A dust collector monitoring method and system

Abstract

A dust collector monitoring method and system relate to the technical field of dust collector monitoring, the method comprising: acquiring a dynamic pressure signal at the inlet of a bag-type dust collector and a dust concentration signal at the outlet; generating a pulse valve action waveform sequence according to the dynamic pressure signal, and identifying the mechanical action state of each pulse valve based on the pressure fluctuation characteristics in the pulse valve action waveform sequence; when the mechanical action state of the target pulse valve is determined to meet the preset action standard according to the pressure fluctuation characteristics, collecting the dust concentration signal corresponding to the blowing timing of the target pulse valve; judging whether dust escape occurs according to the blowing timing and the concentration change trend of the dust concentration signal, and when it is judged that dust escape occurs, time backtracking matching is performed between the concentration change trend of the dust escape and the blowing time of the target pulse valve to locate the area where the fault bag is located. The application can improve the safety and reliability of the operation of the dust collector.

Description

Technical Field

[0001] This application relates to the technical field of dust collector monitoring, and in particular to a dust collector monitoring method and system. Background Technology

[0002] In the field of industrial dust removal, baghouse dust collectors are widely used due to their high dust collection efficiency. To ensure the stable operation of dust collection systems and meet increasingly stringent environmental emission standards, real-time monitoring of the dust collector's operating status is crucial. Traditional monitoring methods typically rely solely on monitoring the inlet and outlet pressure difference or the total dust concentration at the outlet. When emissions exceed standards or equipment malfunctions, they cannot quickly pinpoint the source of the fault. With the development of automation and sensing technologies, intelligent diagnostic methods based on pressure fluctuation analysis and coordinated dust concentration monitoring are gradually becoming mainstream. By extracting features from the pressure response signals generated during the pulse valve's blowing process, the mechanical operation status of the pulse valve can be identified, thereby assessing the operational health of the dust removal system.

[0003] Chinese invention patent application CN118384613A, published on July 26, 2024, discloses a method and system for monitoring the status of a bag filter based on rate of change and distance calculation. This method collects dynamic pressure signals at the inlet and dust concentration signals at the outlet of the bag filter, and uses statistical analysis of the rate of change of pressure fluctuation signals and the fluctuation characteristics of concentration signals to monitor and provide graded early warnings for the operating status of pulse valves. This technology is effective in identifying faults in pulse valves themselves, and can determine the normal, warning, or fault state of the pulse valve by analyzing whether the pressure change rate falls within different preset ranges, providing data support for preventative maintenance of the equipment.

[0004] However, the aforementioned patents can only achieve state identification at the pulse valve level, that is, determine whether a pulse valve has experienced a mechanical failure; the diagnostic conclusion stops at whether the valve is functioning normally. But in actual operation of a dust collector, abnormal pulse valve operation often leads to dust escape from the filter bags within the area covered by its corresponding blowpipe. This area may contain multiple rows and columns of filter bags. Therefore, a pulse valve failure is not equivalent to a filter bag failure, nor does it necessarily mean that the location of the faulty filter bag can be clearly identified. When a dust concentration exceeding the standard occurs, although on-site operators may know which pulse valve might be faulty, they still need to spend a significant amount of time checking each filter bag area covered by that pulse valve to ultimately determine the actual location of the dust escape. Summary of the Invention

[0005] In order to shorten the troubleshooting time and improve the safety and reliability of dust collector operation, this application provides a dust collector monitoring method and system.

[0006] Firstly, this application provides a dust collector monitoring method, which adopts the following technical solution: A dust collector monitoring method includes the following steps: Acquire the dynamic pressure signal at the inlet of the bag filter and the dust concentration signal at the outlet; A pulse valve action waveform sequence is generated based on the dynamic pressure signal, and the mechanical action state of each pulse valve is identified based on the pressure fluctuation characteristics in the pulse valve action waveform sequence. When the mechanical action state of the target pulse valve meets the preset action standard based on the pressure fluctuation characteristics, the dust concentration signal corresponding to the blowing sequence of the target pulse valve is collected. Based on the pulse jet timing and the concentration change trend of the dust concentration signal, it is determined whether dust escape has occurred. When it is determined that dust escape has occurred, the concentration change trend of dust escape is matched with the pulse jet timing of the target pulse valve to locate the area where the faulty filter bag is located.

[0007] This application achieves automated monitoring of the pulse valve's operating status by simultaneously acquiring the dynamic pressure signal at the bag filter inlet and the dust concentration signal at the outlet, and using the dynamic pressure signal to generate a pulse valve action waveform sequence, thereby identifying the mechanical action state of each pulse valve. This application only collects the corresponding time-series dust concentration signal when the target pulse valve's action meets the standard, effectively avoiding interference from invalid data and improving the targeting of data analysis. By performing time-reverse matching of concentration change trends with the pulse-jet timing, this application can locate the area of ​​the filter bag where dust escape occurs, thus providing maintenance personnel with clear fault location guidance, shortening fault diagnosis time, reducing maintenance costs caused by blind inspections, and improving the safety and reliability of the dust collector operation.

[0008] Optionally, identifying the mechanical action state of each pulse valve based on the pressure fluctuation characteristics in the pulse valve action waveform sequence specifically involves: Extract the pressure fluctuation sequence within a predetermined time window after each pulse valve is opened, and obtain the maximum drop amplitude and pressure recovery time in the pressure fluctuation sequence; If the maximum drop value is greater than the preset amplitude threshold and the pressure recovery time is within the preset standard time interval, then the mechanical action state of the pulse valve is determined to be normal operation. If the maximum drop amplitude is less than or equal to the preset amplitude threshold, or if no drop waveform is detected, the mechanical action state of the pulse valve is determined to be non-operational, and a corresponding non-operational alarm signal is generated.

[0009] This application quantitatively evaluates the mechanical performance of a pulse valve by extracting a pressure fluctuation sequence within a predetermined time window after the pulse valve opens and calculating the maximum pressure drop and recovery time. By comparing the maximum pressure drop with a preset threshold and the recovery time with a standard time interval, this application can clearly distinguish between normal operation and inaction, and promptly generate an inaction alarm signal in case of an anomaly. This effectively prevents poor dust removal caused by pulse valve failure and improves dust collection efficiency.

[0010] Optionally, when the mechanical action state of the target pulse valve is determined to meet a preset action standard, the method further includes: Calculate the rate of change between adjacent data points in the pressure fluctuation sequence, and integrate the values ​​with a rate of change less than zero into a set of negative rates of change; The number of data points in the set of negative change rates is counted. If the number of data points falls within a first preset interval, the pulse valve is determined to be in a normal state. If the number of data points falls within a second preset interval, the pulse valve is determined to be in a warning state, and the number of warnings is accumulated. When the accumulated number of warnings reaches a preset threshold, an alarm signal is triggered. If the number of data points falls within a third preset interval, the pulse valve is determined to be in a fault state, and a shutdown maintenance prompt is triggered. The numerical ranges of the first preset interval, the second preset interval, and the third preset interval increase sequentially and do not overlap.

[0011] This application addresses different states by assigning different numbers of statistical data to different preset ranges. In the early warning state, an alarm is triggered when the cumulative number of occurrences reaches a threshold. In the fault state, a shutdown for maintenance is directly prompted. This achieves hierarchical management from minor anomalies to serious faults. By adopting the above scheme, this application can detect the performance degradation trend of pulse valves in advance, effectively avoid sudden shutdowns, provide data support for planned maintenance, and extend the service life of equipment.

[0012] Optionally, the method further includes: The pressure fluctuation sequence collected by the pulse valve under normal operating conditions is used as template data; Calculate the Euclidean distance between the pressure fluctuation sequence to be measured and the template data. If the Euclidean distance is greater than a preset distance threshold, trigger an early warning signal. If the Euclidean distance is greater than the distance threshold for multiple consecutive blowing cycles, an alarm signal is triggered.

[0013] This application uses pressure fluctuation sequences collected under normal operating conditions of a pulse valve as template data. The Euclidean distance between the test sequence and the template is calculated, and the distance is used as an indicator of the pulse valve waveform consistency. This captures overall changes in waveform shape, focusing not only on individual amplitudes or durations but also reflecting anomalies in the pressure recovery curve. A warning is triggered when the distance exceeds the limit once, and an alarm is triggered when the distance exceeds the limit multiple times consecutively. This effectively suppresses false alarms caused by occasional noise, improves diagnostic robustness, and enhances the ability to identify progressive degradation of the pulse valve.

[0014] Optionally, the method further includes: The historical time series data of the dust concentration is obtained, and based on the historical time series data, a time series prediction is performed using a differential autoregressive moving average model to obtain the predicted dust concentration value. The error between the predicted dust concentration and the real-time monitoring value is calculated. If the error exceeds a preset error range, or the real-time monitoring value exceeds a preset emission standard threshold, an abnormal dust concentration alarm signal is generated.

[0015] This application utilizes a differential autoregressive moving average model to predict historical time-series data of dust concentration, obtaining predicted values. The predicted values ​​are then compared with real-time monitoring values ​​to identify errors, and the real-time values ​​are compared with emission standard thresholds. This approach can detect both sudden concentration exceedances and slow-drifting anomalies, improving the comprehensiveness of anomaly detection. Through model prediction, this application can anticipate concentration change trends and issue warnings when the concentration has not yet exceeded emission standards but the trend is abnormal, giving operators more reaction time.

[0016] Optionally, determining whether dust escape has occurred based on the blowing sequence and the concentration change trend of the dust concentration signal specifically involves: Obtain the real-time dust concentration at the outlet of the bag filter; In online dust removal mode, if the real-time dust concentration exceeds the preset escape threshold within the first preset time period after the target pulse valve is activated, it is determined that dust escape has occurred and there is a leak point in the blower bag corresponding to the target pulse valve. In offline mode, if the real-time dust concentration exceeds the escape threshold within the second preset time period after the lift valve is opened, it is determined that dust escape has occurred, and there is a leak in the clean air chamber bag filter corresponding to the lift valve.

[0017] This application defines different dust escape judgment rules for online and offline modes. In online mode, escape is judged when the concentration exceeds a threshold within a first preset time period after the pulse valve actuates; in offline mode, escape is judged when the concentration exceeds a threshold within a second preset time period after the lift valve opens. This application fully considers the differences in airflow and dust removal mechanisms under the two operating conditions, effectively avoiding misjudgments or omissions caused by a uniform threshold.

[0018] This application, through a clearly defined time window and escape threshold, can quickly identify the specific operational stage where a leak occurs (pulse valve blowing or lift valve opening).

[0019] Optionally, the concentration change trend that causes dust escape is matched with the blowing time of the target pulse valve in reverse time to locate the area where the faulty filter bag is located, specifically: Using the moment when the dust concentration signal exceeds the escape threshold as the reference point, trace back according to the preset backtracking time window, and extract the action timestamps of all pulse valves that have performed the blowing action within the backtracking time window; The time difference between the peak time of the dust concentration signal and each action timestamp is calculated, and the pulse valve with the smallest time difference is selected as the target fault pulse valve. Based on the preset installation location information of the target fault pulse valve, the bag area where the faulty bag is located is determined.

[0020] This application uses the moment when the dust concentration exceeds the escape threshold as the benchmark point, traces back according to a preset backtracking time window, extracts the action timestamps of all pulse valves that have performed the blowing action within the window, calculates the time difference between the peak concentration moment and each action timestamp, and selects the pulse valve with the smallest time difference as the target fault valve. This fully considers the time delay from dust leakage from the filter bag to detection by the outlet sensor, effectively correlates the causal relationship between the peak concentration and the induced blowing action, accurately locates the filter bag area where the faulty filter bag is located, effectively avoids mislocation caused by time misalignment, and improves the accuracy and efficiency of fault area investigation.

[0021] Optionally, after acquiring the dynamic pressure signal and the dust concentration signal, the method further includes: By selecting the wavelet basis function and the number of decomposition levels, wavelet decomposition is performed on the original dynamic pressure signal and the dust concentration signal to obtain multi-level wavelet coefficients. Cluster analysis is performed on the wavelet coefficients of each layer, and the coefficients that are furthest from the cluster center in each cluster are set to zero in order to eliminate abnormal noise points; The coefficients of each layer after removing abnormal noise points are quantized using a threshold function; The signal is reconstructed based on the coefficients of each layer after quantization, and the denoised dynamic pressure signal and dust concentration signal are obtained.

[0022] This application performs wavelet decomposition by selecting wavelet basis functions and the number of decomposition levels, and then performs cluster analysis on the wavelet coefficients of each level. Coefficients far from the cluster center are set to zero to eliminate abnormal noise points. Quantization is then performed using a threshold function, and the signal is finally reconstructed. This effectively removes impulse interference, environmental noise, and random fluctuations from sensor signals while preserving key abrupt changes in the signal, effectively avoiding the over-smoothing of effective high-frequency information by traditional filtering methods. The denoised signal has higher quality, reducing the risk of false alarms and missed alarms.

[0023] Optionally, before acquiring the dynamic pressure signal and the dust concentration signal, the method further includes: Sensor components are deployed at the sensor installation location of the bag filter, and the raw electrical signal output by the sensor is initially filtered by a hardware filtering circuit. The Kalman filter algorithm is used to perform secondary filtering on the signal after primary filtering to obtain new dynamic pressure and dust concentration signals.

[0024] This application performs primary filtering on the original electrical signal through a hardware filtering circuit to remove high-frequency interference and power frequency noise at the physical level, thereby reducing gross errors in the signal. Then, a secondary filtering is performed using the Kalman filtering algorithm. By utilizing state estimation and covariance update mechanisms, the signal after primary filtering is optimally estimated, further smoothing random noise and measurement uncertainty, so as to quickly suppress significant interference and improve the signal-to-noise ratio of dynamic pressure and dust concentration signals.

[0025] Secondly, this application provides a dust collector monitoring system, which adopts the following technical solution: A dust collector monitoring system, comprising: The signal acquisition module is used to acquire the dynamic pressure signal at the inlet of the bag filter and the dust concentration signal at the outlet. The waveform analysis module is used to generate a pulse valve action waveform sequence based on the dynamic pressure signal, and to identify the mechanical action state of each pulse valve based on the pressure fluctuation characteristics in the pulse valve action waveform sequence. The escape judgment module is used to collect a dust concentration signal corresponding to the blowing sequence of the target pulse valve when the mechanical action state of the target pulse valve meets the preset action standard based on the pressure fluctuation characteristics, and to determine whether dust escape has occurred based on the blowing sequence and the concentration change trend of the dust concentration signal. The fault location module is used to perform time-retrospective matching between the concentration change trend of dust escape and the blowing time of the target pulse valve when dust escape occurs, so as to locate the area where the faulty filter bag is located.

[0026] In summary, this application includes at least one of the following beneficial technical effects: 1. This application achieves automated monitoring of the working status of pulse valves by simultaneously acquiring the dynamic pressure signal at the inlet and the dust concentration signal at the outlet of the bag filter, and using the dynamic pressure signal to generate a pulse valve action waveform sequence, thereby identifying the mechanical action status of each pulse valve.

[0027] 2. This application only collects the dust concentration signal at the corresponding time sequence when the target pulse valve operates in accordance with the standard, which effectively avoids interference from invalid data and improves the relevance of data analysis.

[0028] 3. By matching the concentration change trend with the pulse jet timing, this application can accurately locate the area of ​​the filter bag where dust escape occurs, thereby providing maintenance personnel with clear fault location guidance, shortening fault diagnosis time, reducing maintenance costs caused by blind inspection, and improving the safety and reliability of dust collector operation. Attached Figure Description

[0029] Figure 1 This is a flowchart of Embodiment 1 of this application; Figure 2 This is a pressure value curve of the pulse valve 3 in Embodiment 1 of this application; Figure 3 This is a rate-of-change distribution diagram of the pulse valve 3 in Embodiment 1 of this application; Figure 4 This is a pressure value curve of the pulse valve 7 in Embodiment 1 of this application; Figure 5 This is a rate-of-change distribution diagram of the pulse valve 7 in Embodiment 1 of this application; Figure 6 This is a pressure value curve of the pulse valve 11 in Embodiment 1 of this application; Figure 7 This is a rate of change distribution diagram of the pulse valve 11 in Embodiment 1 of this application. Detailed Implementation

[0030] The following combination Figures 1 to 7 This application will be described in further detail.

[0031] Example 1: This example discloses a dust collector monitoring method, referring to... Figure 1The method includes: S1 signal acquisition, S2 state recognition, S3 similarity monitoring, S4 abnormal alarm, and S5 dust escape determination. First, a pulse valve action waveform sequence is generated based on the dynamic pressure signal, and the mechanical action state of each pulse valve is identified by the pressure fluctuation characteristics. The action quality of the pulse valve is quantitatively evaluated from multiple dimensions by combining two methods: rate of change statistics and template matching. When the pulse valve action is confirmed to be normal, the dust concentration signal corresponding to the pulse valve's blowing sequence is acquired, and the dust escape is determined by the concentration change trend. Once dust escape is confirmed, the concentration jump moment is spatiotemporally correlated with historical blowing actions by time retrospective matching, thereby accurately locating the area where the faulty filter bag is located.

[0032] Taking a baghouse dust collector with a processing air volume of 50,000 m³ / h, equipped with 12 pulse valves, and 4 pulse valves arranged in each row for a total of 3 rows of baghouse areas as an example, the specific implementation method of this embodiment will be described in detail. In this embodiment, a dynamic pressure sensor is installed at the inlet duct of the dust collector, and a dust concentration detector, which is a light scattering dust meter, is installed at the outlet duct.

[0033] S1 signal acquisition continuously acquires the dynamic pressure signal at the inlet of the bag filter and the dust concentration signal at the outlet.

[0034] The raw electrical signal output from the sensor is conditioned by a process-controlled amplifier and then sent to a hardware filter circuit for primary filtering. This hardware filter circuit is a second-order Butterworth low-pass filter with a cutoff frequency set to 100Hz to filter out high-frequency electromagnetic interference. The signal after primary filtering is then subjected to secondary filtering using a Kalman filter algorithm.

[0035] The state equation and observation equation of the Kalman filter algorithm are shown below:

[0036]

[0037] in, This is the predicted state value at time k. The state prediction value at time k-1; the state transition matrix A is the identity matrix, the control matrix B is the zero matrix, and Q is the process noise covariance matrix, with values... R is the observation noise covariance matrix, with values ​​ranging from 1 to 10. , This is an external control variable, and its value is 0. Let be the prediction error covariance matrix at time k; Let be the estimation error covariance matrix at time k-1; Let A be the transpose of the state transition matrix A.

[0038] After the secondary filtering is completed, this embodiment selects the db4 wavelet as the wavelet basis function and sets the number of decomposition layers to 6. Wavelet decomposition is performed on the dynamic pressure signal and dust concentration signal after preliminary noise reduction to obtain 6 layers of detail coefficients and 1 layer of approximation coefficients.

[0039] K-means clustering analysis was performed on the wavelet coefficients of each layer, with the number of clusters set to 3. The coefficients that are furthest from the cluster center in each cluster were set to zero to remove abnormal noise points. Then, a soft thresholding function was used to quantize the coefficients of each layer after removing abnormal noise points.

[0040] In this embodiment, the calculation model for the soft threshold function is as follows:

[0041]

[0042] in, For threshold; The standard deviation of noise; The logarithm of the detection signal length m; For the first layer of high-frequency detail coefficients of the median.

[0043] The signal is reconstructed based on the coefficients of each layer after quantization, and the denoised dynamic pressure signal and dust concentration signal are obtained.

[0044] S2 status recognition: During the normal pulse cleaning process of the dust collector, each pulse valve operates sequentially according to a preset timing sequence. In this embodiment, the pulse timing sequence is set as follows: pulse valves 1 to 12 are pulsed sequentially, with an interval of 6 seconds between adjacent pulse valves, and the pulse width of each pulse valve is 80 milliseconds.

[0045] Based on the denoised dynamic pressure signal, a pulse valve action waveform sequence is generated. In this embodiment, the rising edge of the PLC output control signal for each pulse valve is taken as the starting point, and the pressure fluctuation sequence within a 500-millisecond time window after each pulse valve opens is extracted and denoted as... Where i is the pulse valve number , .

[0046] For any pulse valve i, extract its pressure fluctuation sequence within a 500-millisecond time window. The maximum decrease in and stress recovery time .

[0047] The calculation model for the maximum decrease is shown below:

[0048] in, This represents the pressure value at the start of the window.

[0049] The pressure recovery time It refers to the time elapsed from the moment the pressure reaches its lowest value until the pressure value recovers to 90% of the pressure value at the beginning of the window.

[0050] In this embodiment, a preset amplitude threshold is used. The Pa is set to 0.15 kPa, and the preset standard time interval is set to [100 ms, 300 ms].

[0051] When pulse valve 1 The pressure is 0.32 kPa. The value is 185ms. Since 0.32kPa > 0.15kPa and 185ms ∈ [100ms, 300ms], the mechanical action state of pulse valve 1 is determined to be normal.

[0052] When pulse valve 5 If the amplitude is 0.08 kPa, which is less than the preset amplitude threshold of 0.15 kPa, the mechanical action state of pulse valve 5 is determined to be non-operational. The system immediately generates a corresponding non-operational alarm signal to prompt the operator to check the solenoid coil or air supply pipeline of pulse valve 5.

[0053] After determining that the mechanical action state of the target pulse valve meets the preset action standard, this embodiment also performs the following refined discrimination process: Taking pulse valve 3 as an example, the pressure fluctuation sequence within its 500-millisecond time window is taken. Calculate the rate of change between adjacent data points in the sequence. The calculation model for the rate of change is as follows:

[0054] in, is the rate of change (i.e., slope or first difference) between the kth adjacent data points. Pressure fluctuation sequence The pressure value at the (k+1)th sampling point; Pressure fluctuation sequence The pressure value at the k-th sampling point; The sampling interval is the sampling interval in this embodiment. The sampling time is 2ms, which means the sampling frequency is 500Hz.

[0055] Reference Figures 2 to 6 All values ​​with a rate of change less than zero are grouped into a set of negative rates of change, and the number of data points in the set of negative rates of change is counted.

[0056] In this embodiment, the first preset interval is set as... This corresponds to the normal state; the second preset interval is set as follows: This corresponds to the warning status; the third preset interval is set as follows: Corresponding to the fault state, the value range of each interval increases sequentially and does not overlap.

[0057] For pulse valve 3, the number of data points in the negative rate of change set is 7, falling into the first preset interval. The pulse valve 3 is determined to be in a normal state. For pulse valve 7, the number of data points in the negative rate of change set is 34, falling within the second preset interval. The system determines that pulse valve 7 is in a warning state. It accumulates the number of warnings for this pulse valve, and triggers an alarm signal when the accumulated number of warnings reaches a preset threshold (set to 3 consecutive warnings in this embodiment). For pulse valve 11, the number of data points in the negative rate of change set is 52, falling within the third preset interval. If the system determines that pulse valve 11 is in a faulty state, it will trigger a shutdown maintenance prompt, notifying maintenance personnel to immediately inspect and repair the pulse valve and its corresponding jetting assembly.

[0058] S3 similarity monitoring involves continuously collecting the pressure fluctuation sequence of each pulse valve during 10 blowing cycles under normal operating conditions, assuming the dust collector is put into operation for the first time and all pulse valves are confirmed to be working properly. The arithmetic mean of these fluctuations is then used as the template data for that pulse valve, denoted as... , .

[0059] The pressure fluctuation sequence to be measured collected during the current injection cycle With template data Euclidean distance between Perform the calculation:

[0060] Where N is the sequence length.

[0061] In this embodiment, the preset distance threshold The value is set to 0.12. In this embodiment, the Euclidean distance between the test sequence of pulse valve 4 and the template sequence is... The value is 0.09, which is less than the distance threshold of 0.12, and is therefore considered normal. (Pulse valve 9) If the value is 0.18, which is greater than 0.12, the system triggers a warning signal. If this occurs within four consecutive blowing cycles (in this embodiment, the consecutive cycle threshold is set to three times),... If the values ​​are all greater than 0.12, the system will trigger an alarm signal, indicating that the pulse valve may have potential faults such as reduced injection pressure or diaphragm wear.

[0062] S4 abnormal alarm: In this embodiment, the differential autoregressive moving average model (ARIMA model) is used to predict the outlet dust concentration over time. Historical time series data of dust concentration over the past 24 hours are obtained, with a sampling interval of 1 minute, for a total of 1440 data points.

[0063] The difference order of the sequence was determined to be 1 by ADF test, and the autoregression order was determined to be 2 and the moving average order to be 1 by autocorrelation function and partial autocorrelation function analysis, and an ARIMA(2,1,1) model was established.

[0064] The model is used to predict the dust concentration at the next time step, and the predicted dust concentration value is obtained. Simultaneously acquire real-time monitoring values. The prediction error is calculated using the following model:

[0065] in, This represents the prediction error.

[0066] The preset error range in this embodiment is ±2mg / m³, and the emission standard threshold is set at 30mg / m³.

[0067] When the prediction error at a certain moment The concentration was 5.6 mg / m³, exceeding the preset error range of ±2 mg / m³. The real-time monitoring value was 35.8 mg / m³, exceeding the emission standard threshold of 30 mg / m³. An abnormal dust concentration alarm signal was immediately generated to alert the operator to pay attention to the abnormal emissions at the outlet.

[0068] S5 Dust Escape Determination: This embodiment provides a dust escape determination method under two working conditions: online cleaning mode and offline mode.

[0069] In online dust removal mode, the system acquires the dust concentration value at the outlet of the bag filter in real time. If the real-time dust concentration value exceeds the preset escape threshold (set to 25mg / m³ in this embodiment) within the first preset time period after the target pulse valve (e.g., pulse valve 2) is activated, it is determined that dust escape has occurred, indicating that there is a leak in the blower bag corresponding to the target pulse valve.

[0070] In this embodiment, 2.3 seconds after the pulse valve 2 is activated, the outlet dust concentration jumps to 42.6 mg / m³, exceeding the escape threshold of 25 mg / m³, and the system determines that dust escape has occurred.

[0071] The execution time is reversed for matching to locate the fault area. In this embodiment, the moment when the dust concentration signal exceeds the escape threshold (i.e., the 2.3rd second) is used as the reference point. According to the preset backtracking time window (set to the first 10 seconds in this embodiment), the process traces backward to extract... to The timestamps of all pulse valves that performed the blowing action within the specified time window. Extracted, the pulse valves that performed the blowing action within this time window include: Pulse valve 12 (action timestamp: ...). (seconds), pulse valve 1 (action timestamp is) (seconds), pulse valve 2 (action timestamp is) Second).

[0072] The peak time of the dust concentration signal (in this embodiment, it is...) The time difference between the second (i.e., the 2.9th second) and the timestamps of each action is calculated: The time difference with pulse valve 12 is 10.4 seconds; the time difference with pulse valve 1 is 4.1 seconds; and the time difference with pulse valve 2 is 2.7 seconds.

[0073] In this embodiment, the pulse valve with the smallest time difference, namely pulse valve 2 (time difference 2.7 seconds), is selected as the target fault pulse valve. According to the preset installation position information of pulse valve 2, the pulse valve is installed in the second row of filter bags, thereby determining that the area where the faulty filter bag is located is the second row of filter bags. Maintenance personnel can directly focus on checking and replacing the filter bags in this row.

[0074] In offline mode, the inlet and outlet valves of the dust collector are closed during the dust removal cycle. After the opening action of the lifting valve is detected, based on the opening time of the lifting valve, if the real-time dust concentration exceeds the escape threshold of 25mg / m³ within the second preset time period after the opening (set to 0-8 seconds in this embodiment), it is determined that dust escape has occurred.

[0075] The clean air chamber bag filter corresponding to the lifting valve has a leakage point. The positioning strategy is similar to that in the online mode, and will not be described again in this embodiment.

[0076] Example 2: This example discloses a dust collector monitoring system, the system comprising: The signal acquisition module is used to acquire the dynamic pressure signal at the inlet of the bag filter and the dust concentration signal at the outlet. The waveform analysis module is used to generate a pulse valve action waveform sequence based on the dynamic pressure signal, and to identify the mechanical action state of each pulse valve based on the pressure fluctuation characteristics in the pulse valve action waveform sequence. The escape judgment module is used to collect a dust concentration signal corresponding to the blowing sequence of the target pulse valve when the mechanical action state of the target pulse valve meets the preset action standard based on the pressure fluctuation characteristics, and to determine whether dust escape has occurred based on the blowing sequence and the concentration change trend of the dust concentration signal. The fault location module is used to perform time-retrospective matching between the concentration change trend of dust escape and the blowing time of the target pulse valve when dust escape occurs, so as to locate the area where the faulty filter bag is located.

[0077] The above are all preferred embodiments of this application, and are not intended to limit the scope of protection of this application. Therefore, all equivalent changes made in accordance with the structure, shape and principle of this application should be covered within the scope of protection of this application.

Claims

1. A method for monitoring a dust collector, characterized in that, include: Acquire the dynamic pressure signal at the inlet of the bag filter and the dust concentration signal at the outlet; A pulse valve action waveform sequence is generated based on the dynamic pressure signal, and the mechanical action state of each pulse valve is identified based on the pressure fluctuation characteristics in the pulse valve action waveform sequence. When the mechanical action state of the target pulse valve meets the preset action standard based on the pressure fluctuation characteristics, the dust concentration signal corresponding to the blowing sequence of the target pulse valve is collected. Based on the pulse jet timing and the concentration change trend of the dust concentration signal, it is determined whether dust escape has occurred. When it is determined that dust escape has occurred, the concentration change trend of dust escape is matched with the pulse jet timing of the target pulse valve to locate the area where the faulty filter bag is located.

2. The dust collector monitoring method according to claim 1, characterized in that, The step of identifying the mechanical action state of each pulse valve based on the pressure fluctuation characteristics in the pulse valve action waveform sequence is as follows: Extract the pressure fluctuation sequence within a predetermined time window after each pulse valve is opened, and obtain the maximum drop amplitude and pressure recovery time in the pressure fluctuation sequence; If the maximum drop value is greater than the preset amplitude threshold and the pressure recovery time is within the preset standard time interval, then the mechanical action state of the pulse valve is determined to be normal operation. If the maximum drop amplitude is less than or equal to the preset amplitude threshold, or if no drop waveform is detected, the mechanical action state of the pulse valve is determined to be non-operational, and a corresponding non-operational alarm signal is generated.

3. The dust collector monitoring method according to claim 2, characterized in that, When the mechanical action state of the target pulse valve is determined to meet the preset action standard, the method further includes: Calculate the rate of change between adjacent data points in the pressure fluctuation sequence, and integrate the values ​​with a rate of change less than zero into a set of negative rates of change; The number of data points in the set of negative change rates is counted. If the number of data points falls within a first preset interval, the pulse valve is determined to be in a normal state. If the number of data points falls within a second preset interval, the pulse valve is determined to be in a warning state, and the number of warnings is accumulated. When the accumulated number of warnings reaches a preset threshold, an alarm signal is triggered. If the number of data points falls within a third preset interval, the pulse valve is determined to be in a fault state, and a shutdown maintenance prompt is triggered. The numerical ranges of the first preset interval, the second preset interval, and the third preset interval increase sequentially and do not overlap.

4. The dust collector monitoring method according to claim 3, characterized in that, The method further includes: The pressure fluctuation sequence collected by the pulse valve under normal operating conditions is used as template data; Calculate the Euclidean distance between the pressure fluctuation sequence to be measured and the template data. If the Euclidean distance is greater than a preset distance threshold, trigger an early warning signal. If the Euclidean distance is greater than the distance threshold for multiple consecutive blowing cycles, an alarm signal is triggered.

5. The dust collector monitoring method according to claim 1, characterized in that, The method further includes: The historical time series data of the dust concentration is obtained, and based on the historical time series data, a time series prediction is performed using a differential autoregressive moving average model to obtain the predicted dust concentration value. The error between the predicted dust concentration and the real-time monitoring value is calculated. If the error exceeds a preset error range, or the real-time monitoring value exceeds a preset emission standard threshold, an abnormal dust concentration alarm signal is generated.

6. The dust collector monitoring method according to claim 3, characterized in that, The step of determining whether dust escape has occurred based on the blowing sequence and the concentration change trend of the dust concentration signal is as follows: Obtain the real-time dust concentration at the outlet of the bag filter; In online dust removal mode, if the real-time dust concentration exceeds the preset escape threshold within the first preset time period after the target pulse valve is activated, it is determined that dust escape has occurred and there is a leak point in the blower bag corresponding to the target pulse valve. In offline mode, if the real-time dust concentration exceeds the escape threshold within the second preset time period after the lift valve is opened, it is determined that dust escape has occurred, and there is a leak in the clean air chamber bag filter corresponding to the lift valve.

7. The dust collector monitoring method according to claim 6, characterized in that, The process of matching the concentration change trend of dust escape with the blowing time of the target pulse valve in reverse time to locate the area where the faulty filter bag is located is specifically as follows: Using the moment when the dust concentration signal exceeds the escape threshold as the reference point, trace back according to the preset backtracking time window, and extract the action timestamps of all pulse valves that have performed the blowing action within the backtracking time window; The time difference between the peak time of the dust concentration signal and each action timestamp is calculated, and the pulse valve with the smallest time difference is selected as the target fault pulse valve. Based on the preset installation location information of the target fault pulse valve, the bag area where the faulty bag is located is determined.

8. The dust collector monitoring method according to claim 1, characterized in that, After acquiring the dynamic pressure signal and the dust concentration signal, the method further includes: By selecting the wavelet basis function and the number of decomposition levels, wavelet decomposition is performed on the original dynamic pressure signal and the dust concentration signal to obtain multi-level wavelet coefficients. Cluster analysis is performed on the wavelet coefficients of each layer, and the coefficients that are furthest from the cluster center in each cluster are set to zero in order to eliminate abnormal noise points; The coefficients of each layer after removing abnormal noise points are quantized using a threshold function; The signal is reconstructed based on the coefficients of each layer after quantization, and the denoised dynamic pressure signal and dust concentration signal are obtained.

9. The dust collector monitoring method according to claim 1, characterized in that, Before acquiring the dynamic pressure signal and the dust concentration signal, the method further includes: Sensor components are deployed at the sensor installation location of the bag filter, and the raw electrical signal output by the sensor is initially filtered by a hardware filtering circuit. The Kalman filter algorithm is used to perform secondary filtering on the signal after primary filtering to obtain new dynamic pressure and dust concentration signals.

10. A dust collector monitoring system, said system being applicable to the method as described in any one of claims 1-9, characterized in that, include: The signal acquisition module is used to acquire the dynamic pressure signal at the inlet of the bag filter and the dust concentration signal at the outlet. The waveform analysis module is used to generate a pulse valve action waveform sequence based on the dynamic pressure signal, and to identify the mechanical action state of each pulse valve based on the pressure fluctuation characteristics in the pulse valve action waveform sequence. The escape judgment module is used to collect a dust concentration signal corresponding to the blowing sequence of the target pulse valve when the mechanical action state of the target pulse valve meets the preset action standard based on the pressure fluctuation characteristics, and to determine whether dust escape has occurred based on the blowing sequence and the concentration change trend of the dust concentration signal. The fault location module is used to perform time-retrospective matching between the concentration change trend of dust escape and the blowing time of the target pulse valve when dust escape occurs, so as to locate the area where the faulty filter bag is located.

Images