A method for monitoring the operation of a filter bag dust collection system

Abstract

This invention relates to the field of dust collector technology and proposes a method for monitoring the operation of a filter bag dust collection system. The method includes: acquiring various operating parameters and production parameter data from multiple production-cleaning processes using the filter bag; performing correlation analysis between operating parameters and production parameters, and constructing a relationship matrix for the production-cleaning process; analyzing similar production-cleaning processes to obtain several types of cleaning behaviors; analyzing the moments when the filter bag resistance changes significantly during the production-cleaning process, as well as the resistance information at the end of the previous production-cleaning process, and combining these with the moments when the parameter data changes significantly, to obtain several short-term deviation time points and several long-term labeled parameters for each production-cleaning process; performing scoring and pruning on neurons to construct a filter bag lifespan prediction model, thereby achieving operational monitoring of the filter bag dust collection system. This invention aims to solve the problem of low efficiency in filter bag lifespan prediction models due to the redundancy of filter bag information and parameters.

Description

Technical Field

[0001] This invention relates to the field of dust collector technology, and specifically to a method for monitoring the operation of a filter bag dust collection system. Background Technology

[0002] Baghouse dust collectors (filter bags) have become one of the most effective pollution control devices for controlling particulate matter in industrial production due to their high dust removal efficiency, strong adaptability, and stable and reliable operation. Therefore, the stability of their operation determines the emission effect. However, the filter bags of baghouse dust collectors are constantly recycled during use and are constantly affected by the pulse-jet cleaning process. At the same time, it is difficult to detect the damage to the filter bags in time, so it is impossible to accurately assess the service life of the filter bags.

[0003] The maintenance method for filter bag resistance in a baghouse dust collector, referenced in patent CN117654175A, involves continuously analyzing various filter bag information during the baghouse dust collector's operation, such as filter bag resistance, cleaning cycle, pulse-jet pressure, and pulse-jet frequency. This comprehensive analysis predicts the entire lifespan of the filter bags, thereby determining their lifespan and enabling operational monitoring of the baghouse dust collection system. However, in actual prediction, filter bag information needs to be input into a trained prediction model. Because filter bag resistance growth is lag-dependent and filter bag lifespan is affected by long-term cycles, LSTM models are widely used for prediction. However, during LSTM model training based on filter bag information, the information and parameters are often redundant, and some parameters do not consistently reflect the impact on filter bag lifespan. This results in an unpruned model with excessively slow inference speed, failing to meet the requirements of filter bag cleaning control response, leading to inaccurate predictions of filter bag lifespan and impacting the overall operational monitoring of the baghouse dust collection system. Summary of the Invention

[0004] This invention provides a method for monitoring the operation of a filter bag dust collection system, to solve the problem of low efficiency in existing filter bag life prediction models due to the redundancy of filter bag information and parameters. The specific technical solution adopted is as follows:

[0005] This invention proposes a method for monitoring the operation of a filter bag dust collection system, which includes the following steps:

[0006] By acquiring various operating parameters and production parameters during multiple production-cleaning processes through filter bags arranged in a bag filter, and recording the pressure data at the inlet and outlet of the filter bags in real time;

[0007] Based on the data change trends of operating parameters during the production-dust cleaning process, the process is segmented. Correlation analysis between operating parameters and production parameters is performed in conjunction with the data changes of production parameters, and a relationship matrix of the production-dust cleaning process is constructed. Based on the relationship matrix constructed by the correlation between operating parameters and production parameters, similar production-dust cleaning processes are analyzed to obtain several types of dust cleaning behaviors.

[0008] Analyze the moments when the filter bag resistance changes significantly during the production-cleaning process, as well as the resistance information at the end of the previous production-cleaning process. Combine these moments with the moments when the parameter data changes significantly to obtain several short-term deviation time points for each production-cleaning process. By analyzing the correlation changes at the same position in the temporally adjacent relationship matrix of similar cleaning behaviors, obtain several long-term labeled parameters for each production-cleaning process.

[0009] Based on the short-term deviation time points and long-term labeled parameters of each production-dust cleaning process in the training dataset, the neurons in the LSTM model are scored and pruned to construct a filter bag life prediction model, thereby realizing the operation monitoring of the filter bag dust removal system.

[0010] Optionally, the specific methods for segmenting the data based on the changing trends of operating parameters during the production-dust cleaning process, performing correlation analysis between operating parameters and production parameters in conjunction with changes in production parameter data, and constructing a relationship matrix for the production-dust cleaning process include:

[0011] Analyze the fluctuation of the slope of the change in the operating parameters during a single production-dust cleaning process, and segment the data change curve of the operating parameters.

[0012] By combining the data change curves of production parameters, a correlation analysis between operating parameters and production parameters is conducted to construct a relationship matrix for each production-dust cleaning process.

[0013] Optionally, the analysis of the fluctuation slope of the change in the operating parameters during a single production-dust cleaning process, and the segmentation of the data change curve of the operating parameters, includes the following specific methods:

[0014] For any production-dust cleaning process, based on the time-series parameter data of each operating parameter and the time-series parameter data of each production parameter in the production-dust cleaning process, obtain the data change curves of each operating parameter and the data change curves of each production parameter.

[0015] For the The data change curves of each operating parameter are obtained, and the slope of each data point is obtained. For any data point, the variance of the slope of the data point and all data points before it is obtained as the variance of the slope before the data point. The variance of the slope of all data points after the data point is obtained as the variance of the slope after the data point.

[0016] Based on the variance of the front slope and the variance of the back slope of each data point in the data change curve, the first and second division points of each operating parameter are obtained.

[0017] The data curve between the first data point and the first dividing point is taken as the first... The initial data change curve of each operating parameter; the data curve between the first data point after the first division point and the second division point is taken as the first... The mid-term data change curve of the first operating parameter; the data curve formed by all data points after the second division point is used as the first... The curves showing the changes in the data of each operating parameter in the later stages.

[0018] Optionally, the specific method for obtaining the first and second division points of each operating parameter is as follows:

[0019] The first The variance of the slope of each data point in the data change curve of each operating parameter is arranged according to the time sequence of the data points to obtain the first... The data change curves of each operating parameter are processed using the preceding slope variance sequence. Several maxima are obtained from the preceding slope variance sequence. Starting from the first maxima, a threshold for the preceding slope variance is set. If the first maxima is greater than the threshold, the data point corresponding to the first maxima is taken as the first... If the first maximum value of the data change curve of each operating parameter is less than or equal to the previous slope variance threshold, the second maximum value is judged until the first dividing point is obtained.

[0020] The variance of the slope of all data points after the first split point is arranged in chronological order to obtain the variance of the slope of the data points after the first split point. The stable slope variance sequence of the data change curves for each operating parameter is obtained. Several maxima are acquired from the stable slope variance sequence. Starting from the first maxima, a threshold for observing the slope is set. If the first maxima is greater than the slope variance threshold, the data point corresponding to the first maxima is taken as the [number of]th [maximums]. If the first maximum value is less than or equal to the previous slope variance threshold, the second maximum value is judged until the second division point is obtained.

[0021] Optionally, the specific method for constructing the relationship matrix of each production-dust cleaning process includes:

[0022] According to the The first and second division points of the _ _ running parameters, for the _ _ _ _ The data change curves for each production parameter were obtained for the initial, middle, and later stages; for the... The first operating parameter and the first Pearson correlation coefficients were obtained for the data change curves of each production parameter in the initial, middle and later stages.

[0023] Preset initial, intermediate, and late-stage weights, corresponding to the data change curves of the initial, intermediate, and late stages, respectively. The Pearson correlation coefficients of these three data change curves are then weighted and summed using these weights. The sum is used as the first value in this production-dust-cleaning process. The first operating parameter and the second The correlation value of each production parameter;

[0024] Obtain the correlation values ​​between each operating parameter and each production parameter during the production-dust cleaning process, and construct a matrix. In the matrix, the same row represents the correlation values ​​between the same operating parameter and each production parameter, and the same column represents the correlation values ​​between each operating parameter and the same production parameter. The resulting matrix is ​​used as the relationship matrix for the production-dust cleaning process.

[0025] Optionally, the specific methods for obtaining several types of dust removal behaviors include:

[0026] Density clustering is performed on all production-dust cleaning processes. The distance metric is the inverse proportional value of the cosine similarity between the relation matrices of each production-dust cleaning process, resulting in several clusters. Production-dust cleaning processes in the same cluster are considered as the same type of dust cleaning behavior.

[0027] Optionally, the method for analyzing the moments when the filter bag resistance changes significantly during the production-cleaning process, and the resistance information at the end of the previous production-cleaning process, combined with the moments when the parameter data of each parameter changes significantly, to obtain several short-term deviation time points for each production-cleaning process includes:

[0028] Based on the pressure difference changes at the inlet and outlet of the filter bag during a single production-cleaning process, and the pressure difference at the end of the previous production-cleaning process, several pressure difference change moments were selected.

[0029] By acquiring several moments of change of various parameters during a single production-dust cleaning process and combining them with the moments of change of pressure difference, a difference analysis is performed to obtain several short-term deviation points in each production-dust cleaning process.

[0030] Optionally, the specific method for obtaining several pressure difference change times through screening includes:

[0031] For the training dataset, the first In the next production-dust cleaning process, the pressure difference between the inlet and outlet is obtained at each moment. The difference between the inlet pressure data and the outlet pressure data is taken as the inlet and outlet pressure difference at each moment. The inlet and outlet pressure difference at the last moment of the next production-dust cleaning process; construct a coordinate system with the sequence value corresponding to each moment as the horizontal axis and the pressure difference as the vertical axis, and then... The inlet and outlet pressure differences at all moments during the secondary production-dust cleaning process are mapped onto a coordinate system according to their corresponding order values ​​to obtain the pressure difference data points at each moment. The inlet and outlet pressure difference at the last moment of the next production-dust cleaning process is used as the pressure difference corresponding to the origin of the horizontal axis. Connecting the pressure difference data points yields the... The pressure difference data curve of the next production-dust cleaning process is used to calculate the slope of the pressure difference data points at each time except time 0, and this slope is used as the slope of the change of the next pressure difference data point.

[0032] A preset slope threshold is used to determine the time point corresponding to the pressure difference data point with a slope greater than the threshold. The timing of pressure difference changes during the secondary production-dust cleaning process.

[0033] Optionally, the specific method for obtaining several short-term deviation time points in each production-dust cleaning process includes:

[0034] For the During the production-dust cleaning process, the data change curve of any operating parameter or production parameter is obtained, and the slope of each data point is acquired. A preset slope threshold is set, and the time corresponding to the data point with a slope greater than the parameter slope threshold is taken as the first... The timing of this parameter change during the next production-dust cleaning process;

[0035] Based on the For each pressure difference change moment and any parameter change moment in the production-dust cleaning process, the absolute value of the difference between the pressure difference change moment and each change moment of the parameter is obtained, and the minimum value is taken as the time offset of the pressure difference change moment.

[0036] Obtain the time offset at each pressure difference change moment. If the time offset at any pressure difference change moment is greater than the time offset threshold, that pressure difference change moment is taken as a short-term deviation time point, and the time offset at that moment is used to obtain the time offset. Secondary dust removal - several short-term deviations from the production process.

[0037] Optionally, the specific method for obtaining several long-term marking parameters for each production-dust cleaning process is as follows:

[0038] For any type of dust removal behavior, several production-dust removal processes are sorted according to time sequence. For adjacent relationship matrices, the absolute value of the difference between the elements at the same position is calculated as the element difference at the corresponding position.

[0039] If the element difference is greater than the aging threshold, the operating parameters and production parameters corresponding to the elements at that position in the production-cleaning process of the two relation matrices are marked as long-term marked parameters in the two production-cleaning processes; aging judgment is performed on adjacent relation matrices in the same type of cleaning behavior to obtain several long-term marked parameters for each production-cleaning process.

[0040] The beneficial effects of this invention are as follows: This invention analyzes the correlation between operating parameters and production parameters during the production-dust cleaning process, while considering the impact of the stability of generated products at different stages on parameter performance. It quantifies correlation coefficients in segments and obtains correlation values, then constructs a relationship matrix to reflect the correlation between operating parameters and production parameters. Similar correlations indicate that the overall parameter operation relationships of the production-dust cleaning process are similar, and they are more likely to correspond to similar dust cleaning behaviors. By performing deviation analysis on filter bag resistance changes and parameter changes, short-term deviation time points are selected to reflect the impact of their unique parameter information on filter bag lifespan, which is then used in neuron pruning in the subsequent model. Furthermore, by analyzing the correlation changes in the relationship matrix of similar dust cleaning behaviors, aging information is analyzed to assess the long-term impact of periodic changes, and corresponding parameters are marked to reflect the long-term aging information of the corresponding operating or production parameters. By using short-term deviation time points and long-term aging information to score neurons and determine pruning, the inference efficiency of the LSTM prediction model is improved, and a filter bag lifespan prediction model is constructed. This enables visualized monitoring of the lifespan of the filter bag dust collection system in baghouse dust collectors, improving the operational monitoring efficiency of the filter bag dust collection system. Attached Figure Description

[0041] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0042] Figure 1 This is a schematic diagram of a filter bag dust collection system operation monitoring method provided in one embodiment of the present invention. Detailed Implementation

[0043] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0044] Please see Figure 1 The diagram illustrates a flowchart of a filter bag dust collection system operation monitoring method according to an embodiment of the present invention, which includes the following steps:

[0045] Step S001: Obtain parameter data of various operating parameters and production parameters during multiple production-cleaning processes by arranging filter bags in the bag filter, and record the pressure data at the inlet and outlet of the filter bags in real time.

[0046] The purpose of this embodiment is to analyze the short-term and long-term periodic effects of filter bag information and various parameters on filter bag life in the process of constructing a filter bag life prediction model using an LSTM prediction model. This allows for the pruning of neurons in the model to improve the inference efficiency of the life prediction model. The pruned LSTM prediction model is then used to train the filter bag life prediction model, thereby enabling the operation and monitoring of the filter bag dust removal system.

[0047] Specifically, data is collected on the filter bags arranged in the bag filter, several production parameters during the production process, and various operating parameters during the dust removal process. The filter bag information includes the resistance information of the filter bags. In this embodiment, the resistance information is reflected by recording the pressure data at the filter bag inlet and outlet, with the sampling time interval set to 1 second. The production parameters include the flue gas outlet temperature, humidity, and other production parameters set according to the production target. The parameter data of each production parameter is collected at the same sampling time interval. The operating parameters of the dust removal process include the blowing pressure, blowing cycle, and number of blowing cycles, etc., which are also recorded at the same sampling time interval to obtain the parameter data of each operating parameter.

[0048] Furthermore, a training dataset is constructed using a large number of operating parameters and production parameters from the production and cleaning process of filter bags of the same model of bag filter, along with recorded filter bag information, for training the subsequent filter bag life prediction model.

[0049] Step S002: Based on the data change trend of operating parameters during the production-dust cleaning process, segment the data, combine the data change of production parameters to conduct correlation analysis between operating parameters and production parameters, and construct a relationship matrix of the production-dust cleaning process; based on the relationship matrix constructed by the correlation between operating parameters and production parameters, analyze similar production-dust cleaning processes to obtain several types of dust cleaning behaviors.

[0050] It should be noted that there is a certain correlation between operating parameters and production parameters. By performing correlation analysis on the data change curves of operating parameters and production parameters, the relationship between operating parameters and production parameters can be reflected. Based on this relationship, similar dust removal behaviors can be analyzed. By constructing a relationship matrix and performing clustering based on the relationship matrix, dust removal behaviors can be classified.

[0051] As an example, this step includes the following specific methods:

[0052] Analyze the fluctuation of the slope of the change in the operating parameters during a single production-dust cleaning process, and segment the data change curve of the operating parameters.

[0053] By combining the data change curves of production parameters, a correlation analysis between operating parameters and production parameters is conducted to construct a relationship matrix for each production-dust cleaning process;

[0054] Cluster analysis of the production-dust cleaning process based on the relationship matrix yielded several types of dust cleaning behaviors.

[0055] It should be noted that a single production-dust removal process includes an initial unstable rising phase, a mid-term stable phase, and a late-term cessation phase. At the beginning of production, the reaction is just starting, and the generated products are unstable, leading to significant deviations in correlation calculations. The late-term cessation phase, however, transitions from stable to unstable, and its data changes are more stable compared to the initial phase. Therefore, segmented analysis based on the slope (trend) of the operating parameters is necessary. Since the primary focus is on monitoring the dust removal system's operation, production parameters are segmented using comparative operating parameters. Due to the initial instability, the slope exhibits significant fluctuations, while the mid-term stable slope shows smaller fluctuations. The slope fluctuations in the late-term cessation phase are smaller than those in the initial phase. Based on this, the variances of the initial and subsequent slopes are constructed for analysis, and threshold values ​​are used to determine the first and second dividing points.

[0056] Preferably, in one embodiment of the present invention, the method for analyzing the fluctuation of the slope of the change in the operating parameters during a single production-dust cleaning process and segmenting the data change curve of the operating parameters includes:

[0057] For any given production-dust cleaning process, which includes several operating parameters and production parameters, based on the time-series parameter data of each operating parameter and each production parameter during that production-dust cleaning process, the data change curves of each operating parameter and each production parameter are obtained; for the th The data change curves of each operating parameter are used to obtain the slope of each data point. Based on the difference between any data point and its adjacent previous data point and the time interval, the slope of the first data point in the data change curve is set as the slope of the second data point. For any data point, the variance of the slopes of the data point and all the data points before it is obtained as the variance of the slope before the data point. Similarly, the variance of the slopes of all the data points after the data point is obtained as the variance of the slope after the data point.

[0058] Furthermore, the first The variance of the slope of each data point in the data change curve of each operating parameter is arranged according to the time sequence of the data points to obtain the first... The data change curves of each operating parameter are analyzed using the preceding slope variance sequence. Several maxima are obtained from the preceding slope variance sequence. Starting from the first maxima, a preceding slope variance threshold is set. In this embodiment, the preceding slope variance threshold is the preceding slope variance corresponding to the top 10% of all preceding slope variances sorted from largest to smallest. If the first maxima is greater than the preceding slope variance threshold, then the data point corresponding to the first maxima is taken as the first... If the first maximum value of the data change curve of each operating parameter is less than or equal to the previous slope variance threshold, then the judgment on the second maximum value continues until the first division point is obtained.

[0059] Furthermore, the variance of the slope of all data points after the first split point is arranged in chronological order to obtain the variance of the slope of the data points after the first split point. The stable slope variance sequence of the data change curves for each operating parameter is obtained. Several maxima are acquired from the stable slope variance sequence. Starting from the first maxima, a slope observation threshold is set. In this embodiment, the slope variance threshold is the number of data points whose slope variances are ranked from largest to smallest. If the first maxima is greater than the slope variance threshold, the data point corresponding to the first maxima is taken as the first... If the first maximum value of the data change curve of each operating parameter is less than or equal to the previous slope variance threshold, then the second maximum value is judged until the second division point is obtained.

[0060] Furthermore, the data curve between the first data point and the first dividing point (inclusive) is taken as the... The initial data change curve of each operating parameter; the data curve between the first data point after the first division point and the second division point (inclusive) is taken as the first... The mid-term data change curve of the first operating parameter; the data curve formed by all data points after the second division point is used as the first... The curves showing the changes in the data of each operating parameter in the later stages.

[0061] Preferably, in one embodiment of the present invention, the correlation analysis between operating parameters and production parameters is performed by combining the data change curves of production parameters to construct a relationship matrix for each production-dust cleaning process. The specific method includes:

[0062] Based on the same first and second division points (corresponding times), the first... The data change curves for each production parameter were obtained for the initial, middle, and later stages; for the... The first operating parameter and the first The Pearson correlation coefficients of the data change curves for each production parameter were obtained for the initial, middle, and later stages. That is, the correlation coefficients of data change curves within the same time period were calculated. In this embodiment, the initial stage was weighted at 0.2, the middle stage at 0.5, and the later stage at 0.3, corresponding to the initial, middle, and later stages of the data change curves, respectively. The Pearson correlation coefficients of the three data change curves were weighted and summed using the corresponding weights, and the sum was used as the first value in the production-dust removal process. The first operating parameter and the second The correlation value of each production parameter.

[0063] It should be noted that the stable changes of the mid-term stabilizer are the most reliable for the correlation analysis of operating parameters and production parameters, and the largest mid-term weight is set. In contrast, the instability of the later stopping period is smaller than the fluctuation range of the initial rising instability period, so the weight of the later period needs to be set larger than the weight of the initial period. Then, by combining the correlation coefficient of the segmented data change curves with the corresponding weights, the correlation value between operating parameters and production parameters in this production-dust cleaning process is quantified.

[0064] Furthermore, the correlation values ​​between each operating parameter and each production parameter in the production-dust cleaning process are obtained according to the above method, and a matrix is ​​constructed. In the matrix, the same row represents the correlation values ​​between the same operating parameter and each production parameter, and the same column represents the correlation values ​​between each operating parameter and the same production parameter. The resulting matrix is ​​used as the relationship matrix of the production-dust cleaning process.

[0065] It should be further explained that if the correlation between the operating parameters and production parameters in similar production-dust cleaning processes is similar, then by calculating the cosine similarity of the relationship matrix, the larger the cosine similarity, the smaller the distance of the corresponding relationship matrix during the clustering process, thus obtaining several clusters, and classifying similar dust cleaning behaviors based on the clusters.

[0066] Preferably, in one embodiment of the present invention, cluster analysis is performed on the production-dust cleaning process based on the relationship matrix to obtain several types of dust cleaning behaviors, including the following specific methods:

[0067] The relationship matrix of all production-cleaning processes in the training dataset is obtained, and density clustering is performed on all production-cleaning processes. In this embodiment, DBSCAN clustering is used. The distance metric is the inverse proportional value of the cosine similarity between the relationship matrices of each production-cleaning process (1 minus the difference obtained by cosine similarity). Several clusters are obtained. The production-cleaning processes in the same cluster are regarded as the same type of cleaning behavior, and each cluster corresponds to a type of cleaning behavior.

[0068] Thus, by conducting correlation analysis between the operating parameters and production parameters in the production-dust cleaning process, and considering the impact of the stability of the generated products at different stages on the parameter performance, the correlation coefficients were quantified in segments and correlation values ​​were obtained. Then, a relationship matrix was constructed to reflect the correlation between the operating parameters and production parameters. Similar correlation performance indicates that the overall parameter operation relationship of the production-dust cleaning process is similar, and it is highly likely to correspond to the same type of dust cleaning behavior.

[0069] Step S003: Analyze the moments when the filter bag resistance changes significantly during the production-cleaning process, as well as the resistance information at the end of the previous production-cleaning process. Combine the moments when the parameter data changes significantly to obtain several short-term deviation time points for each production-cleaning process. Obtain several long-term marked parameters for each production-cleaning process by analyzing the correlation changes at the same position in the temporally adjacent relationship matrix of similar cleaning behaviors.

[0070] It should be noted that the neurons expected to be retained can retain both long-term information and short-term change information. The long-term information represents the aging information of the filter bag during its use (the process in the filter bag of the entire dust removal system is: dust accumulation - formation of dust accumulation layer - dust removal behavior - dust removal and stripping (but dust remains). Therefore, with the continuous repetition of production and dust removal work, there will be some factors that affect the lifespan, which corresponds to the long-term information difference). The short-term difference represents the accumulation of deposition information in a production and dust removal process, that is, the information change from the initial dust accumulation.

[0071] As an example, this step includes the following specific methods:

[0072] Based on the pressure difference changes at the inlet and outlet of the filter bag during a single production-cleaning process, and the pressure difference at the end of the previous production-cleaning process, several pressure difference change moments were selected.

[0073] The changes of various parameters during a single production-dust cleaning process are obtained, and the differences are analyzed in conjunction with the changes in pressure difference to obtain several short-term deviation time points for each production-dust cleaning process.

[0074] By analyzing the changes of elements at the same position in the relation matrix of temporally adjacent production-cleaning processes in the same type of cleaning behavior, several long-term labeled parameters of each production-cleaning process are obtained.

[0075] Preferably, in one embodiment of the present invention, based on the pressure difference change at the inlet and outlet of the filter bag during a single production-cleaning process, and the pressure difference at the end of the previous production-cleaning process, several pressure difference change moments are selected, including the following specific method:

[0076] For the training dataset, the first The next production-dust cleaning process (the first in the sequence) (This is repeated several times). The pressure difference between the inlet and outlet of the filter bag at each time point is obtained to reflect the filter bag's resistance growth rate. Since the inlet pressure is greater than the outlet pressure, the difference between the inlet and outlet pressure data is used as the inlet-outlet pressure difference at each time point. Simultaneously, because the resistance growth rate is affected by the previous production-cleaning process, the pressure difference at the same time point is also obtained. The inlet and outlet pressure difference at the last moment of the next production-dust cleaning process; construct a coordinate system with the sequence value corresponding to each moment as the horizontal axis and the pressure difference as the vertical axis, and then... The inlet and outlet pressure differences at all moments during the secondary production-dust cleaning process are mapped onto a coordinate system according to their corresponding order values ​​to obtain the pressure difference data points at each moment. Simultaneously, the first... The inlet and outlet pressure difference at the last moment of the next production-dust cleaning process is used as the pressure difference corresponding to the origin of the horizontal axis, i.e., the pressure difference at moment 0, and the corresponding pressure difference data point is obtained. Connecting each pressure difference data point yields the pressure difference at the 0th moment. The pressure difference data curve for the production-dust cleaning process is used to calculate the slope of the pressure difference data points at all times except time 0. The slope is obtained by comparing the difference between the inlet and outlet pressure differences at the next time step and the difference between the inlet and outlet pressure differences at the previous time step with the sampling time interval (i.e., the interval between adjacent time steps). This slope is then used as the slope of the next pressure difference data point. Specifically, the slope of the first time step is obtained by comparing the inlet and outlet pressure differences at time 0 with the sampling time interval. A preset slope threshold is used; in this embodiment, the first threshold is used. The ratio of the range of the inlet and outlet pressure differences at all moments during the production-dust cleaning process to the sampling time interval is taken as 1 / 5 of the slope threshold. The time corresponding to the pressure difference data point with a slope greater than the slope threshold is taken as the first... The moment when the pressure difference changes during the secondary production-dust cleaning process is the moment when the resistance information changes significantly.

[0077] It should be noted that the pressure difference data curve corresponds to the dust deposition during the production-cleaning process. Therefore, the pressure difference data curve should correspond to the changes in the operating parameters and production parameters of this process. If the times do not correspond, it indicates that there is some unique information in the short term that affects the filter bag life, and thus it needs to be used as a short-term deviation time point for subsequent pruning.

[0078] Preferably, in one embodiment of the present invention, several change times of various parameters during a single production-dust cleaning process are obtained, and difference analysis is performed in conjunction with the change times of pressure difference to obtain several short-term deviation time points for each production-dust cleaning process. The specific method includes:

[0079] For the The data change curve of any operating parameter or production parameter during the production-dust cleaning process is obtained, and the slope of each data point is acquired. A preset parameter slope threshold is set. In this embodiment, the first... The ratio of the range of the parameter data of this operating parameter or production parameter at all times during the production-dust cleaning process to the sampling time interval is taken as 1 / 5 of the parameter slope threshold. The time corresponding to the data point with a slope greater than the parameter slope threshold is taken as the first... The time of change of this parameter (operating parameter or production parameter) during the next production-dust cleaning process.

[0080] Furthermore, based on the first For each pressure difference change moment and any parameter (operating parameter or production parameter) change moment in the production-dust cleaning process, the absolute value of the difference between each pressure difference change moment and each change moment of the parameter is obtained, and the minimum value is taken as the time offset of the pressure difference change moment. The time offset of each pressure difference change moment is obtained, and a preset time offset threshold is used. In this embodiment, the time offset threshold is described as 10 seconds. If the time offset of any pressure difference change moment is greater than the time offset threshold, the pressure difference change moment is taken as a short-term deviation time point, and the process is repeated to obtain the time offset of the next pressure difference change moment. Secondary dust removal - several short-term deviations from the production process.

[0081] It should be further explained that, in the same type of dust removal behavior, the difference in the change of the elements at the same position in the temporally adjacent relation matrix is ​​obtained. This difference in change represents the loss caused by long-term accumulation, that is, aging information. In other words, the entire production-dust removal process should be analyzed as a whole during the comparison process.

[0082] Preferably, in one embodiment of the present invention, the changes of elements at the same position in the relationship matrix of temporally adjacent production-cleaning processes in the same type of cleaning behavior are analyzed to obtain several long-term labeled parameters for each production-cleaning process. The specific method includes:

[0083] For any type of dust removal behavior involving several production-dust removal processes, the relationship matrices of all production-dust removal processes are sorted according to time sequence. For adjacent relationship matrices, the absolute value of the difference between the elements (correlation values) at the same position is calculated as the element difference at the corresponding position. An aging threshold is preset, and in this embodiment, the aging threshold is described as 0.3. If the element difference is greater than the aging threshold, the operating parameters and production parameters corresponding to the elements at that position in the production-dust removal processes of the two relationship matrices are marked as long-term marked parameters in the two production-dust removal processes. The aging judgment is performed on adjacent relationship matrices in the same type of dust removal behavior according to the above method to obtain several long-term marked parameters for each production-dust removal process.

[0084] Thus, by performing deviation analysis on the changes in filter bag resistance and parameters, short-term deviation time points are selected to reflect the impact of unique parameter information on filter bag life and used in neuron pruning in subsequent models; and by analyzing the correlation changes in the relationship matrix of similar dust removal behaviors, aging information is analyzed to reflect the long-term impact of periodic changes, and then the corresponding parameters are marked to reflect the long-term aging information of the corresponding operating parameters or production parameters.

[0085] Step S004: Based on the short-term deviation time points and long-term labeling parameters of each production-dust cleaning process in the training dataset, the neurons in the LSTM model are scored and pruned to construct a filter bag life prediction model, thereby realizing the operation monitoring of the filter bag dust removal system.

[0086] It should be noted that the goal is to select network branches in the LSTM network that can represent both short-term and long-term information. Therefore, during LSTM training, the activation status of neurons at these time points can determine the importance score of neurons, and then pruning can be performed based on the importance score. The activation status of neurons is represented by the hidden state value and the cell state value of neurons. The hidden state value of neurons stores short-term information, while the cell state value stores long-term information.

[0087] Specifically, for any production-dust cleaning process, for any short-term deviation time point, a preset number of neighborhoods is set. In this embodiment, the number of neighborhoods is described as 5. The time of the number of neighborhoods before and after the short-term deviation time point, together with the short-term deviation time point, constitute the short-term time window of the short-term deviation time point. The average value of the hidden state of each neuron within the short-term time window is used as the short-term activation intensity of the neuron corresponding to the short-term deviation time point.

[0088] Furthermore, for any long-term labeled parameter in the production-cleaning process, the average value of the cell state of all neurons corresponding to the long-term labeled parameter in the production-cleaning process (over the entire cycle) is used as the long-term activation intensity of the corresponding neuron at each time point under the long-term labeled parameter.

[0089] Furthermore, the sum of the short-term and long-term activation intensities of any neuron (a neuron corresponding to a single operating parameter or a moment in a production parameter during a single production-dust cleaning process) is used as the final score for that neuron. A preset screening ratio is used; in this embodiment, the screening ratio is described as 60%. All neurons are arranged from largest to smallest according to their final scores. The neurons in the first screening ratio are retained, and the remaining neurons undergo structured pruning to train an LSTM model, resulting in a trained LSTM model. The input data for the LSTM model is the training dataset, and the output data is the filter bag lifespan as judged by professionals. The root mean square error function is used as the loss function. The hyperparameters set during the parameter update process of the LSTM model depend on the specific implementation situation and are not specifically limited in this embodiment. The trained LSTM model is used as a filter bag lifespan prediction model to predict the lifespan of filter bags in baghouse dust collectors, thereby realizing the operation monitoring of the filter bag dust collection system.

[0090] Thus, by constructing scores for neurons and judging pruning through short-term deviation time points and long-term aging information, the inference efficiency of the LSTM prediction model is improved and a filter bag life prediction model is constructed. This enables the visualization and monitoring of the life of the filter bag dust collection system in the baghouse dust collector, thereby improving the operational monitoring efficiency of the filter bag dust collection system.

[0091] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for monitoring the operation of a filter bag dust collection system, characterized in that, The method includes the following steps: Multiple operating parameters and production parameters are acquired through the filter bags arranged in the baghouse dust collector during the multiple production-cleaning processes, and the pressure data at the inlet and outlet of the filter bags are recorded in real time. Specifically, data is collected on the filter bag information, several production parameters during the production process, and various operating parameters during the cleaning process. The filter bag information includes the filter bag's resistance information, which is reflected by recording the pressure data at the filter bag's inlet and outlet. The production parameters include the flue gas outlet temperature, humidity, and other production parameters set according to production targets. The operating parameters of the cleaning process include the pulse-jet pressure, pulse-jet cycle, and number of pulse-jet cycles. Based on the data change trends of operating parameters during the production-dust cleaning process, the process is segmented. Correlation analysis between operating parameters and production parameters is performed in conjunction with the data changes of production parameters, and a relationship matrix of the production-dust cleaning process is constructed. Based on the relationship matrix constructed by the correlation between operating parameters and production parameters, similar production-dust cleaning processes are analyzed to obtain several types of dust cleaning behaviors. The methods for obtaining several types of dust removal behaviors include: Density clustering is performed on all production-dust cleaning processes. The distance metric is the inverse proportional value of the cosine similarity between the relation matrices of each production-dust cleaning process, resulting in several clusters. Production-dust cleaning processes in the same cluster are considered as the same type of dust cleaning behavior. Analyze the moments when the filter bag resistance changes significantly during the production-cleaning process, as well as the resistance information at the end of the previous production-cleaning process. Combine these moments with the moments when the parameter data changes significantly to obtain several short-term deviation time points for each production-cleaning process. By analyzing the correlation changes at the same position in the temporally adjacent relationship matrix of similar cleaning behaviors, obtain several long-term labeled parameters for each production-cleaning process. The analysis focuses on the moments when the filter bag resistance changes significantly during the production-cleaning process, as well as the resistance information at the end of the previous production-cleaning process. Combined with the moments when the parameter data of each parameter changes significantly, several short-term deviation time points are obtained for each production-cleaning process. The specific methods include: Based on the pressure difference changes at the inlet and outlet of the filter bag during a single production-cleaning process, and the pressure difference at the end of the previous production-cleaning process, several pressure difference change moments were selected. The changes of various parameters during a single production-dust cleaning process are obtained, and the differences are analyzed in conjunction with the changes in pressure difference to obtain several short-term deviation time points for each production-dust cleaning process. The specific method for obtaining several pressure difference change moments through screening is as follows: For the a-th production-dust cleaning process in the training dataset, obtain the pressure difference between the inlet and outlet at each time point. Subtract the outlet pressure from the inlet pressure data to obtain the difference, which is taken as the inlet-outlet pressure difference at each time point. Obtain the inlet-outlet pressure difference at the last time point in the (a-1)-th production-dust cleaning process. Construct a coordinate system with the order value corresponding to each time point on the horizontal axis and the pressure difference on the vertical axis. Map the inlet-outlet pressure differences at all times in the a-th production-dust cleaning process to the coordinate system according to the corresponding order value to obtain the pressure difference data points at each time point. Take the inlet-outlet pressure difference at the last time point in the (a-1)-th production-dust cleaning process as the pressure difference at the corresponding horizontal axis origin. Connect the pressure difference data points to obtain the pressure difference data curve for the a-th production-dust cleaning process. Calculate the slope of the pressure difference data points at each time point except for time point 0, and use it as the slope of the change of the next pressure difference data point. A preset slope threshold is set, and the time corresponding to the pressure difference data point with a slope greater than the slope threshold is taken as the pressure difference change time of the a-th production-dust cleaning process. The specific method for obtaining several short-term deviation time points in each production-dust cleaning process includes: For any operating parameter or production parameter data change curve during the a-th production-dust cleaning process, obtain the slope of each data point, preset the parameter slope threshold, and take the time corresponding to the data point with a slope greater than the parameter slope threshold as the change time of that parameter during the a-th production-dust cleaning process. Based on several pressure difference change times and several parameter change times during the a-th production-dust cleaning process, for any pressure difference change time, obtain the absolute value of the time difference between it and each change time of the parameter, and take the minimum value as the time offset of the pressure difference change time. Obtain the time offset at each pressure difference change moment. If the time offset at any pressure difference change moment is greater than the time offset threshold, take the pressure difference change moment as a short-term deviation time point and obtain several short-term deviation time points in the a-th dust removal-production process. The specific method for obtaining several long-term marker parameters for each production-dust cleaning process is as follows: For any type of dust removal behavior, several production-dust removal processes are sorted according to time sequence. For adjacent relationship matrices, the absolute value of the difference between the elements at the same position is calculated as the element difference at the corresponding position. A preset aging threshold is set. If the element difference is greater than the aging threshold, the running parameters and production parameters corresponding to the elements at that position in the production-dust cleaning process of the two relation matrices are marked as long-term marked parameters in the two production-dust cleaning processes. The aging judgment is performed on adjacent relation matrices in the same type of dust cleaning behavior to obtain several long-term marked parameters for each production-dust cleaning process. Based on the short-term deviation time points and long-term labeled parameters of each production-dust cleaning process in the training dataset, the neurons in the LSTM model are scored and pruned to construct a filter bag life prediction model, thereby realizing the operation monitoring of the filter bag dust removal system.

2. The method for monitoring the operation of a filter bag dust collection system according to claim 1, characterized in that, The method of segmenting the data based on the changing trends of operating parameters during the production-dust cleaning process, performing correlation analysis between operating parameters and production parameters in conjunction with changes in production parameter data, and constructing a relationship matrix for the production-dust cleaning process includes the following specific methods: Analyze the fluctuation of the slope of the change in the operating parameters during a single production-dust cleaning process, and segment the data change curve of the operating parameters. By combining the data change curves of production parameters, a correlation analysis between operating parameters and production parameters is conducted to construct a relationship matrix for each production-dust cleaning process.

3. The method for monitoring the operation of a filter bag dust collection system according to claim 2, characterized in that, The analysis of the fluctuations in the slope of the operating parameters during the single production-dust cleaning process, and the segmentation of the operating parameter data change curves, includes the following specific methods: For any production-dust cleaning process, based on the time-series parameter data of each operating parameter and the time-series parameter data of each production parameter in the production-dust cleaning process, obtain the data change curves of each operating parameter and the data change curves of each production parameter. For the data change curve of the i-th running parameter, obtain the slope of each data point; for any data point, obtain the variance of the slope of the data point and all data points before it as the variance of the slope before the data point, and obtain the variance of the slope of all data points after the data point as the variance of the slope after the data point. Based on the variance of the front slope and the variance of the back slope of each data point in the data change curve, the first and second division points of each operating parameter are obtained. The data curve between the first data point and the first division point is taken as the initial data change curve of the i-th operating parameter; the data curve between the first data point after the first division point and the second division point is taken as the intermediate data change curve of the i-th operating parameter; and the data curve formed by all data points after the second division point is taken as the late data change curve of the i-th operating parameter.

4. The method for monitoring the operation of a filter bag dust collection system according to claim 3, characterized in that, The specific method for obtaining the first and second division points of each operating parameter is as follows: Arrange the variance of the front slope of each data point in the data change curve of the i-th operating parameter according to the temporal order of the data points to obtain the variance sequence of the front slope of the data change curve of the i-th operating parameter. Obtain several maxima from the variance sequence of the front slope. Start judging from the first maxima and set a threshold for the variance of the front slope. If the first maxima is greater than the threshold for the variance of the front slope, take the data point corresponding to the first maxima as the first dividing point of the data change curve of the i-th operating parameter. If the first maxima is less than or equal to the threshold for the variance of the front slope, continue judging the second maxima until the first dividing point is obtained. Arrange the variance of the slope of all data points after the first division point according to the temporal order of the data points to obtain the stable slope variance sequence of the data change curve of the i-th operating parameter. Obtain several maxima from the stable slope variance sequence. Start judging from the first maxima and set a slope observation threshold. If the first maxima is greater than the slope variance threshold, take the data point corresponding to the first maxima as the second division point of the data change curve of the i-th operating parameter. If the first maxima is less than or equal to the slope variance threshold, continue judging the second maxima until the second division point is obtained.

5. The method for monitoring the operation of a filter bag dust collection system according to claim 3, characterized in that, The specific method for constructing the relationship matrix of each production-dust cleaning process is as follows: Based on the first and second division points of the i-th operating parameter, obtain the initial, middle, and later data change curves for the j-th production parameter, respectively; obtain the Pearson correlation coefficients for the initial, middle, and later data change curves for the i-th operating parameter and the j-th production parameter, respectively. Preset initial, intermediate, and late weights, corresponding to the data change curves of the initial, intermediate, and late stages, respectively. The Pearson correlation coefficients of the three data change curves are weighted and summed using the corresponding weights, and the sum is used as the correlation value between the i-th operating parameter and the j-th production parameter in the production-dust cleaning process. Obtain the correlation values ​​between each operating parameter and each production parameter during the production-dust cleaning process, and construct a matrix. In the matrix, the same row represents the correlation values ​​between the same operating parameter and each production parameter, and the same column represents the correlation values ​​between each operating parameter and the same production parameter. The resulting matrix is ​​used as the relationship matrix for the production-dust cleaning process.

Images