A method and system for accurate separation and regulation of complex coal quality

Abstract

The application relates to the technical field of coal quality sorting, and discloses a precision sorting and regulation method and system suitable for complex coal quality, which comprises the following steps: determining the component parameters of each component based on a component analysis model and coal quality parameters, comparing all the component parameters in a historical regulation database, determining the fan rotating speed of a blower based on the similarity of the component parameters when a calling strategy is determined, constructing a wind speed mapping model based on a sample data set and the historical regulation database when a compensation strategy is determined, determining whether the stability of the fan rotating speed is qualified based on wind speed fluctuation frequency, determining the fan rotating speed correlation result based on the valve opening and a correlation rule algorithm, determining the rotating speed regulation coefficient of the fan rotating speed according to the fan rotating speed correlation result, and adjusting the fan rotating speed according to the rotating speed regulation coefficient; and the reliability of sorting is ensured by determining the component parameters and judging the stability of the fan rotating speed.

Description

Technical Field

[0001] This invention relates to the field of coal quality sorting technology, and more specifically, to a method and system for precise sorting and control of complex coal quality. Background Technology

[0002] With the deepening of coal mining and the increasing frequency of coal blending across multiple mining areas, dry separation methods face greater challenges. Dry separation achieves particle stratification through differences in airflow buoyancy and resistance, and the precision of blower control, as the power source for the airflow, directly determines the separation effect for complex coal types.

[0003] Currently, the parameter settings of blowers usually rely on the past experience of operators. Different operators have different judgment standards for the normal fluidization state. Since the composition and particle size of complex coal usually fluctuate, relying on the experience of operators cannot dynamically adjust the blower according to the actual situation of the coal quality, resulting in coal lumps accumulating and forming a "dead bed zone". Due to the lack of feedback adjustment mechanism and the fact that the control is easily affected by human experience, the reliability of sorting is insufficient.

[0004] Therefore, it is necessary to design a precise sorting and control method and system suitable for complex coal quality in order to solve the problems existing in the current technology. Summary of the Invention

[0005] In view of this, the present invention proposes a precise sorting and control method and system suitable for complex coal quality, aiming to solve the problems that rely on the operator's experience judgment to dynamically adjust the blower according to the actual coal quality, lack of feedback adjustment mechanism, and easy influence of human experience on control, resulting in insufficient reliability of sorting.

[0006] In one aspect, this invention proposes a precise sorting and control method suitable for complex coal qualities, comprising:

[0007] Obtain coal quality parameters, and determine the composition parameters of each component based on the composition analysis model and coal quality parameters;

[0008] Based on the comparison of all component parameters in the historical control database, the calling strategy or compensation strategy is determined according to the comparison results. When the calling strategy is determined, the blower speed is determined based on the similarity of component parameters. When the compensation strategy is determined, a wind speed mapping model is constructed based on the sample dataset and the historical control database, and the blower speed is determined according to the wind speed mapping model.

[0009] The wind speed fluctuation frequency during the sorting period is obtained, and the stability of the fan speed is determined based on the wind speed fluctuation frequency. When the stability of the fan speed is determined to be unqualified, the valve opening of the dry separator is obtained, and the fan speed correlation result is determined based on the valve opening and the association rule algorithm.

[0010] The fan speed adjustment coefficient is determined based on the fan speed correlation result, and the fan speed is adjusted according to the fan speed adjustment coefficient. The coal quality is then sorted according to the adjusted fan speed.

[0011] Furthermore, in obtaining coal quality parameters and determining the composition parameters of each component based on the composition analysis model and coal quality parameters, the following steps are included:

[0012] A laser particle size analyzer, an infrared ash analyzer, and a near-infrared spectral camera are installed at the feed end of the dry separator, and the coal quality parameters are obtained based on the laser particle size analyzer, the infrared ash analyzer, and the near-infrared spectral camera.

[0013] Obtain a coal quality dataset and divide the coal quality dataset into a training set and a test set;

[0014] The initial model is constructed by determining the parameters based on grid search, iteratively training the initial model according to the training set, and evaluating the error of the iteratively trained initial model according to the test set to determine the component analysis model.

[0015] Substitute the coal quality parameters into the composition analysis model to determine the composition parameters of each component;

[0016] The initial model is either an RBF neural network model or a BP neural network model.

[0017] Furthermore, when iteratively training the initial model based on the training set, and evaluating the error of the iteratively trained initial model based on the test set to determine the component analysis model, the process includes:

[0018] If the mean squared error of the initial model after the current iteration is less than the mean squared error of the initial model after the previous iteration, and is less than the mean squared error threshold, then stop the iterative training; otherwise, adjust the learning rate of the initial model and continue the iterative training.

[0019] The mean square error threshold is inversely proportional to the amount of data in the coal quality dataset.

[0020] Furthermore, when comparing all component parameters in the historical control database and determining the invocation or compensation strategy based on the comparison results, the following steps are taken:

[0021] All component parameters are constructed into a component dataset. The historical control database includes several historical component datasets and several historical wind turbine speeds, and each historical component dataset corresponds to a historical wind turbine speed.

[0022] The component dataset is compared with each historical component dataset to determine the similarity of component parameters;

[0023] If the similarity of component parameters is greater than 0.8, then the calling strategy is determined.

[0024] When there is no component parameter similarity greater than 0.8, the compensation strategy is determined.

[0025] Furthermore, when the invocation strategy is determined, determining the blower speed based on the component parameter similarity includes:

[0026] The historical component datasets with a component parameter similarity greater than 0.8 are sorted in descending order of component parameter similarity.

[0027] When the maximum component parameter similarity is unique, the historical fan speed corresponding to the maximum component parameter similarity is determined as the fan speed of the blower.

[0028] When the maximum component parameter similarity is not unique, the average of the historical fan speeds corresponding to each maximum component parameter similarity is determined as the fan speed of the blower.

[0029] Furthermore, when the compensation strategy is determined, a wind speed mapping model is constructed based on the sample dataset and historical control database. When determining the blower speed according to the wind speed mapping model, the process includes:

[0030] The sample dataset includes elemental analysis parameters, physical property parameters, and coal and rock composition parameters;

[0031] The sample dataset and the historical regulation database are divided into a second training set and a second test set.

[0032] The wind speed mapping model is constructed based on the second training set and the second test set, and the component dataset is substituted into the wind speed mapping model to determine the blower speed.

[0033] Furthermore, when obtaining the wind speed fluctuation frequency during the sorting period and determining whether the stability of the fan speed is qualified based on the wind speed fluctuation frequency, the process includes:

[0034] The wind speed fluctuation frequency is compared with the wind speed fluctuation frequency threshold.

[0035] When the wind speed fluctuation frequency is greater than the fluctuation frequency threshold, the stability of the wind turbine speed is determined to be unqualified.

[0036] When the wind speed fluctuation frequency is less than or equal to the fluctuation frequency threshold, the stability of the fan speed is determined to be qualified, and the coal quality is sorted at the current fan speed.

[0037] Furthermore, when it is determined that the stability of the fan speed is unqualified, the valve opening of the dry separator is obtained, and the fan speed correlation result is determined based on the valve opening and the correlation rule algorithm, including:

[0038] Obtain the valve opening degree and environmental parameters of the dry separator and construct a related dataset with the fan speed;

[0039] The associated dataset is preprocessed, including data denoising and data normalization;

[0040] The Eclat algorithm is used to generate several candidate itemsets from the preprocessed association dataset. Frequent itemsets are determined based on the support of the candidate itemsets. Association rules are then selected based on the frequent itemsets. The association rules reflect the relationship between environmental conditions, valve opening, and fan speed. The association result of the fan speed is determined based on the association rules.

[0041] Furthermore, when determining the fan speed adjustment coefficient based on the fan speed correlation result, and adjusting the fan speed according to the fan speed adjustment coefficient, the process includes:

[0042] The number of correlation results for the fan speed is counted and recorded as the correlation count;

[0043] Determine the valve opening change value of the dry separator, and determine the speed adjustment coefficient of the fan based on the valve opening change value and the associated quantity;

[0044] The fan speed is directly proportional to the speed adjustment coefficient.

[0045] Compared with existing technologies, the advantages of this invention are as follows: It accurately obtains the component parameters of each coal component through a component analysis model, and matches the data with a historical control database, avoiding decisions based on human experience and reducing interference from human factors. This provides a standardized and regulated control basis for the sorting of complex coal, improving the consistency of control. Given the fluctuating composition and particle size of complex coal, a dual strategy of invocation and compensation is employed. When component parameters are similar, historical fan speeds are directly matched; when differences are significant, a wind speed mapping model is constructed to determine the fan speed. This ensures the flexibility and reliability of fan speed changes with coal quality, avoids coal block accumulation due to coal quality fluctuations, and guarantees the continuity of sorting operations. Based on the frequency of wind speed fluctuations during the sorting period, the stability of the fan speed is judged in real time. If any defects are found, the speed correlation result is determined according to the association rule algorithm, thereby determining the speed adjustment coefficient. This forms a closed-loop control of monitoring, judgment, and adjustment, ensuring that airflow buoyancy and resistance continuously adapt to the needs of particle stratification, thereby improving the clean coal recovery rate and gangue removal rate, and ensuring the reliability of sorting.

[0046] On the other hand, this application also provides a precision sorting and control system suitable for complex coal quality, for applying the above-mentioned precision sorting and control method suitable for complex coal quality, including:

[0047] The data acquisition and processing module is configured to acquire coal quality parameters and determine the composition parameters of each component based on the composition analysis model and the coal quality parameters.

[0048] The sorting and analysis module is configured to compare all component parameters in the historical control database, determine the calling strategy or compensation strategy based on the comparison results, and when the calling strategy is determined, determine the blower speed based on the component parameter similarity. When the compensation strategy is determined, construct a wind speed mapping model based on the sample dataset and the historical control database, and determine the blower speed based on the wind speed mapping model.

[0049] The sorting processing module is configured to acquire the wind speed fluctuation frequency during the sorting time period, determine whether the stability of the fan speed is qualified based on the wind speed fluctuation frequency, and when it is determined that the stability of the fan speed is unqualified, acquire the valve opening of the dry sorting machine, and determine the fan speed correlation result based on the valve opening and the association rule algorithm.

[0050] The control feedback module is configured to determine the speed adjustment coefficient of the fan speed based on the fan speed correlation result, adjust the fan speed according to the speed adjustment coefficient, and complete the coal quality sorting according to the adjusted fan speed.

[0051] It is understandable that the aforementioned method and system for precise sorting and control of complex coal has the same beneficial effects, and will not be elaborated further here. Attached Figure Description

[0052] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings:

[0053] Figure 1 A flowchart illustrating a precise sorting and control method for complex coal quality, provided as an embodiment of the present invention;

[0054] Figure 2 This is a functional block diagram of a precision sorting and control system for complex coal quality provided in an embodiment of the present invention. Detailed Implementation

[0055] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the disclosure to those skilled in the art. It should be noted that, unless otherwise specified, embodiments and features in the embodiments of the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0056] From the perspective of the causes of complex coal quality, on the one hand, as coal mining progresses to deeper levels, the coal quality of different deep coal seams is inherently different (such as the different proportions of combustible components and impurities in different coal seams), and the geological environment at depth is more complex, making it difficult to maintain a stable composition of the mined coal. On the other hand, the normalization of coal blending in multiple mining areas leads to the mixing of coal from different mining areas and coal seams. The particle size distribution (differences in particle size) and compositional properties (such as the proportions of volatile matter and ash) of these coals are already different, and the mixing further disrupts the homogeneity of the coal quality.

[0057] In terms of specific manifestations, the characteristic of complex coal quality is that coal quality parameters, such as component ratio and particle size distribution, fluctuate without a fixed pattern. There may be a sudden increase in the impurity content in a certain batch of coal, or there may be a significant change in the coarseness distribution of coal particles. The principle of dry separation relies on the buoyancy and resistance of airflow to achieve particle stratification. Such fluctuations in coal quality parameters will directly lead to changes in the airflow conditions required for separation. For example, coarser coal requires stronger airflow support. Relying on human experience to adjust the blower is difficult to keep up with such dynamic changes, which can easily lead to problems such as "dead bed zone".

[0058] In some embodiments of this application, see Figure 1 As shown, a precise sorting and control method suitable for complex coal qualities includes:

[0059] S100: Obtain coal quality parameters, and determine the composition parameters of each component based on the composition analysis model and coal quality parameters.

[0060] S200: Based on the comparison of all component parameters in the historical control database, determine the calling strategy or compensation strategy according to the comparison results. When the calling strategy is determined, the blower speed is determined based on the similarity of component parameters. When the compensation strategy is determined, a wind speed mapping model is constructed based on the sample dataset and the historical control database, and the blower speed is determined according to the wind speed mapping model.

[0061] S300: Obtain the wind speed fluctuation frequency during the sorting period, determine whether the stability of the fan speed is qualified based on the wind speed fluctuation frequency, and when it is determined that the stability of the fan speed is not qualified, obtain the valve opening of the dry separator, and determine the fan speed correlation result based on the valve opening and the association rule algorithm.

[0062] S400: Determine the fan speed adjustment coefficient based on the fan speed correlation results, adjust the fan speed according to the adjustment coefficient, and complete the coal sorting based on the adjusted fan speed.

[0063] Specifically, coal quality parameters are obtained. These parameters reflect the fluctuating characteristics of the coal to be sorted, such as its composition and particle size. The coal quality parameters are then processed using a component analysis model, which is a model that can decompose the overall characteristics of coal into individual component characteristics. The component analysis model can extract the specific attributes of each component from the obtained coal quality parameters, that is, the proportion of each component (component parameters). These component parameters meticulously depict the specific characteristics of each coal component, providing accurate data support for subsequent control. After determining the component parameters, these parameters are compared with the historical control database. The historical control database stores experience data on blower control under different coal quality conditions, including matching records of various historical component parameters and corresponding historical blower speeds. By comparing the degree of matching between the current coal quality component parameters and the data in the historical control database, it is possible to determine whether to adopt a calling strategy or a compensation strategy. The calling strategy is suitable for situations where the current component parameters match the existing data in the historical control database to a high degree, while the compensation strategy is for situations where the current component parameters match the existing data in the historical control database to a low degree. Therefore, since historical data cannot be directly matched to the current sorting scenario, a wind speed mapping model is constructed by combining the sample dataset and the historical control database. The sample dataset contains diverse and comprehensive coal quality condition data, effectively making up for the data gaps in the historical control database under special coal quality scenarios. The wind speed mapping model establishes the correspondence between component parameters and blower wind speeds, and the appropriate blower speed for the current component parameters can be determined through the wind speed mapping model. Because different operators have different standards for judging the normal fluidization state, and cannot make dynamic adjustments according to coal quality fluctuations, the coal quality parameters are analyzed by the component analysis model, thereby clarifying the true situation of the coal quality. Then, combined with the historical control database or wind speed mapping model, the blower speed is determined, which reduces the risk of blower speed inaccuracy caused by human experience judgment deviation and coal quality fluctuations, and ensures the reliability of accurate sorting of complex coal quality.

[0064] Specifically, after determining the fan speed, a dynamic feedback adjustment process is initiated to ensure the stability and accuracy of the fan speed during the sorting process. The frequency of wind speed fluctuations during the sorting period is obtained. This frequency refers to the degree to which the fan speed required for dry sorting changes within the set sorting time. It directly reflects the stable state of the fan speed. If the frequency of wind speed fluctuations is large, it indicates that the fan speed deviates from the stable range. The frequency of wind speed fluctuations can be used to determine whether the stability of the fan speed is up to standard. If it is not up to standard, the valve opening of the dry sorting machine is introduced for further analysis. The valve opening of the dry sorting machine is a parameter that affects the airflow state, and its magnitude will indirectly change the influence of airflow on the fan speed. The relationship between valve opening and fan speed is mined using an association rule algorithm. The association rule algorithm is an algorithm that can extract the potential correlation patterns between different parameters from massive data, thereby accurately finding the correspondence between changes in valve opening and fan speed adjustment, and thus forming a fan speed correlation result, that is, clarifying how different valve openings affect the fan speed. Based on the fan speed correlation result, the fan speed adjustment coefficient is determined. The speed adjustment coefficient is a parameter that quantifies the adjustment range of the fan speed and can control the increase or decrease of the fan speed.

[0065] Understandably, the fan speed may deviate from a stable state due to factors such as airflow disturbances and slight fluctuations in coal quality during the sorting process. Traditional control methods lack effective feedback correction mechanisms and cannot respond to such unstable factors in a timely manner. By monitoring the frequency of wind speed fluctuations to establish judgment criteria and combining the correlation between valve opening and fan speed to construct a dynamic control mechanism, the interference of human experience on control is further reduced. This allows the fan speed to be adjusted according to the actual operating conditions during the sorting process, ensuring the reliability of accurate sorting of complex coal.

[0066] In some embodiments of this application, the process of obtaining coal quality parameters and determining the component parameters of each component based on the component analysis model and the coal quality parameters includes: setting a laser particle size analyzer, an infrared ash analyzer, and a near-infrared spectral camera at the feed end of a dry separator; obtaining coal quality parameters based on the laser particle size analyzer, infrared ash analyzer, and near-infrared spectral camera; obtaining a coal quality dataset; dividing the coal quality dataset into a training set and a test set; determining the parameters and constructing an initial model based on grid search; iteratively training the initial model based on the training set; evaluating the error of the iteratively trained initial model based on the test set; determining the component analysis model; and substituting the coal quality parameters into the component analysis model to determine the component parameters of each component. The initial model is an RBF neural network model or a BP neural network model.

[0067] In some embodiments of this application, when iteratively training an initial model based on a training set, evaluating the error of the iteratively trained initial model based on a test set, and determining the component analysis model, the following steps are included: if the mean squared error of the initial model after the current iteration is less than the mean squared error of the initial model after the previous iteration and is less than the mean squared error threshold, then the iterative training is stopped; otherwise, the learning rate of the initial model is adjusted and iterative training continues. The mean squared error threshold is inversely proportional to the amount of data in the coal quality dataset.

[0068] Specifically, the primary step in determining the composition analysis model is to accurately and comprehensively acquire coal quality parameters. Three detection devices are installed at the feed end of the dry separator: a laser particle size analyzer to capture the size distribution of coal particles, providing real-time information on the proportion of particles of different sizes in the coal sample, reflecting the particle size characteristics of the coal; an infrared ash analyzer to detect impurities in the coal, which directly affect the coal quality and sorting difficulty; and a near-infrared spectral camera to collect near-infrared spectral signals from the coal, thereby determining information on multiple components such as volatile matter and fixed carbon, achieving multi-dimensional detection of coal composition. Through the coordinated acquisition of data from these three devices, a coal quality dataset encompassing particle size, ash content, and multi-component information is formed. The coal quality dataset is divided into training and testing sets, typically with 70%-80% of the data allocated to the training set and the remainder to the testing set. This ensures the model's generalization ability. The training set is used for parameter learning and pattern fitting, while the testing set is used to validate the model's generalization ability. Grid search iterates through parameter combinations in the parameter space to select parameters for the initial model, such as the number of hidden layer nodes and activation function type in a neural network. An RBF neural network or a BP neural network is then constructed, both of which excel at handling nonlinear data and can adapt to the complex correspondences between coal quality and composition parameters. The initial model is iteratively trained using the training set, gradually learning data patterns and relationships. Finally, the test set is used for evaluation. The mean squared error (MSE) after iterative training ultimately determines a component analysis model with satisfactory accuracy. The MSE measures the degree of bias in the model's predictions; a smaller MSE indicates more accurate predictions. After each training iteration, the MSE of the current iteration is compared with that of the previous iteration, while also referring to the MSE threshold. The MSE threshold is inversely proportional to the amount of data in the coal quality dataset. This is because a larger amount of data allows the model to learn from a wider range of samples, resulting in higher requirements for prediction accuracy, thus necessitating a corresponding reduction in the MSE threshold. Conversely, when the amount of data is small, the MSE threshold can be appropriately relaxed to prevent the model from failing to achieve stable performance due to insufficient samples, ensuring effective training. Therefore, this embodiment does not impose a fixed MSE threshold; it can be set based on the amount of data in the coal quality dataset.

[0069] Understandably, if the mean squared error (MSE) of the initial model after the current iteration is less than that of the initial model after the previous iteration and is less than the MSE threshold, it indicates that the model's performance is stabilizing, and the iterative training stops. Otherwise, it indicates that the model's performance is still unstable, and the learning rate of the initial model is adjusted. If the MSE is affected by the initial model's oscillation and non-convergence due to an excessively large learning rate, the learning rate is reduced. If the MSE is affected by insufficient training due to an excessively small learning rate, the learning rate is increased. After adjustment, iterative training continues until the condition that "the MSE of the initial model after the current iteration is less than that of the initial model after the previous iteration and is less than the MSE threshold" is met, at which point iterative training stops. By training the initial model, the component analysis model can accurately output the component parameters of each component in the coal quality parameters, i.e., the proportion of each component, ensuring the reliability of accurate sorting of complex coal.

[0070] In some embodiments of this application, when comparing all component parameters in a historical control database and determining the invocation strategy or compensation strategy based on the comparison results, the process includes: constructing a component dataset from all component parameters; the historical control database includes several historical component datasets and several historical wind turbine speeds, with each historical component dataset corresponding to a historical wind turbine speed; comparing the component dataset with each historical component dataset to determine the component parameter similarity; when there is a component parameter similarity greater than 0.8, the invocation strategy is determined; when there is no component parameter similarity greater than 0.8, the compensation strategy is determined.

[0071] Specifically, the component dataset is compared with each historical component dataset to determine the similarity of component parameters. Component parameter similarity is determined using algorithms such as Euclidean distance and cosine similarity; any one algorithm can be selected. Comparing the component dataset with historical component datasets reduces interference and errors from human judgment, enhancing adaptability to different sorting conditions. When the comparison reveals that the similarity of component parameters between a historical component dataset and the current component dataset exceeds a set standard (0.8), it indicates that the current coal quality is highly similar to the historical coal quality, and previously adapted historical blower speeds can be directly reused, thus ensuring the reliability and consistency of the sorting operation. Therefore, a reuse strategy is adopted. If the similarity between all historical component datasets and the current component dataset does not exceed this standard, it indicates that the current coal quality is a type rarely seen in past experience, and existing historical blower speeds cannot be directly adapted; in this case, a compensation strategy is adopted. For changes in sorting operation conditions, data-driven automated adjustments reduce reliance on human experience and intuition, lowering the uncertainty and control risks brought about by human judgment, thereby improving the reliability of sorting.

[0072] In some embodiments of this application, when the invocation strategy is determined, the blower speed is determined based on the component parameter similarity, including: sorting the historical component datasets with a component parameter similarity greater than 0.8 in descending order of component parameter similarity; when the maximum component parameter similarity is unique, the historical blower speed corresponding to the maximum component parameter similarity is determined as the blower speed; when the maximum component parameter similarity is not unique, the average of the historical blower speeds corresponding to each maximum component parameter similarity is determined as the blower speed.

[0073] Specifically, after determining the invocation strategy, the blower speed suitable for the current coal quality is selected from the historical component datasets that meet the conditions. The historical component datasets that meet the requirements of the invocation strategy are arranged in descending order of component parameter similarity. Priority is given to locking the past data that is closest to the current coal quality characteristics. The higher the component parameter similarity, the smaller the component difference between the corresponding historical coal quality and the current coal quality, and the more suitable the historical blower speed is for the current sorting needs. If the maximum parameter similarity is unique after sorting, it means that the matching degree of this data with the current coal quality is the highest among all records that meet the conditions, and there is no other similar data that can compete with it. At this time, the historical blower speed corresponding to the historical component dataset is directly determined as the current blower speed. If multiple historical component datasets have the maximum parameter similarity at the same time, it means that the matching degree of these historical data with the current coal quality is quite similar. Selecting the historical blower speed corresponding to any one of these data may lead to deviation due to accidental factors. Therefore, the average value of the historical blower speeds corresponding to these maximum parameter similarities is determined as the current blower speed. When reusing historical control experience, both accuracy and stability are taken into account, avoiding control deviations caused by improper screening of historical data, reducing the risk of coal block accumulation forming "dead bed zones", reducing reliance on human experience and judgment, reducing human error in the sorting process, and improving the overall level of automation and sorting reliability.

[0074] In some embodiments of this application, when a compensation strategy is determined, a wind speed mapping model is constructed based on a sample dataset and a historical control database. When determining the blower speed according to the wind speed mapping model, the following steps are taken: the sample dataset includes elemental analysis parameters, physical property parameters, and coal and rock composition parameters; the sample dataset and the historical control database are divided into a second training set and a second test set; a wind speed mapping model is constructed based on the second training set and the second test set; and the component dataset is substituted into the wind speed mapping model to determine the blower speed.

[0075] Specifically, under the compensation strategy, a wind speed mapping model is constructed based on the sample dataset and the historical regulation database. The sample dataset is crucial for supplementing the historical control database. It contains three types of parameters: elemental analysis parameters, which include the proportions of chemical elements such as carbon, hydrogen, oxygen, and sulfur in different coal types; physical property parameters, which include physical properties such as particle size, density, and hardness in different coal types, and are important factors affecting the effect of airflow; and coal and petrographic composition parameters, which record the types and proportions of coal and petrographic components such as vitrinite and filamentous group in different coal types, and determine the impact of coal structure characteristics on sorting. After these three types of data form the sample dataset and are integrated with the historical control database, they are divided into a second training set and a second test set. A wind speed mapping model is then constructed based on the second training set and the second test set. The wind speed mapping model is built using a random forest model as its architecture. The random forest model is good at handling multi-dimensional and complexly correlated feature data. The sample dataset contains elemental analysis parameters, physical property parameters, and coal and petrographic composition parameters. There are complex relationships of nonlinearity and cross-influence between different parameters. The random forest, through ensemble learning of multiple decision trees, does not require complex linear transformations or feature selection of the data. Each decision tree can focus on the local regularities of some parameters, while the voting mechanism of multiple trees can integrate global regularities. The process of dividing the second training set and the second test set, as well as the model training process, are consistent with the initial model and will not be repeated here. The blower speed is determined by the wind speed mapping model, which eliminates the reliance on human experience. At the same time, the model construction realizes the accurate mapping between coal quality parameters and blower speed, further reducing the probability of the occurrence of "dead bed zone". This allows dry separation to maintain stable and accurate separation effect under special coal quality scenarios, ensuring the reliability of separation.

[0076] In some embodiments of this application, when obtaining the wind speed fluctuation frequency during the sorting time period and determining whether the stability of the fan speed is qualified based on the wind speed fluctuation frequency, the method includes: comparing the wind speed fluctuation frequency with a wind speed fluctuation frequency threshold; when the wind speed fluctuation frequency is greater than the fluctuation frequency threshold, the stability of the fan speed is determined to be unqualified; when the wind speed fluctuation frequency is less than or equal to the fluctuation frequency threshold, the stability of the fan speed is determined to be qualified, and the coal quality sorting is completed at the current fan speed.

[0077] Specifically, the actual wind speed at the valve of the dry separator is acquired in real time to establish a wind speed curve. This curve is then compared with a preset wind speed range. The number of peak values ​​exceeding the preset range, compared with the corresponding time intervals, is determined as the wind speed fluctuation frequency. The preset wind speed range is the area between two straight lines parallel to the x-axis of the wind speed curve. The wind speeds corresponding to these two lines represent the maximum and minimum allowable fluctuations at the valve. For example, if the designed wind speed at the valve is 5 m / s, with an allowable fluctuation of ±1 m / s, then the wind speeds corresponding to the two lines are 6 m / s and 4 m / s, respectively. If the wind speed fluctuation frequency exceeds the fluctuation frequency threshold, it indicates that the wind speed exceeds the reasonable range too frequently during the separation period, and the airflow frequently deviates from the separation requirements. In this case, the stability of the fan speed is deemed unqualified. If the wind speed fluctuation frequency is less than or equal to the fluctuation frequency threshold, it indicates that the wind speed fluctuation is within a controllable range, and the airflow can stably support the coal particle stratification. In this case, the stability of the fan speed is deemed qualified, and no adjustment of the fan speed is necessary; separation can continue at the current fan speed. When the fan speed stability is within acceptable limits, unnecessary control interventions are avoided, ensuring the continuity and reliability of the sorting process. When the fan speed stability is unacceptable, subsequent control is triggered, preventing the formation of a "dead bed zone" and ensuring that the sorting process always relies on a stable airflow, further improving the reliability of dry sorting.

[0078] In some embodiments of this application, when the stability of the fan speed is determined to be unqualified, the valve opening of the dry separator is obtained, and the fan speed correlation result is determined based on the valve opening and the association rule algorithm. This includes: obtaining the valve opening of the dry separator and environmental correlation parameters and constructing an association dataset with the fan speed; preprocessing the association dataset, including data denoising and data normalization; generating several candidate itemsets from the preprocessed association dataset according to the Eclat algorithm; determining frequent itemsets based on the support of the candidate itemsets; selecting association rules based on the frequent itemsets; the association rules reflecting the correlation between environmental conditions, valve opening, and fan speed; and determining the fan speed correlation result based on the association rules.

[0079] Specifically, when the stability of the fan speed is determined to be unqualified, the direction of fan speed control can be accurately determined by exploring the potential correlation between valve opening, environmental conditions and fan speed. The valve opening of the dry separator reflects the actual flow state of the airflow. The environmental correlation parameters cover external and dynamic factors that affect the airflow characteristics and coal quality separation requirements, including data such as temperature and humidity of the separation environment (both of which will change the air density and indirectly affect the airflow). In addition, the fan speed under the corresponding conditions is recorded simultaneously. The three are integrated to form a correlation dataset, providing a complete data foundation for subsequent correlation analysis. The associated dataset is then preprocessed. Data denoising removes outliers, such as extreme temperature values ​​recorded by environmental sensors, to prevent them from interfering with the discovery of correlation patterns. Data normalization converts parameters of different magnitudes, such as valve opening in percentages, temperature in degrees Celsius, and rotational speed in RPM units, into a uniform numerical range, eliminating algorithm weight bias caused by differences in parameter magnitudes. After preprocessing, the Eclat algorithm is used to generate a candidate set of parameters, which is a set of parameters formed by any combination of environmental correlation parameters, valve opening, and fan speed. The table lists parameter combinations that may have correlations. Next, the support of each candidate itemset is calculated, which indicates the frequency of the parameter combination's occurrence in the correlation dataset, reflecting the prevalence of the combination. Candidate itemsets with support higher than 0.1 are typically identified as frequent itemsets. Frequent itemsets indicate that these parameter combinations frequently appear in actual sorting scenarios. Correlation rules are then selected from these frequent itemsets. These rules explicitly reflect the correspondence between environmental conditions, such as high temperature and low humidity, valve opening within a certain range, and fan speed adjustment to a certain value. Based on these correlation rules, the fan speed correlation result is determined. Traditional control only focuses on the fan speed itself, ignoring the implicit influence of valve opening (directly affecting airflow) and environmental correlation parameters (indirectly affecting airflow) on the stability of fan speed. By integrating multi-dimensional parameters through the correlation dataset, the key factors affecting fan speed can be comprehensively covered. Through data-driven mining, objective and accurate correlation rules are discovered, ensuring the reliability of the fan speed correlation results in relation to actual sorting scenarios. This further improves the feedback loop of control, thereby enhancing the stability and reliability of dry sorting.

[0080] In some embodiments of this application, when determining the fan speed adjustment coefficient based on the fan speed correlation results and adjusting the fan speed based on the fan speed adjustment coefficient, the process includes: counting the number of fan speed correlation results and recording it as the correlation number; determining the valve opening change value of the dry separator; determining the fan speed adjustment coefficient based on the valve opening change value and the correlation number; and the fan speed and the speed adjustment coefficient being directly proportional.

[0081] Specifically, the fan speed correlation results are based on correlation rules selected by the correlation rule algorithm. These rules cover the correspondence between environmental conditions, valve opening of the dry separator, and fan speed. The number of correlations, i.e., the total number of fan speed correlation results, reflects the richness of historical data and correlation rules supporting fan speed adjustment. The more correlations, the more comprehensive the basis for adjustment. The valve opening change value refers to the difference between the current actual valve opening of the dry separator and the valve opening when the fan speed stability was qualified. If the current fan speed stability is unqualified, the actual valve opening will not be equal to the valve opening when the stability was qualified. Therefore, if the valve opening change value is greater than 0, it indicates that the airflow exceeds the stable range, and the fan speed needs to be reduced to ensure the stability of the fan speed. Conversely, if the valve opening change value is less than 0, the fan speed needs to be increased. A first increasing speed adjustment coefficient, a second increasing speed adjustment coefficient, a first decreasing speed adjustment coefficient, and a second decreasing speed adjustment coefficient are set. The first increasing speed adjustment coefficient is preferably 1.8, the second increasing speed adjustment coefficient is preferably 1.6, the first decreasing speed adjustment coefficient is preferably 0.8, and the second decreasing speed adjustment coefficient is preferably 0.6. Taking a valve opening change value greater than 0 as an example, when the number of associated parameters is greater than the number of associated parameters threshold, the first increasing speed adjustment coefficient is used as the speed adjustment coefficient. When the number of associated parameters is less than or equal to the number of associated parameters threshold, the second increasing speed adjustment coefficient is used as the speed adjustment coefficient. The more associated parameters there are, the more comprehensive the basis for adjustment, and the more complex the impact on the fan speed. Therefore, a larger speed adjustment coefficient is needed to adjust the fan speed. Conversely, when the valve opening change value is less than 0, and when the number of associated parameters is greater than the number of associated parameters threshold, the second decreasing speed adjustment coefficient is used as the speed adjustment coefficient. When the number of associated parameters is less than or equal to the number of associated parameters threshold, the first decreasing speed adjustment coefficient is used as the speed adjustment coefficient.

[0082] Understandably, the fan speed is adjusted based on the speed adjustment coefficient. Assuming the fan speed is V and the speed adjustment coefficient is F, the adjusted fan speed is determined to be... When a higher or lower fan speed is required, the speed adjustment coefficient will increase or decrease accordingly based on the number of associated components. By establishing a direct proportional relationship between the fan speed and the speed adjustment coefficient, precise control of the fan speed during the sorting process is achieved, ensuring the reliability of the sorting process.

[0083] In summary, the beneficial effects of this invention are as follows: It accurately obtains the component parameters of each coal component through a component analysis model, and matches the data with a historical control database, avoiding decisions based on human experience and reducing interference from human factors. This provides a standardized and regulated control basis for the sorting of complex coals, improving the consistency of control. Given the fluctuating composition and particle size of complex coals, a dual strategy of invocation and compensation is employed. When component parameters are similar, historical fan speeds are directly matched; when differences are significant, a wind speed mapping model is constructed to determine the fan speed. This ensures the flexibility and reliability of fan speed changes with coal quality, avoids coal block accumulation due to coal quality fluctuations, and guarantees the continuity of sorting operations. Based on the frequency of wind speed fluctuations during the sorting period, the stability of the fan speed is judged in real time. If any defects are found, the speed correlation result is determined according to the association rule algorithm, thereby determining the speed adjustment coefficient. This forms a closed-loop control system of monitoring, judgment, and adjustment, ensuring that airflow buoyancy and resistance continuously adapt to the needs of particle stratification, thereby improving the clean coal recovery rate and gangue removal rate, and ensuring the reliability of sorting.

[0084] In another preferred embodiment based on the above embodiments, see [reference] Figure 2 As shown, this embodiment provides a precision sorting and control system suitable for complex coal qualities, used to apply the above-mentioned precision sorting and control method suitable for complex coal qualities, including:

[0085] The data acquisition and processing module is configured to acquire coal quality parameters and determine the composition parameters of each component based on the composition analysis model and the coal quality parameters.

[0086] The sorting and analysis module is configured to compare all component parameters in the historical control database and determine the calling strategy or compensation strategy based on the comparison results. When the calling strategy is determined, the blower speed is determined based on the similarity of component parameters. When the compensation strategy is determined, a wind speed mapping model is constructed based on the sample dataset and the historical control database, and the blower speed is determined based on the wind speed mapping model.

[0087] The sorting processing module is configured to obtain the wind speed fluctuation frequency during the sorting time period, determine whether the stability of the fan speed is qualified based on the wind speed fluctuation frequency, and when it is determined that the stability of the fan speed is not qualified, obtain the valve opening of the dry sorting machine, and determine the fan speed correlation result based on the valve opening and the association rule algorithm.

[0088] The control feedback module is configured to determine the fan speed adjustment coefficient based on the fan speed correlation result, adjust the fan speed according to the speed adjustment coefficient, and complete the coal quality sorting according to the adjusted fan speed.

[0089] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program goods. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program goods 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 application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program goods according to embodiments of this application. 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... 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] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for precise sorting and control of complex coal, characterized in that, include: Obtain coal quality parameters, and determine the composition parameters of each component based on the composition analysis model and coal quality parameters; Based on the comparison of all component parameters in the historical control database, the calling strategy or compensation strategy is determined according to the comparison results. When the calling strategy is determined, the blower speed is determined based on the similarity of component parameters. When the compensation strategy is determined, a wind speed mapping model is constructed based on the sample dataset and the historical control database, and the blower speed is determined according to the wind speed mapping model. The wind speed fluctuation frequency during the sorting period is obtained, and the stability of the fan speed is determined based on the wind speed fluctuation frequency. When the stability of the fan speed is determined to be unqualified, the valve opening of the dry separator is obtained, and the fan speed correlation result is determined based on the valve opening and the association rule algorithm. The fan speed adjustment coefficient is determined based on the fan speed correlation result, and the fan speed is adjusted according to the fan speed adjustment coefficient. The coal quality is then sorted according to the adjusted fan speed. When obtaining coal quality parameters, and determining the composition parameters of each component based on the compositional analysis model and coal quality parameters, the following steps are included: A laser particle size analyzer, an infrared ash analyzer, and a near-infrared spectral camera are installed at the feed end of the dry separator, and the coal quality parameters are obtained based on the laser particle size analyzer, the infrared ash analyzer, and the near-infrared spectral camera. Obtain a coal quality dataset and divide the coal quality dataset into a training set and a test set; The initial model is constructed by determining the parameters based on grid search, iteratively training the initial model according to the training set, and evaluating the error of the iteratively trained initial model according to the test set to determine the component analysis model. Substitute the coal quality parameters into the composition analysis model to determine the composition parameters of each component; The initial model is an RBF neural network model or a BP neural network model; When the calling strategy is determined, the blower speed is determined based on the similarity of component parameters, including: The historical component datasets with a component parameter similarity greater than 0.8 are sorted in descending order of component parameter similarity. When the maximum component parameter similarity is unique, the historical fan speed corresponding to the maximum component parameter similarity is determined as the fan speed of the blower. When the maximum component parameter similarity is not unique, the average of the historical fan speeds corresponding to each maximum component parameter similarity is determined as the fan speed of the blower. When the compensation strategy is determined, a wind speed mapping model is constructed based on the sample dataset and historical control database. Determining the blower speed according to the wind speed mapping model includes: The sample dataset includes elemental analysis parameters, physical property parameters, and coal and rock composition parameters; The sample dataset and the historical regulation database are divided into a second training set and a second test set. The wind speed mapping model is constructed based on the second training set and the second test set, and the component dataset is substituted into the wind speed mapping model to determine the blower speed. When it is determined that the stability of the fan speed is unqualified, the valve opening of the dry separator is obtained. When determining the fan speed correlation result based on the valve opening and the correlation rule algorithm, the following steps are taken: Obtain the valve opening degree and environmental parameters of the dry separator and construct a related dataset with the fan speed; The associated dataset is preprocessed, including data denoising and data normalization; The Eclat algorithm is used to generate several candidate itemsets from the preprocessed association dataset. Frequent itemsets are determined based on the support of the candidate itemsets. Association rules are then selected based on the frequent itemsets. The association rules reflect the relationship between environmental conditions, valve opening, and fan speed. The association result of the fan speed is determined based on the association rules.

2. The method for precise sorting and control of complex coal quality according to claim 1, characterized in that, When iteratively training the initial model based on the training set, and evaluating the error of the iteratively trained initial model based on the test set to determine the component analysis model, the process includes: If the mean squared error of the initial model after the current iteration is less than the mean squared error of the initial model after the previous iteration, and is less than the mean squared error threshold, then stop the iterative training; otherwise, adjust the learning rate of the initial model and continue the iterative training. The mean square error threshold is inversely proportional to the amount of data in the coal quality dataset.

3. The method for precise sorting and control of complex coal quality according to claim 2, characterized in that, When comparing all component parameters in a historical control database and determining the activation or compensation strategy based on the comparison results, the following steps are included: All component parameters are constructed into a component dataset. The historical control database includes several historical component datasets and several historical wind turbine speeds, and each historical component dataset corresponds to a historical wind turbine speed. The component dataset is compared with each historical component dataset to determine the similarity of component parameters; If the similarity of component parameters is greater than 0.8, then the calling strategy is determined. When there is no component parameter similarity greater than 0.8, the compensation strategy is determined.

4. The method for precise sorting and control of complex coal quality according to claim 3, characterized in that, When acquiring the wind speed fluctuation frequency during the sorting period and determining whether the stability of the fan speed is qualified based on the wind speed fluctuation frequency, the process includes: The wind speed fluctuation frequency is compared with the wind speed fluctuation frequency threshold. When the wind speed fluctuation frequency is greater than the fluctuation frequency threshold, the stability of the wind turbine speed is determined to be unqualified. When the wind speed fluctuation frequency is less than or equal to the fluctuation frequency threshold, the stability of the fan speed is determined to be qualified, and the coal quality is sorted at the current fan speed.

5. The method for precise sorting and control of complex coal quality according to claim 4, characterized in that, When determining the fan speed adjustment coefficient based on the fan speed correlation result, and adjusting the fan speed according to the fan speed adjustment coefficient, the process includes: The number of correlation results for the fan speed is counted and recorded as the correlation count; Determine the valve opening change value of the dry separator, and determine the speed adjustment coefficient of the fan based on the valve opening change value and the associated quantity; The fan speed is directly proportional to the speed adjustment coefficient.

6. A precision sorting and control system suitable for complex coal quality, used to apply the precision sorting and control method for complex coal quality as described in any one of claims 1-5, characterized in that, include: The data acquisition and processing module is configured to acquire coal quality parameters and determine the composition parameters of each component based on the composition analysis model and the coal quality parameters. The sorting and analysis module is configured to compare all component parameters in the historical control database, determine the calling strategy or compensation strategy based on the comparison results, and when the calling strategy is determined, determine the blower speed based on the component parameter similarity. When the compensation strategy is determined, construct a wind speed mapping model based on the sample dataset and the historical control database, and determine the blower speed based on the wind speed mapping model. The sorting processing module is configured to acquire the wind speed fluctuation frequency during the sorting time period, determine whether the stability of the fan speed is qualified based on the wind speed fluctuation frequency, and when it is determined that the stability of the fan speed is unqualified, acquire the valve opening of the dry sorting machine, and determine the fan speed correlation result based on the valve opening and the association rule algorithm. The control feedback module is configured to determine the speed adjustment coefficient of the fan speed based on the fan speed correlation result, adjust the fan speed according to the speed adjustment coefficient, and complete the coal quality sorting according to the adjusted fan speed.

Images