Method and system for controlling ash content of clean coal in dense medium separation process

Abstract

The invention provides a clean coal ash content control method and system in a dense medium separation process. The method comprises the following steps: collecting historical data in the dense medium separation process; preprocessing the data; constructing and training a state estimation model and a system dynamic model; and performing real-time control on the ash content of the clean coal based on the constructed state estimation model and the system dynamic model. According to the method, the modeling problem of the large time-lag characteristic in the dense medium separation process is effectively solved, the influence of the feeding pressure on the clean coal ash content is comprehensively considered, the model can more accurately learn the bottom layer physical characteristics in the dense medium separation process, proper control actions can be adopted in time to compensate the influence of external interference factors, the response speed is high, and the real-time performance is high. The method has strong anti-interference capability and is insensitive to noise of ash content measurement. When the scheme is applied to production, the quality of a coal product can be improved by accurately controlling the ash content of the clean coal; and automatic control can be fully achieved, the labor intensity of operators is relieved, and the production efficiency is improved.

Description

Technical Field

[0001] The invention belongs to the technical field of coal mine production, relates to computer technology, and specifically relates to a clean coal ash control method and system in a heavy medium separation process. Background Art

[0002] Dense medium separation refers to a coal separation method that separates coal particles and gangue in a dense medium suspension with a density greater than that of water. In order to ensure that the clean coal ash content meets the production target during the dense medium separation process, it is necessary to control the clean coal ash content by adjusting the density setting value of the dense medium suspension. At present, the density control of the dense medium suspension in the industry has basically achieved automation.

[0003] However, the response of clean coal ash content to the density adjustment of heavy medium suspension during heavy medium separation is very slow, and there is a large noise in the online measurement results of clean coal ash content. In addition, the clean coal ash content is also affected by the fluctuation of feed pressure. These factors make it difficult for traditional control methods to achieve precise ash content control. Therefore, manual control is still widely used. Summary of the invention

[0004] In order to solve the above problems, the present invention proposes a clean coal ash control method and system in a heavy medium separation process based on a constructed state estimation model and a system dynamic model.

[0005] To achieve the above object, the technical solution of the present invention is as follows:

[0006] A method for controlling ash content of clean coal in a heavy medium separation process comprises the following steps:

[0007] Step 1, collecting historical data of the heavy medium separation process, the historical data including the density of the heavy medium suspension, the feed pressure and the ash content of the clean coal;

[0008] Step 2, performing central average filtering on the clean coal ash data and performing data preprocessing;

[0009] Step 3, construct an Encoder-Decoder model, construct a training data set based on the historical data collected in step 1 and the clean coal ash data preprocessed in step 2, and train the Encoder-Decoder model on the training data set; each training sample in the training set contains: the input of the Encoder network, the input of the Decoder network and the output of the Decoder network; the input of the Encoder network includes: tT 1 Density of heavy medium suspension from time t to time t-1 T 1 Feed pressure from time t to time t-1 and the unfiltered tT 1 Clean coal ash data from time +1 to time t The output value of the Encoder network is input to the Decoder network. The input of the Decoder network also includes: time t to t+T 2 -1 moment density of heavy medium suspension and time t to t+T 2 -1 Feed pressure at time The output of the decoder network includes the center average filtered time t+1 to t+T 2 Clean coal ash content at the time T 1 and T 2 The first preset duration and the second preset duration are respectively;

[0010] Step 4, construct a historical production sequence, generate N alternative input sequences based on the current density of the heavy medium suspension and the feed pressure, and perform real-time control of the clean coal ash content based on the constructed state estimation model and system dynamic model.

[0011] Furthermore, each sample point in the historical data in step 1 is time series data.

[0012] Furthermore, in step 2, the center average filtering is implemented by the following formula:

[0013]

[0014] Among them, yk is the clean coal ash data at time k collected in step 1, is the clean coal ash data at time t after central filtering, 2K+1 is the filter width, and K is a preset value.

[0015] Furthermore, the Encoder model is a state estimation model, which predicts the current state of the system based on the system's recent input and output history sequences, in the following form:

[0016]

[0017] where x t is the estimated value of the current state, and tT 1 The density sequence of the heavy medium suspension and the feed pressure sequence from time t to time t-1, tT 1 The clean coal ash content sequence from time +1 to time t, T 1 is the first preset duration, θ is the model parameter;

[0018] The Decoder model includes a state transfer model and an output model.

[0019] The state transition model describes the state transition characteristics of the system in the following form:

[0020] x t+1 =h φ (x t ,u t ,d t )

[0021] where x t is the system state at time t, u t is the density of the heavy medium suspension at time t, d t is the feed pressure at time t, φ is the model parameter;

[0022] The output model maps the system state output by the state transition model to the controlled output value in the following form:

[0023]

[0024] where x t is the system state at time t, y t is the clean coal ash content output at time t, is the model parameter.

[0025] Furthermore, the step 4 includes the following sub-steps:

[0026] Step 4-1: Construct the historical production sequence. The historical production sequence includes 1 Density of heavy medium suspension from time t to time t-1 and feed pressure And from tT 1 The clean coal ash content sequence from time +1 to time t

[0027] Step 4-2: Based on the constructed historical production sequence, use Encoder to predict the current system state x t ;

[0028] Step 4-3, based on the current density u of the heavy medium suspension t and feed pressure d t Generate N candidate input sequences, where the candidate input sequences are assumed to be in the future T 2 System input within the time range, including from time t to t+T 2 -1 The suspended density of the heavy medium and the feed pressure at time; in each alternative input sequence, the feed pressure at any time is equal to the feed pressure at the current time d t , the density of the heavy medium suspension at any time is equal to a random value, which is from the density interval (u t -a,u t+a) is uniformly sampled, where a is a preset parameter;

[0029] Step 4-4, based on the N candidate input sequences, generate N predicted output sequences. For each candidate input sequence, based on the system state x of the current system obtained in step 4-2 t , using the Decoder model to iteratively obtain a predicted output sequence, indicating that if the future T 2 The predicted system output under the condition that the system input within the time range is the currently selected input sequence;

[0030] Step 4-5, based on the obtained N predicted output sequences, compare with the current ash content set value and calculate the absolute deviation, obtain N absolute deviation sequences, and obtain the optimal heavy medium suspension as the current control action based on the absolute deviation sequence.

[0031] Furthermore, in step 4-5, the step of obtaining the optimal heavy medium suspension based on the absolute deviation sequence as the current control action includes the following process: summing each absolute deviation sequence to obtain N absolute deviation sums; determining the alternative input sequence corresponding to the predicted output sequence with the smallest absolute deviation sum as the optimal input sequence; and using the heavy medium suspension density value corresponding to the optimal input sequence as the current control action.

[0032] The present invention also provides a clean coal ash content control system for a heavy medium separation process, comprising a historical data collection module, a data preprocessing module, a model building and training module, and an ash content control module. The historical data collection module is used to implement the function of step 1 of the clean coal ash content control method for a heavy medium separation process; the data preprocessing module is used to implement the function of step 2 of the clean coal ash content control method for a heavy medium separation process; the model building and training module is used to implement the function of step 3 of the clean coal ash content control method for a heavy medium separation process; and the ash content control module is used to implement the function of step 4 of the clean coal ash content control method for a heavy medium separation process.

[0033] The present invention also provides a computer device, comprising a memory, a processor and a computer program stored in the memory, characterized in that the processor executes the computer program to implement the steps of the method for controlling the ash content of clean coal in the heavy medium separation process provided by the present invention.

[0034] The present invention also provides a computer-readable storage medium having a computer program / instruction stored thereon, characterized in that when the computer program / instruction is executed by a processor, the steps of the method for controlling the ash content of clean coal in the heavy medium separation process provided by the present invention are implemented.

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

[0036] 1. The present invention adopts the Encoder-Decoder network structure to construct a state estimation model and a system dynamic model. The state estimation model estimates the unknown system state based on the short-term historical data of the heavy medium separation process as the initial state of the system dynamic model; the system dynamic model effectively solves the modeling problem of the large time lag characteristics of the heavy medium separation process by introducing state variables. In addition, when training the model, the filtered clean coal ash is used as the output of the system dynamic model, and the influence of feed pressure on clean coal ash is comprehensively considered during modeling, so that the model can more accurately learn the underlying physical characteristics of the heavy medium separation process.

[0037] 2. The present invention adopts model-based predictive control technology. Compared with manual control and traditional PID control and other control schemes, it can take appropriate control actions in time to compensate for the influence of external interference factors. It has a fast response speed, strong anti-interference ability, and is insensitive to the noise of ash measurement.

[0038] 3. Applying the solution of the present invention in production can not only improve the quality of coal products by accurately controlling the ash content of clean coal, but also fully realize automatic control, reduce the labor intensity of operators, and improve production efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 A schematic flow chart of a method for controlling clean coal ash content in a heavy medium separation process provided by the present invention;

[0040] Figure 2 Schematic diagram of the sub-steps of the ash content control step. DETAILED DESCRIPTION

[0041] The technical solution provided by the present invention will be described in detail below in conjunction with specific embodiments. It should be understood that the following specific implementation methods are only used to illustrate the present invention and are not used to limit the scope of the present invention.

[0042] The present invention proposes a method for controlling the ash content of clean coal in a heavy medium separation process, the process of which is as follows: Figure 1 As shown, the following steps are included:

[0043] Step 1: Collect historical data of the heavy medium separation process. Each sample point in the data set is a time series data, including the density of the heavy medium suspension u, the feed pressure d, and the clean coal ash content y.

[0044] Step 2: Data preprocessing: Perform center average filtering on the clean coal ash data to eliminate the measurement noise in the clean coal ash data. The center average filtering method is:

[0045]

[0046] Among them, yk is the clean coal ash data at time k collected in step 1, is the clean coal ash data at time t after central filtering, 2K+1 is the filter width, and K is a preset value.

[0047] Step 3: Model construction and training. The model adopts the Encoder-Decoder structure, where the Encoder model is the state estimation model and the Decoder model is the system dynamic model.

[0048] The state estimation model predicts the current state of the system based on the system's recent input and output history sequences. Its form is as follows:

[0049]

[0050] where x t is the estimated value of the current state, and tT 1 The density sequence of the heavy medium suspension and the feed pressure sequence from time t to time t-1, tT 1 The clean coal ash content sequence from time +1 to time t, T 1 is the first preset duration, and θ is the model parameter.

[0051] The system dynamics model includes two sub-models: state transfer model and output model.

[0052] (a) The state transition model describes the state transition characteristics of the system and is in the following form:

[0053] x t+1 =h φ (x t ,u t ,d t )

[0054] where x t is the system state at time t, u t is the density of the heavy medium suspension at time t, d t is the feed pressure at time t, and φ is the model parameter.

[0055] (b) The output model maps the system state output by the state transition model to the controlled output value, and its form is as follows:

[0056]

[0057] where x t is the system state at time t, y t is the system output at time t, i.e., the ash content of clean coal, is the model parameter.

[0058] The state estimation model is an encoder network that converts a sequence to a state vector, which is implemented by a network structure such as a transformer or LSTM. The system dynamic model is a decoder network that describes the dynamic characteristics of the system, which can be implemented using an LSTM network. When implementing, it is necessary to ensure that the dimensions of the state vectors in the encoder network and the decoder network are equal.

[0059] Model training. Based on the data collected in step 1 and the preprocessed data in step 2, a training data set is constructed. Each training sample in the training set contains three parts: the input of the encoder network, the input of the decoder network, and the output of the decoder network. The input of the encoder network includes and in is the unfiltered clean coal ash data, and the input of the Decoder network is and The output of the Decoder network is in is the clean coal ash content after center average filtering, where T 1 and T 2 The first preset duration and the second preset duration are respectively. The Encoder-Decoder model is trained on the training set.

[0060] Step 4, ash content control. Based on the constructed state estimation model and system dynamic model, the clean coal ash content is controlled in real time. At any control time t, the following steps are specifically included:

[0061] Step 4-1, construct the historical production sequence. The historical production sequence includes 1 Density of heavy medium suspension from time to current time t-1 and feed pressure And from tT 1 The clean coal ash content sequence from time +1 to time t Where T 1 This is the first preset duration.

[0062] Step 4-2: Based on the constructed historical production sequence, use the state estimation model to predict the system state x at the current moment t .

[0063] Step 4-3, based on the current density u of the heavy medium suspension t and feed pressure d t Generate N candidate input sequences. The candidate input sequences are assumed to be in the future T 2 System input within the time range, including from time t to t+T 2-1 time heavy medium suspension density and feed pressure, T 2 is the second preset duration. In each alternative input sequence, let the feed pressure at any time be equal to the feed pressure at the current time d t , let the density of the heavy medium suspension at any time be equal to a random value, which is selected from the density interval (u t -a,u t +a), where a is a preset parameter that determines the size of the density search space at the current control moment.

[0064] Step 4-4, based on the N candidate input sequences, generate N predicted output sequences. For each candidate input sequence, based on the system state x of the current system obtained in step 4-2 t , we can use the system dynamic model to iteratively obtain a prediction output sequence, which means that if the future T 2 The predicted system output within the time range under the condition that the system input is the currently selected input sequence.

[0065] Step 4-5, based on the obtained N predicted output sequences, compare with the current ash content setting value and calculate its absolute deviation to obtain N absolute deviation sequences. Further, sum each absolute deviation sequence to obtain N absolute deviation sums. Determine the candidate input sequence corresponding to the predicted output sequence with the smallest absolute deviation sum as the optimal input sequence.

[0066] After obtaining the optimal input sequence, the density value of the heavy medium suspension is used as the current control action.

[0067] A method for controlling the ash content of clean coal in a heavy medium separation process provided by the present invention should be implemented by computer software. Therefore, the present invention also provides a control system for the ash content of clean coal in a heavy medium separation process, including a historical data collection module, a data preprocessing module, a model construction and training module, and an ash control module, wherein the historical data collection module is used to collect historical data of the heavy medium separation process, specifically realizing the function of step 1 of the method for controlling the ash content of clean coal in the heavy medium separation process; the data preprocessing module is used to perform center average filtering on the clean coal ash data, specifically realizing the function of step 2 of the method for controlling the ash content of clean coal in the heavy medium separation process; the model construction and training module is used to construct an Encoder-Decoder model, and train the model based on the training data set, specifically realizing the function of step 3 of the method for controlling the ash content of clean coal in the heavy medium separation process; the ash control module is used to perform real-time control of the ash content of clean coal based on the constructed state estimation model and system dynamic model, specifically realizing the function of step 4 of the method for controlling the ash content of clean coal in the heavy medium separation process.

[0068] The present invention also provides a computer device, including a memory, a processor and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the method for controlling the ash content of clean coal in the heavy medium separation process provided by the present invention.

[0069] The present invention also provides a computer-readable storage medium having a computer program / instruction stored thereon, which, when executed by a processor, implements the steps of the method for controlling the ash content of clean coal in a heavy medium separation process provided by the present invention.

[0070] It should be noted that the above content only illustrates the technical idea of ​​the present invention and cannot be used to limit the protection scope of the present invention. For ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications all fall within the protection scope of the claims of the present invention.

Claims

1. A method for controlling clean coal ash content in a heavy medium separation process, characterized in that: The steps include: Step 1, collecting historical data of the heavy medium separation process, the historical data including the density of the heavy medium suspension, the feed pressure and the ash content of the clean coal; Step 2, performing central average filtering on the clean coal ash data and performing data preprocessing; Step 3, constructing an Encoder-Decoder model, constructing a training data set based on the historical data collected in step 1 and the clean coal ash data preprocessed in step 2, and training the Encoder-Decoder model on the training data set; each training sample in the training set includes: the input of the Encoder network, the input of the Decoder network, and the output of the Decoder network; The input of the Encoder network includes: the density of the heavy medium suspension from time t-T1 to time t-1 Feed pressure from time t-T1 to time t-1 and the unfiltered clean coal ash data from time t-T1+1 to time t The output value of the Encoder network is input into the Decoder network. The input of the Decoder network includes: the density of the heavy medium suspension from time t to time t+T2-1 and the feed pressure from time t to time t+T2-1 The output of the decoder network includes the clean coal ash content from time t+1 to time t+T2 after center average filtering. T1 and T2 are respectively the first preset duration and the second preset duration; Step 4, construct a historical production sequence, generate N alternative input sequences based on the current density of the heavy medium suspension and the feed pressure, and perform real-time control of the clean coal ash content based on the constructed state estimation model and system dynamic model.

2. The method for controlling clean coal ash content in a heavy medium separation process according to claim 1, characterized in that: Each sample point in the historical data in step 1 is time series data.

3. The method for controlling clean coal ash content in a heavy medium separation process according to claim 1, characterized in that: In step 2, the center average filtering is implemented by the following formula: Among them, y k is the clean coal ash data at time k collected in step 1, is the clean coal ash data at time t after central filtering, 2K+1 is the filter width, and K is a preset value.

4. The method for controlling clean coal ash content in a heavy medium separation process according to claim 1, characterized in that: The Encoder model is a state estimation model. The state estimation model predicts the current state of the system based on the system's recent input and output history sequences. The form is as follows: where x t is the estimated value of the current state, and are the density sequence of the heavy medium suspension and the feed pressure sequence from time t-T1 to time t-1, is the clean coal ash content sequence from time t-T1+1 to time t, T1 is the first preset time length, and θ is the model parameter; The Decoder model includes a state transfer model and an output model. The state transition model describes the state transition characteristics of the system in the following form: x t+1 =h φ (x t ,u t ,d t ) where x t is the system state at time t, u t is the density of the heavy medium suspension at time t, d t is the feed pressure at time t, φ is the model parameter; The output model maps the system state output by the state transition model to the controlled output value in the following form: where x t is the system state at time t, y t is the clean coal ash content output at time t, is the model parameter.

5. The method for controlling clean coal ash content in a heavy medium separation process according to claim 1, characterized in that: The step 4 includes the following sub-steps: Step 4-1, construct a historical production sequence, which includes the density of the heavy medium suspension from time t-T1 to time t-1 and feed pressure And the clean coal ash content sequence from time t-T1+1 to time t Step 4-2: Based on the constructed historical production sequence, use Encoder to predict the current system state x t ; Step 4-3, based on the current density u of the heavy medium suspension t and feed pressure d t Generate N candidate input sequences, which are assumed to be system inputs within the future time range of T2, including the heavy medium suspension density and feed pressure from time t to time t+T2-1; in each candidate input sequence, the feed pressure at any time is equal to the feed pressure dt at the current time, and the heavy medium suspension density at any time is equal to a random value, which is selected from the density interval (u t -a,u t +a) is uniformly sampled, where a is a preset parameter; Step 4-4, based on the N candidate input sequences, generate N predicted output sequences. For each candidate input sequence, based on the system state x of the current system obtained in step 4-2 t , using the Decoder model to iteratively obtain a predicted output sequence, which represents the predicted system output if the system input in the future T2 time range is the currently selected input sequence; Step 4-5, based on the obtained N predicted output sequences, compare with the current ash content set value and calculate the absolute deviation, obtain N absolute deviation sequences, and obtain the optimal heavy medium suspension as the current control action based on the absolute deviation sequence.

6. The method for controlling clean coal ash content in a heavy medium separation process according to claim 5, characterized in that: In step 4-5, the step of obtaining the optimal heavy medium suspension as the current control action based on the absolute deviation sequence includes the following process: summing each absolute deviation sequence to obtain N absolute deviation sums; determining the alternative input sequence corresponding to the predicted output sequence with the smallest absolute deviation sum as the optimal input sequence; and using the heavy medium suspension density value corresponding to the optimal input sequence as the current control action.

7. A clean coal ash control system in a heavy medium separation process, characterized in that: It includes a historical data collection module, a data preprocessing module, a model building and training module, and an ash control module. The historical data collection module is used to implement the function of step 1 of the method for controlling the ash content of clean coal in a heavy medium separation process as described in any one of claims 1-6; the data preprocessing module is used to implement the function of step 2 of the method for controlling the ash content of clean coal in a heavy medium separation process as described in any one of claims 1-6; the model building and training module is used to implement the function of step 3 of the method for controlling the ash content of clean coal in a heavy medium separation process as described in any one of claims 1-6; the ash control module is used to implement the function of step 4 of the method for controlling the ash content of clean coal in a heavy medium separation process as described in any one of claims 1-6.

8. A computer device comprising a memory, a processor and a computer program stored in the memory, characterized in that: The processor executes the computer program to implement the steps of the method for controlling the ash content of clean coal in a heavy medium separation process as described in any one of claims 1-6.

9. A computer-readable storage medium having a computer program / instruction stored thereon, characterized in that: When the computer program / instructions are executed by a processor, the steps of the method for controlling the ash content of clean coal in a heavy medium separation process described in any one of claims 1 to 6 are implemented.

Images