Directional catalytic liquefaction control method for coal tar in high-pressure hydrocracking container
By constructing a real-time parameter matrix and performing state classification and cluster analysis, the problems of reaction instability and catalyst deactivation location in high-pressure hydrocracking vessels were solved, and efficient and safe catalytic liquefaction control was achieved.
Patent Information
- Application Number
- CN202510850797.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-06-24
- Publication Date
- 2025-09-16
AI Technical Summary
In high-pressure hydrocracking vessels, traditional control methods are unable to capture parameter change trends and interactions in real time and comprehensively, resulting in unstable reactions, difficulty in locating catalyst deactivation areas, and increased production costs and safety risks.
By collecting temperature, pressure, coal tar flow rate and catalyst activity parameters, a real-time parameter matrix is constructed, the reaction state levels are divided, stable reaction areas are screened, cluster analysis is performed, abnormal reaction clusters are extracted, reaction abnormality factors and out-of-control risk values are calculated, and the catalyst deactivation area is located.
It achieves comprehensive and accurate monitoring of the reaction process, improves the accuracy and sensitivity of abnormal reaction identification, reduces production costs, and ensures safety and production efficiency.
Smart Images
Figure CN120648498A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the technical field of high-pressure hydrocracking containers, in particular to a method for controlling directional catalytic liquefaction of coal tar in a high-pressure hydrocracking container. Background Art
[0002] Coal tar, a key liquid byproduct of coal processing, has long been a research focus in the coal chemical industry for its efficient conversion and utilization. High-pressure hydrocracking technology, which uses catalysts to crack complex macromolecular organic matter in coal tar into small hydrocarbon compounds under high pressure and temperature conditions, thereby increasing the yield of light oil, has become a key technology for deep processing of coal tar. However, in actual production, the reaction environment within high-pressure hydrocracking vessels is complex and variable, presenting numerous issues that affect reaction efficiency and safety, hindering the further promotion and application of this technology.
[0003] High-pressure hydrocracking reactions involve multiple key parameters, including temperature, pressure, coal tar flow rate, and catalyst activity. The dynamic changes in these parameters are interrelated and mutually influential. Traditional control methods often struggle to fully capture the changing trends and interactions of these parameters in real time, resulting in an inability to accurately judge the reaction state and prone to instability in the reaction area. For example, if small fluctuations in temperature or pressure are not identified and adjusted in a timely manner, they may trigger localized reaction anomalies, affecting the reaction efficiency and product distribution within the entire vessel.
[0004] As the core element of high-pressure hydrocracking reaction, the activity of the catalyst directly determines the reaction rate and selectivity. During long-term operation, the catalyst may lose activity or even become inactivated due to carbon deposition, poisoning, sintering and other reasons. However, existing technologies make it difficult to accurately locate the deactivated area of the catalyst, and it is impossible to achieve targeted maintenance and replacement of the catalyst. This not only increases production costs, but may also cause the reaction to run away due to catalyst deactivation, leading to safety accidents. During the high-pressure hydrocracking process, the emergence of abnormal reaction clusters is often a precursor to runaway reactions. Traditional monitoring methods are usually based on the threshold judgment of a single parameter, and cannot comprehensively consider the coordinated changes of multiple parameters and the entropy characteristics of parameter fluctuations, resulting in delayed identification of abnormal reactions and difficulty in timely detection and handling of problems in the early stages, thereby increasing the risk of the production process. Summary of the Invention
[0005] The object of the present invention is to provide a method for controlling the directional catalytic liquefaction of coal tar in a high-pressure hydrocracking vessel to solve the problems raised in the above-mentioned background technology.
[0006] To achieve the above objectives, the present invention provides the following technical solution: a method for controlling the directional catalytic liquefaction of coal tar in a high-pressure hydrocracking vessel, the method comprising: Collect temperature, pressure, coal tar flow rate and catalyst activity parameters in the high-pressure hydrocracking vessel to build a real-time parameter matrix; Based on the dynamic change trend of each data point in the real-time parameter matrix, the reaction state level is divided and the state level matrix is generated. A sliding detection window is preset, and the stable reaction area is screened according to the standard deviation of the state level in the window to generate a correction matrix. Cluster analysis is performed on the unstable reaction area data in the correction matrix to extract abnormal reaction clusters. The reaction anomaly factor of each data point is calculated by combining the deviation between each data point in the abnormal reaction cluster and the global parameter mean, as well as the parameter fluctuation entropy in the neighborhood window. Analyze the parameter distribution differences between stable reaction regions and abnormal reaction clusters, integrate the parameter entropy change characteristics of stable reaction regions, and generate a sedimentation stability index for each stable reaction region; calculate the shortest Euclidean distance between each stable reaction region and the core point of the abnormal reaction cluster, and combine the reaction anomaly factor and the sedimentation stability index to generate a runaway risk value for each stable reaction region; The coordinates of the center of each stable reaction region in the real-time parameter matrix are mapped to the physical space of the container, and the catalyst deactivation region is located in combination with the out-of-control risk value.
[0007] Preferably, the specific process of dividing the reaction state levels is: The negative exponential function mapping result of the extreme values of all data points in the real-time parameter matrix is used as the grading step size; the minimum data point of the real-time parameter matrix is used as the grading starting benchmark, and the continuous grading interval is divided to assign a status grade to each data point; The state level matrix is constructed by the state levels of all data points.
[0008] Preferably, the process of screening the stable reaction area is specifically as follows: The standard deviation of the data within the sliding detection window is calculated when it moves in the state level matrix, and the continuous area with a standard deviation of 0 is selected as the stable response area.
[0009] Preferably, the correction matrix is a real-time parameter matrix after the data points at corresponding positions in the stable reaction area are set to zero.
[0010] Preferably, the specific process of extracting abnormal reaction clusters is: Set the upper limit of the number of clusters, calculate the parameter mean of the data points in each cluster, and select the cluster with the largest parameter mean as the abnormal response cluster.
[0011] Preferably, the calculation process of the reaction abnormality factor includes: Calculating the parameter mean of all data points in the abnormal reaction cluster; generating a local deviation degree based on the ratio of the parameter value of each data point to the parameter mean; The grayscale run matrix algorithm is used to calculate the fluctuation entropy value of each data point in the neighborhood window; The local deviation degree and the fluctuation entropy value are integrated to obtain a reaction anomaly factor; wherein the reaction anomaly factor is positively correlated with the local deviation degree and negatively correlated with the fluctuation entropy value.
[0012] Preferably, the process of generating the deposition stability index of each stable reaction area is specifically as follows: Calculating the absolute difference between the mean value of the parameter in each stable reaction region and the mean value of the parameter in the abnormal reaction cluster; extracting the maximum value between the absolute difference and zero; Combining this maximum with the parameter entropy within the stable reaction region yields the deposition stability index.
[0013] Preferably, the range of the neighborhood window is determined according to the spatial distribution characteristics of the abnormal reaction cluster: if the abnormal reaction cluster covers more than 3 consecutive sampling points, the neighborhood window is extended to 2 sampling points outside the edge of the abnormal cluster; if it is a discrete single point anomaly, the neighborhood window is limited to 1 sampling point before and after the point.
[0014] Preferably, the specific process of locating the catalyst deactivation area is: Map the real-time parameter matrix to the three-dimensional coordinate system of the container to generate the parameter distribution space; calculate the geometric center coordinates of each stable reaction area data point in the parameter distribution space; Converting the geometric center coordinates into proportional coordinates of the actual position of the container, wherein the abscissa is the ratio of the abscissa of the geometric center to the diameter of the container, and the ordinate is the ratio of the ordinate of the geometric center to the height of the container; When the proportional coordinate falls within the preset boundary of the catalyst bed, the stable reaction area is marked as the inspection area; if the out-of-control risk value of the inspection area exceeds the preset threshold, the corresponding geometric center is determined to be the catalyst deactivation position.
[0015] Preferably, after locating the catalyst deactivation area, an early warning signal is further generated according to the out-of-control risk value of each area: when the out-of-control risk value is higher than the first threshold, a yellow early warning is triggered, prompting the monitoring frequency to be doubled; when it is higher than the second threshold, a red early warning is triggered, and the catalyst replenishment program is automatically started.
[0016] Compared with the prior art, the present invention has the following beneficial effects: The control method and equipment for the directional catalytic liquefaction of coal tar in a high-pressure hydrocracking vessel provided by the present invention construct a real-time parameter matrix through multi-dimensional parameter collection and dynamic analysis, thereby achieving comprehensive and accurate monitoring of the reaction process. Dividing the reaction state levels based on the dynamic change trends of each data point in the real-time parameter matrix can clearly reflect the differences in reaction states, laying the foundation for subsequent screening of stable reaction areas and identification of abnormal reactions. Presetting a sliding detection window and screening stable reaction areas based on the standard deviation of the state levels within the window can quickly and accurately identify stable reaction areas, eliminate the interference of unstable factors, and improve the reliability of subsequent analysis.
[0017] Cluster analysis was performed on the unstable reaction area data in the correction matrix to extract abnormal reaction clusters. The reaction anomaly factor was calculated by combining the deviation of each data point in the abnormal reaction cluster from the global parameter mean and the parameter fluctuation entropy within the neighborhood window. This method can comprehensively assess the degree of abnormality of data points from multiple perspectives, overcoming the limitations of traditional single-parameter judgment and improving the accuracy and sensitivity of abnormal reaction identification. By analyzing the parameter distribution differences between stable reaction areas and abnormal reaction clusters, integrating the parameter entropy change characteristics of stable reaction areas to generate a sedimentation stability index, and calculating the shortest Euclidean distance between stable reaction areas and the core points of abnormal reaction clusters, and combining the reaction anomaly factor and sedimentation stability index to generate a loss of control risk value, a comprehensive and scientific risk assessment system was established. It can accurately predict the loss of control risk in each stable reaction area, providing a strong basis for taking early prevention and control measures.
[0018] The coordinates of the centers of each stable reaction region in the real-time parameter matrix are mapped to the physical space of the vessel. The runaway risk value is then used to locate the catalyst deactivation region, achieving a precise mapping of the parameter distribution to the physical space of the vessel. This allows operators to intuitively understand the specific location of the catalyst deactivation region, facilitating targeted maintenance and replacement, reducing production costs and improving production efficiency. Furthermore, this method can detect abnormal reaction clusters at an early stage. By calculating the reaction anomaly factor and the runaway risk value, it provides an early warning of the risk of runaway reactions, providing a crucial safeguard for the safety of the high-pressure hydrocracking process. BRIEF DESCRIPTION OF THE DRAWINGS
[0019] Figure 1 This is a working principle diagram of the method for controlling the directional catalytic liquefaction of coal tar in a high-pressure hydrocracking vessel according to the present invention; Figure 2 Design diagram for the classification of reaction states; Figure 3 Design diagram for calculation of reaction anomaly factor; Figure 4 Design diagram for locating the catalyst deactivation area. DETAILED DESCRIPTION
[0020] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.
[0021] See also Figure 1-Figure 4 The present invention relates to a method for controlling the directional catalytic liquefaction of coal tar in a high-pressure hydrocracking container, and the specific implementation steps are as follows: Collect temperature, pressure, coal tar flow rate and catalyst activity parameters in the high-pressure hydrocracking vessel to build a real-time parameter matrix; Based on the dynamic change trend of each data point in the real-time parameter matrix, the reaction state level is divided and the state level matrix is generated. A sliding detection window is preset, and the stable reaction area is screened according to the standard deviation of the state level in the window to generate a correction matrix. Cluster analysis is performed on the unstable reaction area data in the correction matrix to extract abnormal reaction clusters. The reaction anomaly factor of each data point is calculated by combining the deviation between each data point in the abnormal reaction cluster and the global parameter mean, as well as the parameter fluctuation entropy in the neighborhood window. Analyze the parameter distribution differences between stable reaction regions and abnormal reaction clusters, integrate the parameter entropy change characteristics of stable reaction regions, and generate a sedimentation stability index for each stable reaction region; calculate the shortest Euclidean distance between each stable reaction region and the core point of the abnormal reaction cluster, and combine the reaction anomaly factor and the sedimentation stability index to generate a runaway risk value for each stable reaction region; The coordinates of the center of each stable reaction region in the real-time parameter matrix are mapped to the physical space of the container, and the catalyst deactivation region is located in combination with the out-of-control risk value.
[0022] Example 1: During the operation of the high-pressure hydrocracking container, the temperature, pressure, coal tar flow rate and catalyst activity parameters are collected in real time by sensors arranged at different positions of the container. The temperature sensor adopts a multi-point thermocouple array, with multiple measuring points evenly distributed along the height direction of the container. Each measuring point has several measuring points evenly arranged in the circumferential direction to form a temperature monitoring network covering the entire internal space of the container. The pressure sensor adopts a high-precision pressure transmitter, which is arranged at the top, middle and bottom of the container respectively to monitor the pressure changes at different positions in the container in real time to ensure a comprehensive grasp of the pressure field inside the container. The coal tar flow rate is measured by an electromagnetic flowmeter, which is installed on the coal tar feed pipeline. It can accurately measure the flow rate of coal tar and provide accurate data support for subsequent flow control and reaction analysis. The catalyst activity parameters are obtained through online analytical instruments, and the catalyst samples are analyzed regularly, and the analysis results are transmitted to the control system in real time.
[0023] The data collected by these sensors is arranged in time sequence according to the order of temperature, pressure, coal tar flow rate, and catalyst activity parameters to form a real-time parameter matrix. Each row of the matrix represents the multi-parameter state at a sampling moment, and each column represents the trend of a specific parameter over time. When constructing the real-time parameter matrix, it is necessary to consider data synchronization and accuracy, ensuring that the data collected by different sensors are aligned in time and that the data accuracy meets the requirements of subsequent analysis. Furthermore, to ensure data reliability, the collected data must be preprocessed, including removing outliers and filtering to reduce noise.
[0024] The construction of a real-time parameter matrix provides a data foundation for subsequent reaction state analysis and the location of catalyst deactivation areas. Analysis of the real-time parameter matrix allows understanding the changing patterns and interrelationships of various parameters during the reaction process, thereby better understanding the evolution of the reaction state. In practical applications, the dimensions and scale of the real-time parameter matrix will vary depending on the specific process requirements and sensor layout. For large-scale high-pressure hydrocracking vessels, the real-time parameter matrix may contain hundreds or even thousands of data points, requiring efficient data storage and processing technologies to ensure the real-time and stability of the system.
[0025] When collecting temperature data, the layout of the multi-point thermocouple array must fully consider the temperature distribution characteristics within the container. Because high-pressure hydrocracking reactions are complex chemical processes, temperatures at different locations within the container can vary significantly. By evenly distributing measurement points in both the vertical and circumferential directions, comprehensive monitoring of temperature changes within the container is possible, enabling timely identification of temperature anomalies and providing accurate temperature data for subsequent reaction status analysis. Furthermore, to improve the accuracy and reliability of temperature measurement, the selection and installation of thermocouples must strictly adhere to relevant standards and specifications.
[0026] The placement of pressure sensors needs to consider the pressure distribution characteristics and flow conditions inside the container. During high-pressure hydrocracking, pressure changes inside the container can have a significant impact on the reaction rate and product distribution. By placing pressure sensors at the top, middle, and bottom of the container, pressure changes at different locations inside the container can be monitored in real time, understanding the distribution characteristics and evolution of the pressure field. At the same time, to ensure the accuracy and reliability of pressure measurement, the selection and installation of pressure sensors also need to fully consider the working environment and medium characteristics inside the container. The appropriate pressure sensor type and installation method must be selected to prevent the pressure sensor from being affected by factors such as high temperature, high pressure, and corrosion inside the container.
[0027] Measuring coal tar flow rate is crucial for controlling the reaction process and ensuring product quality. Real-time coal tar flow rate measurement using an electromagnetic flowmeter allows for accurate control of coal tar feed volume, ensuring the stability and consistency of the reaction process. Furthermore, changes in coal tar flow rate can reflect changes in the reaction state, providing an important reference for subsequent reaction state analysis. When installing an electromagnetic flowmeter, it's important to carefully consider its installation location and piping layout to ensure accurate coal tar flow measurement and avoid measurement errors caused by improper piping layout or installation location.
[0028] Online monitoring of catalyst activity parameters is key to ensuring efficient catalytic reactions. Regular analysis of catalyst samples using online analyzers provides timely insight into catalyst activity and performance changes, providing a scientific basis for catalyst replacement and regeneration. In practical applications, monitoring catalyst activity parameters requires selecting appropriate analytical methods and monitoring indicators based on the specific catalytic reaction type and catalyst characteristics. Furthermore, to ensure the accuracy and reliability of catalyst activity parameter monitoring, the selection and maintenance of online analyzers must strictly adhere to relevant standards and specifications, and regular calibration and maintenance must be performed.
[0029] When constructing a real-time parameter matrix, data synchronization and accuracy are key factors. Because different sensors may have varying sampling frequencies and response times, appropriate data synchronization techniques are necessary to ensure that data collected by different sensors are time-aligned. Furthermore, to improve data accuracy, the collected data must be preprocessed, including outlier removal and filtering for noise reduction. In practical applications, the appropriate method for outlier detection and processing must be selected based on the specific data distribution characteristics and business logic. For data points that significantly deviate from the normal range, a threshold method can be used to detect and remove them. For data points that slightly deviate from the normal range, the context of the data and business logic must be considered to determine whether they are outliers. Filtering and noise reduction can employ various filtering algorithms, such as mean filtering, median filtering, and Kalman filtering, to remove noise from the data.
[0030] The construction of the real-time parameter matrix provides a data foundation for subsequent reaction state analysis and the location of catalyst deactivation areas. By analyzing the real-time parameter matrix, we can understand the changing patterns and interrelationships of various parameters during the reaction process, thereby better understanding the evolution trend of the reaction state. In practical applications, the analysis of the real-time parameter matrix can adopt various data analysis methods and techniques, such as statistical analysis, machine learning, and data mining. Through statistical analysis, we can understand the basic statistical characteristics of each parameter, such as mean, variance, maximum value, minimum value, etc., thus gaining a preliminary understanding of the overall state of the reaction process. Through machine learning and data mining techniques, we can discover the potential relationships and changing patterns between various parameters, thereby conducting more in-depth analysis and prediction of the reaction process.
[0031] In practical applications, the construction and analysis of a real-time parameter matrix must be tailored to specific process requirements and actual production conditions. Different high-pressure hydrocracking processes may have varying reaction characteristics and parameter requirements, necessitating the selection of appropriate sensor layouts and data analysis methods based on the specific circumstances. Furthermore, to ensure the real-time and reliability of the system, a comprehensive data acquisition, transmission, storage, and processing mechanism must be established to ensure the timely and accurate collection and processing of various parameter data.
[0032] During the operation of a high-pressure hydrocracking vessel, the construction of a real-time parameter matrix is an ongoing process. As the reaction progresses and over time, new parameter data must be continuously collected and added to the real-time parameter matrix to ensure its real-time performance and integrity. Furthermore, to improve system performance and efficiency, the real-time parameter matrix must be regularly updated and maintained, including clearing outdated data, optimizing data structures, and updating analysis algorithms.
[0033] The construction of a real-time parameter matrix provides an important data foundation for the control method of coal tar directional catalytic liquefaction in high-pressure hydrocracking vessels. By rationally placing sensors, accurately collecting parameter data, and scientifically constructing a real-time parameter matrix, combined with advanced data analysis methods and technologies, we can fully understand the state changes and parameter relationships of the reaction process, promptly identify abnormalities and potential problems, and provide strong support for locating catalyst deactivation areas and optimizing the reaction process. This will improve the operating efficiency and product quality of high-pressure hydrocracking vessels and achieve precise control and optimized operation of coal tar directional catalytic liquefaction.
[0034] Example 2: During the operation of a high-pressure hydrocracking vessel, the reaction state is classified based on a constructed real-time parameter matrix. This process discretizes the continuous parameter space into a finite number of grade intervals, assigning each data point a corresponding state grade to form a state grade matrix. The state grade matrix intuitively reflects the distribution of reaction states, facilitating subsequent analysis of reaction state stability and anomalies.
[0035] When performing grading, the first step must be determined. This step size requires comprehensive consideration of the distribution characteristics and range of variation of the data in the real-time parameter matrix. By analyzing all data points in the real-time parameter matrix, the maximum and minimum values are calculated, thereby determining the data range. The range reflects the overall range of variation in the data and is an important basis for determining the grading step size. When determining the step size, the severity of parameter fluctuations must be considered. For parameters with large fluctuations, the step size should be appropriately reduced to more accurately reflect the changes. For parameters with smaller fluctuations, the step size can be appropriately increased to improve analysis efficiency.
[0036] Using the minimum data point in the real-time parameter matrix as the starting grade, continuous grade intervals are divided according to the calculated step size. Each grade interval represents a specific range of reaction states, and data points falling into different grade intervals represent corresponding reaction states. When dividing the grade intervals, it is necessary to ensure the continuity and integrity of the intervals to avoid situations where data points cannot be assigned to any grade interval. At the same time, to ensure the rationality of the grade division, it is necessary to appropriately adjust the number and range of grade intervals based on the actual process requirements and reaction characteristics.
[0037] For each data point in the real-time parameter matrix, it is necessary to determine the level interval to which it belongs and assign it a corresponding status level. This process requires iterating over all data points in the real-time parameter matrix, determining the level interval to which they belong based on their numerical values, and then recording the corresponding status level. When determining the level interval to which a data point belongs, it is necessary to consider the accuracy and boundary conditions of the data. For data points that fall exactly on the boundaries of the level interval, their attribution principles must be clearly defined to ensure consistency and accuracy in the level assignment.
[0038] The state levels of all data points form a state level matrix. This matrix has the same dimensions as the real-time parameter matrix, where each element represents the state level of the data point at the corresponding location. The state level matrix discretizes the continuous parameter space into a finite number of levels, facilitating subsequent analysis of reaction state stability and anomalies. By analyzing the state level matrix, we can intuitively understand the distribution of reaction states and identify anomalous areas and changing trends during the reaction process.
[0039] In practical applications, the rationality of grading directly impacts the accuracy and effectiveness of subsequent analysis. If the grading is too coarse, it may mask important changes in reaction states, preventing anomalies from being detected promptly. If the grading is too detailed, it increases data processing complexity, reduces analysis efficiency, and may also introduce excessive noise, affecting the reliability of analytical results. Therefore, when performing grading, it is necessary to select appropriate grading methods and parameters based on specific process requirements and reaction characteristics.
[0040] When grading, the interrelationships and influences between different parameters also need to be considered. During high-pressure hydrocracking, complex interactions and coupling relationships can exist between parameters such as temperature, pressure, coal tar flow rate, and catalyst activity. Therefore, when grading, it is necessary to comprehensively consider the variations in these parameters to avoid the one-sidedness that can result from grading based solely on a single parameter. A multi-parameter comprehensive analysis approach can be employed to rationally weight and integrate the influences of different parameters, resulting in a more comprehensive and accurate classification of reaction states.
[0041] The construction of the state-level matrix provides an important data foundation for subsequent reaction state analysis. By analyzing the state-level matrix, we can understand the distribution and changing trends of the state levels during the reaction process, thereby determining the stability and abnormalities of the reaction state. In practical applications, various data analysis methods and techniques, such as statistical analysis, cluster analysis, and association rule mining, can be used to conduct in-depth analysis of the state-level matrix, uncovering the hidden information and knowledge within it, and providing decision support for the control and optimization of the reaction process.
[0042] When constructing the state-level matrix, the real-time and dynamic nature of the data also needs to be considered. During the operation of the high-pressure hydrocracking vessel, the reaction state is constantly changing, and the data in the real-time parameter matrix is constantly updated. Therefore, the state-level matrix also needs to be updated promptly as the real-time parameter matrix is updated to ensure that it accurately reflects the current reaction state. In practical applications, a real-time data acquisition and processing system can be established to achieve real-time updates of the real-time parameter matrix and dynamic construction of the state-level matrix, thereby supporting real-time monitoring and control of the reaction process.
[0043] The construction of the state level matrix also requires consideration of data visualization and interpretability. The state level matrix is a multidimensional data structure that contains a wealth of information and knowledge. To facilitate operator understanding and analysis of this information, data visualization can be used to present the state level matrix in intuitive graphs and charts. Data visualization allows operators to more intuitively understand the distribution and changing trends of reaction states, enabling more accurate assessment of reaction state stability and anomalies, allowing timely implementation of appropriate adjustments and control measures.
[0044] Example 3: During the operation of the high-pressure hydrocracking vessel, after the state level matrix is constructed, it is necessary to further screen the stable reaction regions and generate a correction matrix.
[0045] Assume that at a certain moment in time, a state-level matrix containing temperature, pressure, coal tar flow rate, and catalyst activity parameters has been obtained. Each element in the matrix represents the state level of the corresponding data point, ranging from 1 to 10, with higher levels indicating more intense reactions. To identify stable reaction regions, a sliding detection window is preset. The window size is determined based on actual production conditions, ensuring that it covers a certain time range and parameter variations, but not so large that it ignores local fluctuations.
[0046] The sliding detection window begins moving within the state-level matrix at a set step size, calculating the standard deviation of the data within the window with each movement. The standard deviation reflects the degree of dispersion of the data within the window; a smaller standard deviation indicates more stable data. For example, when the window moves to a certain area in the matrix, the state-level data contained within the window are 3, 3, 3, 3, 3, and 3. These data are exactly the same, and the calculated standard deviation is 0, indicating that the reaction state in this area is very stable and the parameter fluctuations are minimal.
[0047] Continuing to move the window, when we encounter a region containing state level data of 3, 4, 3, 5, 4, and 3, the data fluctuates somewhat, and the calculated standard deviation is not zero, indicating that the reaction state in this region is relatively unstable and the parameters are fluctuating. In this way, the sliding detection window traverses the entire state level matrix, calculating the standard deviation at each window position.
[0048] Continuous regions with a standard deviation of 0 are selected as stable reaction regions. Within these regions, the data points have identical state levels, indicating that the reaction state remains highly consistent within the time and space corresponding to this region, with no significant fluctuations. For example, in the upper left corner of the state level matrix, a 5×5 submatrix is found where all data points have a state level of 4, and the standard deviation of the surrounding area is not 0. This 5×5 submatrix is then identified as a stable reaction region.
[0049] As the sliding detection window continues to move, a 3×4 submatrix is discovered in the lower right corner of the state-level matrix. All data points in this submatrix have a state level of 6. Similarly, the standard deviation of the area surrounding this submatrix is not zero, so this 3×4 submatrix is also identified as a stable reaction region. In this way, multiple stable reaction regions are screened out from the entire state-level matrix. These regions represent time periods and spatial locations where the reaction process is relatively stable.
[0050] After determining the stable reaction regions, a correction matrix is generated. This is obtained by setting the data points corresponding to the stable reaction regions in the real-time parameter matrix to zero. The real-time parameter matrix is the original matrix containing the temperature, pressure, coal tar flow rate, and catalyst activity parameters, while the state level matrix is obtained by classifying the real-time parameter matrix. To generate the correction matrix, the locations of all stable reaction regions in the state level matrix are first found, and then the data points corresponding to these locations in the real-time parameter matrix are set to zero.
[0051] For example, for the previously determined 5×5 stable response region in the upper left corner, find the corresponding 5×5 submatrix in the real-time parameter matrix and set all data points in this submatrix to zero. Similarly, for the 3×4 stable response region in the lower right corner, find the corresponding 3×4 submatrix in the real-time parameter matrix and set all data points in this submatrix to zero. By performing this process on all stable response regions, we ultimately obtain the correction matrix.
[0052] The correction matrix highlights the data characteristics of unstable reaction regions. Since the data points in the stable reaction regions have been reset to zero, the remaining non-zero data points in the correction matrix represent areas of unstable and fluctuating reaction conditions. These areas may be potential areas of reaction anomalies and require further analysis. For example, in the correction matrix, you may find that the data points in certain areas are large and concentrated, which may indicate obvious reaction anomalies in these areas.
[0053] By screening stable reaction regions and generating a correction matrix, we can effectively distinguish between normal and abnormal reaction states. Stable reaction regions represent areas of relatively stable reaction conditions with minimal parameter fluctuations during the reaction process. These areas are typically areas where the reaction is proceeding normally. Unstable reaction regions, on the other hand, represent areas of unstable reaction conditions with significant parameter fluctuations. These areas may be areas of abnormal reaction and require special attention.
[0054] In practical applications, the process of screening stable reaction regions and generating the correction matrix requires consideration of multiple factors. First, the size of the sliding detection window is crucial. If the window is too small, some large stable regions may be missed; if the window is too large, some areas with local fluctuations may be mistakenly identified as stable regions. Second, the standard deviation threshold setting also needs to be adjusted based on actual conditions. Although the example uses a standard deviation of 0 as the criterion for screening stable regions, in practice, this criterion may need to be relaxed or tightened appropriately based on the characteristics of the reaction and the stability requirements.
[0055] The process of screening for stable reaction regions and generating a correction matrix requires integration with practical process knowledge and experience. Different high-pressure hydrocracking processes may have different reaction characteristics and parameter requirements, necessitating adjustments to the screening methods and parameters based on the specific circumstances. For example, processes with high reaction stability requirements may require more stringent screening for stable regions; whereas processes that tolerate some fluctuations may require more relaxed screening criteria.
[0056] Screening for stable reaction regions and generating a correction matrix are key steps in the control method for directed catalytic liquefaction of coal tar in high-pressure hydrocracking vessels. By presetting a sliding detection window, calculating the standard deviation of the data within the window, and screening continuous regions with a standard deviation of 0 as stable reaction regions, the data points corresponding to these regions are reset to zero in the real-time parameter matrix to generate a correction matrix. This process effectively distinguishes normal from abnormal reaction states, highlighting the data characteristics of unstable reaction regions, and providing a basis for subsequent anomaly detection and processing, thereby improving the monitoring and control capabilities of the reaction process and ensuring the safe and stable operation of the high-pressure hydrocracking vessel.
[0057] Example 4: During the operation of a high-pressure hydrocracking vessel, after the correction matrix is generated, cluster analysis is performed on the unstable reaction region data in the correction matrix to extract abnormal reaction clusters and calculate the reaction anomaly factor. This step is critical for identifying abnormalities during the reaction process, helping operators promptly identify potential problems and take appropriate measures to adjust and control them.
[0058] Assume that at a certain moment in time, a correction matrix has been obtained. This matrix is obtained by setting the data points corresponding to the stable reaction region in the real-time parameter matrix to zero. Therefore, the non-zero data points in the correction matrix represent areas of unstable and fluctuating state during the reaction process. These non-zero data points constitute the unstable reaction region data, which requires cluster analysis.
[0059] First, an upper limit is set for the number of clusters. This limit is determined based on actual production experience and data characteristics. It ensures that a sufficient number of abnormal patterns can be discovered, but not too many, resulting in overly dispersed clustering results. Then, a density clustering algorithm is used to cluster the data in the unstable reaction area. This algorithm is a clustering method based on the density of data points. It divides high-density areas in the data space into clusters, while treating low-density areas as noise points.
[0060] During the clustering process, the parameter mean of the data points within each cluster is calculated. The parameter mean reflects the average characteristics of the reaction state represented by that cluster. For example, for the temperature parameter, a high mean temperature of the data points within a cluster indicates that the temperature of the reaction region represented by that cluster is relatively high; for the pressure parameter, a low mean pressure of the data points within a cluster indicates that the pressure of the reaction region represented by that cluster is relatively low.
[0061] The cluster with the largest parameter mean is selected as the abnormal reaction cluster. This is because in high-pressure hydrocracking, clusters with large parameter means often represent areas of abnormally intense reactions or significant deviations. For example, if the mean temperature of a cluster is significantly higher than that of other clusters, it may indicate overheating in that area, which may be caused by localized excessive catalyst activity, uneven reaction heat release, or cooling system failure.
[0062] For each data point within the abnormal reaction cluster, its deviation from the global parameter mean is calculated. The global parameter mean is the parameter mean of all non-zero data points in the correction matrix, representing the average reaction state of the entire unstable reaction region. The deviation reflects the degree to which the data point deviates from the average reaction state. A larger deviation indicates that the data point deviates further from the normal reaction state. For example, if the temperature value of a data point is much higher than the global temperature mean, it indicates that the reaction region corresponding to that data point is abnormally hot, indicating a potential problem.
[0063] At the same time, the grayscale run-length matrix algorithm is used to calculate the fluctuation entropy of each data point in the neighborhood window. The grayscale run-length matrix algorithm is a method used to analyze image texture features. Here, its application to parameter data analysis effectively captures the local fluctuation characteristics of the data. The neighborhood window refers to a local area centered on a data point. The window size is determined based on the actual situation and is generally selected to cover a certain number of adjacent data points. The fluctuation entropy value reflects the complexity of the parameter fluctuations around the data point. A larger fluctuation entropy value indicates more complex parameter fluctuations around the data point and a more unstable reaction state.
[0064] The reaction anomaly factor is derived by combining the local deviation and the fluctuation entropy. This factor comprehensively considers the deviation of the data point and the fluctuations of the surrounding environment, enabling a more accurate assessment of the degree of reaction anomaly. For example, a large local deviation of a data point indicates that it has deviated significantly from the normal reaction state. Simultaneously, a high fluctuation entropy indicates that the surrounding parameters are also experiencing complex fluctuations. These two factors combined indicate that the reaction region corresponding to that data point is experiencing a significant anomaly.
[0065] The range of the neighborhood window is dynamically adjusted based on the spatial distribution characteristics of the abnormal reaction cluster. If the abnormal reaction cluster covers more than three consecutive sampling points, it indicates that the abnormal situation has a certain spatial extension. In this case, the neighborhood window is extended to two sampling points outside the edge of the abnormal cluster to fully consider the impact of the abnormal area. For example, if the abnormal reaction cluster is manifested in the temperature parameter matrix as five consecutive sampling points with significantly high temperature values, then for the data points in this abnormal cluster, the neighborhood window will be extended to the range of two sampling points before and after these five sampling points, that is, the range of nine sampling points in total. This allows for a more comprehensive analysis of parameter fluctuations in and around the abnormal area.
[0066] If the abnormal reaction cluster is a discrete single-point anomaly, indicating that the anomaly occurs only at a single point with no significant spatial extension, the neighborhood window is limited to one sampling point before and after the point to focus on the characteristics of the anomaly itself. For example, if an isolated data point is found in the pressure parameter matrix with a significantly low pressure value, while the pressure values of the surrounding data points are normal, then for this discrete single-point anomaly, its neighborhood window will be limited to the range of the point and one sampling point before and after it, that is, the range of three sampling points in total. This allows for more accurate analysis of the characteristics of the anomaly point and its relationship with neighboring points.
[0067] By analyzing the data points within the abnormal reaction cluster, key abnormal points in the reaction process can be identified. These key abnormal points typically have high reaction abnormality factors, and their corresponding reaction areas may have serious abnormalities and require special attention. For example, in the temperature parameter matrix, it is found that the temperature value of a data point is much higher than the global temperature mean, and the temperature fluctuations around it are also very complex. The calculated reaction abnormality factor is very high, indicating that the reaction area corresponding to this data point may have local overheating. Further inspection of the catalyst activity and cooling system operation in this area is necessary to determine the cause of the abnormality and take appropriate measures to adjust it.
[0068] Cluster analysis and anomaly factor calculation of the unstable reaction region data in the correction matrix can help operators promptly detect anomalies during the reaction process, determine the location and severity of the anomaly region, and provide a basis for subsequent reaction state adjustment and optimization. In practical applications, this process requires a comprehensive assessment of the analysis results in combination with specific process knowledge and experience to ensure that true anomalies can be accurately identified and effective measures can be taken to address them.
[0069] When performing cluster analysis and calculating outlier factors, data quality and accuracy must also be considered. Noise or outliers in the collected data may affect the accuracy of clustering results and outlier factor calculations. Therefore, before analysis, data preprocessing is often necessary, such as removing noise and correcting outliers, to improve data quality. Furthermore, the choice of clustering algorithm and parameter settings can also affect analysis results and require adjustment and optimization based on specific circumstances.
[0070] Cluster analysis of the unstable reaction region data in the correction matrix, extraction of abnormal reaction clusters, and calculation of the reaction anomaly factor are key steps in the control method for the directional catalytic liquefaction of coal tar in a high-pressure hydrocracking vessel. By setting an upper limit on the number of clusters, a density clustering algorithm is used to cluster the unstable reaction region data. The parameter mean of the data points within each cluster is calculated, and the cluster with the largest parameter mean is selected as the abnormal reaction cluster. The deviation of each data point within the abnormal reaction cluster from the global parameter mean and the fluctuation entropy of the neighborhood window are then calculated. These two factors are combined to derive the reaction anomaly factor. The neighborhood window range is dynamically adjusted based on the spatial distribution characteristics of the abnormal reaction clusters.
[0071] Example 5: During the operation of a high-pressure hydrocracking vessel, after extracting abnormal reaction clusters and calculating reaction anomaly factors, further analysis is required to determine the parameter distribution differences between stable reaction regions and abnormal reaction clusters. This generates a deposition stability index and runaway risk value, locates catalyst deactivation areas, and generates early warning signals. This step, the ultimate goal of the entire control method, helps operators accurately identify catalyst deactivation areas and take timely adjustment and control measures to ensure stable operation of the reaction process.
[0072] Assume that at a certain moment, data for stable reaction regions and abnormal reaction clusters have been obtained. Stable reaction regions are obtained by screening continuous regions with a standard deviation of 0 in the state ranking matrix. These regions represent areas with relatively stable states and minimal parameter fluctuations during the reaction process. Abnormal reaction clusters are obtained by performing cluster analysis on the unstable reaction region data in the correction matrix and selecting the cluster with the largest parameter mean. This cluster represents areas with severe abnormalities or significant deviations during the reaction process.
[0073] First, the absolute difference between the mean parameter value within each stable reaction region and the mean parameter value of the anomalous reaction cluster is calculated. The parameter mean reflects the average characteristics of the reaction state represented by that region, while the absolute difference reflects the degree of difference in parameter characteristics between the stable and anomalous reaction regions. For example, for the temperature parameter, if the mean temperature of a stable reaction region is 300°C, while the mean temperature of the anomalous reaction cluster is 450°C, the absolute difference between the two is 150°C, indicating a significant difference in temperature parameters between the stable reaction region and the anomalous reaction cluster.
[0074] Extract the maximum absolute difference from zero and ensure that the difference is non-negative. This step is to avoid negative values during the calculation process, which may affect the accuracy of subsequent analysis results. For example, if the mean pressure of a stable reaction region is 5 MPa, while the mean pressure of an abnormal reaction cluster is 4 MPa, the absolute difference between the two is 1 MPa, and even after extracting the maximum value from zero, it is still 1 MPa. If the mean catalyst activity of a stable reaction region is 0.8, while the mean catalyst activity of an abnormal reaction cluster is 0.9, the absolute difference between the two is 0.1, and even after extracting the maximum value from zero, it is still 0.1.
[0075] The maximum value is combined with the parameter entropy change within the stable reaction region to produce the deposition stability index. The parameter entropy change reflects the complexity of parameter changes within the stable reaction region. A larger entropy change indicates more complex parameter changes and a more unstable reaction state. The deposition stability index comprehensively considers both parameter differences and the complexity of these changes, effectively assessing the stability and potential risks of stable reaction regions. For example, if a stable reaction region has a large absolute difference in temperature parameters from an anomalous reaction cluster and a large entropy change in the temperature parameters within that region, this indicates that while the region is currently stable, it differs significantly from the anomalous reaction cluster, has complex internal parameter changes, and presents certain potential risks. Therefore, its deposition stability index is relatively low.
[0076] Calculate the shortest Euclidean distance between each stable reaction region and the core point of the abnormal reaction cluster. Euclidean distance refers to the straight-line distance between two points in multidimensional space and is used here to measure the positional relationship between stable reaction regions and abnormal reaction clusters in parameter space. The core point of an abnormal reaction cluster is the geometric center of all data points within an abnormal reaction cluster and represents the average positional characteristics of the abnormal reaction cluster. For example, in a three-dimensional parameter space containing three parameters: temperature, pressure, and coal tar flow rate, the data points of a stable reaction region are distributed near the coordinates (300, 5, 10), while the coordinates of the core point of the abnormal reaction cluster are (450, 4, 12). The shortest Euclidean distance between this stable reaction region and the core point of the abnormal reaction cluster can be calculated by calculating the straight-line distance between these two points.
[0077] Combining the reaction anomaly factor and the sedimentation stability index, a runaway risk value is generated for each stable reaction region. This risk value comprehensively considers the degree of abnormal reaction, the stability of the stable reaction region, and the spatial relationship between the two, providing a comprehensive assessment of the runaway risk of a stable reaction region. For example, a high reaction anomaly factor in a stable reaction region indicates the presence of severe abnormal reactions in its surroundings. Simultaneously, a low sedimentation stability index in this region indicates poor stability. Furthermore, this region is close to the core point of the abnormal reaction cluster, indicating that it is significantly affected by the abnormal reactions. Taking these three factors into account, the runaway risk value of this stable reaction region is relatively high, indicating a high risk of runaway and requiring special attention.
[0078] The real-time parameter matrix is mapped to the container's three-dimensional coordinate system to generate a parameter distribution space. The real-time parameter matrix is the original matrix containing temperature, pressure, coal tar flow rate, and catalyst activity parameters. By mapping it to the container's three-dimensional coordinate system, the distribution of each parameter within the container can be intuitively displayed. For example, in the container's three-dimensional coordinate system, the x-axis represents the container's length, the y-axis represents the container's width, and the z-axis represents the container's height. By mapping the temperature parameters in the real-time parameter matrix to this three-dimensional coordinate system, the three-dimensional distribution of the temperature within the container can be obtained, providing an intuitive understanding of the high and low temperature distribution and changing trends within the container.
[0079] Calculate the geometric center coordinates of each stable reaction region data point in the parameter distribution space. The geometric center coordinates are the average position of all data points within the region and represent the average positional characteristics of the stable reaction region. For example, for a stable reaction region distributed in the three-dimensional coordinate system of a container, whose data point coordinates are (x1, y1, z1), (x2, y2, z2), ..., (xn, yn, zn), the geometric center coordinates of the stable reaction region are ((x1+x2+...+xn) / n, (y1+y2+...+yn) / n, (z1+z2+...+zn) / n).
[0080] Convert the geometric center coordinates to scaled coordinates of the actual container location. Scaled coordinates convert the geometric center coordinates to proportional values relative to the container dimensions, where the horizontal axis represents the ratio of the horizontal coordinate of the geometric center to the container diameter, and the vertical axis represents the ratio of the vertical coordinate of the geometric center to the container height. This provides a more intuitive representation of the actual location of the stable reaction zone within the container. For example, if the geometric center coordinates of a stable reaction zone are (1.5m, 2m, 3m), and the container has a diameter of 4m and a height of 10m, the scaled coordinates of the stable reaction zone are (1.5 / 4=0.375, 2 / 10=0.2, 3 / 10=0.3), indicating that the geometric center of the stable reaction zone is located at 37.5% of the container diameter, 20% of the height, and 30% of the length.
[0081] When the scaled coordinates fall within the preset boundaries of the catalyst bed, the stable reaction region is marked as a pending inspection area. The catalyst bed is the key area where catalytic reactions occur, and its boundaries are usually preset during vessel design. For example, in the three-dimensional coordinate system of the vessel, the preset boundaries of the catalyst bed are 0.2-0.8 in the x-direction, 0.3-0.7 in the y-direction, and 0.4-0.6 in the z-direction. If the scaled coordinates of a stable reaction region are (0.5, 0.5, 0.5), and these coordinates fall within the preset boundaries of the catalyst bed, the stable reaction region is marked as a pending inspection area.
[0082] If the risk of loss of control in the inspection zone exceeds a preset threshold, the corresponding geometric center is identified as the catalyst deactivation location. This threshold is determined based on actual production experience and catalyst performance characteristics and is used to assess the severity of the risk of loss of control. For example, if the preset threshold is 0.7 and the risk of loss of control in a particular inspection zone is 0.8, exceeding the threshold, the geometric center of the inspection zone is identified as the catalyst deactivation location. This indicates that the catalyst at that location may have lost its activity and is no longer able to catalyze the reaction properly, requiring prompt replacement or regeneration.
[0083] After locating the catalyst deactivation area, an early warning signal is generated based on the risk of loss of control in each area. Early warning signals are categorized into two levels: yellow and red, corresponding to different levels of risk. When the risk of loss of control exceeds the first threshold, a yellow warning is triggered, prompting a doubling of monitoring frequency. This first threshold is determined based on actual production experience and is used to determine whether increased monitoring is necessary. For example, if the first threshold is 0.6, and the risk of loss of control in a particular area is 0.65, exceeding the first threshold, a yellow warning is triggered, and the system automatically doubles the monitoring frequency in that area to more promptly detect potential problems.
[0084] When the runaway risk value exceeds the second threshold, a red alert is triggered and the catalyst refill process is automatically initiated. The second threshold is determined based on catalyst performance characteristics and reaction safety requirements and is used to determine whether immediate action is needed. For example, if the second threshold is 0.8 and the runaway risk value in a certain area is 0.85, which is higher than the second threshold, a red alert is triggered and the system automatically initiates the catalyst refill process, replenishing fresh catalyst in that area to restore catalyst activity and ensure normal reaction progress.
[0085] This hierarchical warning mechanism can effectively improve the safety and reliability of the system and reduce production risks. A yellow warning can remind operators to strengthen monitoring of specific areas and promptly identify potential problems; a red warning can automatically initiate emergency measures when the problem is serious to avoid accidents. For example, during the operation of a high-pressure hydrocracking vessel, the system detected an out-of-control risk value of 0.7 in an area, triggering a yellow warning. The operator strengthened monitoring of the area and found that the temperature in the area had a gradual upward trend. The operating parameters of the cooling system were adjusted in time to avoid catalyst deactivation caused by continued temperature increases. Subsequently, in another area, the system detected an out-of-control risk value of 0.85, triggering a red warning. The system automatically started the catalyst replenishment program, replenished fresh catalyst to the area, restored the catalyst activity, and ensured the normal progress of the reaction.
[0086] Analyzing the differences in parameter distribution between stable reaction regions and abnormal reaction clusters, generating a deposition stability index and runaway risk value, locating catalyst deactivation areas, and generating early warning signals are key steps in the control method for directional catalytic liquefaction of coal tar in high-pressure hydrocracking vessels. By calculating the absolute difference between the parameter means of the stable reaction region and the abnormal reaction cluster, extracting the maximum value from zero, and fusing this maximum value with the parameter entropy, the deposition stability index is derived. The shortest Euclidean distance between the stable reaction region and the core point of the abnormal reaction cluster is calculated. The runaway risk value is generated by combining the reaction anomaly factor and the deposition stability index. The real-time parameter matrix is mapped to the three-dimensional coordinate system of the vessel. The geometric center coordinates of the stable reaction region are calculated and converted to proportional coordinates. The inspection area that falls within the preset boundaries of the catalyst bed is marked. The geometric center of the inspection area where the runaway risk value exceeds the preset threshold is determined to be the catalyst deactivation location. Yellow and red warnings are generated based on the runaway risk value, and the monitoring frequency is automatically adjusted and the catalyst replenishment program is initiated.
[0087] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "includes," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus.
[0088] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.
Claims
1. A method for controlling the directional catalytic liquefaction of coal tar in a high-pressure hydrocracking vessel, characterized in that: The method comprises the following steps: Collect temperature, pressure, coal tar flow rate and catalyst activity parameters in the high-pressure hydrocracking vessel to build a real-time parameter matrix; Based on the dynamic change trend of each data point in the real-time parameter matrix, the reaction state level is divided and the state level matrix is generated. A sliding detection window is preset, and the stable reaction area is screened according to the standard deviation of the state level in the window to generate a correction matrix. Cluster analysis is performed on the unstable reaction area data in the correction matrix to extract abnormal reaction clusters. The reaction anomaly factor of each data point is calculated by combining the deviation between each data point in the abnormal reaction cluster and the global parameter mean, as well as the parameter fluctuation entropy in the neighborhood window. Analyze the parameter distribution differences between stable reaction regions and abnormal reaction clusters, integrate the parameter entropy change characteristics of stable reaction regions, and generate a sedimentation stability index for each stable reaction region; calculate the shortest Euclidean distance between each stable reaction region and the core point of the abnormal reaction cluster, and combine the reaction anomaly factor and the sedimentation stability index to generate a runaway risk value for each stable reaction region; The coordinates of the center of each stable reaction region in the real-time parameter matrix are mapped to the physical space of the container, and the catalyst deactivation region is located in combination with the out-of-control risk value.
2. The method for controlling the directional catalytic liquefaction of coal tar in a high-pressure hydrocracking vessel according to claim 1, wherein: The specific process of dividing the reaction state levels is as follows: The negative exponential function mapping result of the extreme values of all data points in the real-time parameter matrix is used as the grading step size; the minimum data point of the real-time parameter matrix is used as the grading starting benchmark, and the continuous grading interval is divided to assign a status grade to each data point; The state level matrix is constructed by the state levels of all data points.
3. The method for controlling the directional catalytic liquefaction of coal tar in a high-pressure hydrocracking vessel according to claim 1, wherein: The process of screening the stable reaction area is specifically as follows: The standard deviation of the data within the sliding detection window is calculated when it moves in the state level matrix, and the continuous area with a standard deviation of 0 is selected as the stable response area.
4. The method for controlling the directional catalytic liquefaction of coal tar in a high-pressure hydrocracking vessel according to claim 1, wherein: The correction matrix is a real-time parameter matrix after the data points at corresponding positions in the stable reaction area are set to zero.
5. The method for controlling the directional catalytic liquefaction of coal tar in a high-pressure hydrocracking vessel according to claim 1, wherein: The specific process of extracting abnormal reaction clusters is as follows: Set the upper limit of the number of clusters, calculate the parameter mean of the data points in each cluster, and select the cluster with the largest parameter mean as the abnormal response cluster.
6. The method for controlling the directional catalytic liquefaction of coal tar in a high-pressure hydrocracking vessel according to claim 1, wherein: The calculation process of the reaction abnormality factor includes: Calculating the parameter mean of all data points in the abnormal reaction cluster; generating a local deviation degree based on the ratio of the parameter value of each data point to the parameter mean; The grayscale run matrix algorithm is used to calculate the fluctuation entropy value of each data point in the neighborhood window; The local deviation degree and the fluctuation entropy value are integrated to obtain a reaction anomaly factor; wherein the reaction anomaly factor is positively correlated with the local deviation degree and negatively correlated with the fluctuation entropy value.
7. The method for controlling the directional catalytic liquefaction of coal tar in a high-pressure hydrocracking vessel according to claim 1, wherein: The process of generating the deposition stability index of each stable reaction area is specifically as follows: Calculating the absolute difference between the mean value of the parameter in each stable reaction region and the mean value of the parameter in the abnormal reaction cluster; extracting the maximum value between the absolute difference and zero; Combining this maximum with the parameter entropy within the stable reaction region yields the deposition stability index.
8. The method for controlling the directional catalytic liquefaction of coal tar in a high-pressure hydrocracking vessel according to claim 6, wherein: The range of the neighborhood window is determined according to the spatial distribution characteristics of the abnormal reaction cluster: if the abnormal reaction cluster covers more than 3 consecutive sampling points, the neighborhood window is extended to 2 sampling points outside the edge of the abnormal cluster; if it is a discrete single point anomaly, the neighborhood window is limited to 1 sampling point before and after the point.
9. The method for controlling the directional catalytic liquefaction of coal tar in a high-pressure hydrocracking vessel according to claim 1, wherein: The specific process of locating the catalyst deactivation area is as follows: Map the real-time parameter matrix to the three-dimensional coordinate system of the container to generate the parameter distribution space; calculate the geometric center coordinates of each stable reaction area data point in the parameter distribution space; Converting the geometric center coordinates into proportional coordinates of the actual position of the container, wherein the abscissa is the ratio of the abscissa of the geometric center to the diameter of the container, and the ordinate is the ratio of the ordinate of the geometric center to the height of the container; When the proportional coordinate falls within the preset boundary of the catalyst bed, the stable reaction area is marked as the inspection area; if the out-of-control risk value of the inspection area exceeds the preset threshold, the corresponding geometric center is determined to be the catalyst deactivation position.
10. The method for controlling the directional catalytic liquefaction of coal tar in a high-pressure hydrocracking vessel according to claim 1, wherein: After locating the catalyst deactivation area, an early warning signal is further generated according to the out-of-control risk value of each area: when the out-of-control risk value is higher than a first threshold, a yellow early warning is triggered, prompting the monitoring frequency to be doubled; when it is higher than a second threshold, a red early warning is triggered, and the catalyst replenishment program is automatically started.