A coal pipeline blockage early warning method for a thermal power plant

Abstract

The application discloses a kind of coal powder pipeline blockage early warning method of thermal power plant, it is related to the field of electric power industry automation and condition monitoring technique, the present application automatically identifies the flow characteristic difference of different coal, and dynamically adjusts early warning threshold, avoid the false alarm caused by coal variety change;Adopt the way of generative adversarial network and computational fluid dynamics simulation combination, realize the virtual monitoring to key area such as elbow pipe, break through the limitation that only can rely on straight pipe section monitoring in the past;Through high-frequency sampling and cumulative sum algorithm, the transient abnormal capture ability is significantly improved, and the precursory characteristics of blockage can be found earlier;The linkage of hierarchical early warning mechanism and disposal suggestion makes the system not only can find problems, but also can provide targeted solution, greatly improves operation and maintenance efficiency, using software algorithm upgrade mainly, maximum utilization of existing hardware resources, reduces the transformation cost and implementation difficulty.

Description

Technical Field

[0001] This invention relates to the field of power industry automation and condition monitoring technology, and in particular to a method for early warning of blockage in pulverized coal pipelines in thermal power plants. Background Technology

[0002] Currently, in order to adapt to changes in the coal market and reduce fuel costs, thermal power plants generally adopt a multi-coal blending operation mode; different coal types have different physicochemical properties.

[0003] Existing coal powder pipeline blockage early warning systems are mostly based on electrostatic sensing technology, which judges the pipeline status by monitoring the charge signal generated by the flow of coal powder. Such systems issue alarms based on preset thresholds and are effective under conditions of single coal type or stable coal quality. However, under conditions of multi-coal blending, the significant differences in the charge characteristics of different coal types cause the signal baseline to drift and the fluctuation characteristics to change. The fixed threshold early warning mechanism is difficult to adapt to such dynamically changing conditions and may result in early warning deviations.

[0004] To address the aforementioned issues, some existing technologies improve early warning accuracy by adjusting thresholds based on human experience or by adding coal quality testing steps. However, these solutions rely on the experience and judgment of operators or require the introduction of additional coal quality analysis equipment, which leads to adjustment lags and increased system complexity in practical applications. Furthermore, due to frequent changes in blending ratios and coal combinations, the adaptability of these methods remains insufficient. Summary of the Invention

[0005] In view of the aforementioned existing problems, the present invention is proposed.

[0006] This invention provides a method for early warning of blockage in pulverized coal pipelines in thermal power plants, which solves the problems of large differences in pulverized coal characteristics caused by the blending of multiple coal types, high false alarm rate and lack of adaptability of existing fixed threshold early warning systems.

[0007] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0008] In a first aspect, embodiments of the present invention provide a method for early warning of blockage in pulverized coal pipelines of thermal power plants, comprising,

[0009] Step S1: The charge signal is collected by an electrostatic sensor array installed in the pulverized coal conveying pipeline;

[0010] Step S2: Under calibrated operating conditions, normal flow signals of different coal types are collected and features are extracted to establish a coal type feature library;

[0011] Step S3: Based on the feature input of the real-time charge signal, the trained pattern recognition model is used to identify the dominant coal type being transported.

[0012] Step S4: Retrieve the corresponding reference feature parameters from the coal type feature library according to the identification results;

[0013] Step S5: Dynamically adjust the warning threshold based on the reference feature parameters;

[0014] Step S6: Perform anomaly detection on the real-time charge signal based on the adjusted threshold;

[0015] Step S7: When an anomaly is determined, output a graded warning message that includes the warning level and handling suggestions.

[0016] As a preferred embodiment of the early warning method for pulverized coal pipeline blockage in thermal power plants according to the present invention, the electrostatic sensor array consists of at least three sets of electrode units arranged in a circular pattern along the circumference. After synchronous acquisition, bandpass filtering, electrostatic drift compensation and time alignment, the sampled data is used to obtain the time-domain waveform and the frequency-domain characteristics obtained by fast Fourier transform.

[0017] As a preferred embodiment of the early warning method for coal pulverized pipeline blockage in thermal power plants according to the present invention, the pattern recognition model uses wavelet packet decomposition when extracting features, selects the energy distribution, spectral entropy and energy ratio of low-frequency and high-frequency sub-bands as the basis for coal type identification, and trains a classifier offline through labeled data to output the dominant coal type.

[0018] As a preferred embodiment of the early warning method for pulverized coal pipeline blockage in a thermal power plant according to the present invention, the dynamic adjustment of the early warning threshold includes:

[0019] Based on the difference in dielectric properties between the current coal type and the reference coal type, as well as the amplitude and fluctuation statistics in the feature library, the threshold scaling factor and upper and lower limits are calculated, and then applied to online threshold decision-making after combining hysteresis and dead zone to suppress jitter.

[0020] The steps for calculating the threshold scaling factor and upper and lower limits are as follows:

[0021] Extract the dielectric parameters, normal flow amplitude quantiles, and volatility indices of the current coal type and the reference coal type;

[0022] The amplitude quantiles are taken from the robust quantiles observed online, and the volatility index is taken from the robust dispersion within the sliding window;

[0023] Mapping dielectric difference, amplitude quantile, and fluctuation ratio to a single scaling factor:

[0024]

[0025] Where, η t Let be the threshold scaling factor at time t, and clip(·,a,b) denotes truncating the input to the interval [a,b]. For the mapping weights, de R is a measure of dielectric difference. q R is the amplitude quantile ratio. v η is the volatility ratio. min ,η max The upper and lower bounds of the scaling factor are given, and t is the index of the discrete time step.

[0026] In the formula:

[0027]

[0028] Where, ε r tanδ represents the relative permittivity and loss tangent of the current coal type. (tanδ) ref For reference parameters corresponding to different coal types, w ε w tan ∈[0,1] difference weighted and w ε +w tan =1; and These are the p-quantiles of the current and reference normal flow amplitudes, respectively, and D. t D is the current volatility indicator. ref Used as a reference volatility scalar;

[0029] Based on the reference quantile and the current volatility, dynamic upper and lower limits are generated and hysteresis is added:

[0030]

[0031] Among them, L t U t Let be the lower and upper thresholds at time t. As the two quantiles of the reference amplitude, 0 <p1<p2<1,c> 0 represents the volatility amplification factor, η t The scaling factor is h ≥ 0, where h is the hysteresis half-width and L is the scaling factor. min L max U min U max For engineering safety boundaries; a dead zone δ0≥0 is added during online decision-making: the state flip is only triggered when the observation exceeds the boundary by more than δ0.

[0032] As a preferred embodiment of the early warning method for pulverized coal pipeline blockage in thermal power plants according to the present invention, the method further includes: training a generative adversarial network on normal flow data, calculating the similarity score between the real-time signal and the generated signal online, and combining the score with a threshold set based on a fixed false alarm rate for anomaly determination.

[0033] As a preferred embodiment of the early warning method for pulverized coal pipeline blockage in thermal power plants according to the present invention, it further includes: aligning samples of the flow field in the bend region obtained from computational fluid dynamics simulation with the monitoring signals of the straight pipe section, training a convolutional long short-term memory network to establish a mapping relationship, and outputting an estimate of the accumulation state in the bend region during operation and fusing it with the anomaly detection results.

[0034] As a preferred embodiment of the early warning method for pulverized coal pipeline blockage in thermal power plants according to the present invention, the anomaly detection includes transient mutation detection based on cumulative summation, and the statistics obtained by high-frequency sampling are accumulated and reset in real time on a single / double-sided basis, wherein the high-frequency sampling frequency is not less than five times the system's preset conventional sampling frequency;

[0035] The steps for performing single / dual-sided real-time accumulation and reset include:

[0036] The sampling frequency was increased to a higher frequency based on the conventional sampling frequency, and the observations were standardized.

[0037] f hf =m s f b m s ∈[5, 16],

[0038] Among them, f hf This is the high-frequency sampling frequency, measured in Hz, f. b This is the standard sampling frequency, measured in Hz (m). s The range is the multiplier. Recommended multiplier;

[0039] Zero drift compensation and scaling for high-frequency sequences:

[0040]

[0041] Where, x t μ is a statistic for high-frequency time t. ref ,σ ref σ represents the reference position and scale under normal flow conditions. ref >0;

[0042] The incremental detection range is constructed according to the design detection range and accumulated online. An alarm is triggered and the system is reset when the threshold is exceeded.

[0043]

[0044] Upper alarm: Lower side alarm:

[0045] Reset after triggering: or And freeze n lockOne high-frequency sample;

[0046] Among them, s t This represents the log-likelihood increment of shifting the mean upward by d0, where d0 > 0 represents the designed detection amplitude. These are the upper and lower CUSUM statistics, A. + A - >0 represents the corresponding decision threshold. To freeze the sample size; use one-sided when only focusing on sudden increases in amplitude, and use two-sided when focusing on both sudden increases and sudden decreases;

[0047] Under the condition of no change in standard normality, the relationship between the threshold and the false alarm probability / ARL can be directly set based on the ordinal ratio test approximation:

[0048]

[0049] Where, α ± The one-sided false alarm probability for each high-frequency sample. The average alarm interval for one side under unchanged conditions is calculated based on the number of samples; when both sides are effective simultaneously, α is divided equally between the two sides or allocated according to risk preference.

[0050] When m s When increasing the threshold leads to an increase in the correlation between samples, it is equivalent to relaxing the threshold:

[0051] Target ARL0 is determined by the effective independent sample ratio ρ eff ∈(0,1] is corrected to ARL0 / ρ eff ;ρ eff Obtained from the first-order autocorrelation estimation of the high-frequency sequence; the freeze time after the alarm is taken as T. lock =n lock / f hf The typical recovery time for a single wind disturbance is covered;

[0052] After the alarm is triggered, a freeze period begins, and the system stops. Accumulate and update only μ ref ,σ ref The slow channel resumes accumulation after the freeze period ends; if a coal type switching event occurs within the window, then d0 and A are refreshed. ± Configured for this type of coal.

[0053] As a preferred embodiment of the early warning method for pulverized coal pipeline blockage in thermal power plants according to the present invention, the graded early warning is classified according to the duration of the abnormality, the degree of deviation and the spatial consistency, and outputs recommended disposal measures including adjusting the primary air volume, changing the coal feeder speed, separator slag discharge and load reduction operation. The early warning information can be connected to the distributed control system through the communication interface.

[0054] In a second aspect, embodiments of the present invention provide a computer device, including a memory and a processor, wherein the memory stores a computer program, and the computer program, when executed by the processor, implements any step of the method for early warning of blockage in pulverized coal pipelines of thermal power plants as described in the first aspect of the present invention.

[0055] Thirdly, embodiments of the present invention provide a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the method for early warning of blockage in pulverized coal pipelines of a thermal power plant as described in the first aspect of the present invention.

[0056] The beneficial effects of this invention are as follows: This invention effectively solves the problem of false alarms under multi-coal blending conditions. It can automatically identify the differences in flow characteristics of different coal types and dynamically adjust the warning threshold, avoiding false alarms caused by changes in coal type. By combining generative adversarial networks and computational fluid dynamics simulation, virtual monitoring of key areas such as bends is realized, breaking through the limitations of relying solely on straight pipe sections for monitoring in the past.

[0057] This invention significantly improves the ability to detect transient anomalies through high-frequency sampling and cumulative summation algorithms, enabling earlier detection of pre-congestion signs. The linkage between the tiered early warning mechanism and handling recommendations allows the system to not only identify problems but also provide targeted solutions, greatly improving operational efficiency. The primarily software algorithm-based upgrade approach maximizes the use of existing hardware resources, reducing modification costs and implementation difficulty.

[0058] This invention establishes a complete adaptive mechanism from coal type identification to dynamic threshold adjustment, enabling the system to automatically optimize operating parameters as coal types change, truly achieving autonomous learning and continuous improvement of the early warning system. This adaptive capability has significant practical value for coal-fired power plants in coping with the complex and ever-changing coal market situation, providing reliable technical support for the safe and stable operation of power plants. Attached Figure Description

[0059] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly described below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation on the scope of this application.

[0060] Figure 1 This is a flowchart illustrating the early warning method for pulverized coal pipeline blockage in a thermal power plant as described in the embodiment. Detailed Implementation

[0061] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0062] All terms used in this application (including technical and scientific terms) have the meanings commonly understood by those skilled in the art, unless otherwise defined. It should be noted that the terms used herein are to be interpreted in a manner consistent with the context of this specification, and not in an idealized or overly rigid way.

[0063] For example, the terms “first” and “second” used in this application are only used to distinguish and describe similar objects, to differentiate the first object from another object, and are not used to describe a specific order or sequence, nor should they be interpreted as indicating or implying relative importance.

[0064] This application proposes a method for early warning of blockage in pulverized coal pipelines of thermal power plants, combined with... Figure 1 As shown, the method includes:

[0065] Step S1: The charge signal is collected by an electrostatic sensor array installed in the pulverized coal conveying pipeline;

[0066] In this embodiment, the electrostatic sensor array refers to a multi-electrode acquisition unit that adheres to the pipe wall and whose disturbance to the flow field inside the pipe is negligible. In engineering practice, a charge amplifier and anti-saturation limiting are used as the basic channel, with shielding and single-point grounding to reduce common-mode interference. To ensure multi-channel time consistency, synchronous acquisition uses the same clock source or backplane bus triggering to achieve alignment accuracy within ±1 sampling point across channels. The default sampling frequency is 2kHz, adjustable from 1-5kHz, set according to the existing device bandwidth and DCS throughput. The default single-shot feature window is 1.0s, adjustable from 0.5-2.0s, determined by a trade-off between abnormal response timescales and statistical stability. In case of missing measurements or single-channel saturation, interpolation of adjacent channels and amplitude clipping are used to maintain channel availability. If the synchronization error exceeds the preset upper limit, the feature calculation for that window is skipped and an invalid sampling status is reported.

[0067] Step S2: Under calibrated operating conditions, normal flow signals of different coal types are collected and features are extracted to establish a coal type feature library;

[0068] Specifically, the calibration condition refers to the stable range where the primary air volume, pulverized coal content, feed coal moisture content, and unit load are within the allowable range of the operation and maintenance procedures, and the duration is no less than 10 times the feature window to cover short-term fluctuations; at least 3 load segment samples are collected for each type of coal to reflect cross-load robustness. Robust quantiles and robust dispersion in feature extraction refer to exponentially weighted quantile estimates and exponentially weighted median absolute deviation indicators, respectively, to avoid the influence of extreme values. The default cumulative effective duration for each coal type is ≥30 minutes, adjustable from 15-120 minutes, set according to the variability within the coal type and the statistical convergence speed. When coal quality test parameters are incomplete, the dielectric parameter field is temporarily filled with online inversion estimates and the source is marked for subsequent backfeeding correction.

[0069] Step S3: Based on the feature input of the real-time charge signal, the trained pattern recognition model is used to identify the dominant coal type being transported.

[0070] For example, the dominant coal type refers to the coal type whose category confidence score in the current window output exceeds the confidence threshold. When all confidence scores are below the threshold, the output is uncertain and maintains the previous stable judgment until the confidence recovers after several consecutive windows. The confidence threshold is 0.6 by default and can be adjusted from 0.5 to 0.8, based on the precision-completeness trade-off of offline cross-validation. The minimum holding time for stable judgments is 20 seconds by default and can be adjusted from 10 to 60 seconds, depending on the blending rhythm and false detection cost. Optionally, when the number of categories increases, a hierarchical judgment is adopted to first coarsely classify coal types and then subdivide them into grades, in order to shorten the convergence time without changing the input and output definitions. The classifier input features and training process are consistent with the aforementioned embodiments and have not been changed.

[0071] Step S4: Retrieve the corresponding reference feature parameters from the coal type feature library according to the identification results;

[0072] Similarly, the reference feature parameters include three types of fields: amplitude quantile vector, volatility scalar, and dielectric parameter. When calling the reference, records matching the current load segment are used first; if none are found, the robust summary value for the entire segment of the same coal type is used. The refresh cycle is pulled every 30 seconds by default, adjustable from 10-120 seconds; when the identification result changes, it is refreshed immediately to reduce latency. If the feature library entry for that coal type is missing or too old, this embodiment reverts to the online steady-state estimate from the most recent hour and marks a temporary reference.

[0073] Step S5: Dynamically adjust the warning threshold based on the reference feature parameters;

[0074] Step S6: Perform anomaly detection on the real-time charge signal based on the adjusted threshold;

[0075] Step S7: When an anomaly is determined, output a graded warning message containing the warning level and handling suggestions;

[0076] In one embodiment, the electrostatic sensor array consists of at least three sets of electrode units arranged in a ring along the circumference. After synchronous acquisition, bandpass filtering, electrostatic drift compensation and time alignment, the sampled data is used to obtain the time-domain waveform and the frequency-domain characteristics obtained by fast Fourier transform.

[0077] Furthermore, the bandpass filter's passband defaults to covering the main frequency band carrying energy and particle collision information, with an initial value of 5-300Hz and an adjustable value of 3-500Hz, determined based on the experimental spectrum and mechanical resonance avoidance band. Electrostatic drift compensation is estimated through the slow channel and subtracted from the observations. The slow channel time constant defaults to 60-180s and is adjusted according to ambient temperature, humidity, and insulation conditions. 3-8 sets of ring electrodes meet the requirements for direction-insensitive acquisition. When the pipe diameter is large and installation is limited, a non-equidistant arrangement can be optionally used to maintain overlap in the field of view of adjacent electrodes, thus maintaining spatial consistency judgment capability. When the noise of a single channel exceeds the preset upper limit or DC saturation occurs, that channel participates in frequency domain statistics but is weighted less in spatial consistency calculations.

[0078] In one embodiment, the pattern recognition model uses wavelet packet decomposition when extracting features, selects the energy distribution, spectral entropy and energy ratio of low-frequency and high-frequency sub-bands as the basis for coal type identification, and trains a classifier offline through labeled data to output the dominant coal type.

[0079] In one embodiment, dynamically adjusting the warning threshold includes:

[0080] Based on the difference in dielectric properties between the current coal type and the reference coal type, as well as the amplitude and fluctuation statistics in the feature library, the threshold scaling factor and upper and lower limits are calculated, and then applied to online threshold decision-making after combining hysteresis and dead zone to suppress jitter.

[0081] The steps for calculating the threshold scaling factor and upper and lower limits are as follows:

[0082] Extract the dielectric parameters (relative permittivity and loss tangent), normal flow amplitude quantiles, and volatility indices of the current coal type and the reference coal type.

[0083] The amplitude quantile is taken from the robust quantile of online observation, and the volatility index is taken from the robust dispersion within the sliding window (such as the exponential weighted form of the median absolute deviation).

[0084] The dielectric difference, amplitude quantile, and fluctuation ratio are mapped to a single scaling factor to unify the stretching or compression threshold range:

[0085]

[0086] Where, η t Let be the threshold scaling factor at time t, and clip(·,a,b) denotes truncating the input to the interval [a,b]. For the mapping weights, d e R is a measure of dielectric difference. q R is the amplitude quantile ratio. v η is the volatility ratio. min ,η max The upper and lower bounds of the scaling factor are given, and t is the index of the discrete time step.

[0087] In the formula:

[0088]

[0089] Where, ε r tanδ represents the relative permittivity and loss tangent of the current coal type. (tanδ) ref For reference parameters corresponding to different coal types, w ε w tan ∈[0,1] difference weighted and w ε +w tan =1; and These are the p-quantiles of the current and reference normal flow amplitudes, respectively, and D. t D is the current volatility indicator. ref Used as a reference volatility scalar;

[0090] Based on the reference quantile and the current volatility, dynamic upper and lower limits are generated and hysteresis is added:

[0091]

[0092] Among them, L t U t Let be the lower and upper thresholds at time t. As the two quantiles of the reference amplitude, 0 <p1<p2<1,c> 0 represents the volatility amplification factor, η t The scaling factor is h ≥ 0, where h is the hysteresis half-width and L is the scaling factor. min L max U min U max For engineering safety boundaries; a dead zone δ0≥0 is added during online decision-making: state reversal is triggered only when the observation exceeds the boundary by more than δ0, in order to suppress jitter;

[0093] Weights α, β, γ, quantiles p, p1, p2, and coefficients c, h, δ0 can be configured hierarchically according to coal type and load segment, and fine-tuned online in small steps. When η t When the value tends to the boundary over a long period, it prompts for backcalibration or updating of the reference quantile and D. ref ;

[0094] Optionally, the recommended amplitude quantile is a combination of the median and high quantiles to balance position and tail sensitivity; the volatility index uses the same robust scaling measure as in the previous embodiment, used only for threshold width adaptation, without changing the definition of the decision statistics. Default parameter values ​​are: quantile pair (p1, p2) ≈ (0.5, 0.9), hysteresis half-width h is adjusted offline based on the target false trigger rate, with an initial value of 5-10% of the reference amplitude; dead zone δ0 is equal to 0.5-1.0 times the current volatility by default. Adjustable parameter range: p1∈[0.3, 0.6], p2∈[0.8, 0.95], h∈[0, 15% reference amplitude], δ0∈[0, 2× volatility]; the setting is based on false alarm control of historical fault-free sections and missed reporting constraints of manually labeled events. When the threshold scaling factor is close to the upper and lower clipping boundaries for a long period, the edge-approaching event is recorded and an offline backtracking calibration prompt is triggered. In abnormal boundary situations (missing or significantly distorted reference parameters), the online quantile and volatility of the most recent stable window are used as substitutes, and parameter fine-tuning is temporarily frozen for no more than 5 minutes.

[0095] Specifically, dielectric response differences, amplitude location, and scale changes are incorporated into the same scaling framework. Dielectric differences reflect statistical drift caused by coal type changes, amplitude quantiles capture the rise or fall of the overall level, and volatility ratio provides scale compensation for short-term disturbances and operating condition fluctuations. These three factors form a single scaling coefficient in the same mapping, facilitating the transfer of thresholds between different coal types and load ranges. The upper and lower limits are based on reference quantiles and linearly extended with the current volatility, then superimposed with hysteresis and dead zones to reduce the triggering frequency of round-trip jitter. This avoids rigid dependence on absolute amplitude and maintains a flexible response to changes in distribution location and scale in a statistical sense. At the same time, through pruning and safety boundary constraints, abnormal thresholds are kept within an acceptable range for engineering. Parameter stratification and online fine-tuning provide an evolutionary path during operation, facilitating the maintenance of discrimination stability during long-term drift or gradual changes in coal quality.

[0096] In one embodiment, the method further includes: training a generative adversarial network on normal flowing data, calculating the similarity score between the real-time signal and the generated signal online, and combining the score with a threshold set based on a fixed false alarm rate for anomaly detection.

[0097] In this embodiment, the similarity score is used to measure the degree to which the real-time window fits the normal distribution, and the threshold is obtained by calibrating the target false alarm rate on replay data without anomalies; the default target false alarm rate is 10. -5 Level, adjustable to 10 -4 -10 -6Level. Training data should cover at least three load segments and two typical coal content ranges, with a recommended effective sample duration of ≥30 minutes for a single coal type. Optionally, when training samples are insufficient, amplitude and scale perturbations can be simulated using data augmentation to supplement boundary conditions, without changing the judgment criteria and output meaning. If the GAN score is temporarily unavailable during the online phase, the system degenerates into independent judgment for the threshold channel.

[0098] In one embodiment, it further includes: aligning samples of the flow field in the bend region obtained from computational fluid dynamics simulation with the monitoring signals of the straight pipe section, training a convolutional long short-term memory network to establish a mapping relationship, and outputting an estimate of the accumulation state in the bend region during operation and fusing it with the anomaly detection results;

[0099] Specifically, the aligned samples achieve a one-to-one correspondence between simulation frames and measured windows by using a unified time base and conditional indexes based on operating condition vectors (air volume, dust content, load segment). The mapping model only outputs the probability or level of pipe bend accumulation, which is fused with the binary / multi-level results of anomaly detection on the same time grid, without generating independent control commands. The default fusion strategy is to take the higher priority; when the two conflict, the accumulation estimate is used to raise the warning level by one level. In boundary cases (CFD boundary condition deviation or out-of-domain operating conditions), the automatically weighted mapping output is reduced to a level no higher than a single threshold channel.

[0100] In one embodiment, anomaly detection includes transient mutation detection based on cumulative summation, which performs single / double-sided real-time accumulation and reset of statistics obtained through high-frequency sampling, wherein the high-frequency sampling frequency is not less than five times the system's preset conventional sampling frequency;

[0101] Optionally, the initial value of the ratio between the regular sampling frequency and the high-frequency sampling frequency is set to 8, and adjusted within the range of 5-16 based on the device bandwidth and CPU usage. The detection amplitude and decision threshold are designed by back-calculating the target average no-alarm sample length after setting historical anomaly-free segments to ensure that excessive alarms are not generated during representative stable periods. To suppress repeated triggering caused by clustered fluctuations, a sample freezing strategy is adopted after triggering, with the freezing duration covering the typical recovery time of a primary wind disturbance, defaulting to 50-200ms. When the autocorrelation of the high-frequency sequence increases significantly, the threshold is effectively relaxed to maintain the target average no-alarm interval. If the high-frequency sampling channel is abnormal or resources are insufficient, it degenerates into one-sided detection of regular sampling and records the performance degradation flag.

[0102] The steps for single / double-sided real-time accumulation and reset include:

[0103] The sampling frequency was increased to a higher frequency based on the conventional sampling frequency, and the observations were standardized.

[0104] f hf =m s f b m s ∈[5, 16],

[0105] Among them, f hf This is the high-frequency sampling frequency, measured in Hz, f. b This is the standard sampling frequency, measured in Hz (m). s The range is the multiplier. Recommended multiplier;

[0106] Zero drift compensation and scaling for high-frequency sequences:

[0107]

[0108] Where, x t μ is a statistical measure at high-frequency time t (which can be the amplitude of the charge after bandpass or its robust envelope). ref , σ ref For the reference position and scale under normal flow conditions (from the feature library of identified coal types or the recent steady-state window), σ ref >0;

[0109] The incremental detection range is constructed according to the design detection range and accumulated online. An alarm is triggered and the system is reset when the threshold is exceeded.

[0110]

[0111] Upper alarm: Lower side alarm:

[0112] Reset after triggering: or And freeze n lock One high-frequency sample;

[0113] Among them, s t This represents the log-likelihood increment of shifting the mean upward by d0, where d0 > 0 represents the designed detection amplitude. These are the upper and lower CUSUM statistics, A. + A - >0 represents the corresponding decision threshold. To freeze the sample size and suppress duplicate reporting, samples covering 50–200 milliseconds are typically selected; when only sudden increases in amplitude are of concern, a one-sided approach is used. Use both sides when focusing on sudden increases and decreases, and maintain simultaneously.

[0114] Under the condition of no change in standard normality, the relationship between the threshold and the false alarm probability / ARL can be directly set based on the ordinal ratio test approximation:

[0115]

[0116] Where, α ±The one-sided false alarm probability for each high-frequency sample. The average alarm interval on one side under unchanged conditions, expressed in terms of sample size; for example, the target (approximately equivalent to α) + ≈10 -5 When ), take A. + ≈11.51; When both sides are in effect simultaneously, α will be equally distributed between the two sides or allocated according to risk preference;

[0117] When m s When increasing the threshold leads to an increase in the correlation between samples, it is equivalent to relaxing the threshold:

[0118] Target ARL0 is determined by the effective independent sample ratio ρ eff ∈(0,1] is corrected to ARL0 / ρ eff ;ρ eff Obtained from the first-order autocorrelation estimation of the high-frequency sequence; the freeze time after the alarm is taken as T. lock =n lock / f hf The typical recovery time for a single wind disturbance is covered;

[0119] After the alarm is triggered, a freeze period begins, and the system stops. Accumulate and update only μ ref , σ ref The slow channel resumes accumulation after the freeze period ends; if a coal type switching event occurs within the window, then d0 and A are refreshed. ± Configured for this type of coal;

[0120] Specifically, this method combines high-frequency sampling with sequential detection, constructs a log-likelihood increment using standardized statistics, performs online one-sided or two-sided accumulation, and reports immediately when the threshold is reached; the increment construction with the designed detection amplitude as the core makes the sensitivity of the statistics to the target amplitude adjustable, while avoiding dependence on unknown scales; the reset adopts zeroing and freezing methods, and the freezing period blocks short-term round-trip triggering, which is suitable for scenarios where the charge signal exhibits clustered fluctuations in a short period of time.

[0121] The correspondence between threshold and false alarm is approximated by the sequential ratio test to give a closed expression. In engineering, the threshold can be directly set by the target average fault-free sample length, which makes it easy to control the alarm frequency within the acceptable range of operation and maintenance. When coal type switching and load fluctuation occur, the design range and threshold of the reference parameters and hierarchical configuration are refreshed to maintain stable triggering characteristics.

[0122] In one embodiment, the graded early warning is classified according to the duration of the abnormality, the degree of deviation, and the spatial consistency, and outputs recommended handling measures including adjusting the primary air volume, changing the coal feeder speed, separator slag discharge, and load reduction operation. The early warning information can be connected to the distributed control system through the communication interface.

[0123] Similarly, the classification criteria are jointly determined by three dimensions: duration threshold, deviation amplitude threshold, and multi-channel consistency threshold. The default duration threshold is set to two levels: 0.5s and 2s. The deviation amplitude is divided into medium / high based on the proportion of deviations exceeding the threshold. Spatial consistency is determined by more than half of the channels in the ring. Specific thresholds can be fine-tuned according to unit characteristics and operation and maintenance experience. The warning information includes a timestamp, level, recommended measures, and a summary of the source channel. The communication interface follows the existing field specifications. When communication is unavailable, this embodiment records the information locally and provides an audio-visual notification, which is then resent after the link is restored.

[0124] This embodiment also provides a computer device, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to realize the early warning method for pulverized coal pipeline blockage in thermal power plants as proposed in the above embodiment.

[0125] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0126] This embodiment also provides a storage medium storing a computer program, which, when executed by a processor, implements a method for early warning of blockage in pulverized coal pipelines of thermal power plants as proposed in the above embodiment. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0127] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application.

[0128] Furthermore, those skilled in the art will understand that although some embodiments herein include certain features included in other embodiments but not others, combinations of features from different embodiments are meant to be within the scope of this application and form different embodiments. For example, all the embodiments above can be used in any combination. The information disclosed in this background section is intended only to enhance the understanding of the general background of this application and should not be construed as an admission or in any way implying that such information constitutes prior art known to those skilled in the art.

Claims

1. A method for early warning of coal powder pipeline blockage in a thermal power plant, characterized in that, The method comprises the following steps: Step S1, collecting charge signals by an electrostatic sensor array arranged on a coal powder conveying pipeline; Step S2, collecting normal flow signals of different coal types under a calibration condition and extracting features to establish a coal type feature library; Step S3, inputting features of real-time charge signals into a trained pattern recognition model to identify a dominant coal type being conveyed; Step S4, calling corresponding reference feature parameters from the coal type feature library according to the identification result; Step S5, dynamically adjusting a warning threshold according to the reference feature parameters; The dynamic adjustment of the warning threshold comprises: Based on the dielectric characteristic difference between the current coal type and the reference coal type and the amplitude and fluctuation statistics in the feature library, a threshold scaling coefficient and upper and lower limits are calculated, and after the hysteresis and dead zone are suppressed, the scaling coefficient and the upper and lower limits are applied to online threshold decision-making; The step of calculating the threshold scaling coefficient and the upper and lower limits comprises: Extracting dielectric parameters, normal flow amplitude quantile, and fluctuation index of the current coal type and the reference coal type; The amplitude quantile is a robust quantile observed online, and the fluctuation index is a robust dispersion within a sliding window; The dielectric difference, amplitude quantile ratio, and fluctuation ratio are mapped into a single scaling coefficient: ， wherein is a threshold scaling coefficient, is a threshold scaling coefficient, denotes truncating the input to the interval , is a mapping weight, is a dielectric difference measure, is an amplitude quantile ratio, is a fluctuation ratio, are lower and upper bounds for the scaling coefficient, is a discrete time index; In the formula: ， ， ， wherein, is the relative permittivity and loss tangent of the current coal type, is the corresponding parameter of the reference coal type, is weighted by the difference and ; is the current and reference normal flow amplitude, respectively, is the current and reference normal flow amplitude, is the current volatility index, is the reference volatility index; On the basis of the reference quantile and the current fluctuation degree, a dynamic upper and lower limit is generated and hysteresis is added: ， ， wherein, is the lower, upper threshold at time is the two end quantile of the reference amplitude, , is the volatility amplification coefficient, is the scaling coefficient, is the hysteresis half-width, is the engineering safety margin; online decision-making with added dead zone : only when the observed out-of-bound amplitude exceeds state flip is triggered;​ Step S6, performing abnormality detection on real-time charge signals based on the adjusted threshold; Step S7, when an abnormality is determined, outputting graded warning information including a warning level and a disposal suggestion.

2. The method for early warning of coal pipe blockage in a thermal power plant according to claim 1, characterized in that, The electrostatic sensor array is at least three groups of electrode units arranged in a circular direction, and after the sampling data are synchronously collected, band-pass filtered, electrostatic drift compensated, and time-aligned, time-domain waveforms and frequency-domain features obtained by fast Fourier transform are obtained.

3. The method of claim 1, wherein the method comprises: The pattern recognition model uses wavelet packet decomposition when extracting features, selects energy distribution, spectral entropy, and energy proportion of low-frequency and high-frequency subbands as the basis for coal type identification, and trains an offline classifier through labeled data to output the dominant coal type.

4. The method for coal pipeline blockage early warning of a thermal power plant according to claim 1, characterized in that, Further comprising: Using a generative adversarial network to train on normal flow data, calculating the similarity score of real-time signals and generated signals online, and combining the score with a threshold set based on a fixed false alarm rate for abnormality determination.

5. The method for coal pipeline blockage early warning of a thermal power plant according to claim 1, characterized in that, Further comprising: Based on the flow field in the elbow region obtained by computational fluid dynamics simulation and the alignment sample of the straight pipe section monitoring signal, a convolutional long short-term memory network is trained to establish a mapping relationship, and the accumulation state estimation of the elbow region is output in operation and fused with the abnormality detection result.

6. The method of claim 1, wherein the method further comprises: determining a coal flow rate of the coal powder in the coal powder pipeline; and determining a coal powder pipeline pressure of the coal powder in the coal powder pipeline. The abnormality detection includes transient mutation detection based on cumulative sum, and the statistics obtained by high-frequency sampling are accumulated and reset in real time on one side or both sides, and the frequency of high-frequency sampling is not less than five times the system preset regular sampling frequency; The step of performing real-time accumulation and reset on one side or both sides comprises: Increasing the regular sampling frequency to a high-frequency sampling frequency and normalizing the observation; ， ， ， wherein, is a high frequency sampling frequency, in Hz, is a regular frequency sampling frequency, in Hz, is a rate interval, is a recommended rate; Zero drift compensation and scale unification are performed on the high-frequency sequence: ， wherein is the high frequency instant is the statistical quantity, is the reference position and scale under normal flow, ; According to the designed detection amplitude, an increment is constructed and accumulated online, and if the threshold is exceeded, an alarm is given and reset: ， ， top side alert: , bottom side alert: , triggered reset: and freeze high frequency samples; where, is the log-likelihood increment of the mean shift, is the design detection amplitude, are the upper and lower CUSUM statistics, respectively, is the corresponding decision threshold, is the number of frozen samples; one-sided is used when only amplitude spikes are of interest, and two-sided is used when both spikes and dips are of interest.​ Under the standard normal distribution, based on the sequential ratio test approximation, the threshold and false alarm probability / ARL corresponding relationship can be directly set: ， ， wherein, is the one-sided false alarm probability for each high frequency sample, is the one-sided average time between alarms in the unaltered situation, in number of samples; when both sides are active at the same time, this is split between the two sides or allocated according to risk appetite; When Increasing the threshold when the correlation between samples rises. Target The effective independent sample ratio Corrected to Obtained from the first-order autocorrelation estimation of the high-frequency sequence; the freeze time after the alarm covers the typical recovery time of the wind disturbance , After alarm, enter freeze period, stop Cumulative and update only Slow channel, freeze period ends and resumes accumulation; if a coal switching event occurs within the window, refresh And Configured for the coal.

7. The method for coal powder pipeline blockage early warning of a thermal power plant according to claim 1, characterized in that, The hierarchical early warning is classified according to abnormal duration, deviation degree and spatial consistency, and outputs recommended treatment measures including adjusting primary air volume, changing coal feeder speed, separator slag discharge and reducing load operation, and early warning information is connected to a distributed control system through a communication interface.

8. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that: The processor executes the computer program to realize the steps of the coal pipe blockage early warning method of the power plant according to any one of claims 1-7.

9. A computer readable storage medium having stored thereon a computer program, characterized in that: The computer program is executed by the processor to realize the steps of the coal pipe blockage early warning method of the power plant according to any one of claims 1-7.

Images