An online monitoring and alarm system and method for chemical plant exhaust gas

Abstract

This invention provides an online monitoring and alarm system and method for chemical plant exhaust gas, relating to the field of monitoring and alarm technology. This invention constructs a multi-dimensional perception and decision-making system that deeply integrates real-time data dynamic characteristics and historical statistical patterns to achieve early and accurate warnings and source differentiation of abnormal exhaust gas emissions. First, by extracting statistical features from the concentration data of each target exhaust gas at multiple time scales, quantitative indicators characterizing abnormal emission distribution patterns and real-time fluctuation deviations are generated and then integrated into a collaborative deviation index, comprehensively reflecting the overall abnormal trend of pollutant emissions. Based on this index, the alarm threshold is dynamically modulated, enabling the alarm sensitivity to adapt to changes in actual operating conditions. Furthermore, by calculating the instantaneous instability energy and comparing it with its health baseline, independent diagnosis of the stability of the production process is achieved. Finally, by collaboratively judging the combination of process stability and concentration exceeding the standard, a graded alarm is triggered.

Description

Technical Field

[0001] This invention relates to the field of monitoring and alarm technology, and in particular to an online monitoring and alarm system and method for chemical plant exhaust gas. Background Technology

[0002] Traditional online monitoring and alarm methods for chemical plant exhaust gases mainly rely on setting fixed concentration alarm thresholds to independently monitor single pollutants. This is a static monitoring method with certain drawbacks. Chemical production processes have complex dynamic characteristics. Exhaust gas emission concentrations are affected by various factors such as raw material fluctuations, catalyst activity changes, and equipment operating condition switching. The data distribution exhibits non-steady-state characteristics, and fixed thresholds cannot adapt to these dynamic changes, leading to frequent false alarms or missed alarms. In addition, traditional methods only focus on whether the instantaneous concentration exceeds the standard, ignoring the inherent statistical regularities and evolutionary trends of the emission data sequence. They cannot identify early abnormal signs in the process. For example, when the concentration data distribution shows changes in statistical characteristics such as sharp peaks and thick tails, or when the volatility continues to increase but has not yet reached the fixed threshold, the system cannot issue an early warning, causing managers to miss the opportunity to intervene. Existing technologies also lack the ability to comprehensively analyze the synergistic emission patterns of multiple pollutants. Each monitoring indicator is judged in isolation, making it difficult to assess the overall operating status of the production system.

[0003] Therefore, it is necessary to provide an online monitoring and alarm system and method for chemical plant exhaust gas to solve the above-mentioned technical problems. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention provides an online monitoring and alarm system and method for chemical plant exhaust gas, achieving the beneficial effects of multi-dimensional dynamic adaptive sensing and accurate alarm decision-making.

[0005] This invention provides a method for online monitoring and alarm of exhaust gas in chemical plants, comprising:

[0006] S1: Simultaneously and continuously collect the concentration data of each target exhaust gas, and calculate the baseline standard deviation and statistical distribution kurtosis of each target exhaust gas within the preset sliding baseline time window, and calculate the short-term standard deviation of each target exhaust gas within the preset sliding short-term time window.

[0007] S2: Based on the statistical distribution kurtosis and standard normal distribution kurtosis of each target exhaust gas, calculate the distribution morphology anomaly factor of each target exhaust gas; based on the short-term standard deviation and baseline standard deviation of each target exhaust gas, calculate the real-time fluctuation deviation factor of each target exhaust gas.

[0008] S3: Based on the preset weighted weights, calculate the weighted Euclidean distance between the distribution anomaly factor and the real-time fluctuation deviation factor of each target exhaust gas to obtain the coordinated deviation index of each target exhaust gas.

[0009] S4: Calculate the dynamic alarm threshold of each target exhaust gas based on the cooperative deviation index of each target exhaust gas and the preset benchmark concentration threshold of each target exhaust gas.

[0010] S5: Obtain a continuous sequence of multiple coordinated deviation indices of each target exhaust gas, and calculate the root mean square of the continuous sequence as the instantaneous instability energy of each target exhaust gas.

[0011] S6: Based on the comparison results of the instantaneous instability energy of each target exhaust gas with the pre-constructed health baseline of each target exhaust gas, and the comparison results of the concentration data of each target exhaust gas with the dynamic alarm threshold of each target exhaust gas, graded alarms and decisions are made.

[0012] Preferably, in step S1, the length of the preset sliding baseline time window is at least 30 times the length of the preset sliding short-term time window.

[0013] Preferably, in step S2, the distribution anomaly factor of each target exhaust gas is obtained by calculating the absolute value of the difference between the statistical distribution kurtosis and the standard normal distribution kurtosis of each target exhaust gas.

[0014] Preferably, in step S2, the real-time fluctuation deviation factor of each target exhaust gas is obtained by calculating the ratio of the short-term standard deviation of each target exhaust gas to the baseline standard deviation.

[0015] Preferably, in step S3, the formula for calculating the cooperative deviation index is:

[0016]

[0017] in, The co-deviation index, This is the distribution morphology anomaly factor. This is the real-time fluctuation deviation factor. and These are the preset weighting weights, and and The sum of is 1.

[0018] Preferably, in step S4, the formula for calculating the dynamic alarm threshold is:

[0019]

[0020] in, For dynamic alarm thresholds, The preset basic alarm threshold, For adjustment coefficients, This is the co-deviation index.

[0021] Preferably, in step S5, the continuous sequence of multiple consecutive co-deviation indices of each target exhaust gas is updated through a sliding window mechanism. The continuous sequence is arranged in the order in which it is calculated. When a new co-deviation index is calculated, it is inserted at the end of the continuous sequence, and the co-deviation index at the beginning of the continuous sequence is removed.

[0022] Preferably, in step S6, the steps for constructing the health baseline of each target exhaust gas include:

[0023] Based on the historical coordinated deviation index sequence of each target exhaust gas under historical stable operating conditions, the first difference of the historical coordinated deviation index sequence of each target exhaust gas is calculated to obtain the historical coordinated deviation index change rate sequence of each target exhaust gas.

[0024] Calculate the mean and standard deviation of the historical co-deviation index change rate sequence for each target exhaust gas, and construct a health baseline for each target exhaust gas based on the mean and standard deviation.

[0025] Preferably, in step S6, the comparison results between the instantaneous instability energy of each target exhaust gas and the pre-constructed health baseline of each target exhaust gas include:

[0026] The control upper limit of each target exhaust gas is calculated based on the health baseline of each target exhaust gas. The process stability state of each target exhaust gas is determined by comparing the instantaneous instability energy of each target exhaust gas with the corresponding control upper limit of each target exhaust gas.

[0027] This invention provides an online monitoring and alarm system for chemical plant exhaust gas, comprising:

[0028] The data acquisition and statistics module is used to synchronously and continuously acquire the concentration data of each target exhaust gas, and calculate the baseline standard deviation and statistical distribution kurtosis of each target exhaust gas within a preset sliding baseline time window, and at the same time calculate the short-term standard deviation of each target exhaust gas within a preset sliding short-term time window.

[0029] The anomaly factor generation module is used to calculate the distribution anomaly factor of each target exhaust gas based on the statistical distribution kurtosis and standard normal distribution kurtosis of each target exhaust gas, and to calculate the real-time fluctuation deviation factor of each target exhaust gas based on the short-term standard deviation and baseline standard deviation of each target exhaust gas.

[0030] The Cooperative Deviation Index Generation Module is used to calculate the weighted Euclidean distance between the distribution anomaly factor and the real-time fluctuation deviation factor of each target exhaust gas based on a preset weighted weight to obtain the cooperative deviation index of each target exhaust gas.

[0031] The dynamic threshold generation module is used to calculate the dynamic alarm threshold of each target exhaust gas based on the cooperative deviation index of each target exhaust gas and the preset benchmark concentration threshold of each target exhaust gas.

[0032] The instability energy calculation module is used to obtain a continuous sequence of multiple coordinated deviation indices of each target exhaust gas, and calculate the root mean square of the continuous sequence as the instantaneous instability energy of each target exhaust gas.

[0033] The graded alarm decision module is used to perform graded alarms and decisions based on the comparison results of the instantaneous instability energy of each target exhaust gas with the pre-constructed health baseline of each target exhaust gas, as well as the comparison results of the concentration data of each target exhaust gas with the dynamic alarm threshold of each target exhaust gas.

[0034] Compared with related technologies, the online monitoring and alarm system and method for chemical plant exhaust gas provided by the present invention have the following beneficial effects:

[0035] This invention constructs a multi-dimensional perception and decision-making system that deeply integrates real-time data dynamic features and historical statistical patterns, achieving early and accurate warning and source differentiation of abnormal exhaust gas emissions. The method first extracts statistical features from the concentration data of each target exhaust gas at multiple time scales, generating quantitative indicators characterizing abnormal emission distribution patterns and real-time fluctuation deviations. These indicators are then integrated into a coordinated deviation index, which comprehensively reflects the overall abnormal trend of pollutant emissions. Based on this index, the alarm threshold is dynamically modulated, enabling the alarm sensitivity to adapt to changes in actual operating conditions and effectively suppressing false alarms caused by normal production fluctuations. Furthermore, the instantaneous instability energy is calculated and correlated with... The baseline comparison of health status enables independent diagnosis of the stability of the production process itself. It can keenly identify early abnormal states caused by abnormal factors such as equipment failure or process imbalance, which have not yet led to concentration exceeding the standard but have deviated from the normal fluctuation pattern. Finally, the system triggers differentiated alarm levels by collaboratively judging the combination of process stability and concentration exceeding the standard. This not only enables the detection of emission exceeding the standard events that have already occurred, but also provides early warning of potential risks. It also provides key decision-making basis for operators to distinguish the nature of events. Overall, it improves the perception capability, early warning accuracy and decision-making intelligence of the monitoring system, and provides reliable technical support for the safe and environmentally friendly operation of chemical plants. Attached Figure Description

[0036] Figure 1 This is a flowchart of an online monitoring and alarm method for chemical plant exhaust gas according to the present invention;

[0037] Figure 2 This is a modular structure diagram of an online monitoring and alarm system for chemical plant exhaust gas according to the present invention. Detailed Implementation

[0038] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and not intended to limit it. Furthermore, it should be noted that, for ease of description, only the parts relevant to the invention are shown in the drawings, not all structures. Moreover, unless otherwise specified, the embodiments and features described herein can be combined with each other.

[0039] It should also be noted that, for ease of description, the accompanying drawings show only the parts relevant to the invention and not all of them. Before discussing exemplary embodiments in more detail, it should be mentioned that some exemplary embodiments are described as processes or methods depicted as flowcharts. Although the flowcharts describe operations (or steps) as being processed sequentially, many of these operations can be performed in parallel, concurrently, or simultaneously. Furthermore, the order of the operations can be rearranged. The process can be terminated when its operation is completed, but it may also have additional steps not included in the drawings. The process may correspond to a method, function, procedure, subroutine, subprogram, etc.

[0040] Example 1

[0041] A method for online monitoring and alarming of exhaust gas in chemical plants, in the specific implementation process, such as... Figure 1 As shown, a flowchart of an online monitoring and alarm method for chemical plant exhaust gas is illustrated, including:

[0042] Step S1: Simultaneously and continuously collect the concentration data of each target exhaust gas, and calculate the baseline standard deviation and statistical distribution kurtosis of each target exhaust gas within the preset sliding baseline time window. At the same time, calculate the short-term standard deviation of each target exhaust gas within the preset sliding short-term time window.

[0043] Specifically, in step S1, the length of the preset sliding baseline time window is at least 30 times the length of the preset sliding short-term time window.

[0044] In specific implementation, for example, an integrated multi-channel gas analyzer group synchronously collects the concentration values ​​of each target exhaust gas, including but not limited to hydrogen sulfide and benzene series compounds, at a frequency of once per second. This forms a concentration data stream corresponding to each target exhaust gas and strictly ordered by timestamp. Two sliding time windows of different scales are independently configured for each target exhaust gas for real-time calculation. The sliding baseline time window is set to a 30-minute time span, containing 1800 continuous sampling points. This window is used to calculate the baseline standard deviation and statistical distribution kurtosis, characterizing the mid-term emission stability. The baseline standard deviation is calculated by measuring the standard deviation of the concentration data from the 1800 continuous sampling points within the sliding baseline time window. The statistical distribution kurtosis is calculated by measuring the standard deviation of the concentration data from the 1800 continuous sampling points within the sliding baseline time window. The fourth central moments of concentration data from 800 consecutive sampling points are used. The kurtosis of the statistical distribution is obtained by dividing the fourth central moments by the square of the variance and then subtracting 3. A sliding short-term time window with a time span of 1 minute is set, which contains 60 consecutive sampling points and is used to calculate the short-term standard deviation characterizing the intensity of instantaneous fluctuations. The sliding baseline time window and the sliding short-term time window slide synchronously, with a sliding step size equal to the length of one sliding short-term time window. In addition, the length of the sliding baseline time window is set to 30 times the length of the sliding short-term time window. This design ensures that the baseline statistics have a sufficient sample size to reflect stable characteristics, while the short-term window can sensitively capture minute-level abnormal fluctuations, realizing minute-level real-time feature extraction and providing timely and multi-dimensional feature input for subsequent anomaly diagnosis.

[0045] Step S2: Based on the statistical distribution kurtosis and standard normal distribution kurtosis of each target exhaust gas, calculate the distribution anomaly factor of each target exhaust gas. Based on the short-term standard deviation and baseline standard deviation of each target exhaust gas, calculate the real-time fluctuation deviation factor of each target exhaust gas.

[0046] Specifically, in step S2, the distribution anomaly factor of each target exhaust gas is obtained by calculating the absolute value of the difference between the statistical distribution kurtosis and the standard normal distribution kurtosis of each target exhaust gas.

[0047] Specifically, in step S2, the real-time fluctuation deviation factor of each target exhaust gas is obtained by calculating the ratio of the short-term standard deviation of each target exhaust gas to the baseline standard deviation.

[0048] In the specific implementation process, firstly, the distribution anomaly factor is calculated based on the kurtosis of the statistical distribution of each target exhaust gas. This is achieved by calculating the absolute value of the difference between the statistical distribution kurtosis of each target exhaust gas and the baseline value of the standard normal distribution kurtosis (3). This transforms the signed original kurtosis value, which characterizes the distribution pattern, into a dimensionless anomaly index that is always non-negative, and the larger the value, the more severe the deviation of the distribution pattern from the normal distribution. This index can effectively identify abnormal patterns in the data distribution, such as sharp peaks, thick tails, or flat peaks. Simultaneously, the real-time fluctuation deviation factor is calculated based on the short-term standard deviation and baseline standard deviation of each target exhaust gas. This is achieved by calculating the ratio of the short-term standard deviation of each target exhaust gas to the baseline standard deviation. This ratio quantifies the degree of deviation of the current instantaneous fluctuation amplitude from the historical normal fluctuation level. When the ratio is significantly greater than 1, it indicates an increase in abnormal fluctuation, while when the ratio is significantly less than 1, it indicates a decrease in abnormal fluctuation. Through the above parallel calculations, two independent factors are finally generated for each target exhaust gas, respectively characterizing the abnormal distribution pattern and the abnormal fluctuation intensity, providing a basis for subsequent collaborative analysis and comprehensive diagnosis.

[0049] Step S3: Based on the preset weighted weights, calculate the weighted Euclidean distance between the distribution anomaly factor and the real-time fluctuation deviation factor of each target exhaust gas to obtain the coordinated deviation index of each target exhaust gas.

[0050] Specifically, in step S3, the formula for calculating the cooperative deviation index is:

[0051]

[0052] in, The co-deviation index, This is the distribution morphology anomaly factor. This is the real-time fluctuation deviation factor. and These are the preset weighting weights, and and The sum of is 1.

[0053] In the specific implementation process, based on the calculated distribution morphology anomaly factor and real-time fluctuation deviation factor of each target exhaust gas, two sets of standardized anomaly indicators are used. A weighted Euclidean distance formula is employed to fuse multi-dimensional anomaly features. First, the two anomaly factors are scaled using preset weighting coefficients, where the weighting weight of the distribution morphology anomaly factor is... The weighted average of the real-time fluctuation deviation factor is: And the weighting coefficients satisfy and The constraint that the sum of the two factors is always 1 is used to ensure the consistency of the feature dimensions after weighted scaling. Then, the sum of squares of the differences between the two factors after weighted scaling is calculated. Finally, the square root of the sum of squares is taken to obtain the final cooperative deviation index of each target exhaust gas. As a comprehensive scalar, the larger the value of the cooperative deviation index, the more significant the overall abnormality of the target exhaust gas in terms of both distribution pattern and fluctuation intensity. Through this fusion calculation based on weighted Euclidean distance, the one-sidedness of single-dimensional evaluation can be overcome, and the cooperative change characteristics under different abnormal modes can be effectively captured, providing a quantitative comprehensive abnormality input for the dynamic threshold modulation in subsequent steps.

[0054] Step S4: Calculate the dynamic alarm threshold of each target exhaust gas based on the cooperative deviation index of each target exhaust gas and the preset benchmark concentration threshold of each target exhaust gas.

[0055] Specifically, in step S4, the formula for calculating the dynamic alarm threshold is:

[0056]

[0057] in, For dynamic alarm thresholds, The preset basic alarm threshold, For adjustment coefficients, This is the co-deviation index.

[0058] In the specific implementation process, based on the calculated coordinated deviation index of each target exhaust gas and the preset benchmark concentration threshold of each target exhaust gas, a linear modulation formula is adopted. That is, the dynamic alarm threshold is equal to the basic alarm threshold multiplied by 1 plus the sum of the product of the adjustment coefficient and the coordinated deviation index. Here, the basic alarm threshold is a fixed concentration limit preset for each type of exhaust gas, and the adjustment coefficient is a positive real number used to control the sensitivity of the threshold to the coordinated deviation index. The coordinated deviation index characterizes the comprehensive abnormality of the exhaust gas emission process. The calculation formula of the dynamic alarm threshold ensures that when the coordinated deviation index increases, the dynamic alarm threshold increases accordingly, thereby automatically relaxing the alarm conditions to reduce false alarms when the process is unstable. When the coordinated deviation index decreases, the dynamic alarm threshold decreases, thereby tightening the alarm conditions to improve detection sensitivity when the process is stable. Through this dynamic modulation mechanism, the system can adapt to changes in process state and achieve more intelligent alarm threshold management.

[0059] Step S5: Obtain a continuous sequence of multiple coordinated deviation indices of each target exhaust gas, and calculate the root mean square of the continuous sequence as the instantaneous instability energy of each target exhaust gas.

[0060] Specifically, in step S5, the continuous sequence of multiple consecutive co-deviation indices of each target exhaust gas is updated through a sliding window mechanism. The continuous sequence is arranged in the order in which it is calculated. When a new co-deviation index is calculated, it is inserted at the end of the continuous sequence, and the co-deviation index at the beginning of the continuous sequence is removed.

[0061] In the specific implementation process, the coordinated deviation index data stream generated by each target exhaust gas in chronological order is continuously acquired. A continuous sequence of coordinated deviation indices of fixed length is maintained for each target exhaust gas. This continuous sequence is updated using a first-in-first-out sliding window mechanism. Whenever a new coordinated deviation index data point is calculated, it is immediately inserted into the end of the continuous sequence in chronological order. At the same time, the earliest data point at the beginning of the continuous sequence is automatically removed, thus keeping the length of the continuous sequence constant and always containing the latest consecutive coordinated deviation indices. Subsequently, the root mean square (RMS) calculation is performed on all coordinated deviation index values ​​in the continuous sequence. First, each coordinated deviation index value in the continuous sequence is squared, then the arithmetic mean of these squared values ​​is calculated, and finally, the square root of the arithmetic mean is taken. The resulting RMS result is the instantaneous instability energy of each target exhaust gas. The instantaneous instability energy comprehensively reflects the average amplitude of recent coordinated deviation index fluctuations and can effectively characterize the instantaneous instability intensity of the exhaust gas emission process. Through this method of combining dynamic sliding window and RMS calculation, the abnormal fluctuation characteristics of process energy can be captured and quantified in real time, providing key energy indicator inputs for subsequent stability assessment.

[0062] Step S6: Based on the comparison results of the instantaneous instability energy of each target exhaust gas with the pre-constructed health baseline of each target exhaust gas, and the comparison results of the concentration data of each target exhaust gas with the dynamic alarm threshold of each target exhaust gas, perform graded alarm and decision-making.

[0063] Specifically, in step S6, the steps for constructing the health baseline of each target exhaust gas include:

[0064] Based on the historical coordinated deviation index sequence of each target exhaust gas under historical stable operating conditions, the first difference of the historical coordinated deviation index sequence of each target exhaust gas is calculated to obtain the historical coordinated deviation index change rate sequence of each target exhaust gas.

[0065] Calculate the mean and standard deviation of the historical co-deviation index change rate sequence for each target exhaust gas, and construct a health baseline for each target exhaust gas based on the mean and standard deviation.

[0066] Specifically, in step S6, the comparison results between the instantaneous instability energy of each target exhaust gas and the pre-constructed health baseline of each target exhaust gas include:

[0067] The control upper limit of each target exhaust gas is calculated based on the health baseline of each target exhaust gas. The process stability state of each target exhaust gas is determined by comparing the instantaneous instability energy of each target exhaust gas with the corresponding control upper limit of each target exhaust gas.

[0068] In the specific implementation process, the health baseline of each target exhaust gas is first constructed. This is achieved by collecting historical coordinated deviation index sequences of each target exhaust gas under historical stable operating conditions, calculating the first difference of these sequences to obtain a historical coordinated deviation index change rate sequence, and then calculating the mean and standard deviation of this sequence. These mean and standard deviation are used as the basic statistics for the health baseline. Next, process stability is determined by calculating the upper control limit based on the health baseline of each target exhaust gas. For example, the mean of the health baseline plus three times the standard deviation is used as the statistical control upper limit. This is achieved by comparing the instantaneous instability energy of each target exhaust gas calculated in real time. The system determines the process stability by comparing the instantaneous instability energy with the corresponding control upper limit. When the instantaneous instability energy continuously exceeds the control upper limit, it is considered a process instability state. For example, if three consecutive instantaneous instability energy values ​​exceed the control upper limit, an instability alarm is triggered. Simultaneously, the system performs concentration exceedance determination in parallel. It compares the real-time concentration data of each target exhaust gas with its corresponding dynamic alarm threshold to determine if the concentration exceeds the limit. In the tiered alarm decision-making stage, for example, a two-level architecture is used for intelligent analysis. The first level involves independent analysis of single pollutants, assigning a preliminary alarm level to each target exhaust gas. If a target exhaust gas is simultaneously in a process instability and concentration exceedance state, it is marked as a level three alarm, indicating that the pollutant... If a serious anomaly exists and the process is unstable, a Level 2 alarm is triggered if the process is only unstable and the concentration does not exceed the standard, indicating an abnormal trend in the production process. If the concentration exceeds the standard but the process is stable, a Level 1 alarm is triggered, indicating a transient disturbance. If there are no anomalies, the system is marked as normal. The second level, system-level collaborative comprehensive decision-making, receives the preliminary alarm levels of all target exhaust gases and performs an overall assessment. When any target exhaust gas is marked as a Level 3 alarm, the system immediately triggers the highest-level global alarm, indicating a definite local serious fault in the production system requiring immediate attention. When there is no Level 3 alarm but more than two target exhaust gases are simultaneously marked as Level 2 alarms, the system triggers... A global systemic alarm indicates that multiple monitoring points are showing abnormal signs simultaneously, suggesting a global risk such as decreased reactor efficiency or utility failure in the production system. When none of the above conditions are met, the system outputs the highest level single-point alarm. If the highest level is a Level 2 alarm, it is considered a local process problem; if the highest level is a Level 1 alarm, it is considered an isolated event or measurement noise. Through this progressive analysis chain from microscopic single-point diagnosis to macroscopic system judgment, the system successfully transforms isolated monitoring indicators into a basis for overall operational status assessment, realizing a comprehensive analysis of multi-pollutant co-emission patterns, and ultimately forming an online graded alarm decision-making system for chemical plant exhaust gas that combines high real-time performance and high reliability.

[0069] The working principle of the online monitoring and alarm method for chemical plant exhaust gas provided by this invention is as follows:

[0070] First, by synchronously and continuously collecting concentration data streams of each target exhaust gas, and employing a dual sliding window mechanism for real-time statistical calculations, baseline standard deviation and statistical distribution kurtosis are obtained to capture the data distribution morphology characteristics. Simultaneously, short-term standard deviation is calculated within a short-term window to quantify instantaneous volatility. Then, an anomaly factor is generated based on the absolute value of the difference between the statistical distribution kurtosis and the standard normal distribution kurtosis, measuring the degree to which the exhaust gas concentration distribution deviates from the normal distribution. A real-time fluctuation deviation factor is generated based on the ratio of the short-term standard deviation to the baseline standard deviation, reflecting the deviation of the current fluctuation from historical normal levels. Next, the two factors are fused into a co-deviation index using a weighted Euclidean distance, characterizing the overall anomaly degree of exhaust gas emissions. Based on the co-deviation index and a preset baseline... The quasi-concentration threshold is used to calculate the dynamic alarm threshold through a linear modulation formula, enabling the threshold to change dynamically with the process state. Furthermore, a time series consisting of multiple consecutive cooperative deviation indices is obtained through a sliding window mechanism, and the root mean square of this series is calculated as the instantaneous instability energy, quantifying the average intensity of recent process fluctuations. Finally, the instantaneous instability energy is compared in parallel with the control upper limit derived from the health baseline to determine the process stability state, while the real-time concentration is compared with the dynamic threshold to determine the concentration exceeding the standard state. Based on dual criteria, a two-level decision logic is adopted, thereby transforming isolated indicators into an overall operational status assessment, achieving intelligent diagnosis from micro to macro levels. The entire method improves alarm accuracy and early warning capability by cascading multi-scale feature extraction, dynamic threshold modulation, and cooperative decision-making mechanisms.

[0071] Example 2

[0072] A chemical plant exhaust gas online monitoring and alarm system, in its specific implementation, such as... Figure 2 As shown, it illustrates a modular structure diagram of an online monitoring and alarm system for chemical plant exhaust gas, including:

[0073] The data acquisition and statistics module 100 is used to synchronously and continuously acquire the concentration data of each target exhaust gas, calculate the baseline standard deviation and statistical distribution kurtosis of each target exhaust gas within a preset sliding baseline time window, and calculate the short-term standard deviation of each target exhaust gas within a preset sliding short-term time window.

[0074] The anomaly factor generation module 200 is used to calculate the distribution anomaly factor of each target exhaust gas based on the statistical distribution kurtosis and standard normal distribution kurtosis of each target exhaust gas, and to calculate the real-time fluctuation deviation factor of each target exhaust gas based on the short-term standard deviation and baseline standard deviation of each target exhaust gas.

[0075] The Cooperative Deviation Index Generation Module 300 is used to calculate the weighted Euclidean distance between the distribution anomaly factor and the real-time fluctuation deviation factor of each target exhaust gas based on a preset weighted weight to obtain the cooperative deviation index of each target exhaust gas.

[0076] The dynamic threshold generation module 400 is used to calculate the dynamic alarm threshold of each target exhaust gas based on the cooperative deviation index of each target exhaust gas and the preset benchmark concentration threshold of each target exhaust gas.

[0077] The instability energy calculation module 500 is used to obtain a continuous sequence of multiple coordinated deviation indices of each target exhaust gas, and calculate the root mean square of the continuous sequence as the instantaneous instability energy of each target exhaust gas.

[0078] The graded alarm decision module 600 is used to perform graded alarms and decisions based on the comparison results of the instantaneous instability energy of each target exhaust gas with the pre-constructed health baseline of each target exhaust gas, and the comparison results of the concentration data of each target exhaust gas with the dynamic alarm threshold of each target exhaust gas.

[0079] The working principle of the online monitoring and alarm system for chemical plant exhaust gas provided by this invention is as follows:

[0080] This invention uses a data acquisition and statistics module 100 to simultaneously acquire raw concentration signals of multiple target exhaust gases. It also incorporates a sliding window calculation engine to process the input data stream in real time, independently calculating the kurtosis and baseline standard deviation of each exhaust gas within a long-term baseline window, and the short-term standard deviation within a short-term window. Subsequently, an anomaly factor generation module 200 receives these statistical features and executes two calculation logics: first, it calculates the difference between the kurtosis of each exhaust gas and the standard normal distribution kurtosis benchmark, taking the absolute value to generate a factor characterizing the anomaly of the distribution pattern; second, it calculates the ratio of the short-term standard deviation to the baseline standard deviation to generate a factor characterizing the real-time fluctuation deviation, thus transforming the raw statistics into dimensionless anomaly indicators. The co-deviation index generation module 300 then introduces preset weighting coefficients to perform weighted fusion calculations on the factors of distribution anomaly and real-time fluctuation deviation. The system employs a weighted Euclidean distance algorithm to output a comprehensive cooperative deviation index. A dynamic threshold generation module 400 uses the cooperative deviation index to linearly modulate the preset baseline concentration thresholds for each exhaust gas, achieving dynamic adaptive adjustment of the alarm threshold to sensitively reflect changes in process state. An instability energy calculation module 500 continuously maintains a sliding time series of the cooperative deviation index, quantifying the instantaneous instability energy of recent processes by calculating the root mean square value of this series, transforming the energy characteristics of the index series into a dynamic scalar. Finally, a graded alarm decision module 600, as the system's decision center, integrates health baseline data, real-time instantaneous instability energy, real-time concentration data, and dynamic alarm thresholds. It performs parallel dual comparison judgments and makes logical decisions based on preset multi-level rules, thereby outputting different levels of signals from low-level warnings to global emergency alarms, realizing an online monitoring and alarm system for chemical plant exhaust gases.

[0081] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0082] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, including read-only memory (ROM), random access memory (RAM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), one-time programmable read-only memory (OTPROM), electrically-Erasable Programmable Read-Only Memory (EEPROM), compactdisc read-only memory (CD-ROM) or other optical disc storage, disk storage, magnetic tape storage, or any other computer-readable medium capable of carrying or storing data.

[0083] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

Claims

1. A chemical plant tail gas on-line monitoring alarm method, characterized in that, The monitoring alarm method comprises the following steps: S1: synchronously and continuously collecting concentration data of each target tail gas, and calculating baseline standard deviation and statistical distribution kurtosis of each target tail gas within a preset sliding baseline time window, and calculating short-term standard deviation of each target tail gas within a preset sliding short-term time window; S2: based on the statistical distribution kurtosis and the standard normal distribution kurtosis of each target tail gas, calculating the distribution abnormality factor of each target tail gas, and based on the short-term standard deviation and the baseline standard deviation of each target tail gas, calculating the real-time fluctuation deviation factor of each target tail gas; S3: based on a preset weighting weight, calculating the weighted Euclidean distance of the distribution abnormality factor and the real-time fluctuation deviation factor of each target tail gas to obtain a cooperative deviation index of each target tail gas; S4: based on the cooperative deviation index of each target tail gas and the preset baseline concentration threshold of each target tail gas, calculating a dynamic alarm threshold of each target tail gas; S5: obtaining a continuous sequence formed by a plurality of cooperative deviation indexes of each target tail gas in sequence, and calculating the root mean square of the continuous sequence as an instantaneous instability energy of each target tail gas; S6: based on the comparison result of the instantaneous instability energy of each target tail gas and the pre-constructed health degree baseline of each target tail gas, and the comparison result of the concentration data of each target tail gas and the dynamic alarm threshold of each target tail gas, performing hierarchical alarm and decision; In step S2, the distribution abnormality factor of each target tail gas is obtained by calculating the absolute value of the difference between the statistical distribution kurtosis and the standard normal distribution kurtosis of each target tail gas; In step S2, the real-time fluctuation deviation factor of each target tail gas is obtained by calculating the ratio of the short-term standard deviation and the baseline standard deviation of each target tail gas; In step S3, the calculation formula of the cooperative deviation index is: wherein, is a synergy deviation index, is a distribution profile abnormality factor, is a real-time volatility deviation factor, and are preset weighting weights, respectively, and and the sum of which is 1. In step S4, the calculation formula of the dynamic alarm threshold is: wherein, is a dynamic alarm threshold, is a preset base alarm threshold, is an adjustment coefficient, is a synergy deviation index.

2. The method according to claim 1, wherein, In step S1, the length of the preset sliding baseline time window is at least 30 times the length of the preset sliding short-term time window.

3. The method according to claim 2, wherein, In step S5, the continuous sequence formed by a plurality of cooperative deviation indexes of each target tail gas in sequence is updated by a sliding window mechanism, wherein the continuous sequence is arranged in the order of the time when the cooperative deviation indexes are calculated, and when a new cooperative deviation index is calculated, it is inserted into the end of the continuous sequence, and the cooperative deviation index at the beginning of the continuous sequence is removed.

4. The method according to claim 3, wherein, In step S6, the construction steps of the health degree baseline of each target tail gas include: Based on the historical cooperative deviation index sequence of each target tail gas under the historical stable working condition, the first-order difference of the historical cooperative deviation index sequence of each target tail gas is calculated to obtain the historical cooperative deviation index change rate sequence of each target tail gas; The mean and standard deviation of the historical cooperative deviation index change rate sequence of each target tail gas are calculated, and the health degree baseline of each target tail gas is constructed based on the mean and standard deviation.

5. The method according to claim 4, wherein the method is characterized by, In step S6, the comparison result of the instantaneous instability energy of each target tail gas and the pre-constructed health degree baseline of each target tail gas includes: Based on the health degree baseline of each target tail gas, the control upper limit of each target tail gas is calculated, and the process stability state of each target tail gas is determined by comparing the instantaneous instability energy of each target tail gas with the corresponding control upper limit of each target tail gas.

6. An on-line monitoring and alarming system for tail gas of a chemical plant, characterized in that, The application is applied to the tail gas online monitoring and alarming method of a chemical plant as claimed in any one of claims 1 to 5, and the monitoring and alarming system comprises: a data acquisition and statistics module, which is used for synchronously and continuously collecting concentration data of each target tail gas, and calculating baseline standard deviation and statistical distribution kurtosis of each target tail gas within a preset sliding baseline time window, and calculating short-term standard deviation of each target tail gas within a preset sliding short-term time window; an abnormal factor generation module, which is used for calculating distribution abnormality degree factor of each target tail gas based on statistical distribution kurtosis and standard normal distribution kurtosis of each target tail gas, and calculating real-time fluctuation deviation factor of each target tail gas based on short-term standard deviation and baseline standard deviation of each target tail gas; a cooperative deviation index generation module, which is used for calculating weighted Euclidean distance of distribution abnormality degree factor and real-time fluctuation deviation factor of each target tail gas based on a preset weighting weight, to obtain cooperative deviation index of each target tail gas; a dynamic threshold generation module, which is used for calculating dynamic alarm threshold of each target tail gas based on cooperative deviation index of each target tail gas and preset reference concentration threshold of each target tail gas; an instability energy calculation module, which is used for obtaining continuous sequence composed of continuous multiple cooperative deviation indexes of each target tail gas, and calculating root mean square of the continuous sequence as instantaneous instability energy of each target tail gas; a hierarchical alarm decision module, which is used for comparing instantaneous instability energy of each target tail gas with pre-constructed health degree baseline of each target tail gas, and comparing concentration data of each target tail gas with dynamic alarm threshold of each target tail gas, to perform hierarchical alarm and decision.

Images