Threshing and redrying production line cloth bag dust removal monitoring method and device, medium and system

Abstract

The invention discloses a threshing and redrying production line cloth bag dust removal monitoring method and device, a medium and a system, relates to the technical field of dust removal, and mainly aims at solving the problem that existing threshing and redrying production line cloth bag dust removal equipment is high in energy consumption. The method mainly comprises the following steps: acquiring multi-dimensional monitoring data, acquired by a plurality of different types of monitoring sensors, of the bag-type dust removal equipment; determining a fault monitoring result of each target monitoring part according to the multi-dimensional monitoring data, and performing prediction processing on the operation environment data through a trained dust hardening degree prediction model to obtain a dust hardening degree; according to the fault monitoring result and the dust hardening degree, blowing dust removal control is carried out on the cloth bag dust removal equipment; and generating monitoring display data according to the fault monitoring result, the dust hardening degree and the injection ash removal execution period, and outputting the monitoring display data to a display terminal. The device is mainly used for monitoring the fault of the bag-type dust removal equipment of the threshing and redrying production line and controlling the blowing period.

Description

Technical Field

[0001] The present invention relates to the technical field of dust removal, and in particular to a bag dust removal monitoring method, device, medium and system for a leaf threshing and redrying production line. Background Art

[0002] The threshing and redrying production line mainly involves the processes of tobacco leaf stem separation, tobacco leaf redrying, and tobacco leaf packing. Due to the crushing of tobacco leaves and dust raised during tobacco leaf transportation, a large amount of dust is generated in the production environment. Each production area is now equipped with a dust removal duct, which sucks the dust away through the dust removal duct and collects it into the dust removal main pipe. The dust is adsorbed to the dust removal equipment through the dust removal main pipe. The dust collector is equipped with a pulse solenoid valve, which controls the compressed air to blow into the bag at a fixed time, so that the dust on the bag is shaken off to the ash discharger. The dust concentration in the production environment is closely related to the quality of tobacco raw materials, the processing technology level of the production line, the temperature and humidity of the production environment, and the wind speed. Therefore, monitoring the bag dust removal of the threshing and redrying production line plays a vital role in ensuring production.

[0003] The existing dust removal system has a single control method, with the bagging frequency and cycle set as unified parameters. This results in energy waste when the dust concentration in the production environment is low, and poor dust removal effect when the dust concentration in the production environment is high. The blower is also damaged by prolonged exposure to large amounts of dust. At the same time, it is impossible to promptly and accurately detect potential equipment failures (such as early abnormalities in the solenoid coil of the pulse valve and minor damage to the diaphragm of the pneumatic diaphragm valve). Furthermore, it is impossible to dynamically adjust the spray cleaning strategy based on the actual operating status of the equipment and the dust compaction situation, resulting in high energy consumption of the dust removal equipment. Summary of the Invention

[0004] In view of this, the present invention provides a bag dust removal monitoring method, device, medium and system for a leaf threshing and redrying production line, the main purpose of which is to solve the problem of high energy consumption of the bag dust removal equipment in the existing leaf threshing and redrying production line.

[0005] According to one aspect of the present invention, a bag dust removal monitoring method for a leaf threshing and redrying production line is provided, comprising:

[0006] Acquire multi-dimensional monitoring data of the bag dust removal equipment collected by multiple different types of monitoring sensors, wherein the multi-dimensional monitoring data includes equipment operation data and operating environment data;

[0007] Determining the fault monitoring results of each target monitoring component based on the multi-dimensional monitoring data, and performing predictive processing on the operating environment data using the trained dust hardening degree prediction model to obtain the dust hardening degree;

[0008] Controlling the dust removal equipment with spraying and cleaning according to the fault monitoring result and the degree of dust compaction;

[0009] Monitoring display data is generated based on the fault monitoring result, the dust compaction degree and the spray cleaning execution cycle, and the monitoring display data is output to a display terminal.

[0010] Furthermore, the fault monitoring results include pulse valve solenoid coil fault monitoring results, pneumatic diaphragm valve diaphragm damage monitoring results, bag damage monitoring results and centrifugal fan fault monitoring results; the equipment operation data includes the pulse signals of all pulse valves, the compressed air storage tank pressure value, the fan vibration frequency and the fan temperature; the operating environment data includes the internal and external pressure difference of the bag and the dust concentration;

[0011] The process of determining the pulse valve electromagnetic coil fault monitoring result includes:

[0012] The pulse valves are sorted according to the order in which their actions are executed, and the pulse signal states of the pulse valves are polled according to their order. For any pulse valve, if within a first preset time period, the pulse signal states of both the preceding and succeeding pulse valves of the pulse valve change, and the pulse signal state of the pulse valve does not change, then it is determined that the pulse valve solenoid coil fault monitoring result is that the solenoid coil of the pulse valve is abnormal.

[0013] The process of determining the pneumatic diaphragm valve diaphragm damage monitoring result includes:

[0014] If the pressure values ​​of all compressed air storage tanks collected within the second preset time period are less than the first preset pressure threshold, and the pressure difference between the inside and outside of all bags is greater than the second preset pressure threshold, then the pneumatic diaphragm valve diaphragm damage monitoring result is determined to be a pneumatic diaphragm valve diaphragm damage;

[0015] The process of determining the bag damage monitoring result includes:

[0016] If all dust concentrations collected within the third preset time period are greater than the preset dust concentration threshold, and all bag internal and external pressure differences are less than the third preset pressure threshold, the bag damage monitoring result is determined to be bag damage;

[0017] The process of determining the centrifugal fan fault monitoring result includes:

[0018] If all dust concentrations collected within the fourth preset time period are greater than the preset dust concentration threshold, all fan vibration frequencies are greater than the preset frequency threshold and / or all fan temperatures are greater than the preset temperature threshold, the centrifugal fan fault monitoring result is determined to be an abnormal fan operating status.

[0019] Furthermore, before performing predictive processing on the operating environment data using the trained dust compaction degree prediction model, the method further includes:

[0020] Constructing an initial dust compaction degree prediction model based on nonlinear kernel function or neural network;

[0021] Based on the subdivided ranges of wind speed, humidity, and dust concentration in the dust removal duct, dust compaction degree data under different parameter combinations are collected in multiple time intervals, and a training sample set is constructed; wherein the different parameter combinations include the same wind speed and humidity with different dust concentrations, the same wind speed and different humidity with the same dust concentration, the same wind speed and different humidity with different dust concentrations, different wind speeds and different humidity with the same dust concentration, and different wind speeds and humidity with the same dust concentration.

[0022] The initial dust compaction degree prediction model is trained using the training sample set to obtain a trained dust compaction degree prediction model.

[0023] Furthermore, the trained dust hardening degree prediction model includes a multi-head attention layer, a convolutional neural network layer, and an activation function layer. The operating environment data includes operating environment data corresponding to each of the multiple bags. The trained dust hardening degree prediction model is used to predict the operating environment data to obtain the dust hardening degree, including:

[0024] Key features are extracted from the operating environment data of each bag, and multiple key features of each bag at different times are obtained. The key features include abnormal pressure drop rate, compressed air pressure decay, and dust concentration correlation features.

[0025] Constructing a multidimensional environmental data time series matrix based on the operating environment data corresponding to each of the plurality of bags and the timestamps of the operating environment data, and identifying the time dimension weight of the multidimensional environmental data time series matrix through the multi-head attention layer to determine the key decision interval;

[0026] Extracting spatial correlation features between the plurality of key features at different time moments through a target filter of a one-dimensional convolutional neural network layer, wherein the target filter has a maximum response to a combined feature of a drop rate of the differential pressure signal and a pressure peak attenuation;

[0027] Through the activation function layer, probability calculation is performed on the spatial correlation features within the key decision interval to obtain the dust compaction probability of different bags, and the degree of dust compaction is determined based on the dust compaction probability.

[0028] Furthermore, the dust compaction degree includes mild, moderate and severe, and the spray cleaning control of the bag dust removal equipment based on the fault monitoring result and the dust compaction degree includes:

[0029] When the fault monitoring result indicates that no fault exists, if the dust compaction degree is mild and the pressure difference between the inside and outside of the bag is greater than or equal to the second preset pressure threshold, the solenoid coil of the pulse valve is triggered to perform the first spray cleaning operation;

[0030] When the fault monitoring result indicates that no fault exists, if the dust compaction degree is severe and the pressure difference between the inside and outside of the bag is greater than or equal to the third preset pressure threshold, the solenoid coil of the pulse valve is triggered to perform the second spray cleaning operation;

[0031] When the fault monitoring result indicates that no fault exists, if the dust compaction degree is moderate and the pressure difference between the inside and outside of the bag is greater than or equal to the fourth preset pressure threshold, the solenoid coil of the pulse valve is triggered to perform the third spray cleaning operation;

[0032] The second preset pressure threshold is greater than the third preset pressure threshold, and the fourth preset pressure threshold is an average of the second preset pressure threshold and the third preset pressure threshold.

[0033] Furthermore, the process of determining the bag damage monitoring result further includes:

[0034] Extracting correlation features between the blowing event and the dust concentration peak value and contradiction features of the differential pressure signal based on the multi-dimensional monitoring data;

[0035] Based on the correlation characteristics between the blowing event and the dust concentration peak and the contradictory characteristics of the differential pressure signal, a deep learning model is used to perform spatiotemporal feature matching to obtain a feature contribution ranking, and feature fusion is performed based on the feature contribution ranking to obtain a fused feature vector;

[0036] According to the feature contribution ranking and fusion feature vectors, the bag breakage posterior probability is calculated through the Bayesian framework;

[0037] A bag damage monitoring result is generated according to the bag damage posterior probability.

[0038] Furthermore, the process of determining the pulse valve solenoid coil fault monitoring result further includes:

[0039] Extracting the pulse valve pressure waveform, pressure difference recovery rate and dust concentration based on the multi-dimensional monitoring data;

[0040] Performing a horizontal comparative analysis of multiple pulse valves based on the pressure rise time in the pulse valve pressure waveform, the pressure difference recovery rate, and the dust concentration to determine the pulse valve with expected failure;

[0041] Performing DTW matching on the pulse valve pressure waveform of the expected fault pulse valve to obtain a fault label of the expected fault pulse valve;

[0042] The failure probability of the expected fault pulse valve is obtained by Bayesian probability calculation according to the pulse valve pressure waveform, the pressure difference recovery rate and the fault label of the expected fault pulse valve.

[0043] According to another aspect of the present invention, a bag dust removal monitoring device for a leaf threshing and redrying production line is provided, comprising:

[0044] A sensor data acquisition module, configured to acquire multi-dimensional monitoring data of the bag dust removal equipment collected by a plurality of different types of monitoring sensors, wherein the multi-dimensional monitoring data includes equipment operation data and operation environment data;

[0045] a data processing module for determining the fault monitoring results of each target monitoring component based on the multi-dimensional monitoring data, and performing predictive processing on the operating environment data using the trained dust compaction degree prediction model to obtain the dust compaction degree;

[0046] A control module, configured to control the dust removal equipment by spraying and cleaning dust based on the fault monitoring result and the degree of dust compaction;

[0047] The display module is used to generate monitoring display data based on the fault monitoring results, the dust compaction degree and the spray cleaning execution cycle, and output the monitoring display data to the display terminal.

[0048] According to another aspect of the present invention, a medium is provided, in which at least one executable instruction is stored, and the executable instruction enables a processor to execute operations corresponding to the bag dust removal monitoring method for the leaf threshing and redrying production line as described above.

[0049] According to another aspect of the present invention, a system is provided, comprising: a plurality of different types of monitoring sensors, a bag dust removal device, a storage server, a monitoring platform server, a controller, a display terminal, and a communication bus, wherein the bag dust removal device comprises a centralized dust collection duct, a pulse valve, an air storage tank, an exhaust dust duct, a centrifugal fan, and dust removal bags; the monitoring sensors comprise a dust concentration sensor, a humidity sensor, and a wind speed sensor installed in the centralized dust collection duct, a pulse valve signal collector installed on the pulse valve, a pressure sensor installed on the air storage tank, a temperature and vibration sensor installed on the centrifugal fan, and a pressure difference sensor installed on the dust removal bag; the multi-dimensional monitoring data comprises equipment operation data and operating environment data;

[0050] The monitoring sensors are respectively connected to the storage servers, and the storage servers, the monitoring platform servers, the controllers and the display terminals communicate with each other via the communication bus;

[0051] The monitoring platform server performs operations corresponding to the above-mentioned bag dust removal monitoring method for the leaf threshing and redrying production line.

[0052] By means of the above technical solution, the technical solution provided by the embodiment of the present invention has at least the following advantages:

[0053] The present invention provides a bag dust removal monitoring method, device, medium and system for a leaf beating and redrying production line. An embodiment of the present invention obtains multi-dimensional monitoring data of the bag dust removal equipment collected by multiple different types of monitoring sensors, wherein the multi-dimensional monitoring data includes equipment operation data and operation environment data; determines the fault monitoring results of each target monitoring component based on the multi-dimensional monitoring data, and predicts and processes the operation environment data through a trained dust compaction degree prediction model to obtain the dust compaction degree; controls the bag dust removal equipment by spraying and cleaning according to the fault monitoring results and the dust compaction degree; generates monitoring display data based on the fault monitoring results, the dust compaction degree and the spray cleaning execution cycle, and outputs the monitoring display data to a display terminal, thereby realizing real-time monitoring of faults and dynamic control of the spray cleaning cycle, greatly reducing the probability of the equipment being exposed to dust for a long time, reducing the equipment energy consumption, and at the same time ensuring the effectiveness of dust removal control, thereby greatly improving the dust removal efficiency.

[0054] The above description is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are specifically listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Various other advantages and benefits will become apparent to those skilled in the art upon reading the detailed description of the preferred embodiment below. The accompanying drawings are for illustration purposes only and are not to be considered as limiting the present invention. The same reference symbols are used throughout the drawings to represent the same components. In the drawings:

[0056] Figure 1 A flow chart of a bag dust removal monitoring method for a leaf threshing and redrying production line provided by an embodiment of the present invention is shown;

[0057] Figure 2 A schematic diagram of the installation position of a monitoring sensor provided by an embodiment of the present invention is shown;

[0058] Figure 3 The present invention provides a block diagram of a bag dust removal monitoring device for a leaf threshing and redrying production line. DETAILED DESCRIPTION

[0059] Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0060] Aiming at the problem of high energy consumption of bag dust removal equipment in existing leaf threshing and redrying production line. The embodiment of the present invention provides a bag dust removal monitoring method for leaf threshing and redrying production line, such as Figure 1 As shown, the method includes:

[0061] 101. Obtain multi-dimensional monitoring data of the bag dust removal equipment collected by multiple different types of monitoring sensors.

[0062] In the embodiment of the present invention, a plurality of different types of monitoring sensors include dust concentration sensors, humidity sensors, wind speed sensors, differential pressure sensors, pressure sensors, temperature vibration sensors, and pulse valve information collectors. The number of sensors of each type is not limited to one. Figure 2 As shown, the bag dust removal equipment includes a centralized dust collection duct 1, a pulse valve 5, an air storage tank 7, an exhaust dust duct 10, a centrifugal fan 11 and a dust bag 13. The dust concentration sensor 3 is installed in the centralized dust collection duct 1 and the exhaust dust duct 10, the humidity sensor 4 and the wind speed sensor 2 are installed in the centralized dust collection duct 1, the pulse valve signal collector 6 is installed on the pulse valve 5, the pressure sensor 8 is installed on the air storage tank, the temperature vibration sensor 12 is installed on the centrifugal fan 11, and the pressure difference sensor 14 is installed on the dust bag 13. Each monitoring sensor is connected to a storage server to transmit the collected data to the storage server for cleaning, storage and preliminary calculation. The monitoring platform server is connected to the storage server and a programmable logic controller (PLC) controller respectively. The PLC controller is used to control the corresponding components in the bag dust removal equipment. The monitoring platform server performs data analysis based on the received multi-dimensional monitoring data to monitor the failure of each component of the equipment and the spray cleaning cycle. Multidimensional monitoring data includes sensor data collected by different monitoring sensors, which can be roughly divided into two categories: equipment operation data and operating environment data.

[0063] 102. Determine the fault monitoring results of each target monitoring component based on the multi-dimensional monitoring data, and perform prediction processing on the operating environment data through the trained dust compaction degree prediction model to obtain the dust compaction degree.

[0064] In an embodiment of the present invention, the monitoring platform server collects data of multiple dimensions of the dust removal equipment through monitoring sensors installed in different positions, such as the dust concentration in the exhaust / dust duct, the humidity of different dust removal ducts, wind speed, the pressure difference between the inside and outside of the bag, the pressure of the compressed air storage tank, etc., and then monitors the fault and degree of compaction based on these data. The monitored faults may specifically include whether the pulse valve is damaged during the operation of the dust removal equipment, whether the pneumatic film valve membrane is damaged, whether the filter bag is damaged, whether the centrifugal fan is faulty, etc. Fault monitoring can quickly locate equipment faults, reduce downtime maintenance costs, and improve equipment maintenance efficiency. By monitoring the degree of dust compaction, the spray cleaning cycle can be controlled according to the degree of dust compaction and the pressure difference between the inside and outside of the bag, effectively improving the bag cleaning effect and reducing energy consumption.

[0065] 103. Control the dust spraying and cleaning of the bag dust removal equipment based on the fault monitoring result and the degree of dust compaction.

[0066] In embodiments of the present invention, fault monitoring results can reflect the operating status of various components of the bag filter equipment in real time. If monitoring results indicate a fault or abnormality in a component, the system can promptly adjust the spray cleaning strategy. For example, the spray frequency can be reduced for the faulty component to prevent further damage, or the spray frequency can be increased to clear potential blockages and ensure normal operation of the equipment. The degree of dust compaction reflects the accumulation of dust on the filter bags. A high degree of dust compaction indicates decreased air permeability and dust removal efficiency. In this case, the frequency and intensity of spray cleaning should be increased to effectively remove dust from the filter bags and restore air permeability. Conversely, when the degree of dust compaction is low, the frequency of spray cleaning can be appropriately reduced to save energy and extend the service life of the filter bags. By combining the fault monitoring results and the degree of dust compaction, the system can intelligently control the spray cleaning process. For example, the spray cleaning frequency and intensity can be set according to the degree of dust compaction, and the spray cleaning strategy can be dynamically adjusted based on the fault monitoring results to achieve optimal cleaning results and equipment protection.

[0067] 104. Generate monitoring display data based on the fault monitoring result, the dust compaction degree and the spray cleaning execution cycle, and output the monitoring display data to a display terminal.

[0068] In embodiments of the present invention, monitoring and display data can include graphical displays of real-time monitoring data, historical monitoring data, monitoring data trend changes, monitoring data analysis results, and early warning data. The interactive interface of the display terminal provides functions such as query, viewing, and parameter setting. These interfaces include monitoring interfaces for pulse valve solenoid coils, pneumatic diaphragm valve diaphragms, filter bags and their position, centrifugal fans, and spray cleaning cycle monitoring. Each interface displays relevant monitoring sensor data, early warning information, analysis results, and parameter setting functions.

[0069] In one embodiment of the present invention, for further explanation and limitation, the fault monitoring results include the pulse valve solenoid coil fault monitoring results, the pneumatic diaphragm valve diaphragm damage monitoring results, the bag damage monitoring results and the centrifugal fan fault monitoring results; the equipment operation data includes the pulse signals of all pulse valves, the compressed air storage tank pressure value, the fan vibration frequency and the fan temperature; the operating environment data includes the pressure difference between the inside and outside of the bag and the dust concentration.

[0070] The process of determining the pulse valve electromagnetic coil fault monitoring result includes:

[0071] The pulse valves are sorted according to the order in which their actions are executed, and the pulse signal states of the pulse valves are polled according to their order. For any pulse valve, if within a first preset time period, the pulse signal states of both the preceding and succeeding pulse valves of the pulse valve change, and the pulse signal state of the pulse valve does not change, then it is determined that the pulse valve solenoid coil fault monitoring result is that the solenoid coil of the pulse valve is abnormal.

[0072] In the embodiment of the present invention, the pulse signal collector converts the physical action into a digital signal to establish a "0 / 1" binary coding system, "0": the pulse valve is in the closed state; "1": the pulse valve has completed the opening action. Each pulse valve is given a unique number H1, H2, ..., H u (u∈N), forming a digital identification system. During the equipment operation cycle T, the pulse valve performs actions in sequence according to the preset process flow, forming a certain time sequence logic chain: H1→H2→H3→…→H u (The actual sequence may be any process sequence such as H1→H3→H2…) According to the actual action timing within the current operation cycle T, the pulse valve execution sequence is dynamically established, and the digital signal status of each pulse valve is detected in sequence according to the sorted order to form an inspection queue. Set the first preset time length Δt (need to meet Δt <T),作为状态变化观测窗口,对任一目标脉冲阀H i , the fault judgment is triggered when the following conditions are met: a) Leading constraint: within Δt, its preceding pulse valve H i-1The signal state of the 0→1 jump occurs b) Subsequent constraint: within Δt, the subsequent pulse valve H i+1 The signal state of the 0→1 jump occurs c) Self-state: During the Δt detection period, H i The signal state of the target valve always remains constant (0→0 or 1→1). When the adjacent valves in the process timing chain are operating normally, the signal state of the target valve should also jump. If it does not, it is determined that the electromagnetic coil of the current pulse valve is abnormal.

[0073] The process of determining the pneumatic diaphragm valve diaphragm damage monitoring result includes:

[0074] If the pressure values ​​of all compressed air storage tanks collected within the second preset time period are less than the first preset pressure threshold, and the pressure difference between the inside and outside of all bags is greater than the second preset pressure threshold, then the pneumatic diaphragm valve diaphragm damage monitoring result is determined to be a pneumatic diaphragm valve diaphragm damage;

[0075] In an embodiment of the present invention, the first preset pressure threshold is the minimum allowable pressure value, which serves as the safety lower limit of the pressure of the compressed air storage tank. When the actual pressure is lower than this value, it indicates that there is a leak in the compressed air storage tank. The second preset pressure threshold is the maximum allowable pressure difference value, which serves as the safety upper limit of the pressure difference between the inside and outside of the bag. When the actual pressure difference exceeds this value, it indicates that the dust cleaning effect does not meet expectations. The second preset time length is the data acquisition window, which is the observation window for status judgment. This time length must cover at least one complete blowing cycle, and the specific value can be customized according to needs. If the compressed air system continues to be under low pressure within the second preset time length, and the pressure difference between the inside and outside of the bag continues to exceed the limit, it indicates that there is an air leakage and the valve cannot be fully opened. It can be determined that the diaphragm of the pneumatic diaphragm valve is damaged.

[0076] The process of determining the bag damage monitoring result includes:

[0077] If all dust concentrations collected within the third preset time period are greater than the preset dust concentration threshold, and all bag internal and external pressure differences are less than the third preset pressure threshold, the bag damage monitoring result is determined to be bag damage;

[0078] In the embodiment of the present application, the preset dust concentration threshold is the maximum allowable concentration value, which is the safety upper limit of dust emission of the exhaust duct. When the actual concentration exceeds this value, it indicates that the dust removal equipment has an abnormal leakage. The third preset pressure threshold is the minimum allowable pressure difference value, which is the safety lower limit of the pressure difference inside and outside the bag. When the actual pressure difference is lower than this value, it indicates that the filtration resistance of the bag is abnormally low. The third preset time length is the observation window for state judgment, which should cover the complete filtration-ash cleaning cycle of the bag, and the specific value can be customized according to the requirements. If the dust concentration continues to exceed the standard and the pressure difference inside and outside the bag continues to be low within the third preset time length, it indicates that there is both air leakage short circuit and dust layer cannot be attached, and it can be determined that the bag body is damaged.

[0079] The determination process of the centrifugal fan fault monitoring result includes:

[0080] If all the dust concentrations collected within the fourth preset time length are greater than the preset dust concentration threshold, all the fan vibration frequencies are greater than the preset frequency threshold, and / or all the fan temperatures are greater than the preset temperature threshold, it is determined that the centrifugal fan fault monitoring result is abnormal running state of the fan.

[0081] In the embodiment of the present application, the preset frequency threshold is the maximum allowable vibration frequency value, and the preset temperature threshold is the maximum allowable temperature value. The fourth preset time length is the observation window for state judgment, which should cover the complete filtration-ash cleaning cycle of the bag, and the specific value can be customized according to the requirements. When the fan vibration frequency continues to be greater than the preset frequency threshold, the fan temperature continues to be greater than the preset temperature threshold, and the dust concentration continues to be greater than the preset dust concentration threshold within the fourth preset time length, it indicates that the centrifugal fan may have a dust accumulation condition.

[0082] In one embodiment of the present application, in order to further illustrate and limit, before the step of performing prediction processing on the running environment data by the trained dust agglomeration degree prediction model, the method further includes:

[0083] An initial dust agglomeration degree prediction model is constructed based on a nonlinear kernel function or a neural network;

[0084] Dust agglomeration degree data under different parameter combinations in multiple time intervals are collected based on the subdivided ranges of the dust removal air pipe air speed, humidity and dust concentration, and a training sample set is constructed;

[0085] The initial dust agglomeration degree prediction model is trained using the training sample set to obtain a trained dust agglomeration degree prediction model.

[0086] In an embodiment of the present invention, the initial dust hardening degree prediction model can be constructed based on a nonlinear kernel function or a neural network. The nonlinear kernel function can be a support vector regression model, a decision tree regression model, etc., and the neural network can be a multi-layer perceptron model, a convolutional neural network model, etc. The construction of the training sample first performs small-scale segmentation based on the historical wind speed range, humidity range, and dust concentration range, and then collects data based on the segmented range, including the dust hardening degree in different time intervals with the same wind speed and humidity in the dust removal duct; the dust hardening degree in different time intervals with the same wind speed and humidity in the dust removal duct; the dust hardening degree in different time intervals with the same wind speed and humidity in the dust removal duct; the dust hardening degree in different time intervals with the same wind speed and humidity in the dust removal duct; the dust hardening degree in different time intervals with the same dust removal duct and the same dust concentration; the dust hardening degree in different time intervals with different wind speeds and humidity in the dust removal duct and the same dust concentration. The same means being in the same segmented range. That is, different parameter combinations include the same wind speed and humidity but different dust concentrations, the same wind speed and different humidity but the same dust concentration, the same wind speed and different humidity but different dust concentrations, different wind speeds and different humidity but the same dust concentration, and the same wind speed and humidity but the same dust concentration.

[0087] In one embodiment of the present invention, for further explanation and limitation, the step of predicting the operating environment data using the trained dust hardening degree prediction model to obtain the dust hardening degree includes:

[0088] Key features are extracted from the operating environment data of each bag to obtain multiple key features of each bag at different times;

[0089] Constructing a multidimensional environmental data time series matrix based on the operating environment data corresponding to each of the plurality of bags and the timestamps of the operating environment data, and identifying the time dimension weight of the multidimensional environmental data time series matrix through the multi-head attention layer to determine the key decision interval;

[0090] Extracting spatial correlation features between the multiple key features at different time moments through the target filter of the one-dimensional convolutional neural network layer;

[0091] Through the activation function layer, probability calculation is performed on the spatial correlation features within the key decision interval to obtain the dust compaction probability of different bags, and the degree of dust compaction is determined based on the dust compaction probability.

[0092] In an embodiment of the present invention, the trained dust compaction degree prediction model is constructed based on a convolutional neural network. The convolutional neural network is preceded by a key feature extraction layer and a multi-head attention layer. Key features include abnormal pressure drop rate, compressed air pressure decay, and dust concentration correlation features. Among them, the determination process of abnormal pressure drop rate specifically includes: based on the pressure difference change data inside and outside each bag for a period of time after the blowing operation, the pressure difference drop rate can be calculated. Based on the normal operating parameters of the equipment and experience summary, it is concluded that the pressure difference after blowing should drop by 200-500Pa within 5 seconds, and the pressure difference drop rate threshold is -200Pa / s. When the pressure difference drop rate of a certain bag is less than the pressure difference drop rate threshold, such as only dropping by 50Pa within 5 seconds, the pressure difference drop rate is -10Pa / s, indicating that the pressure difference drop speed of the current bag after blowing is obviously too slow, it is determined that the pressure difference drop rate is abnormal, and there may be a problem affecting the bag cleaning effect. Compressed air pressure decay can be caused by problems such as blockage in the compressed air pipeline, valve failure, or undersized pipe diameter. It can also be caused by a malfunction in the injection device (e.g., nozzle), which can affect the compressed air injection effect and prevent the effective transfer of energy to the bag surface. If the peak pressure of a bag during injection is only 0.68 MPa, lower than the normal 0.75 MPa, and the energy integral decreases by 20%, then this bag exhibits the key characteristic of compressed air pressure decay. A damaged bag allows some dust to pass through without being effectively filtered. Improper bag installation can also result in gaps, affecting filtration efficiency and leading to increased dust concentration at the outlet. Therefore, dust concentration serves as a key feature for bag damage identification, namely, a dust concentration correlation feature. After extracting each key feature, timestamp synchronization is used to align the key features representing the anomaly with the injection time of each bag for subsequent spatiotemporal feature correlation.

[0093] The critical decision interval is the key time period for determining anomalies. The multidimensional environmental data time series matrix can be sensor data including four dimensions: pressure, inlet differential pressure, outlet differential pressure, and dust concentration, arranged in time series to form a 60×4 matrix. The multi-head attention layer calculates the time dimension weights of different time periods in the multidimensional environmental data time series matrix, identifies which time period after the injection has the largest weight, and determines the time period with the largest weight as the critical decision interval. For example, if the weight of the time period from 2 to 4 seconds after the injection reaches 65% (maximum), then this time period is determined as the critical decision interval.

[0094] Then, spatial features are extracted through a one-dimensional convolutional neural network layer, wherein the one-dimensional convolutional neural network layer includes a target filter, which has the maximum response to the combined features of the differential pressure signal drop rate and the pressure peak attenuation. Among them, the target filter can be the 23rd filter of the first Conv1D layer (filter = 64, kernel_size = 5), so that it has the strongest response to the combined features of the differential pressure signal drop rate and the pressure peak attenuation. This target filter outputs different activation values ​​for different bags. When the activation value of a bag is significantly higher than the activation values ​​of other bags, it indicates that the characteristics of the current bag are spatially specific. Finally, through the Softmax output layer, multi-classification probability calculation is performed in combination with the key decision interval and spatial correlation features to obtain the failure probability of different bags. For example, the output layer contains 12 neurons, corresponding to the failure probabilities of 12 bags, among which the activation value Z5 of bag No. 5 is 3.2, and the activation values ​​Z of other bags are ≤3. j <1.0, the final calculated probability of bag 5 being blocked is: P bag5 =e 3.2 ÷(e 3.2 +9×e 1.0 ≈92.7%; it can be determined that bag No. 5 is compacted. This probability is then matched with the probability ranges corresponding to different degrees of compaction to determine the degree of compaction. The probability ranges corresponding to different degrees of compaction can be customized according to specific application requirements.

[0095] In one embodiment of the present invention, for further explanation and limitation, the spray cleaning control of the bag dust removal equipment based on the fault monitoring result and the dust compaction degree includes:

[0096] When the fault monitoring result indicates that no fault exists, if the dust compaction degree is mild and the pressure difference between the inside and outside of the bag is greater than or equal to the second preset pressure threshold, the solenoid coil of the pulse valve is triggered to perform the first spray cleaning operation;

[0097] When the fault monitoring result indicates that no fault exists, if the dust compaction degree is severe and the pressure difference between the inside and outside of the bag is greater than or equal to the third preset pressure threshold, the solenoid coil of the pulse valve is triggered to perform the second spray cleaning operation;

[0098] When the fault monitoring result indicates that no fault exists, if the dust compaction degree is moderate and the pressure difference between the inside and outside of the bag is greater than or equal to the fourth preset pressure threshold, the solenoid coil of the pulse valve is triggered to perform the third spray cleaning operation;

[0099] In an embodiment of the present invention, the degree of dust compaction includes mild, moderate and severe. According to the different degrees of dust compaction and the comparison result of the pressure difference between the inside and outside of the bag and the fourth preset pressure threshold, it is jointly judged whether it is necessary to trigger the pulse valve solenoid coil to operate. Of course, the premise of the judgment is that the current equipment needs to be fault-free, that is, the fault monitoring result is that there is no fault. The specific process of joint judgment includes: when the degree of compaction is mild and the pressure difference between the inside and outside of the bag reaches the second preset pressure threshold, the pulse valve solenoid coil is triggered to operate through the heavy sensor module; when the degree of compaction is moderate and the pressure difference between the inside and outside of the bag reaches the fourth preset pressure threshold, the pulse valve solenoid coil is triggered to operate through the heavy sensor module; when the degree of compaction is severe and the pressure difference between the inside and outside of the bag reaches the third preset pressure threshold, the pulse valve solenoid coil is triggered to operate through the heavy sensor module. Among them, the second preset pressure threshold is greater than the third preset pressure threshold, and the fourth preset pressure threshold is the average of the second preset pressure threshold and the third preset pressure threshold. The first spray cleaning operation, the second spray cleaning operation, and the third spray cleaning operation can be cleaning operations of the same intensity or cleaning operations of different intensities. They can be customized according to actual needs and are not specifically limited in the embodiments of the present invention.

[0100] In one embodiment of the present invention, for further explanation and limitation, the process of determining the bag damage monitoring result further includes:

[0101] Extracting correlation features between the blowing event and the dust concentration peak value and contradiction features of the differential pressure signal based on the multi-dimensional monitoring data;

[0102] Based on the correlation characteristics between the blowing event and the dust concentration peak and the contradictory characteristics of the differential pressure signal, a deep learning model is used to perform spatiotemporal feature matching to obtain a feature contribution ranking, and feature fusion is performed based on the feature contribution ranking to obtain a fused feature vector;

[0103] According to the feature contribution ranking and fusion feature vectors, the bag breakage posterior probability is calculated through the Bayesian framework;

[0104] A bag damage monitoring result is generated according to the bag damage posterior probability.

[0105] In an embodiment of the present invention, the process of extracting the correlation features between the blowing event and the dust concentration peak includes 1) timing alignment, that is, synchronizing the sensor clock through the NTP protocol to ensure that the time error between the dust concentration peak and the blowing signal of the No. 12 pulse valve is less than 0.1 seconds. 2) Cross-correlation verification: Calculate the cross-correlation function R(τ) of the blowing signal V(t) and the concentration signal C(t), quantify the temporal consistency between the blowing action and the concentration surge, establish a causal chain of "blowing → crack leakage → concentration surge", and convert the temporal correlation into a characteristic value. The contradictory characteristics of the differential pressure signal require determining the pressure drop rate. If it is within the normal range, the blockage fault is preliminarily ruled out. At this time, if the steady-state pressure difference is lower than the normal value, it indicates that the filtration resistance is abnormally reduced, and the airflow may be "short-circuited" due to damage. Then, the deep learning model is used to match the temporal and spatial characteristics of the correlation features between the blowing event and the dust concentration peak and the contradictory characteristics of the differential pressure signal to obtain a multi-dimensional feature contribution ranking and a fusion feature vector. Specifically, it includes: assigning 72% weight to the period of 0-2 seconds after the injection through the self-attention layer, capturing the transient coupling of the concentration surge and the pressure difference change, and determining the feature contribution through the gradient weighted class activation mapping (Grad-CAM). This contribution is used for the subsequent feature fusion weight. Features such as pressure P, vibration V, and differential pressure D are mapped to a unified latent space and weighted by learnable weights, where the weights are optimized by back propagation to reflect the difference in contribution of different modes to damage diagnosis. Finally, the feature contribution ranking and the fused feature vector are used as the input of the Bayesian algorithm. By converting the feature contribution ranking into a conditional probability product, and then using the product of the conditional probability product and the prior probability setting value as the posterior probability result, the posterior probability of bag damage is obtained.

[0106] In one embodiment of the present invention, for further explanation and limitation, the process of determining the pulse valve solenoid coil fault monitoring result further includes:

[0107] Extracting the pulse valve pressure waveform, pressure difference recovery rate and dust concentration based on the multi-dimensional monitoring data;

[0108] Performing a horizontal comparative analysis of multiple pulse valves based on the pressure rise time in the pulse valve pressure waveform, the pressure difference recovery rate, and the dust concentration to determine the pulse valve with expected failure;

[0109] Performing DTW matching on the pulse valve pressure waveform of the expected fault pulse valve to obtain a fault label of the expected fault pulse valve;

[0110] The failure probability of the expected fault pulse valve is obtained by Bayesian probability calculation according to the pulse valve pressure waveform, the pressure difference recovery rate and the fault label of the expected fault pulse valve.

[0111] In the embodiment of the present invention, the pulse valve pressure waveform can be further extracted to obtain the pressure rise time delay, pressure waveform kurtosis decrease (diaphragm failure will cause the waveform to flatten), and injection energy integral (area under the pressure curve). The pressure difference recovery rate and dust concentration can be combined to make a preliminary judgment on the pulse valve failure and obtain contradictory characteristics. For example, after a pulse valve is injected, the pressure difference only drops by 18% (normal should drop by 35%), but the dust concentration does not increase abnormally (normal <10mg / m 3 ), rule out bag blockage. If the bag is clogged, the pressure difference should continue to increase, and the dust concentration increases due to the decrease in filtration efficiency. The current situation is that the pressure difference recovery is insufficient but the dust concentration is normal. Therefore, it can be determined that the cleaning force is insufficient, and the pulse valve fault can be inferred. Then, the pressure rise time delay, pressure difference recovery rate and dust concentration of multiple pulse valves are horizontally compared. If a pulse valve has both pressure rise delay and pressure difference recovery abnormalities, and the dust concentration is normal, the current pulse valve is locked as an expected fault. At the same time, the fault label is determined by waveform matching, specifically including: aligning the pulse valve pressure waveform of the expected fault pulse valve with the waveform samples in the historical fault waveform library (diaphragm aging, diaphragm rupture, etc.) by dynamic time warping (DTW). If the minimum DTW distance corresponds to the "diaphragm aging" mode, the fault label is determined to be aging. If the minimum DTW distance corresponds to "diaphragm rupture", the fault label is determined to be rupture. Finally, for the pulse valve with expected fault, the failure probability is obtained by calculating the delay probability through Bayesian probability calculation based on the set prior probability and the previously determined pressure rise time delay, pressure waveform kurtosis decrease, injection energy integral and fault label.

[0112] The present invention provides a bag dust removal monitoring method for a leaf threshing and re-drying production line. An embodiment of the present invention obtains multi-dimensional monitoring data of the bag dust removal equipment collected by multiple different types of monitoring sensors, wherein the multi-dimensional monitoring data includes equipment operation data and operation environment data; determines the fault monitoring results of each target monitoring component based on the multi-dimensional monitoring data, and predicts and processes the operation environment data through a trained dust compaction degree prediction model to obtain the dust compaction degree; controls the bag dust removal equipment by spraying and cleaning according to the fault monitoring results and the dust compaction degree; generates monitoring display data based on the fault monitoring results, the dust compaction degree and the spray cleaning execution cycle, and outputs the monitoring display data to a display terminal, thereby realizing real-time monitoring of faults and dynamic control of the spray cleaning cycle, greatly reducing the probability of the equipment being exposed to dust for a long time, reducing the equipment energy consumption, and at the same time ensuring the effectiveness of dust removal control, thereby greatly improving the dust removal efficiency.

[0113] Furthermore, as a response to the above Figure 1The embodiment of the present invention provides a bag dust removal monitoring device for a leaf threshing and redrying production line, such as Figure 3 As shown, the device includes:

[0114] A sensor data acquisition module 31 is used to acquire multi-dimensional monitoring data of the bag dust removal equipment collected by multiple different types of monitoring sensors, wherein the multi-dimensional monitoring data includes equipment operation data and operation environment data;

[0115] A data processing module 32 is configured to determine the fault monitoring results of each target monitoring component based on the multi-dimensional monitoring data, and to predict the operating environment data using the trained dust hardening degree prediction model to obtain the degree of dust hardening;

[0116] A control module 33 is configured to control the dust removal equipment by spraying and cleaning according to the fault monitoring result and the degree of dust compaction;

[0117] The display module 34 is used to generate monitoring display data based on the fault monitoring result, the dust compaction degree and the spray cleaning execution cycle, and output the monitoring display data to the display terminal.

[0118] Furthermore, the data processing module includes:

[0119] A pulse valve solenoid coil fault monitoring unit is used to sort the pulse valves according to the order of action execution between the pulse valves, and poll the pulse signal status of the pulse valves according to the sequence of the pulse valves; for any pulse valve, if within a first preset time period, the pulse signal status of the pulse valve in the preceding sequence and the pulse valve in the succeeding sequence of the pulse valve change, and the pulse signal status of the pulse valve does not change, then the pulse valve solenoid coil fault monitoring result is determined to be an abnormality of the solenoid coil of the pulse valve;

[0120] a pneumatic diaphragm valve diaphragm damage monitoring result unit, configured to determine that the pneumatic diaphragm valve diaphragm damage monitoring result is pneumatic diaphragm valve diaphragm damage if all compressed air storage tank pressure values ​​collected within a second preset time period are less than a first preset pressure threshold, and all bag internal and external pressure differentials are greater than a second preset pressure threshold;

[0121] a bag damage monitoring unit, configured to determine that the bag damage monitoring result is bag damage if all dust concentrations collected within a third preset time period are greater than a preset dust concentration threshold and all internal and external pressure differences of the bags are less than a third preset pressure threshold;

[0122] The centrifugal fan fault monitoring unit is used to determine that the centrifugal fan fault monitoring result is an abnormal fan operating status if all dust concentrations collected within the fourth preset time period are greater than the preset dust concentration threshold, all fan vibration frequencies are greater than the preset frequency threshold and / or all fan temperatures are greater than the preset temperature threshold.

[0123] Furthermore, the device further comprises:

[0124] A model building module is used to build an initial dust compaction degree prediction model based on a nonlinear kernel function or a neural network;

[0125] A sample construction module is used to collect dust compaction degree data under different parameter combinations in multiple time intervals based on the subdivided ranges of wind speed, humidity and dust concentration of the dust removal air duct, and to construct a training sample set; wherein the different parameter combinations include the same wind speed and the same humidity but different dust concentrations, the same wind speed and different humidity but the same dust concentration, the same wind speed and different humidity but different dust concentrations, different wind speeds and different humidity but the same dust concentration, and different wind speeds and the same humidity but the same dust concentration;

[0126] The training module is used to train the initial dust compaction degree prediction model using the training sample set to obtain a trained dust compaction degree prediction model.

[0127] Furthermore, the data processing module further includes:

[0128] A key feature extraction unit is used to extract key features from the operating environment data of each bag, and obtain multiple key features of each bag at different times, wherein the key features include abnormal pressure drop rate, compressed air pressure decay and dust concentration correlation features;

[0129] a determination unit, configured to construct a multidimensional environmental data time series matrix based on the operating environment data corresponding to each of the plurality of bags and the timestamps of the operating environment data, and identify the time dimension weight of the multidimensional environmental data time series matrix through the multi-head attention layer to determine a key decision interval;

[0130] a spatial correlation feature extraction unit, configured to extract spatial correlation features between the plurality of key features at different time moments through a target filter of a one-dimensional convolutional neural network layer, wherein the target filter has a maximum response to a combined feature of a drop rate of the differential pressure signal and a pressure peak attenuation;

[0131] The calculation unit is used to perform probability calculation on the spatial correlation features within the key decision interval through the activation function layer, obtain the dust compaction probability of different bags, and determine the degree of dust compaction based on the dust compaction probability.

[0132] Furthermore, the control module includes:

[0133] a first control unit, configured to, when the fault monitoring result indicates that no fault exists, trigger the solenoid coil of the pulse valve to operate so as to perform a first spray cleaning operation if the dust compaction degree is mild and the pressure difference between the inside and outside of the bag is greater than or equal to a second preset pressure threshold;

[0134] a second control unit, configured to, when the fault monitoring result indicates that no fault exists, trigger the solenoid coil of the pulse valve to operate so as to perform a second spray cleaning operation if the dust compaction degree is severe and the pressure difference between the inside and outside of the bag is greater than or equal to a third preset pressure threshold;

[0135] a third control unit, configured to, when the fault monitoring result indicates that no fault exists, trigger the solenoid coil of the pulse valve to operate so as to perform a third spray cleaning operation if the dust compaction degree is moderate and the pressure difference between the inside and outside of the bag is greater than or equal to a fourth preset pressure threshold;

[0136] The second preset pressure threshold is greater than the third preset pressure threshold, and the fourth preset pressure threshold is an average of the second preset pressure threshold and the third preset pressure threshold.

[0137] Furthermore, the device further comprises:

[0138] A first extraction module is used to extract the correlation characteristics between the blowing event and the dust concentration peak and the contradiction characteristics of the differential pressure signal based on the multi-dimensional monitoring data;

[0139] A feature matching module is used to perform spatiotemporal feature matching based on the correlation features between the blowing event and the dust concentration peak and the contradictory features of the differential pressure signal through a deep learning model to obtain a feature contribution ranking, and perform feature fusion based on the feature contribution ranking to obtain a fused feature vector;

[0140] A first calculation module is used to sort and fuse feature vectors according to the feature contribution, and calculate the bag damage posterior probability through a Bayesian framework;

[0141] A generation module is used to generate a bag damage monitoring result based on the bag damage posterior probability.

[0142] Furthermore, the device further comprises:

[0143] A second extraction module is used to extract the pulse valve pressure waveform, pressure difference recovery rate and dust concentration based on the multi-dimensional monitoring data;

[0144] a determination module, configured to perform a horizontal comparative analysis of multiple pulse valves based on the pressure rise time in the pulse valve pressure waveform, the pressure difference recovery rate, and the dust concentration, to determine the pulse valve with an expected fault;

[0145] A waveform matching module is used to perform DTW matching on the pulse valve pressure waveform of the expected fault pulse valve to obtain a fault label of the expected fault pulse valve;

[0146] The second calculation module is used to obtain the failure probability of the expected fault pulse valve through Bayesian probability calculation based on the pulse valve pressure waveform, the pressure difference recovery rate and the fault label of the expected fault pulse valve.

[0147] The present invention provides a bag dust removal monitoring device for a leaf beating and redrying production line. An embodiment of the present invention obtains multi-dimensional monitoring data of the bag dust removal equipment collected by multiple different types of monitoring sensors, wherein the multi-dimensional monitoring data includes equipment operation data and operation environment data; determines the fault monitoring results of each target monitoring component based on the multi-dimensional monitoring data, and predicts and processes the operation environment data through a trained dust compaction degree prediction model to obtain the dust compaction degree; controls the bag dust removal equipment for spray cleaning based on the fault monitoring results and the dust compaction degree; generates monitoring display data based on the fault monitoring results, the dust compaction degree and the spray cleaning execution cycle, and outputs the monitoring display data to a display terminal, thereby realizing real-time monitoring of faults and dynamic control of the spray cleaning cycle, greatly reducing the probability of the equipment being exposed to dust for a long time, reducing the equipment energy consumption, and at the same time ensuring the effectiveness of dust removal control, thereby greatly improving the dust removal efficiency.

[0148] According to one embodiment of the present invention, a medium is provided, wherein the medium stores at least one executable instruction, and the computer executable instruction can execute the bag dust removal monitoring method for the leaf threshing and redrying production line in any of the above method embodiments.

[0149] According to one embodiment of the present invention, a system is provided, which may include: a plurality of different types of monitoring sensors, a bag dust removal device, a storage server, a monitoring platform server, a controller, a display terminal and a communication bus, wherein the bag dust removal device includes a centralized dust collection duct, a pulse valve, an air storage tank, an exhaust dust duct, a centrifugal fan and a dust removal bag; the monitoring sensors include a dust concentration sensor, a humidity sensor and a wind speed sensor installed in the centralized dust collection duct, a pulse valve signal collector installed on the pulse valve, a pressure sensor installed on the air storage tank, a temperature and vibration sensor installed on the centrifugal fan, and a pressure difference sensor installed on the dust removal bag; the multi-dimensional monitoring data includes equipment operation data and operation environment data;

[0150] The monitoring sensors are respectively connected to the storage servers, and the storage servers, the monitoring platform servers, the controllers and the display terminals communicate with each other via the communication bus;

[0151] The monitoring platform server performs operations corresponding to the above-mentioned bag dust removal monitoring method for the leaf threshing and redrying production line.

[0152] Obviously, those skilled in the art will appreciate that the various modules or steps of the present invention described above can be implemented using a general-purpose computing device, centralized on a single computing device, or distributed across a network of multiple computing devices. Alternatively, they can be implemented using program code executable by a computing device, which can then be stored in a storage device and executed by the computing device. In some cases, the steps shown or described can be performed in a different order than that shown, or can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. Thus, the present invention is not limited to any particular combination of hardware and software.

[0153] The foregoing description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Those skilled in the art will readily appreciate that various modifications and variations of the present invention are possible. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention shall be included within the scope of protection of the present invention.

Claims

1. A bag dust removal monitoring method for a leaf threshing and redrying production line, characterized in that: include: Acquire multi-dimensional monitoring data of the bag dust removal equipment collected by multiple different types of monitoring sensors, wherein the multi-dimensional monitoring data includes equipment operation data and operating environment data; Determining the fault monitoring results of each target monitoring component based on the multi-dimensional monitoring data, and performing predictive processing on the operating environment data using the trained dust hardening degree prediction model to obtain the dust hardening degree; Controlling the dust removal equipment with spraying and cleaning according to the fault monitoring result and the degree of dust compaction; Monitoring display data is generated based on the fault monitoring result, the dust compaction degree and the spray cleaning execution cycle, and the monitoring display data is output to a display terminal.

2. The method according to claim 1, characterized in that The fault monitoring results include pulse valve solenoid coil fault monitoring results, pneumatic diaphragm valve diaphragm damage monitoring results, bag damage monitoring results and centrifugal fan fault monitoring results. The equipment operation data includes the pulse signals of all pulse valves, the compressed air storage tank pressure value, the fan vibration frequency and the fan temperature. The operating environment data includes the internal and external pressure difference of the bag and the dust concentration; The process of determining the pulse valve electromagnetic coil fault monitoring result includes: The pulse valves are sorted according to the order in which their actions are executed, and the pulse signal states of the pulse valves are polled according to their order. For any pulse valve, if within a first preset time period, the pulse signal states of both the preceding and succeeding pulse valves of the pulse valve change, and the pulse signal state of the pulse valve does not change, then it is determined that the pulse valve solenoid coil fault monitoring result is that the solenoid coil of the pulse valve is abnormal. The process of determining the pneumatic diaphragm valve diaphragm damage monitoring result includes: If the pressure values ​​of all compressed air storage tanks collected within the second preset time period are less than the first preset pressure threshold, and the pressure difference between the inside and outside of all bags is greater than the second preset pressure threshold, then the pneumatic diaphragm valve diaphragm damage monitoring result is determined to be a pneumatic diaphragm valve diaphragm damage; The process of determining the bag damage monitoring result includes: If all dust concentrations collected within the third preset time period are greater than the preset dust concentration threshold, and all bag internal and external pressure differences are less than the third preset pressure threshold, the bag damage monitoring result is determined to be bag damage; The process of determining the centrifugal fan fault monitoring result includes: If all dust concentrations collected within the fourth preset time period are greater than the preset dust concentration threshold, all fan vibration frequencies are greater than the preset frequency threshold and / or all fan temperatures are greater than the preset temperature threshold, the centrifugal fan fault monitoring result is determined to be an abnormal fan operating status.

3. The method according to claim 1, characterized in that Before performing prediction processing on the operating environment data using the trained dust compaction degree prediction model, the method further includes: Constructing an initial dust compaction degree prediction model based on nonlinear kernel function or neural network; Based on the subdivided ranges of wind speed, humidity, and dust concentration in the dust removal duct, dust compaction degree data under different parameter combinations are collected in multiple time intervals, and a training sample set is constructed; wherein the different parameter combinations include the same wind speed and humidity with different dust concentrations, the same wind speed and different humidity with the same dust concentration, the same wind speed and different humidity with different dust concentrations, different wind speeds and different humidity with the same dust concentration, and different wind speeds and humidity with the same dust concentration. The initial dust compaction degree prediction model is trained using the training sample set to obtain a trained dust compaction degree prediction model.

4. The method according to claim 1, wherein The trained dust compaction degree prediction model includes a multi-head attention layer, a convolutional neural network layer, and an activation function layer. The operating environment data includes operating environment data corresponding to each of the multiple bags. The trained dust compaction degree prediction model is used to predict the operating environment data to obtain the dust compaction degree, including: Key features are extracted from the operating environment data of each bag, and multiple key features of each bag at different times are obtained. The key features include abnormal pressure drop rate, compressed air pressure decay, and dust concentration correlation features. Constructing a multidimensional environmental data time series matrix based on the operating environment data corresponding to each of the plurality of bags and the timestamps of the operating environment data, and identifying the time dimension weight of the multidimensional environmental data time series matrix through the multi-head attention layer to determine the key decision interval; Extracting spatial correlation features between the plurality of key features at different time moments through a target filter of a one-dimensional convolutional neural network layer, wherein the target filter has a maximum response to a combined feature of a drop rate of the differential pressure signal and a pressure peak attenuation; Through the activation function layer, probability calculation is performed on the spatial correlation features within the key decision interval to obtain the dust compaction probability of different bags, and the degree of dust compaction is determined based on the dust compaction probability.

5. The method according to claim 1, wherein The dust compaction degree includes mild, moderate and severe. The spray cleaning control of the bag dust removal equipment based on the fault monitoring result and the dust compaction degree includes: When the fault monitoring result indicates that no fault exists, if the dust compaction degree is mild and the pressure difference between the inside and outside of the bag is greater than or equal to the second preset pressure threshold, the solenoid coil of the pulse valve is triggered to perform the first spray cleaning operation; When the fault monitoring result indicates that no fault exists, if the dust compaction degree is severe and the pressure difference between the inside and outside of the bag is greater than or equal to the third preset pressure threshold, the solenoid coil of the pulse valve is triggered to perform the second spray cleaning operation; When the fault monitoring result indicates that no fault exists, if the dust compaction degree is moderate and the pressure difference between the inside and outside of the bag is greater than or equal to the fourth preset pressure threshold, the solenoid coil of the pulse valve is triggered to perform the third spray cleaning operation; The second preset pressure threshold is greater than the third preset pressure threshold, and the fourth preset pressure threshold is an average of the second preset pressure threshold and the third preset pressure threshold.

6. The method according to claim 2, characterized in that The process of determining the bag damage monitoring result also includes: Extracting correlation features between the blowing event and the dust concentration peak value and contradiction features of the differential pressure signal based on the multi-dimensional monitoring data; Based on the correlation characteristics between the blowing event and the dust concentration peak and the contradictory characteristics of the differential pressure signal, a deep learning model is used to perform spatiotemporal feature matching to obtain a feature contribution ranking, and feature fusion is performed based on the feature contribution ranking to obtain a fused feature vector; According to the feature contribution ranking and fusion feature vectors, the bag breakage posterior probability is calculated through the Bayesian framework; A bag damage monitoring result is generated according to the bag damage posterior probability.

7. The method according to claim 2, characterized in that The process of determining the pulse valve electromagnetic coil fault monitoring result further includes: Extracting the pulse valve pressure waveform, pressure difference recovery rate and dust concentration based on the multi-dimensional monitoring data; Performing a horizontal comparative analysis of multiple pulse valves based on the pressure rise time in the pulse valve pressure waveform, the pressure difference recovery rate, and the dust concentration to determine the pulse valve with expected failure; Performing DTW matching on the pulse valve pressure waveform of the expected fault pulse valve to obtain a fault label of the expected fault pulse valve; The failure probability of the expected fault pulse valve is obtained by Bayesian probability calculation according to the pulse valve pressure waveform, the pressure difference recovery rate and the fault label of the expected fault pulse valve.

8. A bag dust removal monitoring device for a leaf threshing and redrying production line, characterized in that: include: A sensor data acquisition module, configured to acquire multi-dimensional monitoring data of the bag dust removal equipment collected by a plurality of different types of monitoring sensors, wherein the multi-dimensional monitoring data includes equipment operation data and operation environment data; a data processing module for determining the fault monitoring results of each target monitoring component based on the multi-dimensional monitoring data, and performing predictive processing on the operating environment data using the trained dust compaction degree prediction model to obtain the dust compaction degree; A control module, configured to control the dust removal equipment by spraying and cleaning dust based on the fault monitoring result and the degree of dust compaction; The display module is used to generate monitoring display data based on the fault monitoring results, the dust compaction degree and the spray cleaning execution cycle, and output the monitoring display data to the display terminal.

9. A medium storing at least one executable instruction, wherein the executable instruction enables a processor to execute operations corresponding to the bag dust removal monitoring method for a leaf threshing and redrying production line according to any one of claims 1 to 7.

10. A system comprising: Multiple different types of monitoring sensors, bag-type dust removal equipment, storage servers, monitoring platform servers, controllers, display terminals, and communication buses, wherein the bag-type dust removal equipment includes a centralized dust collection duct, a pulse valve, an air storage tank, an exhaust dust duct, a centrifugal fan, and dust collection bags; the monitoring sensors include a dust concentration sensor, a humidity sensor, and a wind speed sensor installed in the centralized dust collection duct, a pulse valve signal collector installed on the pulse valve, a pressure sensor installed on the air storage tank, a temperature and vibration sensor installed on the centrifugal fan, and a pressure difference sensor installed on the dust collection bag; the multi-dimensional monitoring data includes equipment operation data and operating environment data; The monitoring sensors are respectively connected to the storage servers, and the storage servers, the monitoring platform servers, the controllers and the display terminals communicate with each other via the communication bus; The monitoring platform server performs operations corresponding to the bag dust removal monitoring method for the leaf threshing and redrying production line as described in any one of claims 1 to 7.

Images