Real-time measurement method for particulate matter dust removal efficiency of wet dust collector

Abstract

The invention discloses a method for measuring the particulate matter dust removal efficiency of a wet dust collector in real time. The method comprises the following steps: selecting the liquid level height, the inlet airflow velocity, the inlet air pressure and the inlet dust concentration in the wet dust collector as input parameters; acquiring a data set in the dust removal process of particulate matters with different particle sizes; constructing an initial calculation model through a data driving method, and respectively constructing respective calculation models according to particles with different particle sizes; the particulate matters with different particle sizes respectively adopt respective data sets to train respective calculation models, and respective dust removal efficiency calculation formulas of the particulate matters with different particle sizes are obtained after training is completed; during real-time measurement, a corresponding dust removal efficiency calculation formula is selected according to the particle size of particles in the environment needing dust removal, corresponding input parameters are continuously obtained in the dust removal process and substituted into the formula, and finally a real-time dust removal efficiency value is obtained. According to the method, on the premise that the measurement real-time performance is guaranteed, the measurement precision and robustness can be guaranteed, and online continuous dust removal efficiency monitoring is achieved.

Description

Technical Field

[0001] This invention belongs to the field of dust collector performance monitoring technology, specifically a real-time measurement method for particulate matter dust removal efficiency of a wet dust collector. Background Technology

[0002] Wet scrubbers are widely used for controlling particulate pollution emissions from indoor industrial buildings and workshops. Real-time and accurate measurement of their dust removal efficiency is crucial for ensuring indoor air quality and the occupational health of workers. Currently, methods for measuring dust removal efficiency mainly include two categories: direct measurement and indirect measurement. Direct measurement methods include filter membrane weighing and online sensor monitoring. While filter membrane weighing is accurate, it requires offline sampling and drying, which is time-consuming and cannot meet real-time monitoring needs. Online sensor monitoring (such as optical sensors) is affected by the high humidity and droplet interference at the dust collector outlet, requiring frequent sensor calibration to ensure measurement accuracy.

[0003] Indirect measurement methods include those based on high-frequency pressure signal analysis (such as wavelet analysis and power spectral density estimation) or image processing techniques (such as high-speed dynamic imaging). These methods indirectly infer dust removal efficiency by identifying the gas-liquid two-phase flow pattern, but they suffer from problems such as multiple intermediate steps, limited data extraction, and large errors. Furthermore, methods based on optical principles are easily affected by ambient light. While feasible in laboratory environments, they face challenges in terms of economy, system complexity, and robustness in industrial applications. In addition, existing data-driven methods (such as traditional regression and neural networks) suffer from a "black box" problem; the models are uninterpretable, making them difficult for engineers to understand and trust, which is detrimental to on-site debugging and optimization. Moreover, the gas-liquid-dust three-phase coupling mechanism inside wet scrubbers is complex, with strong coupling and multicollinearity among input variables, making it difficult for existing models to establish accurate and robust input-output mapping relationships. At the same time, measurement results are affected by factors such as the complex internal gas-liquid two-phase flow and high moisture content at the outlet, leading to the scrubber often operating in an inefficient and uncontrollable state.

[0004] Therefore, the research direction of this invention is to provide a new real-time measurement method that can ensure measurement accuracy and robustness while guaranteeing real-time measurement, thereby achieving continuous online monitoring. Summary of the Invention

[0005] To address the problems existing in the prior art, this invention provides a real-time measurement method for particulate matter dust removal efficiency of a wet scrubber. By selecting specific input parameters and establishing calculation formulas for particles of different sizes, the method can ensure both real-time measurement accuracy and robustness, thereby achieving continuous online monitoring of dust removal efficiency.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is: a method for real-time measurement of particulate matter dust removal efficiency of a wet scrubber, comprising the following steps: Step 1: Determine the input parameters: Select the liquid level height, inlet airflow velocity, inlet air pressure, and inlet dust concentration inside the wet scrubber as the input parameters. This combination effectively avoids information redundancy between variables and comprehensively characterizes the core state of the gas-liquid-dust coupling process with the fewest sensing dimensions.

[0007] Step 2: Obtain training data: Test the dust removal efficiency of the wet scrubber in environments with particles of different sizes, and collect the input parameters determined in Step 1 in real time, thereby obtaining a dataset showing the correspondence between the real-time input parameters and the real-time dust removal efficiency during the dust removal process of particles of different sizes.

[0008] Step 3: Construct computational models for particles of different sizes: Construct an initial computational model using a data-driven approach, and set the model complexity according to the different particle sizes to construct computational models for particles of different sizes.

[0009] Step 4: Determine the dust removal efficiency calculation formula for different particle sizes: First, select a particle size range, obtain the corresponding calculation model through Step 3 and the corresponding dataset through Step 2, and divide the dataset into training and testing sets. Then, set the model training parameters and train the initial calculation model using the training set. After training, verify the model using the testing set. If the model meets the requirements, determine the final calculation model and obtain the dust removal efficiency calculation formula for that particle size composed of input parameters. If the model does not meet the requirements, adjust the model training parameters and model complexity and repeat this step until the verification meets the requirements. Repeat this step for all other particle sizes to obtain the dust removal efficiency calculation formula for each particle size. The calculation formulas reveal the complex relationship between input parameters and dust removal efficiency (i.e., the output predicted value).

[0010] Step 5: Real-time measurement of particulate matter dust removal efficiency: First, determine the particle size of the particulate matter in the required dust removal environment. Select the corresponding dust removal efficiency calculation formula from Step 4. Then, place the wet scrubber in the required dust removal environment for dust removal. During the dust removal process, continuously acquire the real-time input parameters determined in Step 1 and substitute them into the selected dust removal efficiency calculation formula to finally obtain the real-time dust removal efficiency prediction value.

[0011] Furthermore, in step two, the initial computational model is constructed using the multigene genetic programming (MGGP) method. Other data-driven methods can also be used, but MGGP is preferred because the computational formulas generated after training the model produced by it have the highest accuracy.

[0012] Furthermore, the different particle sizes in step two are: PM1 particle size particles, PM2.5 ... 2.5 Particulate matter, PM 10 The model considers both particulate matter and total suspended particulate matter; and the model complexity decreases as the particulate matter size increases. For particles with smaller sizes, the dust removal efficiency fluctuates more widely and is harder to predict, thus requiring a higher complexity setting. Conversely, for particles with larger sizes, the dust removal efficiency fluctuates less widely and is relatively easier to predict, requiring a lower complexity setting (because setting it too high can lead to overfitting and increased computation).

[0013] Furthermore, in step three, the dataset consists of data units from all acquisition times, and each data unit from the acquisition time consists of the real-time input parameters obtained at the current acquisition time and the dust removal efficiency corresponding to that acquisition time.

[0014] Furthermore, in step four, the dataset is divided into a training set and a test set. Specifically, the number of data units collected at each time point in the dataset is divided into a training set and a test set in a ratio of 8:2.

[0015] Furthermore, the calculation formulas for the dust removal efficiency of particles of different sizes in step four are as follows: (1) (2) (3) (4) In the formula, η PM1 The dust removal efficiency for PM1 particle size; η PM2.5 For PM 2.5 Dust removal efficiency for particles of a certain size; η PM10 For PM 10 Dust removal efficiency for particles of a certain size; η TSP denoted as the total suspended particulate matter dust removal efficiency; h as the liquid level height; v as the inlet airflow velocity; p as the inlet air pressure; and c as the inlet dust concentration.

[0016] Furthermore, the specific parameters for setting the model training in step four include: population size, number of generations, tournament size, elite ratio, termination value, crossover rate, mutation rate, direct replication rate, and function set.

[0017] Compared with the prior art, the present invention has the following advantages: 1. Unity of accuracy and interpretability: For the first time in the field of wet dust collector efficiency monitoring, this invention provides a solution that combines the high accuracy of a "black box model" with the high interpretability of a "white box model". The dust removal efficiency formulas for particles of different sizes obtained by the invention enable engineers to intuitively understand the influence between variables, which greatly improves the engineering credibility and practical value of the model.

[0018] 2. Precise dust removal efficiency calculation: This invention selects liquid level height h, inlet airflow velocity v, inlet air pressure p, and inlet dust concentration c as input parameters. This combination has been tested for collinearity and model performance verification, achieving the best estimation effect with the fewest variables. That is, the dust removal efficiency can be accurately determined with only the above input parameters.

[0019] 3. Excellent edge adaptability and real-time performance: This invention uses the multi-gene genetic programming (MGGP) method to construct an initial computational model. After training with selected input parameters to form a dataset and model structure constraints, a dust removal efficiency calculation formula for particles of different sizes is obtained. During subsequent monitoring, the dust removal efficiency calculation formula has a very small computational load, which is perfectly adapted to the resource-constrained industrial edge computing environment, realizing millisecond-level real-time efficiency estimation, laying the foundation for closed-loop control.

[0020] 4. Good robustness and low uncertainty: The method of this invention has undergone comprehensive sensitivity and uncertainty analysis using existing methods, which shows that the model is not sensitive to changes in key parameters, and the prediction coverage of the 90% confidence interval is close to the ideal value. The average prediction interval width is within the acceptable range for engineering, which proves its reliability and stability under complex working conditions in industrial sites. Attached Figure Description

[0021] Figure 1 This is the overall flowchart of the present invention. Detailed Implementation

[0022] The present invention will be further described below.

[0023] like Figure 1 As shown, this embodiment includes the following steps: Step 1: Determine the input parameters: Select the liquid level height h, inlet airflow velocity v, inlet air pressure p, and inlet dust concentration c inside the wet scrubber as the input parameters. This combination effectively avoids information redundancy between variables and comprehensively characterizes the core state of the gas-liquid-dust coupling process with the fewest sensing dimensions.

[0024] Step 2: Obtain training data: Test the dust removal efficiency of the wet scrubber in environments with particles of different sizes, and collect the input parameters determined in Step 1 in real time to obtain a dataset showing the correspondence between the real-time input parameters and the real-time dust removal efficiency during the dust removal process of particles of different sizes. The dataset consists of data units at all collection times, and each data unit at a collection time consists of the real-time input parameters obtained at the current collection time and the dust removal efficiency corresponding to that collection time.

[0025] Step 3: Constructing computational models for particulate matter of different sizes: An initial computational model is constructed using the polygenic genetic programming (MGGP) method. The different particulate matter sizes are: PM1 particles, PM... 2.5 Particulate matter, PM 10 The model considers both particle size and total suspended particulate matter; and sets the model complexity to decrease as particle size increases, thereby constructing computational models for particles of different sizes. In this embodiment, a maximum tree depth of 4 is set as a constraint to actively control model complexity and generate compact, explicit mathematical equations suitable for edge deployment.

[0026] Step 4: Determine the dust removal efficiency calculation formula for particles of different sizes: First, select a particle size range, obtain the corresponding calculation model through Step 2, and obtain the corresponding dataset through Step 3. Divide the number of data units at the collection time in the dataset into a training set and a test set according to an 8:2 ratio. Then, set the model training parameters, including: population size, generation, tournament size, elite ratio, termination value, crossover rate, mutation rate, direct replication rate, and function set. The setting values ​​of the training parameters in this embodiment are shown in Table 1. Table 1: The initial computational model is trained using a training set, and then validated using a test set. If the model meets the requirements, the final computational model is determined, and the dust removal efficiency calculation formula for that particle size, composed of input parameters, is obtained. If the model does not meet the requirements, the training parameters and model complexity are adjusted, and this step is repeated until validation is successful. This step is repeated for all other particle sizes to obtain the dust removal efficiency calculation formulas for different particle sizes, i.e., four explicit and analytical mathematical equations. The complex relationship between input parameters and dust removal efficiency (i.e., the output predicted value) is revealed through each calculation formula, specifically: (1) (2) (3) (4) In the formula, η PM1The dust removal efficiency for PM1 particle size; η PM2.5 For PM 2.5 Dust removal efficiency for particles of a certain size; η PM10 For PM 10 Dust removal efficiency for particles of a certain size; η TSP denoted as the total suspended particulate matter dust removal efficiency; h as the liquid level height; v as the inlet airflow velocity; p as the inlet air pressure; and c as the inlet dust concentration.

[0027] Step 5: Real-time measurement of particulate matter dust removal efficiency: First, determine the particle size of the particulate matter in the required dust removal environment. Select the corresponding dust removal efficiency calculation formula from Step 4 and store it in the edge computing device. Then, place the wet scrubber in the required dust removal environment for dust removal. During the dust removal process, continuously acquire the real-time input parameters determined in Step 1 and input them into the edge computing device. The device substitutes the input parameters into the selected dust removal efficiency calculation formula and finally calculates the real-time dust removal efficiency prediction value of different particle sizes under the current operating conditions within milliseconds.

[0028] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for real-time measurement of particulate matter removal efficiency in a wet scrubber, characterized in that, Includes the following steps: Step 1: Determine the input parameters: Select the liquid level height, inlet airflow velocity, inlet air pressure, and inlet dust concentration inside the wet scrubber as the input parameters; Step 2: Obtain training data: Test the dust removal efficiency of the wet scrubber in environments with particles of different sizes, and collect the input parameters determined in Step 1 in real time to obtain a dataset showing the correspondence between the real-time input parameters and the real-time dust removal efficiency during the dust removal process of particles of different sizes. Step 3: Construct computational models for particles of different sizes: Construct an initial computational model using a data-driven approach, and set the model complexity according to the different particle sizes to construct computational models for particles of different sizes. Step 4: Determine the dust removal efficiency calculation formula for particles of different sizes: First, select a particle size range. Obtain the corresponding calculation model through Step 3 and the corresponding dataset through Step 2. Divide the dataset into a training set and a test set. Then, set the model training parameters and train the initial calculation model using the training set. After training, verify the model using the test set. If the model meets the requirements, determine the final calculation model and obtain the dust removal efficiency calculation formula for that particle size composed of the input parameters. If the model does not meet the requirements, adjust the model training parameters and model complexity and repeat this step until the model meets the requirements. Repeat this step for all other particle sizes to obtain the dust removal efficiency calculation formula for each particle size. Step 5: Real-time measurement of particulate matter dust removal efficiency: First, determine the particle size of the particulate matter in the required dust removal environment. Select the corresponding dust removal efficiency calculation formula from Step 4. Then, place the wet scrubber in the required dust removal environment for dust removal. During the dust removal process, continuously acquire the real-time input parameters determined in Step 1 and substitute them into the selected dust removal efficiency calculation formula to finally obtain the real-time dust removal efficiency prediction value.

2. The real-time measurement method for particulate matter dust removal efficiency of a wet scrubber according to claim 1, characterized in that, In step three, a multi-gene genetic programming method is selected to construct an initial computational model.

3. The method for real-time measurement of particulate matter dust removal efficiency of a wet scrubber according to claim 2, characterized in that, The different particle sizes in step two are: PM1 particle size particles, PM... 2.5 Particulate matter, PM 10 The model complexity is set to decrease as the particle size increases.

4. The method for real-time measurement of particulate matter removal efficiency of a wet scrubber according to claim 3, characterized in that, In step three, the dataset consists of data units from all acquisition times. Each data unit from acquisition times consists of the real-time input parameters obtained at the current acquisition time and the dust removal efficiency corresponding to that acquisition time.

5. The method for real-time measurement of particulate matter removal efficiency of a wet scrubber according to claim 4, characterized in that, In step four, the dataset is divided into a training set and a test set. Specifically, the number of data units collected at each time point in the dataset is divided into a training set and a test set in a ratio of 8:

2.

6. The method for real-time measurement of particulate matter removal efficiency of a wet scrubber according to claim 5, characterized in that, The specific formulas for calculating the dust removal efficiency of particles of different sizes in step four are as follows: (1) (2) (3) (4) In the formula, η PM1 η is the dust removal efficiency for PM1 particle size particles. PM2.5 For PM 2.5 Dust removal efficiency for particles of a certain size; η PM10 For PM 10 Dust removal efficiency for particles of a certain size; η TSP denoted as the total suspended particulate matter dust removal efficiency; h as the liquid level height; v as the inlet airflow velocity; p as the inlet air pressure; and c as the inlet dust concentration.

7. The method for real-time measurement of particulate matter dust removal efficiency of a wet scrubber according to claim 5, characterized in that, The specific parameters for setting the model training in step four include: population size, number of generations, tournament size, elite ratio, termination value, crossover rate, mutation rate, direct replication rate, and function set.

Images