, while microwave method produces electromagnetic distortion when metal impurities exceed the standard. Existing technologies attempt to integrate data from multiple sensors, but lack a dynamic algorithm selection mechanism, and cannot achieve accurate matching of "detection scenario - optimal algorithm".

[0005] III. Uncompensated Environmental Interference: Petroleum coke is often stored in the open air, where ambient temperature and humidity cause moisture migration. Experiments show that for every 10°C increase in temperature, the microwave detection value drifts by +0.8%; for every 20% increase in humidity, the near-infrared detection value drifts by -1.2%. The existing system does not have a built-in real-time compensation module.

[0006] Therefore, to address the above problems, this invention provides a high-precision petroleum coke moisture content analysis system that integrates dynamic algorithm matching and real-time environmental correction to solve the problem of deteriorated detection accuracy under multi-source interference. Summary of the Invention

[0007] To address the shortcomings of existing technologies, this invention provides a high-precision petroleum coke moisture content analysis system, which solves the problems mentioned in the background section.

[0008] To achieve the above objectives, the present invention provides the following technical solution: a high-precision petroleum coke moisture content analysis system, the system comprising a petroleum coke multispectral feature processing module, a moisture analysis algorithm matching module, and a moisture content execution module; The petroleum coke multispectral feature processing module collects static physical feature data and dynamic spectral feature data of petroleum coke, including: particle size classification data, near-infrared spectral data, microwave dielectric feature data and surface texture image data; The moisture analysis algorithm matching module constructs a petroleum coke multispectral feature fusion matrix, aligns and combines particle size classification data, near-infrared spectral data, microwave dielectric feature data and surface texture image data according to sample number, constructs a petroleum coke multispectral feature fusion matrix, and matches it with a pre-stored standard moisture analysis algorithm feature library based on a biomimetic optimization algorithm, adaptively selecting the optimal moisture analysis algorithm type. The moisture content execution module calls a matching algorithm to calculate the moisture content and incorporates the environmental temperature and humidity compensation coefficient correction result. The system achieves a moisture content detection accuracy error of less than 0.5% through a four-level collaborative mechanism of particle size classification, multispectral fusion, algorithm adaptive matching, and environmental compensation.

[0009] Preferably, the petroleum coke multispectral feature processing module includes: The particle size classification unit uses a three-layer vibrating screen to classify petroleum coke into three grades based on particle size: less than 0.5 mm, 0.5 mm, and greater than 3 mm; and generates a particle size distribution vector. The static near-infrared acquisition unit uses a near-infrared spectrometer to scan the surface of petroleum coke. The microwave dielectric sensing unit measures the dielectric constant of petroleum coke using a dual-frequency microwave sensor. The surface texture analysis unit extracts surface texture features based on an industrial camera.

[0010] Preferably, the particle size classification unit performs: Set the screen tilt angle to 15° and the vibration frequency to 25Hz; The formula for calculating the mass percentage of each particle size is:

[0011] in For the sample quality at the i-th level of granularity, This refers to the total mass of the three particle sizes.

[0012] Preferably, the moisture analysis algorithm matching module includes: The feature fusion unit, which integrates granular vector features, integrates granular vector features. Near-infrared matrix Dielectric vector Texture vectors Construct a fusion matrix aligned by sample number ; Algorithm matching unit, based on Osprey optimization algorithm With standard algorithm library match, , These correspond to the drying and weighing method, near-infrared regression, and microwave phase difference method, respectively.

[0013] Preferably, the algorithm matching unit performs: Initialize osprey population location ; The formula for updating position during the exploration phase is:

[0014] in To explore the updated location during the exploration phase. This is the current position of the Osprey. For a range of random numbers, This is the current optimal solution position. It is a constant; During the development phase, new locations are generated and Euclidean distance fitness is calculated. Choose the smallest The corresponding algorithm type.

[0015] Preferably, the moisture content execution module includes: Algorithm execution unit: Matching drying and weighing method Then calculate the drying quality loss rate and add the particle size compensation coefficient; Matched near-infrared regression method Then the PLSR model is called to calculate the moisture content; Matching microwave phase difference method The moisture content is then calculated based on the dielectric eigenvector.

[0016] Preferably, the moisture content execution module further includes: The environmental compensation unit acquires environmental parameters through temperature and humidity sensors, calculates compensation factors, and corrects the moisture content results.

[0017] Preferred options also include: The database module stores the feature library of standard algorithms and historical data; The feedback learning unit dynamically updates the algorithm model parameters for samples with a deviation greater than 1%.

[0018] Preferably, the steps include: S1: Collect data on petroleum coke particle size, near-infrared spectroscopy, microwave dielectric properties, and surface texture. S2: Construct a multispectral feature fusion matrix; S3: The Osprey optimization algorithm is used to match the optimal moisture analysis algorithm; S4: Perform algorithm calculations and introduce environmental compensation; S5: Model parameter updates are triggered based on detection bias.

[0019] Preferably, the implementation process of the Osprey optimization algorithm in S3 includes the following steps: Step 10.1: Algorithm parameter initialization: The osprey population size was set at 50 birds, the maximum number of iterations was 100, the search space dimension was 3 corresponding to three candidate algorithm types: drying and weighing method, near-infrared regression method, and microwave phase difference method. The position boundary was set in the interval [0.1, 0.9], the step size scaling factor was 0.03, and the Levy flight index was fixed at 1.5. Step 10.2: Population location generation: The initial position of each osprey is generated using a random initialization strategy; Step 10.3: Levy Flight Development Phase During each iteration, the Osprey's position is updated using the Levy flight mechanism:

[0020] in Let i be the position of the osprey after t iterations. Step scaling factor, Levy is the dot product operator. To meet The Levy distribution has a random step size; This mechanism enables the algorithm to perform a fine-grained local search with a 90% probability and a large-scale jump with a 10% probability. Step 10.4: Boundary constraint handling: Reflection correction is applied to positions outside the [0.1, 0.9] boundary to avoid boundary clustering effects; Step 10.5: Fitness Assessment and Selection: Calculate the fitness value corresponding to the position:

[0021] in Let represent the fitness of osprey i at iteration t. This is a multispectral feature fusion matrix. Let be the feature vector of the h-th class of algorithms.

[0022] Preserve the best historical position; Step 10.6: Iterative Position Update: Execute the position update rules for the exploration and development phases until the maximum number of iterations is reached; Step 10.7: Output of the optimal algorithm type: Map the global optimal position to the algorithm type index: The interval [0.1, 0.4] is mapped to the drying and weighing method; The [0.4, 0.7] interval is mapped to the near-infrared regression method; The [0.7,0.9] interval is mapped using the microwave phase difference method.

[0023] Compared with existing technologies, the present invention provides a high-precision petroleum coke moisture content analysis system, which has the following advantages: 1. Overcoming the limitations of particle size adaptability: Traditional methods neglect the differences in water absorption rates across different particle sizes, leading to severe distortions in the detection of samples with high fine powder content. This invention innovatively establishes a particle size stratification analysis system: (1) Three-level particle size independent analysis channel: By finely dividing petroleum coke into three independent analytical levels—fine powder, medium powder, and coarse powder—based on particle size, and dynamically allocating detection weights based on the mass ratio of each particle size, the system automatically increases the detection weight of fine powder when the proportion of fine powder exceeds the critical threshold, thus capturing the high water absorption characteristics of fine powder caused by the surge in specific surface area.

[0024] (2) Intelligent compensation for capillary effect: To address the capillary moisture adsorption phenomenon caused by fine powder enrichment, the system generates a compensation coefficient that is positively correlated with the fine powder concentration in real time, eliminating the structural bias caused by neglecting the particle size-moisture correlation in traditional methods, thus achieving a new level of accuracy in detecting high-fine powder samples.

[0025] 2. Addressing the rigidity issue in algorithm selection: Traditional fixed-weight fusion strategies fail when faced with multi-source disturbances such as sulfur interference and metal contamination. This invention proposes a biomimetic optimization dynamic decision-making architecture: (1) Osprey Optimized Intelligent Decision Engine: Through a biomimetic population location update mechanism, the system can detect sudden changes in sulfur content and excessive metal impurities in real time. It can dynamically switch the dominant algorithm within milliseconds. Under high sulfur conditions, it can automatically shield near-infrared data and increase microwave weight. In metal pollution scenarios, it can seamlessly switch to the drying method as the dominant mode, avoiding systemic misjudgments caused by the superposition of multiple sources of interference.

[0026] (3) Multimodal sensing fault-tolerant mechanism: When oil film contamination causes optical detection to fail, the system automatically activates the surface texture analysis alternative channel to invert the moisture content through micro-texture features. This mechanism enables the detection system to have self-healing capabilities under working conditions and maintain stable output even in extremely complex environments.

[0027] 3. Systematically eliminate environmental interference defects: Traditional methods, lacking a coupled model between environmental disturbances and moisture migration, result in fluctuating detection values ​​in open-air scenes. This invention constructs an environmental adaptive compensation system: (1) Temperature and humidity linkage compensation mechanism: The system collects ambient temperature and humidity parameters in real time. Through the dual effects of temperature deviation compensation and humidity deviation compensation, it offsets the detection drift caused by changes in environmental parameters. At the same time, it combines infrared thermal imaging technology to dynamically track the moisture gradient distribution of the material pile and capture surface-migrating moisture.

[0028] (2) Dynamic modeling of water migration: A three-dimensional moisture distribution map is constructed by a distributed sensor network. When the redistribution of moisture due to environmental changes is detected, the system automatically activates a hierarchical compensation model. This model dynamically corrects the detection value based on the rate of change of dielectric properties of different depth layers, thus solving the inherent defect of traditional methods that treat materials as static homogeneous systems. Attached Figure Description

[0029] Figure 1 This is a schematic diagram of the overall system architecture of the present invention. Detailed Implementation

[0030] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0031] Please see Figure 1 This is a high-precision petroleum coke moisture content analysis system, which includes a petroleum coke multispectral feature processing module, a moisture analysis algorithm matching module, and a moisture content execution module. The petroleum coke multispectral feature processing module collects static physical feature data and dynamic spectral feature data of petroleum coke, including: particle size classification data, near-infrared spectral data, microwave dielectric feature data and surface texture image data; The moisture analysis algorithm matching module constructs a multispectral feature fusion matrix for petroleum coke. It aligns and combines particle size classification data, near-infrared spectral data, microwave dielectric feature data, and surface texture image data according to sample numbers to construct the multispectral feature fusion matrix for petroleum coke. Based on a biomimetic optimization algorithm, it matches the matrix with a pre-stored standard moisture analysis algorithm feature library and adaptively selects the optimal moisture analysis algorithm type. The moisture content execution module calls the matching algorithm to calculate the moisture content and incorporates the results of the environmental temperature and humidity compensation coefficient correction. The system achieves a moisture content detection accuracy error of less than 0.5% through a four-level collaborative mechanism of granular classification, multispectral fusion, algorithm adaptive matching, and environmental compensation.

[0032] The petroleum coke multispectral feature processing module includes: The particle size classification unit uses a three-layer vibrating screen to classify petroleum coke into three grades based on particle size: less than 0.5 mm, 0.5 mm, and greater than 3 mm; and generates a particle size distribution vector. The static near-infrared acquisition unit uses a near-infrared spectrometer to scan the surface of petroleum coke. The microwave dielectric sensing unit measures the dielectric constant of petroleum coke using a dual-frequency microwave sensor. The surface texture analysis unit extracts surface texture features based on an industrial camera.

[0033] Particle size classification unit execution: Set the screen tilt angle to 15° and the vibration frequency to 25Hz; The formula for calculating the mass percentage of each particle size is:

[0034] in For the sample quality at the i-th level of granularity, This refers to the total mass of the three particle sizes.

[0035] The moisture analysis algorithm matching module includes: The feature fusion unit, which integrates granular vector features, integrates granular vector features. Near-infrared matrix Dielectric vector Texture vectors Construct a fusion matrix aligned by sample number ; Algorithm matching unit, based on Osprey optimization algorithm With standard algorithm library match, , These correspond to the drying and weighing method, near-infrared regression, and microwave phase difference method, respectively.

[0036] Algorithm matching unit execution: Initialize osprey population location ; The formula for updating position during the exploration phase is:

[0037] in To explore the updated location during the exploration phase. This is the current position of the Osprey. For a range of random numbers, This is the current optimal solution position. It is a constant; During the development phase, new locations are generated and Euclidean distance fitness is calculated. Choose the smallest The corresponding algorithm type.

[0038] The moisture content execution module includes: Algorithm execution unit: Matching drying and weighing method Then calculate the drying quality loss rate and add the particle size compensation coefficient; Matched near-infrared regression method Then the PLSR model is called to calculate the moisture content; Matching microwave phase difference method The moisture content is then calculated based on the dielectric eigenvector.

[0039] The moisture content execution module also includes: The environmental compensation unit acquires environmental parameters through temperature and humidity sensors, calculates compensation factors, and corrects the moisture content results.

[0040] Also includes: The database module stores the feature library of standard algorithms and historical data; The feedback learning unit dynamically updates the algorithm model parameters for samples with a deviation greater than 1%.

[0041] Including the following steps: S1: Collect data on petroleum coke particle size, near-infrared spectroscopy, microwave dielectric properties, and surface texture. S2: Construct a multispectral feature fusion matrix; S3: The Osprey optimization algorithm is used to match the optimal moisture analysis algorithm; S4: Perform algorithm calculations and introduce environmental compensation; S5: Model parameter updates are triggered based on detection bias.

[0042] The implementation process of the Osprey optimization algorithm in S3 includes the following steps: Step 10.1: Algorithm parameter initialization: The osprey population size was set at 50 birds, the maximum number of iterations was 100, the search space dimension was 3 corresponding to three candidate algorithm types: drying and weighing method, near-infrared regression method, and microwave phase difference method. The position boundary was set in the interval [0.1, 0.9], the step size scaling factor was 0.03, and the Levy flight index was fixed at 1.5. Step 10.2: Population location generation: The initial position of each osprey is generated using a random initialization strategy; Step 10.3: Levy Flight Development Phase During each iteration, the Osprey's position is updated using the Levy flight mechanism:

[0043] in Let i be the position of the osprey after t iterations. Step scaling factor, Levy is the dot product operator. To meet The Levy distribution has a random step size; This mechanism enables the algorithm to perform a fine-grained local search with a 90% probability and a large-scale jump with a 10% probability. Step 10.4: Boundary constraint handling: Reflection correction is applied to positions outside the [0.1, 0.9] boundary to avoid boundary clustering effects; Step 10.5: Fitness Assessment and Selection: Calculate the fitness value corresponding to the position:

[0044] in Let represent the fitness of osprey i at iteration t. This is a multispectral feature fusion matrix. Let be the feature vector of the h-th class of algorithms.

[0045] Preserve the best historical position; Step 10.6: Iterative Position Update: Execute the position update rules for the exploration and development phases until the maximum number of iterations is reached; Step 10.7: Output of the optimal algorithm type: Map the global optimal position to the algorithm type index: The interval [0.1, 0.4] is mapped to the drying and weighing method; The [0.4, 0.7] interval is mapped to the near-infrared regression method; The [0.7,0.9] interval is mapped using the microwave phase difference method.

[0046] Example 1: Verification of particle size compensation under extreme fine powder conditions: In a petroleum coke micronization production line, where fine powder accounts for more than 60%, this system is implemented, and the particle size compensation mechanism is verified through the following steps: Step 1: Granularity stratification data collection: Deploy a high-speed pneumatic sorting device, combined with a laser diffraction particle size analyzer, to separate and measure the three-level particle size in real time: Fine powder grade: Precision electronic balance under nitrogen protection, with an accuracy of +0.01g; Medium-grained: A machine vision-based volume-mass conversion model; Coarse-grained: 3D scanning and reconstruction of geometric volume; Each batch can process 5 tons, and the sampling cycle is less than 3 minutes.

[0047] Step 2: Dynamic compensation for capillary effect: When the proportion of fine powder reaches 65%: The system automatically activates the compensation coefficient. ,in This is the capillary effect compensation coefficient; The weight of the fine powder channel is increased to 80%, medium powder to 15%, and coarse powder to 5%. Samples with a true moisture content of 9.8% were tested: Traditional methods, which ignore the enrichment effect of fine powders, yield a detection value of only 7.2%. In this scheme: the fine powder channel measured 10.5%, the medium powder 9.1%, and after weighted compensation, the output was 9.85%. Step 3: Continuous Production Validation: 30 days of continuous operation showed that: When the proportion of fine powder fluctuates between 55% and 75%, the detection stability remains at +0.6%. The improved precision has reduced quality claims by 250,000 yuan per month.

[0048] Example 2: Verification of temperature and humidity compensation in high-altitude and cold environments: At a northern storage and transportation base where winter temperatures reach -30°C, the environmental adaptive system was tested. Step 1: Real-time tracking of environmental parameters: Temperature monitoring: Distributed thermocouple array; Humidity monitoring: Anti-icing capacitive sensor; Moisture migration: Insertion TDR probe network; Step 2: Execution of the composite compensation model: At 3:00 AM, the temperature was -25°C and the humidity was 70%. Temperature compensation term: 0.085 × (-25 - 25) = 4.25% Humidity compensation: 0.012 × (70 - 50) = +0.24% Infrared thermal imaging shows the surface freezing moisture gradient:

[0049] in This is the amount of compensation for the freezing moisture gradient. For integration operators, The percentage change in moisture content per unit depth. For depth infinitesimal elements; Final revision:

[0050] in To compensate for the final moisture content, This is the original calculated moisture value. This is the compensation value calculated using the temperature and humidity compensation formula described above.

[0051] Example 3: Smart Factory End-to-End Integration: End-to-end deployment at a refining and chemical base with a capacity of tens of millions of tons: 1. Three-level perception architecture: Physical property layer: Laser diffraction and machine vision are linked to achieve precise metering of 80-micron fine powder; Chemical composition layer: Miniature mass spectrometer coupled with laser spectroscopy to simultaneously detect the concentration of sulfur and metal elements; Environmental perception layer: Weather stations and buried sensor networks construct a three-dimensional environmental field; 2. Dynamic Decision-Making Center: The edge computing unit executes the Osprey optimization algorithm to achieve three core capabilities: The dominant algorithm is automatically switched when the sulfur content exceeds 3%. When metal contamination exceeds 500 ppm, the drying method fault-tolerant mode is activated; Real-time compensation lag is less than 10 seconds under conditions of drastic temperature and humidity changes. 3. Economic benefit transformation: The time required for single-sample testing has been reduced from 4 hours to 2 minutes; Annual quality incident losses reduced by 90%; The factory's revenue increased by over 8 million yuan due to more precise moisture testing. Example 4: Dynamic correction under high sulfur and high humidity combined operating conditions: At a coastal coking plant, with a sulfur content of 4.2% and an ambient humidity of 85%, the system's ability to resist complex interference was verified. Step 1: Multi-source interference sensing: Laser spectroscopy detected that the intensity of the sulfur characteristic peak exceeded the standard by 150%, triggering a red alert for sulfur interference. The capacitive humidity sensor monitors humidity levels that are consistently above 80% in real time. The microwave dielectric sensor showed an abnormal fluctuation of 30% in the imaginary part of the dielectric constant; Step 2: Three-level dynamic response: The system performs triple correction within 100 milliseconds: Algorithm switching: The Osprey optimization algorithm sets the near-infrared weight to zero, reduces the microwave weight to 20%, and increases the drying method to 80% dominance; Humidity Compensation: Activate the humidity correction module to compensate for 0.12% moisture content for every 10% humidity deviation. Sulfur interference shielding: A digital filter is implanted at 2200nm to eliminate parasitic absorption of SH bonds. Step 3: Verification under combined operating conditions: Traditional method: Sulfur interference causes near-infrared values ​​to be artificially high by 2.8%, and humidity drift causes microwave values ​​to be low by 1.5%, resulting in a final error of 4.3%. This system: The baseline value measured by the drying method is 6.5%, the humidity compensation is +0.42%, and the output is 6.92%; Continuous operation for 30 days showed that the stability under high sulfur and high humidity conditions remained at 0.4%. Example 5: Heavy metal contamination tolerance verification: For pollution conditions where vanadium-containing catalyst residues exceed 500 ppm: Step 1: Pollution Feature Identification Laser-induced breakdown spectroscopy captures characteristic signals of vanadium in real time, and a microwave sensor detects an abnormal increase in the imaginary part of the dielectric constant to three times the normal value. (1) The system determines the heavy metal pollution level to be severe; (2) Automatically reduce the weight of microwave data to 10%; Step 2: Multi-algorithm collaborative decision-making: The Osprey optimization algorithm completes dynamic switching within 95 milliseconds: (1) The near-infrared weight remains at 30%; (2) The weight of the drying method has been increased to 60%, making it the dominant detection method; (3) Shield microwave data channels that are susceptible to metal interference; The actual measured contaminated sample had an output value of 6.6%, with the error controlled within 1.5%.

[0052] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0053] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A high-precision petroleum coke moisture content analysis system, characterized in that, The system includes a petroleum coke multispectral feature processing module, a moisture analysis algorithm matching module, and a moisture content execution module. The petroleum coke multispectral feature processing module collects static physical feature data and dynamic spectral feature data of petroleum coke, including: particle size classification data, near-infrared spectral data, microwave dielectric feature data and surface texture image data; The moisture analysis algorithm matching module constructs a petroleum coke multispectral feature fusion matrix, aligns and combines particle size classification data, near-infrared spectral data, microwave dielectric feature data and surface texture image data according to sample number, constructs a petroleum coke multispectral feature fusion matrix, and matches it with a pre-stored standard moisture analysis algorithm feature library based on a biomimetic optimization algorithm, adaptively selecting the optimal moisture analysis algorithm type. The moisture content execution module calls a matching algorithm to calculate the moisture content and incorporates the environmental temperature and humidity compensation coefficient correction result. The system achieves a moisture content detection accuracy error of less than 0.5% through a four-level collaborative mechanism of particle size classification, multispectral fusion, algorithm adaptive matching, and environmental compensation.

2. The system according to claim 1, characterized in that, The petroleum coke multispectral feature processing module includes: The particle size classification unit uses a three-layer vibrating screen to classify petroleum coke into three grades based on particle size: less than 0.5 mm, 0.5 mm, and greater than 3 mm; and generates a particle size distribution vector. The static near-infrared acquisition unit uses a near-infrared spectrometer to scan the surface of petroleum coke. The microwave dielectric sensing unit measures the dielectric constant of petroleum coke using a dual-frequency microwave sensor. The surface texture analysis unit extracts surface texture features based on an industrial camera.

3. The system according to claim 2, characterized in that, The particle size classification unit performs the following: Set the screen tilt angle to 15° and the vibration frequency to 25Hz; The formula for calculating the mass percentage of each particle size is: in For the sample quality of the i-th level granularity, This refers to the total mass of the three particle sizes.

4. The system according to claim 1, characterized in that, The moisture analysis algorithm matching module includes: The feature fusion unit, which integrates granular vector features, integrates granular vector features. Near-infrared matrix Dielectric vector Texture vectors Construct a fusion matrix aligned by sample number ; Algorithm matching unit, based on Osprey optimization algorithm With standard algorithm library match, , These correspond to the drying and weighing method, near-infrared regression, and microwave phase difference method, respectively.

5. The system according to claim 1, characterized in that, The algorithm matching unit executes: Initialize osprey population location ; The formula for updating position during the exploration phase is: in To explore the updated location during the exploration phase. This is the current position of the Osprey. For a range of random numbers, This is the current optimal solution position. It is a constant; During the development phase, new locations are generated and Euclidean distance fitness is calculated. Choose the smallest The corresponding algorithm type.

6. The system according to claim 1, characterized in that, The moisture content execution module includes: Algorithm execution unit: Matching drying and weighing method Then calculate the drying quality loss rate and add the particle size compensation coefficient; Matched near-infrared regression method Then the PLSR model is called to calculate the moisture content; Matching microwave phase difference method The moisture content is then calculated based on the dielectric eigenvector.

7. The system according to claim 1, characterized in that, The moisture content execution module also includes: The environmental compensation unit acquires environmental parameters through temperature and humidity sensors, calculates compensation factors, and corrects the moisture content results.

8. The system according to claim 1, characterized in that, Also includes: The database module stores the feature library of standard algorithms and historical data; The feedback learning unit dynamically updates the algorithm model parameters for samples with a deviation greater than 1%.

9. A method for analyzing the moisture content of petroleum coke, characterized in that, Including the following steps: S1: Collect data on petroleum coke particle size, near-infrared spectroscopy, microwave dielectric properties, and surface texture. S2: Construct a multispectral feature fusion matrix; S3: The Osprey optimization algorithm is used to match the optimal moisture analysis algorithm; S4: Perform algorithm calculations and introduce environmental compensation; S5: Model parameter updates are triggered based on detection bias.

10. The method according to claim 9, characterized in that, The implementation process of the Osprey optimization algorithm in S3 includes the following steps: Step 10.1: Algorithm parameter initialization: The osprey population size was set at 50 birds, the maximum number of iterations was 100, the search space dimension was 3 corresponding to three candidate algorithm types: drying and weighing method, near-infrared regression method, and microwave phase difference method. The position boundary was set in the interval [0.1, 0.9], the step size scaling factor was 0.03, and the Levy flight index was fixed at 1.

5. Step 10.2: Population location generation: The initial position of each osprey is generated using a random initialization strategy; Step 10.3: Levy Flight Development Phase During each iteration, the Osprey's position is updated using the Levy flight mechanism: in Let i be the position of the osprey after t iterations. Step scaling factor, Levy is the dot product operator. To meet The Levy distribution has a random step size; This mechanism enables the algorithm to perform a fine-grained local search with a 90% probability and a large-scale jump with a 10% probability. Step 10.4: Boundary constraint handling: Reflection correction is applied to positions outside the [0.1, 0.9] boundary to avoid boundary clustering effects; Step 10.5: Fitness Assessment and Selection: Calculate the fitness value corresponding to the position: in Let represent the fitness of osprey i at iteration t. This is a multispectral feature fusion matrix. This represents the feature vector of the h-th class of algorithms. The historical best position is preserved. Step 10.6: Iterative Position Update: Execute the position update rules for the exploration and development phases until the maximum number of iterations is reached; Step 10.7: Output of the optimal algorithm type: Map the global optimal position to the algorithm type index: The interval [0.1, 0.4] is mapped to the drying and weighing method; The [0.4, 0.7] interval is mapped to the near-infrared regression method; The [0.7,0.9] interval is mapped using the microwave phase difference method.

Images