Residual life estimation method of filter bag for bag type dust collector

Abstract

The invention discloses a residual life estimation method of a filter bag for a bag type dust collector. The method comprises the following steps: acquiring sample data including environmental conditions and flue gas parameters of actual application working conditions, basic properties and mechanical properties of the filter bag before and after use, and wear conditions and use time of the filter bag; a BP neural network model is established, the sample data obtained in the first step are input into the first layer of the BP neural network model, and the output result of the last layer of the BP neural network model is the estimated filter bag remaining life; and performing iterative calculation and correction on the BP neural network model through a gradient descent algorithm until an absolute error is within a preset range, and then estimating an optimal value of the residual life of the filter bag through the BP neural network model after iteration. Based on the BP neural network technology and the gradient descent algorithm, a high-efficiency and high-accuracy estimation method is provided for estimating the residual life of the filter bag for the bag type dust collector, and the problem of estimating the residual life of the filter bag is solved.

Description

Technical Field

[0001] The invention relates to the technical field of bag dust collectors, and in particular to a method for estimating the remaining life of a filter bag for a bag dust collector. Background Art

[0002] With the continuous development of the bag filter industry, the use environment of bag filter bags has become increasingly harsh, and their service life has become the focus of attention in the industry. In existing technologies, the remaining life of filter bags is mainly calculated and estimated using calculation models in industry standards. However, the data calculated in this way is relatively subjective and cannot accurately estimate the remaining life of filter bags. In addition, since the research and development of filter bags is still ongoing, the remaining life estimation method also needs to be improved and optimized.

[0003] In the field of data analysis, the application of artificial neural networks has been widely developed. In practical applications, artificial neural networks transfer the weights of each layer backward, continuously adjust the weights of each layer, and finally obtain the optimal model parameters and complete the algorithm solution. Therefore, it has become a demand to estimate the remaining life of filter bags through big data and improve the accuracy of the estimation results. Summary of the invention

[0004] The purpose of the present invention is to provide a method for estimating the remaining life of filter bags for bag dust collectors. Based on BP neural network technology and gradient descent algorithm, a high-efficiency and high-accuracy estimation method is provided for estimating the remaining life of filter bags for bag dust collectors, which solves the problem of estimating the remaining life of filter bags.

[0005] In order to achieve the above object, the solution of the present invention is: A method for estimating the remaining life of a filter bag for a bag filter, comprising: Step 1. Obtain sample data including environmental conditions and flue gas parameters of actual application conditions, basic performance and mechanical properties of filter bags before and after use, and wear conditions and usage time of filter bags; Step 2. Create a BP neural network model, input the sample data obtained in step 1 into the first layer of the BP neural network model, and the output result of the last layer of the BP neural network model is the estimated remaining life of the filter bag; Step 3. Iteratively calculate and correct the BP neural network model through the gradient descent algorithm until the absolute error is within a preset range, and then estimate the optimal value of the remaining life of the filter bag through the iterative BP neural network model.

[0006] In step 1, the environmental conditions and flue gas parameters of the actual application conditions include system operating pressure difference, cleaning pressure, inlet dust concentration, outlet dust concentration, sulfur dioxide concentration and nitrogen oxide concentration, and the basic performance of the filter bag before and after use includes the thickness and air permeability of the filter bag.

[0007] In the step 1, the wear of the filter bag is evaluated and the rating is divided into 0 to 3 levels; wherein, 0 means no signs of wear, 1 means the filter bag is worn to the point where the fibers on the dust-facing surface are loose, 2 means the filter bag is worn to the point where the base fabric is exposed, and 3 means the filter bag is severely worn and has penetrating holes.

[0008] In step 2, the BP neural network model includes 3 hidden layers, and the activation function is logistic; according to the 3 hidden layers, the nodes are set to 32 nodes, 16 nodes, and 1 node respectively, and the corresponding BP neural network model is created accordingly, and then the sample data obtained in step 1 is input into the BP neural network model.

[0009] In step 3, the gradient descent algorithm randomly extracts a set of sample data and updates the relevant parameters of the BP neural network model to train the sample data and continuously iterate and correct it, and finally learns the mapping rule between the input sample data and the expected data; when training the sample data, the expression of each iteration is ,in, represents the parameters of the model when the tth batch is updated, is the learning rate at step t, represents the gradient of the tth batch update, which satisfies , is the sample size of this set of sample data, are model parameters, represents the i-th sample in the sample data set, Represents the target value corresponding to the i-th sample.

[0010] Preferably, in step 3, the absolute error is introduced as an indicator, and the absolute error is defined as ,in For estimated value, is the expected value.

[0011] Preferably, in step 3, the mean square error is used as the Loss function to measure the difference between the remaining life estimated by the BP neural network model and the remaining life of the expected sample; the mean square error refers to the expected value of the square of the difference between the estimated value and the true value, and the calculation formula is: ,in, is the sample size of this set of sample data, is the estimated value of the i-th sample, is the expected value of the ith sample.

[0012] Preferably, in step 3, when the absolute error error≤3 and the mean square error MSE≤1.3, the iterative calculation and correction of the BP neural network model are terminated.

[0013] After adopting the above technical solution, the present invention has the following technical effects: The present invention combines the BP neural network model with the gradient descent algorithm to iteratively calculate all the acquired sample data and correct the estimated value, thereby avoiding the defect of the relatively subjective calculation model in the industry standard, obtaining a more scientific estimation result, and utilizing the advantages of big data to make the estimation result more accurate, thereby realizing a high-efficiency and high-accuracy estimation method, solving the difficult problem of estimating the remaining life of the filter bag. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 is a flow chart of a specific embodiment of the present invention; Figure 2 This is an experimental data table of a specific embodiment of the present invention; Figure 3 This is a diagram showing the effect of the remaining life training of a specific embodiment of the present invention; Figure 4 This is a diagram showing the absolute error training effect of a specific embodiment of the present invention. DETAILED DESCRIPTION

[0015] In order to further explain the technical solution of the present invention, the present invention is described in detail below through specific embodiments.

[0016] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.

[0017] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the present invention to be protected, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of the present invention to be protected.

[0018] The disclosure below provides many different embodiments or examples to implement different structures of the present invention. In order to simplify the disclosure of the present invention, the parts and settings of specific examples are described below. Of course, they are only examples, and the purpose is not to limit the present invention. In addition, the present invention can repeat reference numbers and / or reference letters in different examples, and this repetition is for the purpose of simplification and clarity, which itself does not indicate the relationship between the various embodiments and / or settings discussed.

[0019] Furthermore, the present invention provides examples of various specific processes and materials, but a person skilled in the art may recognize the application of other processes and / or the use of other materials.

[0020] refer to Figure 1 As shown, the present invention discloses a method for estimating the remaining life of a filter bag for a bag filter, comprising: Step 1. Obtain sample data including environmental conditions and flue gas parameters of actual application conditions, basic performance and mechanical properties of filter bags before and after use, and wear conditions and usage time of filter bags; Step 2. Create a BP neural network model, input the sample data obtained in step 1 into the first layer of the BP neural network model, and the output result of the last layer of the BP neural network model is the estimated remaining life of the filter bag; Step 3. Iteratively calculate and correct the BP neural network model through the gradient descent algorithm until the absolute error is within the preset range, and then estimate the optimal value of the remaining life of the filter bag through the iterative BP neural network model.

[0021] Specific embodiments of the present invention are shown below.

[0022] In the above step 1, when collecting data information, the environmental conditions and flue gas parameters of the actual application conditions include system operating pressure difference, cleaning pressure, inlet dust concentration, outlet dust concentration, sulfur dioxide concentration and nitrogen oxide concentration. The basic performance of the filter bag before and after use includes the thickness and air permeability of the filter bag.

[0023] In the above step 1, the wear of the filter bag is evaluated and the rating is divided into 0 to 3 levels; among them, level 0 means no signs of wear, level 1 means the filter bag is worn to the point where the fibers on the dust-facing surface are loose, level 2 means the filter bag is worn to the point where the base fabric is exposed, and level 3 means the filter bag is severely worn and has penetrating holes.

[0024] In the above step 2, the BP neural network model includes 3 hidden layers, and the activation function is logistic; according to the 3 hidden layers, the nodes are set to 32 nodes, 16 nodes, and 1 node respectively, and the corresponding BP neural network model is created accordingly, and then the sample data obtained in step 1 is input into the created BP neural network model.

[0025] In step 3 above, the gradient descent algorithm randomly extracts a set of sample data and updates the relevant parameters of the BP neural network model to train the sample data and iterate and correct it continuously, and finally learns the mapping rule between the input sample data and the expected data; when training the sample data, the expression of each iteration is ,in, represents the parameters of the model when the tth batch is updated, is the learning rate at step t, represents the gradient of the tth batch update, which satisfies , is the number of samples in this set of sample data, are model parameters, represents the i-th sample in the sample data set, Represents the target value corresponding to the i-th sample.

[0026] Furthermore, in order to accurately measure the deviation between the estimated value and the actual value (of remaining life), the absolute error indicator is introduced. The absolute error is defined as ,in For estimated value, is the expected value.

[0027] Secondly, the algorithm uses the mean square error (MSE) as the loss function to measure the difference between the remaining life estimated by the BP neural network model and the remaining life of the expected sample; in mathematical statistics, the mean square error refers to the expected value of the square of the difference between the estimated value and the true value, and the calculation formula is: ,in, is the number of samples in this set of sample data, is the estimated value of the i-th sample, is the expected value of the ith sample.

[0028] Furthermore, in the above step 3, when the output absolute error error≤3 and the mean square error MSE≤1.3, the estimated remaining life of the filter bag is closest to the true value; otherwise, the expected value is re-estimated and the calculation and correction are repeated.

[0029] Hereinafter, the method steps of the present invention are used to perform an estimation calculation on 100 sample data: Figure 2 The figure shows part of the data in the sample data set obtained. This set of sample data is input into the BP neural network model for training. Figure 3 The expected value of 100 sample data and its corresponding estimated value are shown. Figure 4The absolute error of 100 sample data is shown. After iterative calculation and correction by the random gradient descent algorithm, the estimated value of 100 sample data is very close to the expected value, with an average absolute error of 1.02 and an average mean square error of 1.26.

[0030] Through the above scheme, the present invention combines the BP neural network model with the gradient descent algorithm to achieve iterative calculation of all acquired sample data and correct the estimated value, avoiding the subjective defects of the calculation model in the industry standard, and obtaining more scientific estimation results. It also utilizes the advantages of big data to make the estimation results more accurate, thereby realizing a high-efficiency and high-accuracy estimation method, solving the problem of estimating the remaining life of the filter bag.

[0031] The above clearly shows and describes the preferred embodiment of the present invention, and it should be appreciated that the present invention is not limited to the form disclosed herein, nor is it limited to the application of other embodiments, but can be applied to many other combinations, variations and environments, and can be modified by the above examples or the technology or knowledge in the relevant field within the scope of the present invention. However, changes and modifications made by those skilled in the art are within the scope of the protection of the appended claims of the present invention without departing from the spirit and scope of the present invention.

Claims

1. A method for estimating the remaining life of a filter bag for a bag filter, characterized in that include: Step 1. Obtain sample data including environmental conditions and flue gas parameters of actual application conditions, basic performance and mechanical properties of filter bags before and after use, and wear conditions and usage time of filter bags; Step 2. Create a BP neural network model, input the sample data obtained in step 1 into the first layer of the BP neural network model, and the output result of the last layer of the BP neural network model is the estimated remaining life of the filter bag; Step 3. Iteratively calculate and correct the BP neural network model through the gradient descent algorithm until the absolute error is within a preset range, and then estimate the optimal value of the remaining life of the filter bag through the iterative BP neural network model.

2. The method for estimating the remaining life of a filter bag for a bag filter according to claim 1, characterized in that: In step 1, the environmental conditions and flue gas parameters of the actual application conditions include system operating pressure difference, cleaning pressure, inlet dust concentration, outlet dust concentration, sulfur dioxide concentration and nitrogen oxide concentration, and the basic performance of the filter bag before and after use includes the thickness and air permeability of the filter bag.

3. The method for estimating the remaining life of a filter bag for a bag filter according to claim 1, characterized in that: In the step 1, the wear of the filter bag is evaluated and the rating is divided into 0 to 3 levels; wherein, 0 means no signs of wear, 1 means the filter bag is worn to the point where the fibers on the dust-facing surface are loose, 2 means the filter bag is worn to the point where the base fabric is exposed, and 3 means the filter bag is severely worn and has penetrating holes.

4. The method for estimating the remaining life of a filter bag for a bag filter according to claim 1, characterized in that: In step 2, the BP neural network model includes 3 hidden layers, and the activation function is logistic; according to the 3 hidden layers, the nodes are set to 32 nodes, 16 nodes, and 1 node respectively, and the corresponding BP neural network model is created accordingly, and then the sample data obtained in step 1 is input into the BP neural network model.

5. The method for estimating the remaining life of a filter bag for a bag filter according to claim 1, characterized in that: In step 3, the gradient descent algorithm randomly extracts a set of sample data and updates the relevant parameters of the BP neural network model to train the sample data and continuously iterate and correct it, and finally learns the mapping rule between the input sample data and the expected data; when training the sample data, the expression of each iteration is ,in, represents the parameters of the model when the tth batch is updated, is the learning rate at step t, represents the gradient of the tth batch update, which satisfies , is the number of samples in this set of sample data, are model parameters, represents the i-th sample in the sample data set, Represents the target value corresponding to the i-th sample.

6. The method for estimating the remaining life of a filter bag for a bag filter according to claim 5, characterized in that: In step 3, the absolute error is introduced, and the absolute error is defined as ,in For estimated value, is the expected value.

7. The method for estimating the remaining life of a filter bag for a bag filter according to claim 6, characterized in that: In step 3, the mean square error is used as the Loss function to measure the difference between the remaining life estimated by the BP neural network model and the remaining life of the expected sample; the mean square error refers to the expected value of the square of the difference between the estimated value and the true value, and the calculation formula is: ,in, is the number of samples in this set of sample data, is the estimated value of the i-th sample, is the expected value of the ith sample.

8. The method for estimating the remaining life of a filter bag for a bag filter according to claim 7, characterized in that: In step 3, when the absolute error error≤3 and the mean square error MSE≤1.3, the iterative calculation and correction of the BP neural network model are terminated.

Images