WO2023221473A1

WO2023221473A1 - Online soft measurement method for clean coal ash content during lump coal shallow trough sorting process

Info

Publication number: WO2023221473A1
Application number: PCT/CN2022/138267
Authority: WO
Inventors: 桂夏辉; 邢耀文; 王兰豪; 代世琦; 曹亦俊; 刘炯天
Original assignee: 中国矿业大学
Priority date: 2022-05-18
Filing date: 2022-12-12
Publication date: 2023-11-23
Also published as: CN114758196A

Abstract

An online soft measurement method for clean coal ash content during a lump coal shallow trough sorting process, relating to the technical field of coal sorting and processing, and solving the problem that an existing clean coal ash content prediction method during a lump coal shallow trough sorting process is complex and has poor adaptability and high cost. The method comprises: collecting data in real time, the data comprising clean coal production, tail coal production, qualified medium density and a clean coal dual-energy X-ray image; inputting an R value of each pixel point in the clean coal dual-energy X-ray image collected in real time into a trained clean coal ash content prediction main model to be processed, so as to obtain a clean coal ash content prediction value; processing the clean coal production and the tail coal production which are collected in real time to obtain clean coal yield; inputting the clean coal yield and the normalized qualified medium density into a trained clean coal ash content compensation model to be processed, so as to obtain a clean coal ash content error prediction value; and compensating the clean coal ash content prediction value with the clean coal ash content error prediction value, so as to obtain a final clean coal ash content prediction result.

Description

An online soft measurement method of clean coal ash content in the shallow slot separation process of lump coal

Technical field

The invention relates to the technical field of coal sorting and processing, and in particular to an online soft measurement method of clean coal ash content in the shallow slot sorting process of lump coal.

Background technique

With the continuous advancement of the "dual carbon" strategy, low-carbon and clean utilization of coal resources has become a top priority in the energy economy. However, the traditional lump coal shallow slot sorting and detection technology has serious lag and cannot guide the adjustment of production indicators in a timely manner, resulting in a decline in product quantity and a large loss of coal resources.

Shallow trough sorting of lump coal is often used for pre-discharge and grading selection. The existing sorting process is: raw coal enters the shallow trough sorter through the raw coal belt, and the two separated products are dehydrated and deinterposed respectively, and then enter the clean coal belt and The tailing coal belt becomes clean coal product and tailing coal product. The existing method for testing the ash content of clean coal is sample collection and preparation class testing. This method has serious lag. The operating status of production equipment, production process parameters and product quality cannot be perceived in real time. Quality is often ensured by losing output, resulting in a large amount of clean coal. The loss is in the tailing coal, resulting in low ash content, high calorific value, and waste of coal resources.

Therefore, there is an urgent need to develop an online prediction method for clean coal ash content that is simple, widely adaptable, low-cost, and suitable for the shallow slot separation process of lump coal.

Contents of the invention

In view of the above analysis, embodiments of the present invention aim to provide an online soft measurement method for the ash content of clean coal in the shallow slot separation process of lump coal, so as to solve the problem of the complex and complicated ash content prediction method of clean coal in the existing shallow slot separation process of lump coal. Problems of poor adaptability and high cost.

The invention discloses an online soft measurement method of clean coal ash content in the shallow slot separation process of lump coal, which includes:

Collect data in real time, including clean coal production, tailing coal production, qualified medium density and dual-energy X-ray images of clean coal;

Input the R value of each pixel in the dual-energy X-ray image of clean coal collected in real time into the trained clean coal ash content prediction main model, and obtain the clean coal ash content prediction value after processing;

Process the clean coal yield and tailing coal yield collected in real time to obtain the clean coal yield; input the clean coal yield and the normalized qualified medium density into the trained clean coal ash compensation model, and obtain after processing Clean coal ash error prediction value;

The clean coal ash content prediction value is compensated with the clean coal ash content error prediction value to obtain the final clean coal ash content prediction result.

On the basis of the above solution, the present invention also makes the following improvements:

Further, the clean coal ash content prediction main model is trained in the following way:

Obtaining a first sample set, each set of first sample data in the first sample set includes: a dual-energy X-ray image of clean coal, and an actual measured value of ash content of clean coal;

Using the R value of each pixel in the dual-energy X-ray image of the clean coal in each set of first sample data as input and the measured value of the ash content of the clean coal as the label, the main model for predicting the ash content of the clean coal is trained to determine The structure and parameters of the clean coal ash content prediction master model are used to obtain a trained clean coal ash content prediction master model.

Further, the clean coal ash compensation model is trained in the following way:

Obtaining a second sample set, each set of second sample data in the second sample set includes: clean coal production, tail coal production, qualified medium density, dual-energy X-ray image of clean coal, and actual measured value of clean coal ash;

Process the clean coal production and tailing coal production in each set of second sample data to obtain the clean coal production rate; calculate the R value of each pixel in the dual-energy X-ray image of the clean coal in each set of second sample data. Input it into the trained clean coal ash content prediction main model to obtain the clean coal ash content prediction value;

The clean coal yield and normalized qualified medium density corresponding to each set of second sample data are used as input, and the difference between the corresponding measured value of clean coal ash and the predicted value of clean coal ash is used as a label. The clean coal ash compensation model is trained to obtain the trained clean coal ash compensation model.

Furthermore, each set of second sample data satisfies: the deviation between the theoretical value of clean coal ash obtained by matching the clean coal production, tailing coal production and raw coal selectivity curve, and the actual measured value of clean coal ash of the current second sample data does not exceed the set deviation threshold.

Further, the method also includes:

During the process of real-time data collection, actual measured values of ash content of clean coal are also collected regularly;

If the error between the actual measured value of clean coal ash collected regularly and the final predicted result of clean coal ash based on the actual collected data at that time is lower than the set error, then the actual collected data at that time and the actual measured value of clean coal ash will be value as a set of correction data, multiple sets of correction data form a correction data set, and the clean coal ash compensation model is corrected online based on the correction data set.

Further, the R value of each pixel in the dual-energy X-ray image of the clean coal:

Among them, μl represents the attenuation coefficient of the pixels in the dual-energy X-ray image under low-energy X-rays, and μh represents the attenuation coefficient of the pixels in the dual-energy X-ray image under high-energy X-rays.

Further, the clean coal production and tailing coal production collected in real time are processed to obtain the clean coal production rate, including:

Among them, r1 represents the clean coal yield, Q1 represents the clean coal production, and Q2 represents the tailing coal production.

Furthermore, the main model for predicting the ash content of clean coal is implemented based on a convolutional neural network.

Furthermore, the clean coal ash compensation model is implemented based on least squares support vector regression.

Further, the clean coal output is measured by a clean coal belt scale;

The tailing coal output is measured by a tailing coal belt scale;

The density of the qualified medium is measured by a density meter in the qualified medium barrel;

The dual-energy X-ray image of the clean coal is collected by an industrial X-ray machine installed above the clean coal belt.

Compared with the prior art, the present invention can achieve at least one of the following beneficial effects:

The invention discloses an online soft measurement method of clean coal ash content in the shallow slot separation process of lump coal, which has the following advantages:

First, the predicted value of clean coal ash is determined based on the R value of each pixel in the dual-energy X-ray image of clean coal collected in real time, and the ash content error of clean coal is determined based on the real-time collected clean coal production, tailing coal production and qualified medium density. The prediction value is then used to compensate the clean coal ash prediction value with the clean coal ash error prediction value to obtain the final clean coal ash prediction result; this method can achieve continuous measurement and can realize the lump coal shallow slot sorting process. Effective, accurate and rapid detection of ash content in clean coal, in order to guide the setting of feed volume, circulating suspension volume and suspension density based on the prediction results, reduce system fluctuations and increase clean coal production.

Second, it can realize real-time online continuous detection of ash content of sorted clean coal, and has certain adaptive capabilities, and can correct the predicted value according to production conditions. According to the predicted value, the sorting working conditions can be adjusted in time to meet the new working conditions.

Description of the drawings

The drawings are for the purpose of illustrating specific embodiments only and are not to be construed as limitations of the invention. Throughout the drawings, the same reference characters represent the same components.

Figure 1 is a schematic flowchart of the online soft measurement process of clean coal ash in the lump coal shallow slot separation process disclosed in the embodiment of the present invention;

Figure 2 is a schematic diagram of the main model training process of clean coal ash content prediction in the sorting process provided by the embodiment of the present invention;

Figure 3 is a schematic diagram of the training process of the clean coal ash content prediction and compensation model in the sorting process provided by the embodiment of the present invention.

Detailed ways

The preferred embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

The embodiment of the present invention discloses an online soft measurement method of clean coal ash content in the shallow slot separation process of lump coal. The schematic flow chart is shown in Figure 1 , including:

Step S1: Collect data in real time, including clean coal production, tail coal production, qualified medium density and dual-energy X-ray images of clean coal;

Step S2: Input the R value of each pixel in the dual-energy X-ray image of clean coal collected in real time into the trained clean coal ash content prediction main model, and obtain the clean coal ash content prediction value after processing;

Step S3: Process the clean coal yield and tailing coal yield collected in real time to obtain the clean coal yield; input the clean coal yield and the normalized qualified medium density into the trained clean coal ash compensation model, After processing, the error prediction value of clean coal ash content is obtained;

Step S4: Use the clean coal ash error prediction value to compensate the clean coal ash prediction value to obtain the final clean coal ash prediction result.

Specifically, the data in this embodiment are obtained in the following ways: the clean coal output is measured by a clean coal belt scale; the tail coal output is measured by a tail coal belt scale; the qualified medium density is measured by a density meter in the qualified medium barrel ; The dual-energy X-ray image of clean coal is collected by an industrial X-ray machine installed above the clean coal belt.

There is a direct relationship between the R value of dual-energy X-ray images of clean coal and the prediction of clean coal ash content. Specifically, the R value of material properties is related to the effective atomic number of the material. The R value of each pixel in the dual-energy X-ray image of clean coal is calculated according to formula (1):

Among them, μl represents the attenuation coefficient of the pixels in the dual-energy X-ray image under low-energy X-rays, and μh represents the attenuation coefficient of the pixels in the dual-energy X-ray image under high-energy X-rays;

The Z value of material properties satisfies:

Z＝-6.596×105×e-9.815R+4.685×e0.6783R(2)

And the relationship between Z value and substance classification is shown in Table 1:

Table 1 Relationship between Z value and substance classification

In the sorted clean coal, the higher the ash content of the clean coal, the more inorganic substances (such as gangue, etc.) it contains; conversely, the lower the ash content of the clean coal, the more organic substances it contains; therefore, the dual properties of clean coal are There is a direct relationship between the R value of each pixel in the dual-energy X-ray image and the ash content of clean coal. The ash content of clean coal can be predicted based on the R value of each pixel in the dual-energy X-ray image of clean coal.

In addition, the clean coal yield and qualified medium density also have varying degrees of impact on the prediction of clean coal ash content, and the prediction results of clean coal ash content can be compensated based on the clean coal yield and qualified medium density.

Before implementing the above solution, it is necessary to complete the training of the clean coal ash content prediction main model and the clean coal ash content prediction compensation model. The specific implementation method is introduced as follows:

(1) Train the clean coal ash content prediction main model in the following way. The training process is shown in Figure 2 :

Step A1: Obtain a first sample set. Each set of first sample data in the first sample set includes: a dual-energy X-ray image of clean coal, and an actual measured value of ash content of clean coal;

Step A2: Use the R value of each pixel in the dual-energy X-ray image of the clean coal in each set of first sample data as input and the actual measured value of the ash content of the clean coal as a label to perform the main model for predicting the ash content of the clean coal. Training is performed to determine the structure and parameters of the clean coal ash content prediction master model, and obtain a trained clean coal ash content prediction master model.

In the process of training the main model for prediction of clean coal ash content, a mapping relationship between the R value of each pixel in the dual-energy X-ray image of clean coal and the measured value of clean coal ash content can be established. This mapping relationship is achieved through clean coal The structure and parameters of the main gray prediction model are reflected. Therefore, during real-time processing, the R value of each pixel in the dual-energy X-ray image of clean coal collected in real time can be input into the trained clean coal ash content prediction master model, and the clean coal ash content prediction value can be obtained after processing.

Preferably, the main model for predicting the ash content of clean coal is a main model for predicting the ash content of clean coal based on a convolutional neural network. During the training process of the main model for prediction of clean coal ash content, the convolutional neural network backpropagates the error obtained through gradient descent, updates the parameters of each layer of the convolutional neural network layer by layer, and finally determines the clean coal ash content after multiple rounds of iterative training. Predict the structure and parameters of the main model.

The main model for prediction of clean coal ash content based on convolutional neural network can make a preliminary prediction of the ash content of sorted clean coal. However, due to factors such as complex sorting conditions, errors in the dual-energy X-ray image acquisition process, and limited information reflected in the image, high accuracy cannot be obtained. Accurate prediction results. At the same time, other influencing factors of the ash content of sorted clean coal have also been analyzed previously. Therefore, based on these influencing factors of the ash content of sorted clean coal, the clean coal ash content prediction compensation model is trained to predict the clean coal ash content of the clean coal ash content prediction main model. deviation in predicted values.

(2) Train the clean coal ash content prediction and compensation model in the following way. The training process is shown in Figure 3 :

Step B1: Obtain a second sample set. Each set of second sample data in the second sample set includes: clean coal output, tail coal output, qualified medium density, dual-energy X-ray image of clean coal, and actual measurement of clean coal ash content. value;

Step B2: Process the clean coal yield and tailing coal yield in each set of second sample data to obtain the clean coal yield; specifically,

Step B3: Input the R value of each pixel in the dual-energy X-ray image of the clean coal in each set of second sample data into the trained clean coal ash content prediction main model to obtain the clean coal ash content prediction value;

Step B4: Use the clean coal yield and the normalized qualified medium density corresponding to each set of second sample data as input, and use the difference between the corresponding measured value of clean coal ash content and the predicted value of clean coal ash content as a label. The clean coal ash compensation model is trained to obtain a trained clean coal ash compensation model. Preferably, the maximum and minimum normalization method is used to normalize the qualified medium density. After normalization, all data are converted into the [0,1] interval, so that each indicator belongs to the same magnitude.

Exemplarily, the clean coal ash compensation model is implemented based on least squares support vector regression.

It should be noted that, in order to ensure the accuracy of the training model data, each set of second sample data meets: the theoretical value of clean coal ash obtained by matching the clean coal output, tail coal output and raw coal selectivity curve, and the current The deviation between the actual measured values of the clean coal ash content of the second sample data does not exceed the set deviation threshold.

Preferably, the method in this embodiment also includes:

If the error between the actual measured value of clean coal ash collected regularly and the final predicted result of clean coal ash based on the actual collected data at that time is lower than the set error, then the actual collected data at that time and the actual measured value of clean coal ash will be value as a set of correction data, multiple sets of correction data form a correction data set, and the clean coal ash compensation model is corrected online based on the correction data set to achieve self-correction of model parameters and enhance the performance of the model under various working conditions. adaptive ability.

In summary, compared with the existing technology, the online soft measurement method of clean coal ash in the lump coal shallow slot separation process provided by this embodiment has the following advantages: First, dual-energy X-ray measurement of clean coal based on real-time collection The R value of each pixel in the radiographic image determines the predicted value of the ash content of the clean coal, and the predicted value of the error of the ash content of the clean coal is determined based on the real-time collected production of clean coal, the yield of tailing coal and the density of the qualified medium, and then the predicted value of the error of the ash content of the clean coal is used The predicted value of the clean coal ash content is compensated to obtain the final clean coal ash content prediction result; this method can realize continuous measurement and can achieve effective, accurate and rapid detection of the clean coal ash content in the shallow slot separation process of lump coal, so as to Based on the prediction results, the setting of feed volume, circulating suspension volume and suspension density is guided to reduce system fluctuations and increase clean coal production. Second, it can realize real-time online continuous detection of ash content of sorted clean coal, and has certain adaptive capabilities, and can correct the predicted value according to production conditions. According to the predicted value, the sorting working conditions can be adjusted in time to meet the new working conditions.

Those skilled in the art can understand that all or part of the process of implementing the method of the above embodiments can be completed by instructing relevant hardware through a computer program, and the program can be stored in a computer-readable storage medium. Wherein, the computer-readable storage medium is a magnetic disk, an optical disk, a read-only memory or a random access memory, etc.

Claims

An online soft measurement method of clean coal ash content in the shallow slot separation process of lump coal, which is characterized by including:

Collect data in real time, including clean coal production, tailing coal production, qualified medium density and dual-energy X-ray images of clean coal;

Input the R value of each pixel in the dual-energy X-ray image of clean coal collected in real time into the trained clean coal ash content prediction main model, and obtain the clean coal ash content prediction value after processing;

Process the clean coal yield and tailing coal yield collected in real time to obtain the clean coal yield; input the clean coal yield and the normalized qualified medium density into the trained clean coal ash compensation model, and obtain after processing Clean coal ash error prediction value;

The clean coal ash content prediction value is compensated with the clean coal ash content error prediction value to obtain the final clean coal ash content prediction result.
The online soft measurement method of clean coal ash content in the lump coal shallow slot separation process according to claim 1, characterized in that the clean coal ash content prediction main model is trained in the following manner:

Obtaining a first sample set, each set of first sample data in the first sample set includes: a dual-energy X-ray image of clean coal, and an actual measured value of ash content of clean coal;

Using the R value of each pixel in the dual-energy X-ray image of the clean coal in each set of first sample data as input and the measured value of the ash content of the clean coal as the label, the main model for predicting the ash content of the clean coal is trained to determine The structure and parameters of the clean coal ash content prediction master model are used to obtain a trained clean coal ash content prediction master model.
The online soft measurement method of clean coal ash content in the shallow slot separation process of lump coal according to claim 1, characterized in that the clean coal ash content compensation model is trained in the following manner:

Obtaining a second sample set, each set of second sample data in the second sample set includes: clean coal production, tail coal production, qualified medium density, dual-energy X-ray image of clean coal, and actual measured value of clean coal ash;

Process the clean coal production and tailing coal production in each set of second sample data to obtain the clean coal production rate;

Input the R value of each pixel in the dual-energy X-ray image of the clean coal in each set of second sample data into the trained clean coal ash content prediction main model to obtain the clean coal ash content prediction value;

The clean coal yield and normalized qualified medium density corresponding to each set of second sample data are used as input, and the difference between the corresponding measured value of clean coal ash and the predicted value of clean coal ash is used as a label. The clean coal ash compensation model is trained to obtain the trained clean coal ash compensation model.
The online soft measurement method of clean coal ash content in the shallow slot separation process of lump coal according to claim 3, characterized in that each set of second sample data satisfies: clean coal output, tail coal output and raw coal selectivity curve The deviation between the theoretical value of clean coal ash obtained by matching and the actual measured value of clean coal ash of the current second sample data does not exceed the set deviation threshold.
The online soft measurement method of clean coal ash content in the shallow slot separation process of lump coal according to claim 3, characterized in that the method further includes:

During the process of real-time data collection, actual measured values of ash content of clean coal are also collected regularly;

If the error between the actual measured value of clean coal ash collected regularly and the final predicted result of clean coal ash based on the actual collected data at that time is lower than the set error, then the actual collected data at that time and the actual measured value of clean coal ash will be value as a set of correction data, multiple sets of correction data form a correction data set, and the clean coal ash compensation model is corrected online based on the correction data set.
The online soft measurement method of clean coal ash content in the shallow slot separation process of lump coal according to any one of claims 1 to 5, characterized in that the R of each pixel in the dual-energy X-ray image of the clean coal value:

Among them, μl represents the attenuation coefficient of the pixels in the dual-energy X-ray image under low-energy X-rays, and μh represents the attenuation coefficient of the pixels in the dual-energy X-ray image under high-energy X-rays.
The online soft measurement method of clean coal ash content in the lump coal shallow slot separation process according to any one of claims 1 to 5, characterized in that the clean coal output and tailing coal output collected in real time are processed to obtain clean coal. Yields, including:

Among them, r1 represents the clean coal yield, Q1 represents the clean coal production, and Q2 represents the tailing coal production.
The online soft measurement method of clean coal ash content in the lump coal shallow slot sorting process according to any one of claims 1 to 5, characterized in that the clean coal ash content prediction main model is implemented based on a convolutional neural network.
The online soft measurement method of clean coal ash content in the lump coal shallow slot separation process according to any one of claims 1 to 5, characterized in that the clean coal ash content compensation model is implemented based on least squares support vector regression.
The online soft measurement method of clean coal ash content in the shallow slot separation process of lump coal according to any one of claims 1 to 5, characterized by:

The clean coal output is measured by a clean coal belt scale;

The tailing coal output is measured by a tailing coal belt scale;

The density of the qualified medium is measured by a density meter in the qualified medium barrel;

The dual-energy X-ray image of the clean coal is collected by an industrial X-ray machine installed above the clean coal belt.