Intelligent monitoring and accurate positioning method for bag breaking of dust collector based on big data time series analysis

Abstract

This invention discloses a method for intelligent monitoring and precise positioning of dust collector bag breakage based on big data time series analysis, specifically relating to the field of industrial dust monitoring technology. It involves simultaneously collecting temperature, humidity, airflow velocity, and dust concentration at the flue gas inlet and outlet and writing them into a multimodal time series dataset. The method records the operating condition control sequence, constructs a coupling feature vector containing pairwise and three-source interaction terms, recursively calculates the coupling contribution parameter vector, applies projection operator constraints, and outputs a decoupled dust concentration sequence, a residual sequence, and a coupling dominant marker. Within the coupling dominant marker indication segment, the coupling contribution is gated and converged to generate an interaction influence map and path weights. Based on the path weights and residual sequence, an interference coupling map and a sequence of handling action instructions are generated. The handling actions are written into the handling event marker sequence and handling fragment records, and an index is established for reuse, thus solving the problems of misjudgment and difficulty in closing the loop due to multi-source interference coupling.

Description

Technical Field

[0001] This invention relates to the field of industrial dust collection monitoring technology, and more specifically, to a method for intelligent monitoring and precise positioning of dust collector bag breakage based on big data time-series analysis. Background Technology

[0002] In high-temperature, high-dust industrial settings such as lithium battery material production and cement manufacturing, baghouse dust collectors are used to purify dust particles in flue gas and ensure emissions meet standards, while also protecting downstream process equipment such as denitrification systems for stable operation. In current operation and maintenance, when filter bags are damaged or partially leaking, on-site inspection typically relies on shutdown for maintenance and manual chamber-by-chamber checks to pinpoint the damage location. This not only causes production interruptions and increased maintenance costs but may also lead to excessive dust emissions and wear risks to downstream equipment due to delayed response. To achieve online monitoring, some systems use outlet dust concentration, pressure differential, and cleaning signals for anomaly detection or matching with cleaning sequences. However, under complex operating conditions, false alarms and inaccurate location are prone to occur, making it difficult to stably and quickly indicate the damaged chamber or section without shutting down the system.

[0003] A key reason for the aforementioned problems lies in the coupling effect of multi-source physical interference during dust collector operation. In conditions such as the roasting section and kiln tail, fluctuations in raw material moisture content and fuel combustion cause interactive changes in flue gas temperature, humidity, and airflow velocity. Humidity and temperature changes can induce particle agglomeration and condensation due to moisture absorption. Uneven airflow distribution amplifies the settling and resuspension of agglomerates, resulting in irregular composite peaks in the dust concentration detection signal, and the stable one-to-one correspondence between the signal and the dust removal action is no longer maintained. Under these circumstances, relying solely on a single dust concentration or a simple threshold and time-series matching method can easily misjudge the composite fluctuations caused by the coupling of operating conditions as filter bag abnormalities, or fail to distinguish between coupled disturbances and concentration deviations caused by actual bag rupture when abnormalities occur. This leads to unstable fault location results, frequent repeated troubleshooting, and affects the reliable operation and continuous production of the dust collection system. Summary of the Invention

[0004] To overcome the aforementioned deficiencies of existing technologies, this invention provides a method for intelligent monitoring and precise positioning of dust collector bag breakage based on big data time series analysis. By aligning multimodal time series datasets with operating condition control sequences using unified timestamps, a coupled feature vector containing interaction terms is constructed, and the coupling contribution parameter vector is recursively obtained and a projection operator constraint is applied. Based on the coupling-dominant marker gating convergence, an interaction influence map and path edge weights are generated, and an interference coupling map and a sequence of handling action instructions are generated and the index is written back for reuse, thereby solving the problems mentioned in the background technology.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a method for intelligent monitoring and precise positioning of dust collector bag breakage based on big data time-series analysis, comprising: Step S1: Simultaneously collect temperature, humidity, airflow velocity and dust concentration at the flue gas inlet and outlet of the bag filter and write them into a multimodal time series dataset. At the same time, the operating condition control sequence is formed by the dust removal control quantity, the induced draft fan control quantity and the baffle control quantity. Step S2: Construct a coupled feature vector from the multimodal time series dataset containing the interaction terms of temperature and humidity, temperature and airflow velocity, humidity and airflow velocity, and the interaction terms of the three sources of temperature, humidity, airflow velocity. Recursively obtain the coupling contribution parameter vector and apply projection operator constraints to it to generate the decoupled dust concentration sequence, residual sequence and coupling dominant label. Step S3: Gated convergence of coupling contributions within the segment indicated by the coupling dominance marker, outputting the interaction influence map and path edge weights; the gated convergence filters valid sampling points under a unified timestamp using the coupling dominance marker, and performs segmented convergence of temperature and humidity interaction contributions, temperature-airflow-velocity interaction contributions, humidity-airflow-velocity interaction contributions, and temperature-humidity-airflow-velocity three-source interaction contributions on the valid sampling points, and normalizes them according to the number of valid sampling points to generate the path edge weights; the interaction influence map uses temperature, humidity, and airflow velocity as nodes, the edge set corresponding to the above interaction contributions as the structure, and outputs the path edge weights as edge attributes; Step S4: Generate an interference coupling map and a sequence of handling action instructions based on the path edge weights and residual sequences, and write the handling actions into the handling event mark sequence and handling fragment records. In addition, establish an index for handling fragment records according to the switching mode of the dominant path mark and the operating condition control sequence for subsequent generation of handling action instruction sequences.

[0006] Preferably, in step S1, the temperature sensor and humidity sensor are set at the same sampling position or adjacent sampling positions, the airflow velocity probe is set in the straight pipe section upstream of the dust concentration detector, and the multimodal time series dataset is formed by the same data acquisition unit writing to the same timestamp according to the same sampling period; in step S1, the operating condition control sequence is obtained by normalizing the dust removal control quantity, the induced draft fan control quantity and the baffle control quantity respectively and then summing them, and the summing result is aligned and saved with the multimodal time series dataset under the same timestamp.

[0007] Preferably, before constructing the coupled feature vector in step S2, the missing measurement segments of the multimodal time series dataset are filled in using piecewise cubic Hermite interpolation, wherein the endpoint derivative is determined by the difference between adjacent effective samples, and the endpoint derivative is set to zero when the derivative direction is inconsistent with the endpoint increment direction; the coupled feature vector in step S2 includes the corresponding quantities of the temperature sensor, the corresponding quantities of the humidity sensor, the corresponding quantities of the airflow velocity probe, the pairwise interaction terms of the above three, the common interaction terms of the three, and the operating condition control sequence, and each component except the constant term is normalized according to the upper and lower bounds of the commissioning and setting period (min-max normalization).

[0008] Preferably, the coupling feature vector is obtained in the following way: The multimodal time series dataset is reconstructed into an aligned sequence with a uniform sampling step size. For missing measurement points, piecewise cubic Hermit interpolation is used for continuous processing, so that temperature data, humidity data, airflow velocity data, and dust concentration data correspond point by point on the same index to obtain the basic items. Multiply the corresponding values ​​of the temperature sensor and humidity sensor in the basic item point by point to form a temperature-humidity interaction item. Multiply the corresponding values ​​of the temperature sensor and airflow velocity probe point by point to form a temperature-airflow velocity interaction item. Multiply the corresponding values ​​of the humidity sensor and airflow velocity probe point by point to form a humidity-airflow velocity interaction item. Multiply the corresponding values ​​of the temperature sensor, humidity sensor, and airflow velocity probe point by point to form a three-source interaction item of temperature, humidity, airflow velocity. The basic term, interaction term, and operating condition control sequence are concatenated in a predetermined order to obtain a coupled feature vector, and amplitude normalization is applied to each non-zero channel of the coupled feature vector.

[0009] Preferably, step S2 uses recursive least squares to update the coupled contribution parameter vector, and after each update, the projection operator restricts the pairwise interaction terms and the common interaction terms of the three to be non-negative and have limited amplitudes, removes the sign constraints of the parameters of the corresponding terms of the operating condition control sequence, and restricts their amplitudes to the boundary calculated and fixed by the commissioning and setting period.

[0010] Preferably, the coupling contribution ratio is determined by the ratio of the magnitude of the interaction term contribution under the same unified timestamp to the magnitude of the model output after deducting the constant term (deducting the constant term means deducting the predicted contribution formed by the parameter components corresponding to the constant term from the model output to eliminate the influence of the fixed baseline on the calculation of the coupling contribution ratio), and the obtained ratio is smoothed by a preset window to avoid single-point jitter causing label jumps; the coupling dominant label is a binary label obtained by comparing the smoothed coupling contribution ratio with the commissioning tuning threshold; the magnitude is taken as the absolute value, and the smoothing is taken as the moving average.

[0011] Preferably, in step S3, the residual edge weights and interaction influence graph are obtained in the following way: Within the segment indicated by the coupling dominant marker, the temperature and humidity interaction contributions, temperature-airflow-velocity interaction contributions, humidity-airflow-velocity interaction contributions, and temperature-humidity-airflow-velocity three-source interaction contributions corresponding to the coupling contribution parameter vector are gated and converged and normalized respectively. The four types of contributions are used as the edge set of the interaction influence graph and the path edge weight is used as the edge attribute to output the interaction influence graph. At the same time, the residual sequence is gated and converged within the same segment and the residual edge weight is output. The residual edge weight is used to handle the generation of priority markers.

[0012] Preferably, in step S4, the method for obtaining the interference coupling map and the sequence of handling action instructions is as follows: In the operating condition control sequence, the switching point is identified, and the time range between adjacent switching points is defined as the candidate segment corresponding to the switching mode of the operating condition control sequence. Within each candidate segment, the coupling dominant marker indication segment is obtained by filtering based on the coupling dominant marker, and the starting point of the segment where the coupling dominant marker first enters after the operating condition control sequence is switched is determined as the trigger index for the action. Within each dominant coupling marker indication segment, the edge weights of four types of paths are read and the dominant path marker is determined by comparison. Then, the edge weights of the four types of paths are written into the edge attributes of the interference coupling map according to the pre-fixed display fields and unit conversion rules. The residual sequence in the same segment is used to generate residual segment attributes and written into the segment attribute column of the map. The interference coupling map uses temperature, humidity, airflow speed and dust concentration as display objects, and marks the edges, switching points and segment boundaries corresponding to the dominant path marker. The sequence of action instructions is selected and sorted from the action library by calling the pre-fixed branch rule table according to the dominant path mark. When the subsequent monitoring meets the requirement that the dominant path mark and the working condition control sequence switching mode are consistent, the action segment record index is hit and the corresponding action segment record is read to generate the action instruction sequence and write it to the reference mark.

[0013] Preferably, the disposal segment records are indexed according to the switching mode of the dominant path marker and the operating condition control sequence. The disposal segment records include decoupled dust concentration sequence, residual sequence, and window segments that couple the dominant marker and the operating condition control sequence. When the same index conditions are met in subsequent monitoring, the corresponding disposal segment records are read to generate a disposal action instruction sequence, and a reference mark is written in the disposal event marker sequence to associate the current disposal with historical disposal segments.

[0014] Preferably, the gating convergence in step S3 is performed as follows: Valid point screening: Valid sampling points are screened using composite gating conditions. The composite gating conditions require that the coupled dominant marker is valid and that the dust removal control quantity in the working condition control sequence under the same time stamp is not in the dust removal triggering section. Interactive contribution processing: For the selected valid sampling points, calculate the interactive contributions of temperature and humidity, temperature and airflow velocity, humidity and airflow velocity, and the three-source interactive contributions of temperature, humidity, airflow velocity, and temperature, humidity, airflow velocity, and then perform segment aggregation after squaring the contribution values ​​of each type. Normalization and robust output: Divide the convergence result by the sum of the number of effective sampling points and the pre-fixed stable small quantities to generate the final path edge weights; Consistency verification: Calculate the differential indicator sequence based on the decoupled dust concentration sequence, and use this sequence to perform intra-segment consistency verification on the selected valid sampling points. Only output the interaction influence map and path edge weights corresponding to the segments that pass the verification.

[0015] Preferably, the method further includes: In step S1, the dust concentration detector is configured as a heterogeneous dual-channel system, with the two channels being a triboelectric channel and an optical channel, respectively. The generation of the coupling dominance marker can only enter the effective coupling dominance section when the data of the two channels are consistent. The consistency of the heterogeneous channels is compared by comparing the concentration change trends of the two channels of the dust concentration detector. If the responses of the two channels to the dust concentration change are consistent, the generation of the coupling dominance marker is allowed. If the responses of the two channels are inconsistent, it is determined to be a measurement disturbance, and the coupling dominance marker corresponding to that section is forcibly set to invalid.

[0016] Preferably, the method further includes: In step S2, a false peak confidence coefficient is constructed and used as a prerequisite for determining the generation of the coupling dominant marker. The calculation of the false peak confidence coefficient takes the corresponding values ​​of the temperature sensor, humidity sensor, airflow velocity probe, and operating condition control sequence under a unified timestamp as inputs. First, a thermal condensation tendency index and a flow field phase mismatch index are formed. The thermal condensation tendency index is obtained by converting the corresponding values ​​of the temperature sensor and humidity sensor to the dew point value. Then, the dew point proximity value is constructed based on the difference between the corresponding values ​​of the temperature sensor and the dew point value. The minimum value of the dew point proximity value and the decrease amplitude of the dew point proximity value are extracted within the sliding time window and then monotonically saturated mapping is performed to obtain the flow field. The phase mismatch index is determined by identifying the dust removal trigger time as the event anchor point in the dust removal control quantity in the operating condition control sequence. Within a preset search window for the dust removal event, the response lag of the corresponding quantity from the airflow velocity probe to the dust removal event is calculated. The response lags of multiple dust removal events are statistically distributed according to lag levels to obtain an empirical distribution, which is then mapped by its dispersion degree. The thermal and moisture condensation tendency index and the flow field phase mismatch index are used as two-dimensional inputs. Based on a mapping table solidified during the commissioning and tuning period, the false peak reliability coefficient is determined. The mapping table consists of a two-dimensional grid composed of the thermal and moisture condensation tendency index level interval and the flow field phase mismatch index level interval. The false peak reliability coefficient is generated by reading the coefficient values ​​of the corresponding grid cells. The technical effects and advantages of this invention are as follows: (1) This invention collects temperature, humidity, airflow velocity and dust concentration at the flue gas inlet and outlet of a bag filter and writes them into a multimodal time series dataset. At the same time, the working condition control sequence is formed by the dust removal control quantity, the induced draft fan control quantity and the baffle control quantity. This enables traceable alignment of multiple physical quantities and working conditions under the same unified time stamp, effectively solving the problems of inconsistent monitoring data sources and difficulty in distinguishing fault signals and working condition disturbances under high temperature, high humidity and airflow fluctuation conditions.

[0017] (2) This invention constructs a coupling feature vector containing temperature and humidity interaction terms, temperature and airflow velocity interaction terms, humidity and airflow velocity interaction terms and the interaction terms of the three sources of temperature, humidity and airflow velocity, recursively obtains the coupling contribution parameter vector and applies projection operator constraints, generates decoupled dust concentration sequence, residual sequence and coupling dominant marker, and gates and converges the coupling contribution output interaction influence map and path weights in the coupling dominant marker indication section, and then generates interference coupling map and disposal action instruction sequence based on path weights and residual sequence, writes them into disposal event marker sequence and disposal fragment record and establishes index reuse, realizes closed-loop collaboration of identification-interpretation-disposal-writeback, effectively solves the problem of frequent false alarms caused by multi-source interference coupling, lack of basis for positioning and disposal and difficulty in reuse. Attached Figure Description

[0018] Figure 1 For reference in this invention Figure 1 A simplified flowchart of the intelligent monitoring and precise positioning method for bag breakage in dust collectors.

[0019] Figure 2 This is a simplified flowchart of the multi-source interference coupling decoupling analysis process of the present invention. Detailed Implementation

[0020] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

[0021] At the same time, it should be understood that, for ease of description, the dimensions of the various parts shown in the accompanying drawings are not drawn according to actual scale.

[0022] The following description of at least one exemplary embodiment is merely illustrative and is in no way intended to limit the scope of this application and its application or use.

[0023] Techniques, methods, and equipment known to those skilled in the art may not be discussed in detail, but where appropriate, such techniques, methods, and equipment should be considered part of the specification.

[0024] Example 1, see Figure 1 A simplified flowchart of a method for intelligent monitoring and precise positioning of dust collector bag breakage is provided in this invention. Figure 1 The method for intelligent monitoring and precise positioning of dust collector bag breakage based on big data time series analysis, as shown, includes: Step S1 (Multi-source data acquisition and operating condition sequence formation): Temperature, humidity, airflow velocity and dust concentration are collected simultaneously at the flue gas inlet and outlet of the bag filter and written into a multi-modal time series dataset. At the same time, the operating condition control sequence is formed by the dust removal control quantity, the induced draft fan control quantity and the baffle control quantity. The explanation is that this step establishes a multi-physics synchronous sensing array in the flue gas channel of the bag filter, so that the temperature, humidity, airflow velocity and dust concentration of the same flue gas at the same time are collected consistently and formed into multimodal time series data that can be used for subsequent time series analysis. This provides a traceable data foundation for identifying the composite peak mode caused by the coupling of temperature, humidity and airflow unevenness. In scenarios such as spodumene roasting sections and cement kiln tails, flue gas typically carries high concentrations of particulate matter, along with water vapor and acidic gas components. Fluctuations in fuel combustion and differences in raw material moisture content can cause rapid temperature jumps in the flue gas. Simultaneously, condensation and secondary carryover may occur in local cold-end areas within the flue. Combined with the flow field deflection caused by inlet guides and box structures, this easily leads to an airflow state where laminar and turbulent flows intermingle. At this time, dust particles are affected by changes in particle thermal state due to temperature variations and by hygroscopic agglomeration due to humidity variations during transport. Furthermore, the settling and resuspension of agglomerates are further amplified by uneven airflow velocity distribution, resulting in composite peak values ​​from dust concentration detectors that do not correspond one-to-one with the dust removal actions. If only a single-point dust concentration or a single physical quantity is used as input, subsequent time-series analysis will find it difficult to distinguish between measurement disturbances caused by operating condition coupling and emission anomalies caused by filter bag damage. Therefore, it is necessary to record multiple physical quantities in parallel within the same spatial segment and at the same time reference to preserve process evidence of coupling. Therefore, the processing logic of the array deployment is to first determine the spatial segment where coupling occurs, and then transcribe the flue gas state within that segment into data in an alignable manner. Specifically, the main deployment lines are the dust collector inlet flue, the main airflow channel above the ash hopper, the confluence section of each clean air chamber, and the dust collector outlet flue. Temperature sensors, humidity sensors, airflow velocity probes, and dust concentration detectors are simultaneously installed at each deployment point, and the signals from each sensor are connected to the same data acquisition unit. To avoid sensor drift caused by high flue gas temperature and dust erosion, temperature and humidity sensors are preferentially placed in sampling chambers or protective tubes with protective sleeves. The dust concentration detector is placed in the outlet straight pipe section, ensuring its detection direction is consistent with the mainstream direction. The airflow velocity probe is placed at a position that represents the mainstream velocity. Avoid obvious backflow areas; when there is a significant risk of cold-end condensation on site, the sampling pipeline at the humidity sensor location is equipped with a heat tracing and condensation deflection structure, and the status of the heat tracing switch is written as an operating condition marker into the data acquisition unit to ensure that the source status of the humidity data can be traced; after the array is installed, the dust concentration detector is first zero-point calibrated during the shutdown or low-load phase, the temperature sensor and humidity sensor are checked for consistency, and the airflow velocity probe is calibrated in the direction; then, after production enters a stable feeding phase, the data acquisition unit is started, and it continuously writes multi-channel data according to the preset acquisition cycle, and writes each ash cleaning control signal, induced draft fan frequency conversion command, or baffle opening change and other flow field-related control quantities into the extended channel with the same timestamp to form the output of a multimodal time series dataset.

[0025] To facilitate understanding, one implementation method is illustrated as follows: At the site of a bag filter in a spodumene roasting section, the construction personnel first install a short flanged pipe with a sampling port on the maintenance short section of the inlet flue of the dust collector, and insert the temperature sensor and humidity sensor into the same sampling chamber. The heat tracing cable is fixed on the outside of the sampling chamber and connected to the local control box. Subsequently, an installation seat is set at the straight pipe position of the clean air chamber confluence section, the dust concentration detector is fixed in the mainstream direction, and an airflow velocity probe is installed on the side wall of the same cross section upstream of it. After mechanical fixing, the electrical personnel led the four types of sensor cables to the terminal block of the data acquisition unit, and powered on each channel in sequence. They first checked the stability of the temperature, humidity, airflow velocity, and dust concentration displays on the control box panel, and then confirmed through the data acquisition unit's recording interface that the four curves were scrolling synchronously at the same time scale. When the furnace briefly increased the fuel supply, causing the flue gas temperature to rise, the temperature curve on the field screen changed first, followed by fluctuations in the humidity curve. The airflow velocity curve fluctuated accordingly after the induced draft fan command changed, and the dust concentration curve showed a peak-shaped change within the same time window. The data acquisition unit simultaneously wrote the induced draft fan command and the dust removal control signal. At this point, field personnel could obtain the recorded results of multiple physical quantities changing simultaneously in the same spatial segment without disassembling the dust collector. Subsequently, they only needed to export this multimodal time series dataset to the host computer or local storage to proceed to the next step of data fusion and decoupling processing.

[0026] Step S2 (Multi-source interference coupling decoupling analysis): Construct a coupling feature vector containing temperature and humidity interaction terms, temperature-airflow-velocity interaction terms, humidity-airflow-velocity interaction terms and the interaction terms of the three sources of temperature, humidity and airflow-velocity from the multi-modal time series dataset. Recursively obtain the coupling contribution parameter vector and apply projection operator constraints to it to generate decoupled dust concentration sequence, residual sequence and coupling dominant label. The initial coupling contribution parameter vector is obtained as follows: during the commissioning and tuning period, a continuous multimodal time series dataset is selected, and the coupling feature vector is constructed according to step S2 and paired with the corresponding quantity of the dust concentration detector. The initial coupling contribution parameter vector estimate is obtained by one-time constrained least squares solution. At the same time, the covariance state quantity is initialized as a diagonal matrix and its diagonal elements are set according to the channel normalization coefficient. The initial coupling contribution parameter vector estimate and the covariance state quantity are used as the starting state of the recursive constrained least squares for subsequent time-stamp updates. Background Summary: Step S2 of this invention takes the multimodal time series dataset output from step S1 as input and transforms it into a coupled feature vector with interaction terms. Specifically, it establishes a computable coupling characterization around the interaction perturbation of dust concentration by temperature, humidity, and airflow velocity, separates the components attributable to operating condition coupling in dust concentration from the original observations, solves the coupling contribution parameters, and outputs the decoupled dust concentration sequence. At the same time, it forms the decoupled dust concentration sequence, the coupling dominant label sequence, and the coupling contribution parameter group that can be called in step S3. This ensures that the subsequent interaction influence map is based on consistent data objects and is not misled by composite peaks. It records the composite peaks caused by operating condition coupling and the concentration deviations caused by filter bag anomalies separately at the data level, avoiding composite peaks from influencing subsequent judgments.

[0027] Technical Motivation: In baghouse dust collectors at the spodumene roasting section and cement kiln tail, flue gas carries dust particles along with water vapor and corrosive gases. Fluctuations in fuel combustion and differences in raw material moisture content can trigger temperature jumps. The cold end of the flue and the structure of the casing may induce condensation and local eddies, causing the airflow velocity distribution to deflect and reflux at different cross sections. At this time, the peak value observed by the dust concentration detector is often not the result of a single cleaning action, but rather the superposition of temperature-induced hygroscopic agglomeration, irregular settling and resuspension of agglomerates in a non-uniform flow field. If the dust concentration sequence is directly used for time-series matching, the composite peak value will be mistakenly taken as a sign of bag breakage. Therefore, it is necessary to explicitly write the interaction of temperature, humidity, and airflow velocity into the features, quantify their contribution to dust concentration using a unified solution process, and then separate this contribution from the observation, retaining the remaining deviation for subsequent positioning, and separating the coupled disturbance from the abnormal deviation for recording.

[0028] See Figure 2 A simplified flowchart of the multi-source interference coupling decoupling analysis process is provided. The implementation process of step S2 includes: Step 201: Time Alignment and Interpolation The multimodal time series is reconstructed into an aligned sequence with a uniform sampling step size. The missing measurement points in each channel are processed by piecewise cubic Hermit interpolation to obtain the basic items under the unified time index. Step 202: Construction of Coupled Feature Vectors Based on the basic terms, temperature and humidity, temperature and airflow velocity, humidity and airflow velocity, and three-source interaction terms are constructed point by point, and then spliced ​​with the operating condition control sequence to form a coupled feature vector. Amplitude normalization is applied to the non-zero channels. Step 203: Recursive Parameter Estimation and Constraints The coupled contribution parameter vector is updated by time-stamping using recursive constrained least squares, and projection operator constraints are performed after each update to ensure that the parameters meet the sign and magnitude boundary requirements. Step 204: Decoupling and Residual Generation The model output is generated based on the coupled feature vector and constrained parameters. The contribution of the interaction term is subtracted to obtain the decoupled dust concentration sequence. At the same time, the difference between the model output and the measured concentration is calculated to form the residual sequence. Step 205: Coupling Dominant Marker Formation The contribution ratio of the interaction item is calculated as the coupling contribution ratio, which is compared with the commissioning and setting threshold to generate a coupling dominant marker. Then, the effective marker points that are continuous in time are grouped into the coupling dominant marker indication segment.

[0029] In one possible embodiment, step S2 is implemented as follows: Step 201: Based on the multi-channel records of the same time and spatial segment, first solve the problems of time axis alignment and continuity of missing segments (that is, reconstruct the multimodal time series dataset into an aligned sequence according to a unified sampling step size, and use piecewise cubic Hermit interpolation to maintain the shape consistency of monotonic segments and avoid misconstructing missing segments into extra peak shapes. The interpolation is performed separately in each channel, but aligned after output with a unified time index), so that temperature data, humidity data, airflow velocity data, and dust concentration data correspond point by point on the same index to obtain the basic items; Step 202: Construct a coupled feature sequence containing interaction terms, so that the combined effects of temperature and humidity, temperature and airflow velocity, and humidity and airflow velocity are explicitly expressed; wherein each pairwise interaction term and the common interaction term of the three are formed by multiplying the corresponding quantities point by point under a unified timestamp; the basic terms, interaction terms and operating condition control sequence are concatenated in a predetermined order to obtain the coupled feature vector. This explains why, when increased temperature promotes humidity-induced agglomeration and uneven airflow amplifies agglomerate settling, the impact on dust concentration typically exhibits a structure that is difficult to explain with univariate variables but is better explained by interaction terms. Therefore, this step constructs coupled feature vectors based on the aligned sequences. Write the second-order and third-order interaction terms into the same feature space: ; Coupled feature vectors Indicates the first Points are used to interpret the dust concentration sequence. The feature set, with values ​​being real number vectors, is used to carry the interaction effects; the constant term 1 is the intercept term, with a constant value, used to absorb the fixed bias; the interaction term... This represents the interaction between temperature and humidity, taking non-negative real numbers, and is used to characterize the combined effects of thermally induced condensation and hygroscopic aggregation; the interaction term... This represents the interaction between temperature and airflow velocity, taking non-negative real numbers to characterize the combined effect of thermal state changes and transport conditions; the interaction term... This represents the interaction between humidity and airflow velocity, taking non-negative real numbers to characterize the combined effects of viscous aggregation and resuspension conditions; the interaction term... This is a three-source interactive representation, with values ​​being non-negative real numbers, used to characterize the common driving term of the composite peak; operating condition control sequence. This step is used to include dust removal and airflow adjustment in interpretable terms, avoiding treating control actions as abnormalities; to avoid solution bias caused by different dimensions, this step... Amplitude normalization is applied to each non-zero channel. Normalization only changes the numerical scale and does not change the construction relationship of the interaction term. The normalization coefficients are written into the parameter group for use in step S3. Step 203: Use recursive constrained least squares with a forgetting factor to solve for the coupling contribution parameters, so that the parameters can be updated with the change of working conditions and maintain the consistency of the physical direction; the specific operation is as follows: Using the coupled feature vector as the independent variable and the corresponding quantity of the dust concentration detector as the dependent variable, a recursive constrained least squares algorithm is used to update the coupled contribution parameter vector time-stamp by time stamp to obtain the parameter estimation result under the current unified time stamp. Immediately after each parameter update, a projection operator constraint is applied. This projection operator constraint is executed in a component-by-component interval projection manner, that is, for each parameter component of the coupled contribution parameter vector, it is determined whether it falls within the corresponding amplitude boundary interval. If it exceeds the interval, it is truncated to the corresponding boundary value. Specifically, the parameter components corresponding to pairwise interactions and common interactions among the three are limited to non-negative values ​​and have restricted amplitudes. The parameter components corresponding to the operating condition control sequence are limited to allow positive and negative variations and have restricted amplitudes. The amplitude boundaries are calculated from historical data under the same device and sensor configuration during the commissioning and tuning period and are fixed as configuration parameters, ensuring consistent reference to the boundaries during subsequent operation and preventing drift due to sudden changes in operating conditions. A recursive algorithm is used to update the parameters using the parameters and covariance state quantities of the previous time step, combined with the data of the current time step, and the projection operator constraint is executed after the update. Motivation Explanation: When the calcination load changes, the ash removal cycle time switches, or the induced draft fan command changes, the coupling relationship will shift with the operating conditions, and a single fit with fixed parameters is prone to mismatch. Therefore, this step uses recursive constrained least squares with a forgetting factor to ensure that the coupling contribution parameters are updated over time and maintain a physically consistent sign direction; dust concentration sequence With coupling feature vector The relationship is represented as: ; Coupling contribution parameter vector For the first Point parameter estimation, taking values ​​as a real number vector, serves to quantify each feature pair. Contribution; residual sequence The remaining term, which is not explained by the coupling feature, takes a real number and serves to carry out possible abnormal deviations of the filter bag and is used as the input for step S3; In one possible embodiment, the recursive constrained least squares can be implemented in the following or equivalent form: Under a unified timestamp, using the estimated value of the coupling contribution parameter vector at the previous time step and the covariance state quantity at the previous time step as the initial state, the coupling feature vector at the current time step and the corresponding quantity of the dust concentration detector are read. First, the gain vector is calculated from the coupling feature vector, the covariance state quantity, and the forgetting factor. Then, the prediction error is obtained by subtracting the inner product of the current coupling feature vector and the estimated value of the coupling contribution parameter vector at the previous time step from the corresponding quantity of the dust concentration detector. The coupling contribution parameter vector is then recursively corrected using the gain vector and the prediction error to obtain a more accurate coupling contribution parameter vector at the current time step. New value; then the covariance state quantity is recursively updated according to the gain vector, coupling feature vector and forgetting factor to obtain the covariance state quantity used in the next time step; after obtaining the updated value of the coupling contribution parameter vector at the current time step, the projection operator constraint is immediately executed, and the parameter components are pruned according to the component interval projection, wherein the parameter components corresponding to the pairwise interaction terms and the common interaction terms of the three are pruned to be non-negative and do not exceed their amplitude boundaries, and the parameter components corresponding to the operating condition control sequence are pruned to allow positive and negative changes and do not exceed their amplitude boundaries. The amplitude boundaries are calculated and fixed by the commissioning and tuning period, thereby forming a constrained coupling contribution parameter vector for the next unified timestamp calculation; The forgetting factor is a positive number between 0 and 1, used to reduce the weight of historical sampling on the current parameter update; the covariance state variable is a symmetric positive definite matrix with the same dimension as the coupling contribution parameter vector, used to adjust the gain vector and suppress ill-conditioned amplification in each update. Step 204: After completing the projection operator constraints, generate the model output (i.e., predict the dust concentration detector corresponding quantity) based on the coupled feature vector and coupled contribution parameter vector under the current unified timestamp, and decompose the model output into interaction term contribution and non-interaction term contribution; the generation of the decoupled dust concentration sequence is achieved by subtracting the interaction term contribution from the dust concentration detector corresponding quantity (i.e., for each unified timestamp, subtract the interaction term contribution formed by the pairwise interaction terms and the common interaction term of the three to obtain the decoupled dust concentration sequence, so that the decoupled dust concentration sequence retains the part closer to the single source change without abrupt change with the superposition of interaction terms); the generation of the residual sequence is achieved by subtracting the dust concentration detector corresponding quantity from the model output (i.e., for each unified timestamp, subtract the model output from the dust concentration detector corresponding quantity to obtain the residual sequence), and write the decoupled dust concentration sequence and the residual sequence into the output set of step S2 according to the unified timestamp; The explanation is that the contribution of the interaction term is only obtained by multiplying the parameter components corresponding to the interaction terms of temperature and humidity, temperature and airflow velocity, humidity and airflow velocity, and the interaction terms of the three sources of temperature, humidity and airflow velocity with the corresponding interaction terms point by point and summing them up. It does not include constant terms, single variable terms and terms corresponding to the operating condition control sequence. Step 205: Under each unified timestamp, calculate the coupling contribution ratio based on the coupling feature vector and the coupling contribution parameter vector. The coupling contribution ratio characterizes the proportion or dominance of the interaction term contribution in the current model output. Compare the coupling contribution ratio with the threshold determined and fixed during the commissioning and tuning period to generate a coupling dominance marker. The coupling dominance marker provides a segment indication result at the unified timestamp granularity, and the set of unified timestamps that continuously meet the conditions are merged into the coupling dominance marker indication segment. Output the coupling dominance marker and its indication segment to step S3. That is, the coupling dominance marker indication segment is generated by determining the coupling contribution ratio at each unified timestamp (compared with the commissioning setting threshold), and the continuous time segment is formed by automatically merging the time-continuous valid marker points after combining the operating condition control sequence switching rules; the commissioning setting threshold is determined by analyzing the statistical distribution of the coupling contribution ratio under normal operating conditions during the equipment commissioning setting period, and is solidified as a configuration parameter after the setting is completed, which is used to determine the coupling contribution ratio in the subsequent operation phase; In one possible embodiment, the coupling feature vector in step S2, while maintaining the reference relationships of the corresponding quantities of the temperature sensor, humidity sensor, dust concentration detector, and operating condition control sequence, further includes the corresponding quantity of the inlet and outlet temperature difference formed by the corresponding quantities of the temperature sensors at the inlet and outlet of the bag filter flue gas. This corresponding quantity of the inlet and outlet temperature difference participates in the construction of the pairwise interaction terms or the common interaction terms of the three. Furthermore, the corresponding quantity of the airflow velocity probe can be replaced by the differential pressure sequence collected at the corresponding location of the bag filter. After replacement, the pairwise interaction terms and the common interaction terms of the three are still formed according to a unified timestamp. At the same time, the recursive least squares with constraints can be replaced by a windowed quadratic programming solution method to update the coupling contribution parameter vector. After each update, the constraints of the projection operator are still executed, and the generation objects and reference relationships of the decoupled dust concentration sequence, the residual sequence, and the coupling dominant marker remain unchanged.

[0030] Explanation: The component-wise interval projection applies to any parameter component. The processing is ; where the components corresponding to the interaction items are taken and To solidify the amplitude boundaries, for the corresponding components of the operating condition control sequence, a is taken as the lower limit of solidification and b is taken as the upper limit of solidification.

[0031] Step S3 (Gated Convergence and Interaction Influence Map Generation): Within the coupling dominant marker indication segment (referring to the continuous time range where the coupling dominant marker generated in Step S2 is valid on a unified timestamp sequence), the coupling contribution is gated and converged to output the interaction influence map and path edge weights. The gated convergence uses the coupling dominant marker to filter valid sampling points under a unified timestamp, and performs segmented convergence on the temperature and humidity interaction contribution, temperature and airflow velocity interaction contribution, humidity and airflow velocity interaction contribution, and temperature, humidity and airflow velocity three-source interaction contribution on the valid sampling points, and normalizes them according to the number of valid sampling points to generate the path edge weights. The interaction influence map uses temperature, humidity, and airflow velocity as nodes, the edge set corresponding to the above interaction contribution as the structure, and outputs the path edge weights as edge attributes. Among them, the coupling contribution gating convergence refers to first filtering (gating) the time series with the coupling dominant marker in step S3, retaining only the coupling dominant marker as a valid unified timestamp point, and then summarizing (converging) the contributions of various interaction items on the retained point set at the segment level; the coupling contribution corresponds to the interaction item contribution that can be decomposed in step S2, including temperature and humidity interaction contribution, temperature airflow velocity interaction contribution, humidity airflow velocity interaction contribution, and temperature, humidity airflow velocity three-source interaction contribution. Background: Step S3 takes the decoupled dust concentration sequence, residual sequence, coupling dominant marker, and coupling contribution parameter vector output from step S2 as input. It establishes an interaction influence map around the interaction relationship between temperature, humidity, and airflow velocity on dust concentration. Its core purpose is to transcribe the parameters and sequences obtained in step S2 into an interaction influence map, converging the interactions from a temporal dimension into readable path relationships and edge weights. This ensures that the coupling strength between temperature, humidity, and airflow velocity, the time intervals in which coupling occurs, and the correlation direction of residual deviations are characterized by the same set of edge weight rules. In this way, step S4 can distinguish the dominant mechanism sources of coupling dominant segments and abnormal deviation segments before triggering the positioning response, providing a traceable chain of evidence for subsequent positioning responses.

[0032] Technical Motivation: In baghouse dust collectors at the spodumene roasting section and cement kiln tail, temperature jumps alter the thermal state of particles carried by the flue gas, humidity fluctuations promote particle agglomeration due to moisture absorption, and deflection of airflow velocity distribution changes the settling and resuspension conditions of the agglomerates. These three factors often interact to drive a composite peak in dust concentration. Although step S2 has already coupled the contribution parameter vector... With coupling dominant marker By separating the contributions attributable to coupling from the observations, a decoupled dust concentration sequence can be obtained. With residual sequence However, without a process to organize these quantities into a relational structure, field personnel will only see multiple curves and will find it difficult to determine which interaction is dominant under what conditions, or whether residual deviation occurs synchronously with a particular interaction path. Therefore, it is necessary to organize the coupling contribution parameter vector... The weight and The contributions of the interaction terms are aggregated on the timeline and then expressed in a graphical manner, thereby transforming the multi-source interference coupling effect from a curve description to a readable path description.

[0033] Processing logic: This step first uses coupling as the primary marker. As a gating condition, the interaction contribution is limited to the coupling-dominant section to avoid mixing uncoupled fluctuations caused by dust removal actions or short-term disturbances into the relationship calculation. Subsequently, edge weights are calculated for the three two-source interaction paths (temperature-humidity, temperature-airflow velocity, humidity-airflow velocity) and the three-source co-driven path (temperature-humidity-airflow velocity), ensuring that each edge in the interaction influence map corresponds to a computable contribution convergence rule. Then, the residual sequence... With decoupled dust concentration sequence The temporal variation is juxtaposed with the aforementioned edge weights, causing segments where the coupling contribution increases but the residual does not increase accordingly to present different readings on the graph compared to segments where both the coupling contribution and the residual increase simultaneously. The final output is an interaction influence graph and its calculation basis, including edge weight calculation results, the time segments participating in the convergence, and their corresponding... The component reference relationship enables step S4 to reference the same set of edge weight sources when generating the interference coupling map and disposal suggestions.

[0034] Furthermore, the specific implementation process of step S3 includes: Step 301 (Gated Segment Determination and Segment Boundary Fixation): Read the coupling dominant marker output in step S2, and filter point by point using the unified timestamp as the index to obtain the set of effective gated points; merge the temporally continuous effective gated points into the coupling dominant marker indicating segments, and record the start and end unified timestamps of each segment as the segment boundary; when the operating condition control sequence changes, use the switching point as the segment segmentation reference to ensure that the same indicated segment does not cross the switching boundary of the operating condition control sequence; Step 302 (Interaction Item Contribution Extraction and Amplification Processing): Within each coupling dominant marker indication segment, the coupling contribution parameter vector and corresponding interaction items generated in step S2 are read according to a unified timestamp, and the interaction contributions of temperature and humidity, temperature and airflow velocity, humidity and airflow velocity, and the three-source interaction contributions of temperature, humidity, airflow velocity, and temperature are calculated point by point. To avoid the interaction item contributions canceling each other out in the sign direction, resulting in a smaller edge weight, the amplitude processing of each type of interaction item contribution is performed first, that is, the interaction item contribution is converted into a non-negative amplitude sequence before participating in the convergence. The amplitude processing rules are fixed as configuration during the commissioning and tuning period and remain consistent in subsequent operations. In one basic implementation, the amplitude reduction process involves taking the absolute value of the contribution of the interaction item before participating in the aggregation; in an optional implementation, the amplitude reduction process involves taking the square of the contribution of the interaction item before participating in the aggregation, which is used to emphasize the peak-type interaction contribution; both methods solidify the amplitude reduction process method identifier as a configuration parameter and keep it unchanged during the operation of the same device; the stable small quantity is set and solidified as a configuration parameter during the commissioning and tuning period, and is loaded as the lower limit term of the normalization denominator along with the configuration of the data acquisition unit or the host computer.

[0035] Step 303 (Gated Convergence and Path Weight Normalization Generation): For each coupled dominant marker indication segment, gated convergence is performed on the contribution amplitude sequences of the above four types of interaction items. The gated convergence obtains the convergence amount of each type of contribution by accumulating point by point within the segment, and the convergence amount is normalized using the number of effective gated points as the normalization benchmark. When the number of effective gated points is too small, a stable small quantity is introduced to participate in the normalization denominator to avoid numerical divergence. The stable small quantity is a pre-fixed positive number configuration. The four convergence amounts after normalization are output as temperature and humidity path weights, temperature and airflow velocity path weights, humidity and airflow velocity path weights, and three-source common driving path weights, respectively, and are bound and stored with the corresponding segment boundaries. Step 304 (Assembling and Outputting Interactive Influence Map): Using temperature, humidity, and airflow velocity as map nodes, and three two-source interaction edges and one three-source co-driving edge as the map edge set, the path edge weights generated in Step 303 are written into the corresponding edges as edge attributes to form an interactive influence map. Within the same coupling dominant marker indication segment, the residual sequence is subjected to gating convergence and normalization consistent with Step 303 to obtain residual edge weights. The residual edge weights and the interactive influence map are output together. The interactive influence map and path edge weights serve as inputs for generating the interference coupling map and the disposal action instruction sequence in Step S4. The residual edge weights are used as reference quantities required for generating the disposal priority marker in Step S4.

[0036] Furthermore, the gating convergence in step S3 is executed using composite gating conditions. These composite gating conditions include the coupling dominant marker being valid and the dust removal control quantity in the operating condition control sequence not being in the dust removal triggering segment. This further eliminates unified timestamp points corresponding to dust removal triggering within the coupling dominant marker indication segment. When generating path edge weights, the temperature and humidity interaction contributions, temperature and airflow velocity interaction contributions, humidity and airflow velocity interaction contributions, and the temperature, humidity, airflow velocity, and three-source interaction contributions are first squared and then converged into segments. During normalization, a pre-solidified stable small quantity is introduced into the denominator. A differential indication sequence is formed from the decoupled dust concentration sequence. This differential indication sequence is calculated by point-by-point difference between adjacent unified timestamps and is used to screen the valid sampling points under the composite gating conditions for consistency within the segment before outputting the interaction influence map and path edge weights.

[0037] Step S4 (Intelligent Decision Making and Historical Experience Reuse): Generate an interference coupling map and a sequence of handling action instructions based on the path edge weights and residual sequences, and write the handling actions into the handling event mark sequence and handling fragment records. In addition, establish an index for handling fragment records according to the switching mode of the dominant path mark and the operating condition control sequence for subsequent generation of handling action instruction sequences. The dominant path marker is determined by comparing the edge weights of various paths within the same coupled dominant marker indication segment and selecting the dominant one. The sequence of disposal action instructions is determined by calling pre-fixed branch rules according to the dominant path marker among fuel input adjustment, drying input adjustment, induced draft fan control quantity adjustment and ash cleaning cycle adjustment, and written with a unified timestamp bound to the disposal action trigger index. The disposal segment record index uses the dominant path marker and the operating condition control sequence switching mode as a matching key, and in subsequent monitoring, when the matching key is consistent, the corresponding disposal segment record is read to generate the disposal action instruction sequence and a reference mark is written into the disposal event marker sequence.

[0038] Background Summary: Step S4 takes the interaction influence graph edge weight convergence result output from Step S3, the decoupled dust concentration sequence, residual sequence, coupling dominance marker output from Step S2, and the operating condition control sequence output from Step S1 as inputs to form an expression of the interference coupling map without opening the dust collector maintenance door. Its core purpose is to construct the interference coupling map based on the interaction influence graph and sequence markers, unify the interference source, handling actions, and write-back markers on the same time axis, and provide a handling process corresponding to specific operating conditions. This allows on-site personnel to handle coupled disturbances and abnormal deviations in segments and write back the handling results as traceable event markers, thereby forming a closed-loop record of on-site positioning response and process handling.

[0039] Technical Motivation: In the spodumene roasting section and cement kiln tail scenarios, the interaction of flue gas temperature, humidity, and airflow velocity can cause a composite peak in dust concentration. Steps S2 and S3 have already structured the output of coupling contributions and interaction paths, but on-site handling still faces two challenges: First, on-site personnel need to know which type of coupling path is dominant and map it to executable operating actions; otherwise, coupling disturbances may be mistaken for filter bag damage, triggering shutdown for investigation. Second, even if the dominant coupling section is identified, if the handling actions and the sequence changes after handling are not written back and archived, the same warnings will be repeatedly triggered when the same operating conditions reappear, resulting in repeated interventions. Therefore, it is necessary to map the edge weight results of the interaction influence spectrum, the dominant coupling marker, and the residual sequence operating condition control sequence together into an interference coupling map, and clearly define the on-site handling actions and write-back marker rules corresponding to each type of dominant coupling situation in the map, so that the handling process has traceability and reusability.

[0040] Processing Logic: This step follows the gating concept of step S3. First, it divides the coupled dominant but residual-inactive segments and coupled dominant and residual-inactive segments based on the amplitude relationship between the coupling dominant marker and the residual sequence, thereby determining the handling priority. Then, based on the edge weights of the two-source interaction path and the three-source common-drive path obtained in step S3, it determines the dominant coupling path category and maps this category to the dominant edge on the disturbance coupling map. Next, it aligns the disturbance coupling map with the operating condition control sequence to generate a targeted handling action sequence. The content of the handling action sequence includes the sequential relationship of fuel input adjustment, drying input adjustment, induced draft fan command adjustment, and ash cleaning cycle adjustment, and writes the action execution time into the handling event marker sequence.

[0041] Specifically, the implementation process of step S4 includes: Step 401 (Determination of Disposal Trigger Index and Anchoring of Disposal Segment): Read the path edge weights output in Step S3 and the residual sequence output in Step S2, and align them with a unified timestamp; identify switching points in the operating condition control sequence, and define the time range between adjacent switching points as candidate segments corresponding to the switching mode of the operating condition control sequence; within each candidate segment, filter the segments indicated by the coupling dominant marker based on the coupling dominant marker, and determine the starting point of the segment where the coupling dominant marker first enters after the operating condition control sequence switches as the disposal action trigger index, so that the subsequent issuance, writing back, and segment truncation of disposal actions are all bound to the same unified timestamp axis; when there are multiple coupling dominant marker indication segments in the same candidate segment, generate the corresponding disposal action trigger indexes in chronological order and process them sequentially. Step 402 (Dominant Path Marker Determination and Interference Coupling Map Generation): Within each dominant coupling marker indication segment, four types of path weights are read, and the dominant path marker is determined by comparing these weights. The dominant path marker corresponds one-to-one with the path weight categories from Step S3. Simultaneously, the four types of path weights are written into the edge attributes of the interference coupling map using the same display method. The residual sequence within the same segment is used to generate residual segment attributes, which are then written into the segment attribute column of the map using the same gated aggregation method as in Step S3. The interference coupling map is generated in a structured output format, using temperature, humidity, airflow velocity, and dust concentration as map display objects. Edges corresponding to the dominant path markers are marked on the map, and the switching points of the operating condition control sequence and the boundaries of the dominant coupling marker indication segments are also marked on the same map. This allows operators to see the correspondence between the dominant path marker, path weights, residual segment attributes, and switching points on the same interface. Map data is output in segments and bound to segment boundaries for later use in generating and calling subsequent handling action command sequences. Step 403 (Generating Disposal Priority Markers and Arranging Disposal Action Instruction Sequences): Within each coupled dominant marker indication segment, disposal priority marks are generated based on the residual sequence. These priority marks distinguish between two disposal paths: those requiring only operational condition handling and those requiring operational condition handling with additional troubleshooting preparation. Subsequently, using the dominant path mark as the entry point for action orchestration, the order of disposal actions is determined among fuel input adjustment, dryer input adjustment, induced draft fan control quantity adjustment, and ash removal cycle adjustment. The action trigger index is written into each action instruction to achieve unified timestamp alignment. The action orchestration employs pre-defined branching rules: when the dominant path mark corresponds to temperature and humidity interaction contributions, first... The process involves scheduling adjustments for drying inputs followed by adjustments for fuel inputs. When the dominant path marker corresponds to an interaction of temperature, airflow, and velocity, adjustments for induced draft fan control are scheduled first, followed by adjustments for fuel inputs. When the dominant path marker corresponds to an interaction of humidity, airflow, and velocity, adjustments for drying inputs are scheduled first, followed by adjustments for induced draft fan control. When the dominant path marker corresponds to an interaction of temperature, humidity, airflow, and velocity, adjustments are scheduled in the order of fuel input, drying input, and induced draft fan control. When a priority indicator indicates the need for additional investigation preparation, an investigation preparation instruction is appended to the end of the action instruction sequence and bound to the trigger index of the same action, thus forming an executable action instruction sequence. Step 404 (Write-back the event marker sequence, archive the event fragment record, and reuse the index): When an action in the event action instruction sequence is executed, the action number corresponding to the action is written into the event marker sequence, using the event action trigger index as the anchor point. The execution order of the action and the dominant path marker are also written into the auxiliary field of the event marker sequence, so that the event action can be verified on a unified timestamp axis. At the same time, with the event action trigger index as the center, the window segments before and after the action are extracted from the decoupled dust concentration sequence, residual sequence, and coupled dominant marker and operating condition control sequence to form an event fragment record. The event fragment record is indexed according to the dominant path marker + operating condition control sequence switching mode. In subsequent monitoring, when the same dominant path marker appears again and the operating condition control sequence switching mode is consistent, the hit event fragment record is read first. The event action instruction sequence is generated according to the recorded action order and action number. A reference marker is written into the event marker sequence to associate the current event with the historical event fragment, thereby ensuring that the generation of the event action instruction sequence has a traceable and reusable basis.

[0042] Furthermore, in step S4, the dominant path marker is determined by comparing the edge weights of various paths within the same coupled dominant marker indication segment and selecting the category corresponding to the path with the largest edge weight; the sequence of disposal action instructions determines the order of actions such as fuel input adjustment, drying input adjustment, induced draft fan control quantity adjustment, and ash cleaning cycle adjustment by calling a pre-fixed branch rule table based on the dominant path marker, and binds each action instruction with the disposal action trigger index and writes it into the disposal event marker sequence; the disposal segment record index uses the dominant path marker and the operating condition control sequence switching mode as a matching key, and when the matching key matches in subsequent monitoring, it reads the corresponding disposal segment record to generate a disposal action instruction sequence and writes it into the reference marker.

[0043] Furthermore, the operating condition control sequence switching mode is formed by taking a time window before and after the switching point, and symbolically encoding the change direction and trigger state of the dust removal control quantity, induced draft fan control quantity, and baffle control quantity; the encoding rules are solidified into configuration parameters during the commissioning and setting period; the branch rule table is written into the configuration file and solidified during the commissioning and setting period by the correspondence between the dominant path marker and the order of the disposal actions, and subsequent operation reads the corresponding action order according to the dominant path marker index; when the matching key is consistent in subsequent monitoring, the action number and action order in the disposal segment record are directly transcribed into the disposal action instruction sequence, and the current disposal event marker sequence is written into the reference marker to associate with the referenced disposal segment record index entry.

[0044] Example 2: Based on Example 1, Example 2 of the present invention includes a heterogeneous dual-channel consistency determination step and a pseudo-peak credibility gating step. First, to address the issue that dust concentration detectors using different measurement principles exhibit varying sensitivities to humidity condensation, particle size drift, and electrostatic adsorption under the same flue gas conditions; when measurement disturbances caused by temperature-humidity-fluid coupling occur, they often show a strong response from one principle while another responds weakly, resulting in spurious peaks; and when actual bag breakage or sudden increases in actual emissions are more likely to cause anomalies simultaneously across different principles, the method differs from Example 1 in that it further includes a heterogeneous dual-channel consistency determination step, specifically: In step S1, the dust concentration detector is configured as a heterogeneous dual-channel system, with the two channels being a triboelectric channel and an optical channel, respectively. The generation of the coupling dominance marker can only enter the effective coupling dominance section when the data of the two channels are consistent. The consistency of the heterogeneous channels is compared by comparing the concentration change trends of the two channels of the dust concentration detector. If the responses of the two channels to the dust concentration change are consistent, the generation of the coupling dominance marker is allowed. If the responses of the two channels are inconsistent, it is determined to be a measurement disturbance, and the coupling dominance marker corresponding to that section is forcibly set to invalid and does not participate in the processing of step S3 and subsequent steps.

[0045] Specifically, the dust concentration detector at the outlet flue of the bag filter adopts a heterogeneous dual-channel configuration including a triboelectric channel and an optical channel. The two channels perform parallel measurements on the dust-laden gas at the same sampling location and write them into the multimodal time-series dataset at a unified timestamp. The multimodal time-series dataset includes corresponding values ​​from the temperature sensor, humidity sensor, airflow velocity probe, and dust concentration detector, as well as corresponding values ​​from the triboelectric channel and optical channel of the dust concentration detector. A consistency comparison is performed before generating the dominant coupling marker. The consistency comparison uses consistent trends and waveform similarity as dual conditions to distinguish between single-channel spurious peaks caused by wet dust adhesion, condensation film formation, and short-term disturbances in the sampling link, and common disturbances from synchronous changes in both channels. The specific implementation is as follows: 1) Select a sliding window on a unified timestamp sequence, and convert the corresponding quantities of the triboelectric channel and the optical channel within the window into sequences of change directions within the window respectively; when the change directions of the two channels are in the same direction most of the time within the same window, the trend is considered consistent. 2) After removing the baseline offset within the window from the two channel window sequences, calculate the waveform similarity. If the similarity reaches the threshold determined by the commissioning and setting period, the waveform similarity is determined to be consistent. If both conditions are met, mark the time segment covered by the window as the consistency pass segment. If either condition is not met, mark the segment as the measurement disturbance segment.

[0046] To facilitate direct implementation by those skilled in the art, a similarity calculation method independent of specific dimensions is given by the following expression; assuming the triboelectric channel concentration sequence... (The output of the triboelectric channel at index k), the optical channel concentration sequence is as follows: (Optical channel output at index k), i represents the sequential sequence, in the set of indices. Within the window, first take the window mean and form a mean-reduced sequence, then calculate the cosine similarity as the waveform similarity: ; ; Among them, the triboelectric channel concentration sequence and the optical channel concentration sequence correspond to the window mean, respectively. Used to remove baseline bias within the window, the corresponding values ​​of the i-th sequence are: , Similarity The value range is [-1, 1], and it is used to measure the synchronicity and co-directionality of the two channels within the window.

[0047] The threshold (i.e., the commissioning and setting threshold) is determined by the commissioning and setting period and fixed as a configuration item to avoid confusion with the source of thresholds in subsequent steps; the consistency tag is defined as the consistency tag. The generation method is as follows: when the trend is consistent and the similarity is high. If both threshold conditions are met, the segment is passed; otherwise, it is failed. The processing of measurement disturbance segments involves either segment elimination or segment weight reduction, both of which are compatible with the recursive solution of the original step S2: when eliminating segments, the coupling feature vectors within the measurement disturbance segment do not enter the recursive update; when reducing segment weight, the recursive update within the measurement disturbance segment uses a smaller update weight. Regardless of the processing method, the consistency flag serves as a prerequisite constraint for generating the coupling dominance flag in step S2: only when the consistency flag passes is the coupling contribution ratio allowed to be compared with the threshold to generate the coupling dominance flag, thus ensuring that the segments entering the gated convergence step S3 are not dominated by single-channel spurious peaks. In the implementation visible on-site, the two channels of the dust concentration detector are sampled from the same sampling port. An anti-blocking structure is set at the front end of the sampling port. During measurement, the triboelectric channel curve and the optical channel curve can be seen simultaneously on the central control screen. When the flue gas condenses or adheres, causing the optical channel to rise briefly while the triboelectric channel does not change accordingly, the consistency mark is set to fail as the window scrolls. Subsequently, this segment will no longer trigger the coupling dominant mark to enter the effective segment, and the interaction influence spectrum will not accumulate path edge weights in this segment. When the curves of the two channels rise and fall synchronously, the consistency mark is considered to pass, and subsequent steps are run according to the predetermined link and the writing of the disposal event mark sequence and disposal segment record is completed.

[0048] Second, if the path weights of the interaction influence map are directly obtained by interaction contribution gating convergence, they may still be distorted by sudden shifts in particle size distribution or temporary dust layers adhering to the probe surface; and these distortions are often unrelated to actual bag breaking. Based on this, the method also includes a pseudo-peak confidence gating step, specifically including: In step S2, a false peak confidence coefficient is constructed and used as a prerequisite for determining the generation of the coupling-dominant marker. The false peak confidence coefficient is calculated jointly by the thermal and moisture condensation tendency index and the flow field phase mismatch index. The thermal and moisture condensation tendency index is calculated by calculating the relative rate of change of temperature and humidity, and the flow field phase mismatch index is calculated by the hysteresis characteristics of the dust removal control quantity and the airflow velocity probe response. The stringency of the coupling contribution ratio and the threshold comparison is adjusted according to the false peak confidence coefficient. If the false peak confidence coefficient is low (below the preset requirement), a stricter threshold is required for the generation of the coupling-dominant marker. If the false peak confidence coefficient is high (above the preset requirement), the threshold can be relaxed for marker generation, thereby reducing mis-marking caused by measurement disturbances.

[0049] Specifically, in step S2, in addition to calculating the coupling contribution ratio and generating the coupling dominant marker based on the coupling contribution parameter vector and the coupling feature vector, a pseudo-peak confidence coefficient is also introduced as a prerequisite for the generation of the coupling dominant marker. In step S2, a pseudo-peak confidence coefficient is constructed and used as a prerequisite for determining the generation of the coupling dominant marker. The calculation of the pseudo-peak confidence coefficient takes the corresponding values ​​of the temperature sensor, humidity sensor, airflow velocity probe, and operating condition control sequence under a unified timestamp as inputs. First, a thermal condensation tendency index and a flow field phase mismatch index are formed. The thermal condensation tendency index is obtained by converting the corresponding values ​​of the temperature sensor and humidity sensor to the dew point value. Then, the dew point proximity value is constructed based on the difference between the corresponding values ​​of the temperature sensor and the dew point value. The minimum value of the dew point proximity value and the decrease rate of the dew point proximity value are extracted within the sliding time window, and then a monotonically saturated mapping is performed. The flow field phase mismatch index is obtained by identifying the dust removal trigger time as the event anchor point in the dust removal control quantity in the operating condition control sequence. Within the preset search window of the dust removal event, the response lag of the corresponding quantity of the airflow velocity probe to the dust removal event is calculated. The response lags of multiple dust removal events are statistically distributed according to the lag level, and then mapped by their dispersion degree. The thermal and moisture condensation tendency index and the flow field phase mismatch index are used as two-dimensional inputs. The false peak confidence coefficient is determined according to the mapping table solidified during the commissioning and tuning period. The mapping table is composed of a two-dimensional grid consisting of the thermal and moisture condensation tendency index level interval and the flow field phase mismatch index level interval. The false peak confidence coefficient is generated by reading the coefficient value of the corresponding grid cell. .

[0050] The explanation is as follows: the thermal condensation tendency index is calculated from the dew point proximity correlation feature using a monotonically bounded saturated mapping function. The mapping process applies a normalization constraint to the output value, ensuring that the index remains bounded and comparable across all operating conditions. The flow field phase mismatch index is obtained by bounded mapping of the response hysteresis feature under the dust removal event triggering condition after dispersion measurement. This mapping also applies a normalization constraint to the output value to ensure the consistency of the index scale under different operating conditions. The two-dimensional mapping table constructed based on the two normalized indices is set and solidified during the commissioning and tuning period. The coefficient values ​​corresponding to the level intervals of each thermal and wet condensation tendency index and the level interval of the flow field phase mismatch index are all subject to normalization constraints, so that the generated pseudo-peak confidence coefficient has a unified dimension and numerical scale, which is suitable as a decision scheduling factor to participate in the subsequent threshold comparison process. At the same time, by introducing the dew point proximity feature and the dust removal response hysteresis feature into the sliding time window and the preset search window respectively for extraction, the pseudo-peak confidence coefficient reflects the stable evolution characteristics of the operating condition over a time scale, avoiding the judgment bias caused by instantaneous fluctuations or single events.

[0051] Furthermore, the credibility coefficient of the pseudo-peak The value ranges from 0 to 1, where the closer the value is to 1, the higher the probability that the current dust concentration fluctuation is dominated by a real anomaly, and the closer the value is to 0, the higher the probability that the current dust concentration fluctuation is caused by false peaks due to operating conditions such as temperature and humidity condensation and dust removal-flow field phase mismatch. The false peak credibility coefficient is used to characterize the credibility of the dust concentration anomaly signal at the current moment and serves as the basis for effective threshold scheduling. By adjusting the strictness of the judgment threshold, the probability of misjudgment under false peak conditions is reduced.

[0052] Furthermore, after obtaining the pseudo-peak credibility coefficient, the commissioning setting threshold is scheduled to generate an effective threshold based on the pseudo-peak credibility coefficient, and the effective threshold is used to replace the commissioning setting threshold in the threshold comparison of the coupling contribution ratio; the coupling dominant flag is generated when the consistency flag passes and the coupling contribution ratio meets the effective threshold condition; in order to maintain consistency with the data consistency constraints formed in step S1, the generation of the coupling dominant flag is also constrained by the consistency flag, that is, only when the consistency flag passes and the coupling contribution ratio meets the effective threshold condition, the corresponding unified timestamp segment is marked as the coupling dominant flag indication segment, thereby providing a clear segment input for the gating convergence in step S3.

[0053] For ease of understanding, let the coupling contribution ratio be . The commissioning setting threshold is The effective threshold is : Among them, the scheduling coefficient To configure the non-negative parameters, Compare and replace and The comparison serves as the entry point for determining the coupling-dominant marker; simultaneously, in conjunction with the aforementioned consistency marker, only when the consistency marker passes and The coupling-dominant marker is generated only when the effective threshold condition is met. This arrangement incorporates the temperature and humidity condensation mechanism and the dust removal-flow field phase mechanism into the coupling-dominant determination, without changing the main structure of the recursive band constraint least squares update and the projection operator constraint.

[0054] In the implementation methods visible on-site, the commissioning and setting period is selected by the operator when the filter bag is confirmed to be in normal condition and the operating conditions are relatively stable. During operation, if the flue gas temperature and humidity combination causes the dew point to change rapidly and the airflow velocity response corresponding to the dust removal event is delayed and dispersed, the false peak confidence coefficient decreases accordingly, and the effective threshold is raised, making it less likely for the coupling dominant marker to be triggered by the false peak. When the temperature, humidity and flow field phases return to stability, the false peak confidence coefficient increases, the effective threshold falls back, and the coupling dominant marker is generated according to the conventional judgment and enters the subsequent interaction influence spectrum construction and processing fragment record indexing process.

[0055] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for intelligent monitoring and precise positioning of dust collector bag breakage based on big data time series analysis, characterized in that, include: Step S1: Simultaneously collect temperature, humidity, airflow velocity and dust concentration at the flue gas inlet and outlet of the bag filter and write them into a multimodal time series dataset. The operating condition control sequence is formed by the dust removal control quantity, the induced draft fan control quantity and the baffle control quantity. Step S2: Construct a coupled feature vector from the multimodal time series dataset containing the interaction terms of temperature and humidity, temperature and airflow velocity, humidity and airflow velocity, and the interaction terms of the three sources of temperature, humidity, airflow velocity. Update the coupling contribution parameter vector using recursive constrained least squares, and apply the projection operator constraint to the coupling contribution parameter vector after each update to generate the decoupled dust concentration sequence, residual sequence, and coupling dominant label. Step S3: Gated aggregation filters valid sampling points under a unified timestamp using the coupling-dominant marker, and aggregates the interaction contributions of temperature and humidity, temperature and airflow velocity, humidity and airflow velocity, and the three-source interaction contributions of temperature, humidity, airflow velocity, and temperature, humidity, airflow velocity, respectively, and normalizes them according to the number of valid sampling points to generate path edge weights; the interaction influence map uses temperature, humidity, and airflow velocity as nodes, the edge set corresponding to the above interaction contributions as the structure, and outputs the path edge weights as edge attributes; Step S4: Generate an interference coupling map and a sequence of handling action instructions based on the path edge weights and residual sequences, and write the handling actions into the handling event mark sequence and handling fragment records. In addition, establish an index for handling fragment records according to the switching mode of the dominant path mark and the operating condition control sequence for subsequent generation of handling action instruction sequences.

2. The intelligent monitoring and precise positioning method for dust collector bag breakage based on big data time series analysis according to claim 1, characterized in that, The coupling feature vector is obtained as follows: The multimodal time series dataset is reconstructed into an aligned sequence with a uniform sampling step size. For missing measurement points, piecewise cubic Hermit interpolation is used for continuous processing, so that temperature data, humidity data, airflow velocity data, and dust concentration data correspond point by point on the same index to obtain the basic items. Multiply the corresponding values ​​of the temperature sensor and humidity sensor in the basic item point by point to form a temperature-humidity interaction item. Multiply the corresponding values ​​of the temperature sensor and airflow velocity probe point by point to form a temperature-airflow velocity interaction item. Multiply the corresponding values ​​of the humidity sensor and airflow velocity probe point by point to form a humidity-airflow velocity interaction item. Multiply the corresponding values ​​of the temperature sensor, humidity sensor, and airflow velocity probe point by point to form a three-source interaction item of temperature, humidity, airflow velocity. The basic term, interaction term, and operating condition control sequence are concatenated in a predetermined order to obtain a coupled feature vector, and amplitude normalization is applied to each non-zero channel of the coupled feature vector.

3. The intelligent monitoring and precise positioning method for dust collector bag breakage based on big data time series analysis according to claim 2, characterized in that, Step S2 uses recursive least squares with constraints to update the coupled contribution parameter vector. After each update, the projection operator restricts the pairwise interaction terms and the common interaction terms of the three to be non-negative and have limited amplitudes. The sign constraints of the parameters of the corresponding terms of the operating condition control sequence are removed, and their amplitudes are restricted to the boundary calculated and fixed by the commissioning and setting period.

4. The intelligent monitoring and precise positioning method for dust collector bag breakage based on big data time series analysis according to claim 1, characterized in that, The coupling contribution ratio is determined by the ratio of the magnitude of the contribution of the interaction term under the same unified timestamp to the magnitude of the model output after deducting the constant term, and the obtained ratio is smoothed by a preset window to avoid single-point jitter causing label jumps. The coupling dominance label is a binary label obtained by comparing the smoothed coupling contribution ratio with the commissioning setting threshold.

5. The intelligent monitoring and precise positioning method for dust collector bag breakage based on big data time series analysis according to claim 1, characterized in that, In step S3, the residual edge weights and interaction influence graph are obtained as follows: Within the segment indicated by the coupling dominant marker, the temperature and humidity interaction contributions, temperature-airflow-velocity interaction contributions, humidity-airflow-velocity interaction contributions, and temperature-humidity-airflow-velocity three-source interaction contributions corresponding to the coupling contribution parameter vector are gated and converged and normalized respectively. The four types of contributions are used as the edge set of the interaction influence graph and the path edge weight is used as the edge attribute to output the interaction influence graph. At the same time, the residual sequence is gated and converged within the same segment and the residual edge weight is output. The residual edge weight is used to handle the generation of priority markers.

6. The intelligent monitoring and precise positioning method for dust collector bag breakage based on big data time series analysis according to claim 1, characterized in that, In step S4, the method for obtaining the interference coupling map and the sequence of handling action instructions is as follows: In the operating condition control sequence, the switching point is identified, and the time range between adjacent switching points is defined as the candidate segment corresponding to the switching mode of the operating condition control sequence. Within each candidate segment, the coupling dominant marker indication segment is obtained by filtering based on the coupling dominant marker, and the starting point of the segment where the coupling dominant marker first enters after the operating condition control sequence is switched is determined as the trigger index for the action. Within each dominant coupling marker indication segment, the edge weights of four types of paths are read and the dominant path marker is determined by comparison. The edge weights of the four types of paths are written into the edge attributes of the interference coupling map according to the pre-fixed display fields and unit conversion rules. The residual sequence in the same segment is used to generate residual segment attributes and written into the segment attribute column of the map. The interference coupling map uses temperature, humidity, airflow speed and dust concentration as display objects, and marks the edges, switching points and segment boundaries corresponding to the dominant path marker. The sequence of action instructions is selected and sorted from the action library by calling the pre-fixed branch rule table according to the dominant path mark. When the subsequent monitoring meets the requirement that the dominant path mark and the working condition control sequence switching mode are consistent, the action segment record index is hit and the corresponding action segment record is read to generate the action instruction sequence and write it to the reference mark.

7. The intelligent monitoring and precise positioning method for dust collector bag breakage based on big data time series analysis according to claim 6, characterized in that, The disposal segment records are indexed according to the dominant path marker and the switching mode of the operating condition control sequence. The disposal segment records contain decoupled dust concentration sequences, residual sequences, and window segments that couple the dominant marker and the operating condition control sequence. When the same index conditions are met in subsequent monitoring, the corresponding disposal segment records are read to generate a disposal action instruction sequence, and a reference marker is written into the disposal event marker sequence to associate the current disposal with historical disposal segments.

8. The intelligent monitoring and precise positioning method for dust collector bag breakage based on big data time series analysis according to claim 5, characterized in that, The gated aggregation in step S3 is performed as follows: Valid point screening: Valid sampling points are screened using composite gating conditions. The composite gating conditions require that the coupled dominant marker is valid and that the dust removal control quantity in the working condition control sequence under the same time stamp is not in the dust removal triggering section. Interactive contribution processing: For the selected valid sampling points, calculate the interactive contributions of temperature and humidity, temperature and airflow velocity, humidity and airflow velocity, and the three-source interactive contributions of temperature, humidity, airflow velocity, and temperature, humidity, airflow velocity, and then perform segment aggregation after squaring the contribution values ​​of each type. Normalization and robust output: Divide the convergence result by the sum of the number of effective sampling points and the pre-fixed stable small quantities to generate the final path edge weights; Consistency verification: Calculate the differential indicator sequence based on the decoupled dust concentration sequence, and use this sequence to perform intra-segment consistency verification on the selected valid sampling points. Only output the interaction influence map and path edge weights corresponding to the segments that pass the verification.

9. The intelligent monitoring and precise positioning method for dust collector bag breakage based on big data time series analysis according to claim 1, characterized in that, The method further includes: In step S1, the dust concentration detector is configured as a heterogeneous dual-channel system, with the two channels being a triboelectric channel and an optical channel, respectively. The generation of the coupling dominance marker can only enter the effective coupling dominance section when the data of the two channels are consistent. The consistency of the heterogeneous channels is compared by comparing the concentration change trends of the two channels of the dust concentration detector. If the responses of the two channels to the dust concentration change are consistent, the generation of the coupling dominance marker is allowed. If the responses of the two channels are inconsistent, it is determined to be a measurement disturbance, and the coupling dominance marker corresponding to that section is forcibly set to invalid.

10. The intelligent monitoring and precise positioning method for dust collector bag breakage based on big data time series analysis according to any one of claims 3 or 4, characterized in that, The method further includes constructing a pseudo-peak confidence coefficient in step S2 and using it as a prerequisite for determining the generation of the coupling dominant marker; The calculation of the false peak confidence coefficient takes the corresponding values ​​of temperature sensor, humidity sensor, airflow velocity probe, and operating condition control sequence under the same timestamp as input. First, a thermal condensation tendency index and a flow field phase mismatch index are formed. The thermal condensation tendency index is obtained by converting the corresponding values ​​of temperature sensor and humidity sensor to dew point. Then, the dew point proximity value is constructed based on the difference between the corresponding values ​​of temperature sensor and dew point. The minimum value of the dew point proximity value and the decrease rate of the dew point proximity value are extracted within the sliding time window and then monotonically saturated mapping is performed to obtain the result. The flow field phase mismatch index is obtained by identifying the dust removal trigger time as the event anchor point in the dust removal control quantity in the working condition control sequence. The response lag of the corresponding quantity of the airflow velocity probe to the dust removal event is calculated within the preset search window of the dust removal event. The response lag of multiple dust removal events is statistically distributed according to the lag level and then mapped by its dispersion degree. The thermal condensation tendency index and the flow field phase mismatch index are used as two-dimensional inputs. The false peak confidence coefficient is determined based on the mapping table solidified during the commissioning and tuning period. The mapping table is composed of a two-dimensional grid consisting of the thermal condensation tendency index level range and the flow field phase mismatch index level range. The false peak confidence coefficient is generated by reading the coefficient value of the corresponding grid cell.

Images