Abnormal monitoring and early warning method and system for fly ash melting processing system

Abstract

The present application relates to the field of intelligent monitoring, and provides a method and system for monitoring and warning of abnormalities in a fly ash melting treatment system. By adopting advanced sensor technology and a central control system, the temperature and pressure in the fly ash melting process can be monitored in real time. When abnormal temperature and pressure are detected, the temperature reduction and emergency pressure relief procedures are executed for processing. In this way, the abnormal conditions of the fly ash melting treatment system can be accurately captured, and measures can be taken in time to deal with them, thereby improving the safety and reliability of the fly ash melting treatment system.

Description

Technical Field

[0001] The present application relates to the field of intelligent monitoring, and more specifically, to an abnormal monitoring and early warning method and system for a fly ash melting processing system. Background Art

[0002] With the rapid development of urbanization, domestic and industrial waste has increased dramatically, and incineration has become one of the main methods for dealing with these wastes. However, this process will produce fly ash containing heavy metals, which will cause serious pollution to the environment if not handled properly. Fly ash melting technology converts it into a stable glassy substance through high temperature, effectively reducing the risk of environmental pollution. This technology requires precise temperature and pressure control to ensure process safety and efficiency. Therefore, accurate monitoring and control of temperature and pressure during fly ash melting is the key to ensuring environmental safety and operational efficiency.

[0003] However, traditional monitoring systems rely on manual inspections or regular manual checks, which not only have a long response time, but may also completely miss abnormal conditions during non-working hours or when personnel are negligent, leading to potential escalation of risks. In addition, intermittent sampling means that there are obvious time intervals between data acquisition, and any abnormalities that occur during these intervals cannot be captured in time, which may lead to the loss of critical information. This data discontinuity greatly limits the accurate analysis of equipment status change trends, makes preventive maintenance difficult, and increases the risk of sudden failures.

[0004] Therefore, an optimized abnormal monitoring and early warning scheme for the fly ash melting processing system is desired. Summary of the invention

[0005] In view of the shortcomings of the prior art, the present application provides a method and system for abnormal monitoring and early warning of a fly ash melting processing system.

[0006] According to one aspect of the present application, a method for abnormal monitoring and early warning of a fly ash melting processing system is provided, which includes: collecting temperature data and pressure data at a predetermined sampling frequency by temperature sensors and pressure sensors installed at key positions, and sending the temperature data and the pressure data to a central control system; in the central control system, comparing the temperature data with a temperature safety threshold to obtain a temperature comparison result; in the central control system, when the temperature comparison result exceeds a set range, triggering a temperature abnormality early warning; in the central control system, after triggering the temperature abnormality early warning, arranging the received pressure data according to the time dimension to obtain a time series of pressure data; in the central control system, determining whether to trigger a pressure abnormality early warning based on the time series of the pressure data; in the central control system, if only the temperature abnormality warning is triggered, executing a cooling procedure; if both the temperature abnormality early warning and the pressure abnormality early warning are triggered at the same time, executing an emergency pressure relief procedure.

[0007] According to another aspect of the present application, a fly ash melting processing system abnormality monitoring and early warning system is provided, which is used to execute the fly ash melting processing system abnormality monitoring and early warning method as described above, and includes: a pressure data time series feature extraction module, which is used to perform local time domain implicit association on the time series of the pressure data to obtain a sequence of local time domain implicit association feature vectors of the pressure data; a pressure time series feature context association module, which is used to perform spatial attention enhancement and temporal attention enhancement on the sequence of local time domain implicit association feature vectors of the pressure data to obtain a sequence of pressure data time series context enhancement coding feature vectors; a pressure data time series significant propagation aggregation representation module, which is used to perform multi-gated time series feature dynamic propagation aggregation on the sequence of pressure data time series context enhancement coding feature vectors to obtain a pressure data time series significant propagation aggregation representation vector; a pressure abnormality early warning module, which is used to obtain a recognition result based on the pressure data time series significant propagation aggregation representation vector, and the recognition result is used to indicate whether a pressure abnormality warning is triggered.

[0008] The present application has significant technical effects due to the adoption of the above technical solutions: the fly ash melting treatment system abnormal monitoring and early warning method and system provided by the present application can monitor the temperature and pressure in the fly ash melting process in real time by adopting advanced sensor technology and central control system, and when abnormal temperature and pressure are detected, the temperature reduction and emergency pressure relief procedures are executed for processing. In this way, the abnormal situation of the fly ash melting treatment system can be accurately captured, and measures can be taken in time to deal with it, thereby improving the safety and reliability of the fly ash melting treatment system. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0010] Figure 1 The present invention is a flowchart of a method for monitoring and warning abnormalities in a fly ash melting processing system according to an embodiment of the present application.

[0011] Figure 2 This is a flow chart of step S150 in the abnormal monitoring and early warning method for a fly ash melting processing system according to an embodiment of the present application.

[0012] Figure 3 This is a flow chart of step S153 in the abnormal monitoring and early warning method for a fly ash melting processing system according to an embodiment of the present application.

[0013] Figure 4 It is a system block diagram of the abnormal monitoring and early warning system of the fly ash melting processing system according to an embodiment of the present application. DETAILED DESCRIPTION

[0014] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0015] Based on the problems in the above-mentioned background technology, the present application proposes a method for abnormal monitoring and early warning of a fly ash melting processing system, which can perform real-time monitoring and effectively solve the problems of traditional manual work by adopting advanced sensor technology and a central control system, thereby improving the safety, reliability and efficiency of the system. Figure 1 Flow chart of the abnormal monitoring and early warning method of the fly ash melting processing system according to the embodiment of the present application. Figure 1 As shown, the abnormal monitoring and early warning method of the fly ash melting processing system according to the embodiment of the present application includes: S110, collecting temperature data and pressure data at a predetermined sampling frequency through temperature sensors and pressure sensors installed at key positions, and sending the temperature data and the pressure data to a central control system; S120, in the central control system, comparing the temperature data with a temperature safety threshold to obtain a temperature comparison result; S130, in the central control system, when the temperature comparison result exceeds a set range, triggering a temperature abnormality early warning; S140, in the central control system, after triggering the temperature abnormality early warning, arranging the received pressure data according to the time dimension to obtain a time series of pressure data; S150, in the central control system, determining whether to trigger a pressure abnormality early warning based on the time series of the pressure data; S160, in the central control system, if only the temperature abnormality warning is triggered, executing a cooling procedure; if both the temperature abnormality early warning and the pressure abnormality early warning are triggered at the same time, executing an emergency pressure relief procedure.

[0016] In step S110, temperature data and pressure data are collected at a predetermined sampling frequency by temperature sensors and pressure sensors installed at key positions, and the temperature data and pressure data are sent to the central control system. It should be understood that during the operation of the fly ash melting treatment system, temperature and pressure are two key operating parameters, which may change at any time. By installing sensors at key positions and collecting data at a predetermined sampling frequency, the dynamic situation of temperature and pressure can be captured in real time. The collected temperature data includes real-time temperature values ​​at different key positions of the fly ash melting treatment system, such as the temperature of different parts in the furnace, which reflects the heating condition of the fly ash during the melting process. Normal fly ash melting at different stages should be in a specific temperature range. These real-time temperature values ​​can intuitively reflect whether the current melting process is carried out in accordance with the predetermined process requirements. Once the temperature exceeds a reasonable range, it may mean that the heat transfer is abnormal or other problems that affect the stable melting of the fly ash occur. The collected pressure data involves the pressure condition of the container where the fly ash is melted. When the fly ash melting treatment system operates normally, the pressure will be maintained within a certain safety range. Its data can reflect the gas generation, circulation and equipment pressure conditions inside the system. If the pressure rises or falls abnormally, it is likely to indicate blockage, leakage or other process abnormalities. In general, accurate and real-time temperature and pressure data are the most direct basis for judging whether there is an abnormality in the fly ash melting treatment system. In order to centrally manage and analyze temperature data and pressure data, the collected temperature data and pressure data need to be transmitted to the central control system. In the central control system, these data can be efficiently analyzed to improve the efficiency of abnormal monitoring and early warning of the fly ash melting treatment system.

[0017] The key steps to realize intelligent monitoring and early warning of fly ash melting treatment system are to collect temperature and pressure data at a predetermined sampling frequency through temperature sensors and pressure sensors installed at key positions, and send these data to the central control system. The following is a detailed description of its implementation process: First, it is necessary to select appropriate temperature sensors and pressure sensors according to the actual working environment and specific needs. These sensors must have the characteristics of high precision, fast response, and high temperature and high pressure resistance to adapt to extreme working conditions. For example, thermocouples or RTD resistance thermometers can be used as temperature sensors, while pressure sensors can be strain gauge, piezoresistive or capacitive. After selecting the sensor, install it at key positions of the system, such as inside the furnace, at pipe connections, etc., so that the temperature and pressure changes during the entire fly ash melting process can be accurately reflected.

[0018] Next, to ensure that the data collected by the sensor can be smoothly transmitted to the central control system, it is necessary to properly arrange the signal cables or use wireless transmission technology. The wired method usually uses shielded cables to reduce the impact of electromagnetic interference, while the wireless method may use Bluetooth, Wi-Fi or other industrial-grade wireless communication protocols. During the installation process, it is also necessary to configure the sampling frequency, data format and other parameters of each sensor to meet the needs of real-time monitoring. In addition, the power supply of the sensor also needs to be carefully considered to ensure that it can work stably for a long time.

[0019] During the data acquisition phase, the sensor continuously collects temperature and pressure information at a predetermined sampling frequency. Since the acquired data may be affected by noise or have irregular fluctuations, simple filtering is usually performed locally before being transmitted to the central control system to remove outliers or smooth the data curve, thereby improving the effectiveness of subsequent analysis. This step helps to ensure the quality of the data and avoid misjudgments caused by noise interference.

[0020] Subsequently, the temperature and pressure data after preliminary processing are encapsulated into data packets in a specific format and transmitted to the server of the central control system through a wired or wireless network. This process requires the communication link to be highly stable and secure to prevent data loss or tampering. Sometimes encryption measures are taken to protect sensitive information and ensure the security and reliability of data transmission. At the same time, in order to cope with emergencies, the system must also have a redundant backup mechanism to enhance the overall robustness. When the central control system receives the data sent from each sensor, it will store it in the database for further analysis and support historical record query.

[0021] Through the above steps, accurate and reliable temperature and pressure data can be collected, thereby laying a good data foundation for subsequent data analysis steps.

[0022] In step S120, in the central control system, the temperature data is compared with the temperature safety threshold to obtain a temperature comparison result. It should be understood that in the fly ash melting process, temperature is an intuitive and crucial parameter. If the feed rate is too fast or the amount of combustion aid added is too much, the combustion will be too intense, causing a sharp rise in the temperature inside the melting furnace. Once the temperature exceeds the set safety limit, it usually indicates that the system may be at risk. In this case, the temperature abnormality warning must be triggered quickly to prevent possible accidents. Therefore, by real-time monitoring and comparing the current temperature with the safety limit, this immediate response mechanism can quickly identify whether the temperature is beyond the normal range, which can significantly shorten the response time. This is a key step in preliminarily determining whether the system is in an abnormal state.

[0023] In the central control system, the process of comparing temperature data with the temperature safety threshold to obtain the temperature comparison result is an important part of ensuring the safe operation of the fly ash melting treatment system. The following is a detailed description of its implementation process: The setting of the temperature safety threshold is the basis of the entire comparison process, which requires comprehensive consideration of many factors. First, the basic threshold range is determined according to the inherent requirements of the fly ash melting process. The composition of fly ash is complex and diverse, and its melting process requires specific temperature conditions to promote its transformation into a stable glassy substance. Through in-depth analysis of the chemical composition, physical properties and combustion reaction characteristics of fly ash, combined with a large amount of experimental data and actual operation experience, the appropriate temperature range at different process stages is determined. This range must ensure that the fly ash can be fully melted, achieve stable solidification of heavy metals, and reduce the risk of environmental pollution, and avoid a series of adverse consequences caused by excessive temperature. For example, in the melting stage, if the temperature is too low, the fly ash may not be completely melted, resulting in poor treatment effect, and residual harmful substances may seep out, posing a potential threat to the environment; while too high a temperature may accelerate the loss of refractory materials in the furnace, shorten the service life of the equipment, and even cause damage to the furnace structure, leading to safety accidents.

[0024] At the same time, the bearing capacity of the equipment plays a decisive role in the setting of the temperature safety threshold. All kinds of equipment in the fly ash melting treatment system, including furnaces, heat exchangers, connecting pipes, etc., are made of specific materials and have their own thermophysical properties and structural strength limits. The lining material of the furnace is usually made of special ceramics or refractory bricks that are resistant to high temperatures and corrosion, but there is still a risk of performance degradation under long-term high temperatures. If the temperature exceeds the upper limit it can withstand, the lining material may soften, deform or chemically react, causing damage to the furnace structure, thereby affecting the sealing and stability of the entire system. The heat exchange efficiency and service life of the heat exchanger are also closely related to the temperature. Excessive temperature may reduce the performance of the heat exchange tube material, reduce the heat exchange efficiency, and even cause the pipeline to rupture and leak. Therefore, when setting the temperature safety threshold, the technical parameters provided by the equipment manufacturer must be fully considered, and the threshold must be accurately adjusted in combination with the monitoring data during actual operation to ensure that sufficient safety margin is reserved for the equipment to prevent equipment failures due to abnormal temperature and ensure the continuous and stable operation of the system.

[0025] In order to achieve real-time temperature comparison, the central control system is usually equipped with a data buffer. The newly received temperature data is first stored in the buffer and processed in real time according to the predetermined time interval and processing algorithm. The design of the buffer needs to balance storage capacity and processing efficiency, and adopt reasonable data management strategies such as first-in-first-out (FIFO) to ensure that the system always processes the latest temperature data and avoids processing delays due to data backlog. In the data processing process, by optimizing the algorithm and selecting a high-performance processor, the data processing speed is increased, the processing delay is reduced, and the temperature comparison can be carried out in a timely manner.

[0026] The design of temperature comparison logic directly affects the accuracy of anomaly detection. Simple threshold comparison is the most basic method, which directly determines whether the temperature exceeds a preset single threshold. It is suitable for situations where there are clear requirements for the upper or lower temperature limits. Interval judgment is more flexible. Different temperature intervals can be set according to different process stages to determine whether the temperature is within a reasonable range. The trend analysis algorithm starts from a dynamic perspective. By monitoring the changing trends of multiple consecutive temperature data points, it can detect abnormal temperature increases or decreases in advance. Even if the current temperature has not yet exceeded the threshold, if the trend shows that it is about to exceed the safe range, an early warning can be issued in time. In practical applications, a variety of comparison algorithms are often used in combination according to the specific characteristics and needs of the fly ash melting treatment system to improve the sensitivity and reliability of temperature anomaly detection.

[0027] The entire comparison process is carried out in real time and dynamically. When each new temperature data point arrives, it is immediately compared with the temperature safety threshold and the corresponding result is output. The comparison result can be presented as an intuitive digital signal or status mark, such as "normal", "above threshold", "below threshold", "temperature rise trend abnormal", etc., which can provide operators with clear system operation status information. In order to facilitate the subsequent in-depth analysis of the system operation status, the central control system will record the results of each temperature comparison in detail. The record content includes key information such as the comparison timestamp, temperature data value, corresponding temperature safety threshold, and comparison result.

[0028] Through the above steps, temperature abnormalities can be identified in the first place, thereby providing a basis for subsequent temperature warning measures.

[0029] In step S130, in the central control system, when the temperature comparison result exceeds the set range, the temperature abnormality warning is triggered. It should be understood that the normal operation of the fly ash melting treatment system depends on the temperature being maintained within a reasonable range. Once the temperature comparison result shows that it exceeds the set range, it means that the system has deviated from the normal operating state and may face various risks such as equipment damage and process failure. By triggering the temperature abnormality warning, this potential risk can be immediately prompted to the automatic abnormality detection and early warning system, so that timely measures can be taken to adjust the operation of the early warning system to avoid the continuous abnormal temperature causing irreversible damage to key equipment such as the melting furnace, thereby ensuring the safe, stable and continuous operation of the entire fly ash melting treatment system.

[0030] In the central control system, the process of triggering a temperature anomaly warning in response to a temperature comparison result exceeding a set range involves steps such as data analysis and activation of an early warning mechanism. The following is a detailed description of its implementation process: When the temperature data is found to exceed the safety threshold, the central control system will immediately generate a specific temperature anomaly event report. This report not only records the time point when the anomaly occurred, but also includes detailed information such as specific temperature values. This information is crucial for subsequent targeted measures, and helps to quickly locate problems and formulate effective response strategies. At the same time, based on the generated abnormal event report, the central control system will automatically initiate a series of predetermined emergency measures, the most direct and important of which is to trigger a temperature anomaly warning.

[0031] In order to ensure that relevant personnel can receive alarm information in the first time, the central control system usually adopts a variety of notification methods, such as sound and light alarms, SMS notifications, email reminders, etc. These multi-channel alarm mechanisms can significantly improve the visibility and urgency of alarms, thereby speeding up response. In some cases, in order to further enhance the effect of the alarm, the system may adopt a graded warning method, that is, the warning level is determined according to the degree to which the temperature deviates from the safe range. For example, a slight deviation may only be prompted by the flashing indicator light on the control panel, while a serious deviation will trigger a stronger sound and light alarm and send an emergency notification to the designated manager. This graded warning method helps to distinguish different levels of risks, allowing staff to prioritize the most urgent issues according to the situation.

[0032] In addition to issuing an immediate alarm, the central control system also records all abnormal temperature events and their handling process. These records not only help track the root cause of the problem, but also provide a basis for optimizing process parameters. By accumulating and analyzing historical data, the system can identify the root cause of the temperature anomaly and take measures to prevent similar problems from happening again. In addition, the system can also configure automated response measures, such as starting a cooling program in the case of abnormal temperature to restore the temperature to a safe range as soon as possible. This automated response mechanism can reduce the time of human intervention and improve the system's response speed, thereby effectively reducing potential risks.

[0033] Through the above steps, an abnormal temperature warning can be issued in time, which is conducive to taking timely action against the abnormal temperature, thereby ensuring the safe and stable operation of the fly ash melting treatment system.

[0034] In step S140, after the temperature abnormality warning is triggered, the central control system arranges the received pressure data according to the time dimension to obtain the time series of pressure data. It should be understood that in the fly ash melting treatment system, the two parameters of temperature and pressure are not independent of each other, but are closely related and affect each other. When the temperature is abnormal and the warning is triggered, it is very likely to cause the pressure to change accordingly. For example, a sharp increase in temperature may cause the state of the material in the system to change, the gas to expand, and so on, thereby increasing the pressure. Once the temperature abnormality is detected, it is necessary to analyze the pressure situation closely related to it in a timely manner. Arranging the pressure data according to the time dimension to form a time series of pressure data is a basic step for subsequent accurate analysis of the pressure change trend and judgment of whether the pressure is also abnormal. And in the abnormal monitoring and early warning of the entire system, only based on the temperature abnormality warning, it can only be preliminarily known that the system has deviated from the normal operating state in terms of temperature, but it is impossible to determine the overall condition of the entire system. The stable operation of the fly ash melting treatment system depends on the coordination of multiple parameters in a reasonable range. As one of the key parameters, the dynamic change of pressure is crucial to judging the safety and stability of the system. By arranging the pressure data after the temperature anomaly warning is triggered by time, the evolution of pressure can be fully presented from the perspective of time change, which helps the abnormal warning system to comprehensively evaluate whether the pressure is also affected when the temperature anomaly occurs, and then more accurately grasp the overall operating status of the fly ash melting treatment system, providing sufficient basis for taking appropriate response measures in the future.

[0035] In step S150, the central control system determines whether to trigger a pressure abnormality warning based on the time series of the pressure data. It should be understood that abnormal temperature rise is often accompanied by abnormal changes in pressure. Specifically, the temperature rise may cause the gas in the furnace to expand. If the exhaust system fails to adjust in time at this time, pressure accumulation will occur. The traditional pressure warning judgment mainly relies on the method of setting a fixed threshold, that is, when the detected pressure value exceeds the normal operating range (safety upper limit), the alarm is triggered. However, the pressure abnormality is not only affected by the current working conditions, but also by the time series synergy of historical data. The change in pressure may indicate potential problems. The traditional method only considers the relationship between the current pressure reading and the preset threshold, ignoring the historical trend of pressure changes and the potential problems reflected by the long-term accumulated data, which may lead to the failure to timely discover slowly developing faults or abnormal conditions. In addition, pressure abnormality is usually the result of gradual changes. For those risks that do not immediately reach a dangerous level but gradually accumulate, the traditional method may not issue a warning until the last moment, missing the best time for early intervention, increasing the cost of processing and even the possibility of accidents.

[0036] Based on this, in the step of determining whether to trigger a pressure anomaly warning based on the time series of the pressure data in the central control system, the technical concept of the present application is to use data analysis and coding technology based on deep learning to perform local time domain correlation on the pressure data, and then perform time series enhancement in the spatial and time dimensions on the local time domain implicit correlation features of each pressure data, so as to intelligently determine whether to trigger a pressure anomaly warning based on the multi-gated time series dynamic aggregation representation between the time series context enhancement coding features of each pressure data. In this way, complex patterns and trends in pressure data can be learned and potential abnormal behaviors can be identified, which helps to discover potential problems at an early stage. And through continuous learning and optimization, its detection strategy is automatically adjusted to adapt to different working conditions and environmental changes, thereby significantly improving the accuracy and intelligence level of pressure anomaly warning in the fly ash melting processing system.

[0037] Figure 2 FIG. 1 is a flow chart of step S150 in the abnormal monitoring and early warning method for the fly ash melting processing system according to an embodiment of the present application. Figure 2 As shown, the step S150 includes: S151, performing local time domain implicit association on the time series of the pressure data to obtain a sequence of local time domain implicit association feature vectors of the pressure data; S152, performing spatial attention enhancement and temporal attention enhancement on the sequence of local time domain implicit association feature vectors of the pressure data to obtain a sequence of temporal context enhancement coding feature vectors of the pressure data; S153, performing multi-gated temporal feature dynamic propagation aggregation on the sequence of temporal context enhancement coding feature vectors of the pressure data to obtain a pressure data temporal significant propagation aggregation representation vector; S154, obtaining a recognition result based on the pressure data temporal significant propagation aggregation representation vector, and the recognition result is used to indicate whether a pressure abnormality warning is triggered.

[0038] In step S151, the time series of the pressure data is implicitly associated in the local time domain to obtain a sequence of local time domain implicit association feature vectors of the pressure data. Specifically, in an embodiment of the present application, the step S151 includes: inputting the time series of the pressure data into a sequence encoder based on a 1D-CNN model to obtain a sequence of local time domain implicit association feature vectors of the pressure data. It should be understood that the pressure data contains different local time series patterns and structural information, such as periodic changes in pressure, fluctuations in local time, or sudden outliers. Based on this, in order to capture and mine the implicit pressure time series feature information in different local time periods, in the technical solution of the present application, the time series of the pressure data is input into a sequence encoder based on a 1D-CNN model to obtain a sequence of local time domain implicit association feature vectors of the pressure data. It can be understood that the 1D-CNN model is particularly suitable for processing time series or sequence data, and can automatically learn useful features from the original pressure data efficiently. By sliding the convolution kernel on the time series, 1D-CNN can not only capture local patterns and subtle changes, but also identify potential temporal correlations, thereby extracting feature vectors rich in information. This ability makes 1D-CNN an ideal choice for analyzing the state of dynamic systems and detecting abnormal behaviors, providing a solid data foundation for subsequent intelligent early warnings.

[0039] In step S152, the sequence of local temporal implicitly associated feature vectors of the pressure data is subjected to spatial attention enhancement and temporal attention enhancement to obtain a sequence of temporal context enhanced coding feature vectors of the pressure data. Specifically, in an embodiment of the present application, step S152 includes: inputting the sequence of local temporal implicitly associated feature vectors of the pressure data into a spatial attention module based on a converter structure to obtain a sequence of local temporal attention enhanced feature vectors of the pressure data; inputting the sequence of local temporal attention enhanced feature vectors of the pressure data into a temporal attention module based on a converter structure to obtain a sequence of temporal context enhanced coding feature vectors of the pressure data.

[0040] It should be understood that each local time-domain implicit correlation feature vector of pressure data represents the local characteristics of pressure data within a certain time period in the original time series, such as a local peak, short-term fluctuation or a stable segment, etc. And different local time-domain implicit correlation features have different importance and influence in the entire sequence. Therefore, in the technical solution of the present application, each local time-domain implicit correlation feature vector of pressure data in the sequence of local time-domain implicit correlation feature vectors of pressure data is input into a spatial attention module based on a converter structure to obtain a sequence of local time-series attention enhancement feature vectors of pressure data. In particular, here, "space" refers to the relationship within the feature vector, that is, the degree of correlation of pressure data between each key node, and adjusts its importance accordingly.

[0041] Specifically, in the spatial attention module based on the converter structure, the local temporal implicit correlation feature vector of the pressure data is first converted into three different representations through a specific weight matrix: a query vector (a representation of the current focus or demand), a key vector (the type of information or characteristics that each feature can provide) and a value vector (the specific content actually to be passed to the output). Next, the self-attention mechanism in the converter structure is used to process these three vectors to obtain the re-evaluated importance according to the correlation between each feature vector and other feature vectors, and adjust its contribution in the final output accordingly, so as to obtain a sequence of local temporal attention-enhanced feature vectors of the pressure data. Through this process, the spatial attention module based on the converter structure can automatically identify and emphasize those pressure data feature information that are more critical to the current task, while reducing the influence of less important features, so as to improve the model's ability to capture complex patterns and enhance sensitivity to abnormal situations, so that the system can more accurately monitor and warn abnormalities.

[0042] Then, in order to capture and strengthen the contextual relationship of the sequence of local temporal attention-enhanced feature vectors of the pressure data in the time dimension, so as to obtain richer pressure features with temporal context information, in the technical solution of the present application, the sequence of local temporal attention-enhanced feature vectors of the pressure data is input into the temporal attention module based on the converter structure to obtain a sequence of temporal context-enhanced encoded feature vectors of the pressure data. In other words, through the temporal attention module based on the converter structure, the model can use the self-attention mechanism to automatically adjust the attention according to the correlation between each feature, so as to better understand and model the complex dependencies between different time points in the pressure data, which helps to identify trends, patterns or anomalies that appear over time. In this way, the feature vector processed by temporal attention not only retains the original spatial correlation information, but also integrates the relevant information in the time dimension, forming a more comprehensive and richer feature representation.

[0043] In step S153, the sequence of the pressure data temporal context enhanced encoding feature vectors is subjected to multi-gated temporal feature dynamic propagation aggregation to obtain a pressure data temporal significant propagation aggregation representation vector. Specifically, Figure 3 FIG. 1 is a flow chart of step S153 in the abnormal monitoring and early warning method for the fly ash melting processing system according to an embodiment of the present application. Figure 3As shown, the step S153 includes: S1531, mapping the sequence of the pressure data temporal context enhanced coding feature vectors to the Poincare space to obtain the sequence of the pressure data temporal context enhanced feature heterogeneous modulation coding feature vectors; S1532, forward LSTM sequence encoding the sequence of the pressure data temporal context enhanced feature heterogeneous modulation coding feature vectors to obtain the pressure data temporal context enhanced feature heterogeneous propagation representation vector; S1533, based on the pressure data temporal context enhanced feature heterogeneous propagation representation vector, calculating the sequence of the pressure data temporal context enhanced feature heterogeneous modulation coding feature vector The semantic association and semantic space jump value of each pressure data temporal context enhanced feature heterogeneous modulation coding feature vector in the column are used to obtain a sequence of pressure data temporal context semantic associations and a sequence of pressure data temporal context semantic space jump values; S1534, based on the sequence of pressure data temporal context semantic associations and the sequence of pressure data temporal context semantic space jump values, the pressure data temporal context enhanced feature heterogeneous propagation representation vector and the sequence of pressure data temporal context enhanced feature heterogeneous modulation coding feature vector are compensated and dynamically aggregated to obtain the pressure data temporal significant propagation aggregated representation vector.

[0044] It should be understood that the sequence of the pressure data temporal context enhancement encoding feature vectors expresses the temporal correlation characteristics between the contexts of the pressure data at different time points. Considering that the pressure data at different time points not only has significant temporal correlation characteristics, but also there are different degrees of correlation between the pressure data at each time point and the overall pressure characteristics of the entire sequence. Therefore, in order to capture and utilize the complex patterns and dynamic changes of pressure data in the time dimension, the pressure data at each time point is linked to the data before and after it to form a richer feature representation, so as to help the model better understand the change pattern of pressure data over time. In the technical solution of the present application, the sequence of the pressure data temporal context enhancement encoding feature vectors is subjected to multi-gated temporal feature dynamic propagation aggregation to obtain a pressure data temporal significant propagation aggregation representation vector.

[0045] In detail, the sequence of the temporal context enhancement coding feature vectors of the pressure data is first mapped to the Poincare space to obtain a sequence of temporal context enhancement feature heterogeneous modulation coding feature vectors of the pressure data. A person skilled in the art should know that the Poincare space is a hyperbolic space that can effectively reduce the curse of dimensionality while maintaining the distance relationship between nodes (i.e., the semantic association between feature vectors at different temporal nodes). Through this step, the unique geometric properties of the Poincare space can be used to capture the nonlinear relationship and hierarchical structure between the temporal context enhancement coding features of the pressure data. The above process can be expressed as: ;in, is a sequence of temporal context-enhanced encoding feature vectors of the pressure data, , , and are the first, second, and third in the sequence of the temporal context enhanced encoding feature vector of the pressure data respectively. and The temporal context-enhanced encoding feature vector of pressure data, and are the first weight matrix and the second weight matrix respectively, , , and are the first, second, and third in the sequence of the temporal context enhanced feature of the pressure data heterogeneous modulation coding feature vector. and The temporal contextual enhancement feature of the pressure data is modulated and coded by a feature vector. It is a sequence of the temporal context-enhanced feature heterogeneous modulation coding feature vectors of the pressure data.

[0046] Next, the sequence of the temporal contextual reinforcement feature heterogeneous modulation coding feature vectors of the pressure data is forward LSTM sequence encoded to obtain the temporal contextual reinforcement feature heterogeneous propagation representation vector of the pressure data. Here, the long-term dependency in the sequence of the input temporal contextual reinforcement feature heterogeneous modulation coding feature vectors of the pressure data can be captured through the forward long short-term memory (LSTM) network, and the selective memory and update of the pressure feature information can be achieved while avoiding overfitting through the synergy of the forget gate, input gate and output gate. The above process can be expressed as: ;in, is a sequence of temporal contextual reinforcement feature heterogeneous modulation coding feature vectors of the pressure data, Encode the forward LSTM. It is the representation vector of the cross-domain propagation of the temporal context-enhanced features of pressure data.

[0047] Specifically, in an embodiment of the present application, the step S1533 includes: calculating the semantic association between each pressure data timing context enhancement feature foreign domain modulation coding feature vector in the sequence of the pressure data timing context enhancement feature foreign domain modulation coding feature vector and the pressure data timing context enhancement feature foreign domain propagation representation vector to obtain a sequence of the pressure data timing context semantic associations; calculating the semantic space jump value of each pressure data timing context enhancement feature foreign domain modulation coding feature vector in the sequence of the pressure data timing context enhancement feature foreign domain modulation coding feature vector relative to the pressure data timing context enhancement feature foreign domain propagation representation vector to obtain a sequence of the pressure data timing context semantic space jump values. The processing process of the above step S1533 can be expressed as: ;in, is the first in the sequence of the temporal contextual reinforcement feature of the pressure data, the heterogeneous modulation coding feature vector The temporal contextual enhancement feature of the pressure data is modulated and coded by a feature vector. is the representation vector of the temporal context-enhanced feature propagation in different domains of pressure data, To calculate the square of the length of a vector, is the inverse hyperbolic cosine function, , , and are the first, second, and third in the sequence of the semantic association of the temporal context of pressure data. and The semantic correlation of the temporal context of pressure data, is a sequence of temporal context semantic associations of the pressure data, , , and are the first, second, and third jump values ​​in the sequence of the semantic space of the temporal context of the pressure data. and The semantic space jump value of the pressure data temporal context, is a sequence of jump values ​​in the semantic space of the temporal context of the pressure data.

[0048] That is, by calculating the semantic correlation between each pressure data time series context enhanced feature heterodomain modulation coding feature vector in the sequence of the pressure data time series context enhanced feature heterodomain modulation coding feature vector and the pressure data time series context enhanced feature heterodomain propagation representation vector, the semantic intrinsic correlation between the pressure data time series propagation feature and each pressure data time series context enhanced feature heterodomain modulation coding feature vector in the Poincare space is expressed, so as to facilitate the subsequent introduction of the attention mechanism to enhance the model's attention to specific pressure time series features. Next, by calculating the semantic space jump value of each pressure data time series context enhanced feature heterodomain modulation coding feature vector in the sequence of the pressure data time series context enhanced feature heterodomain modulation coding feature vector relative to the pressure data time series context enhanced feature heterodomain propagation representation vector, the changing trend and rate of the feature expression of the pressure data time series feature over time are measured. In this way, not only the dynamic characteristics of the pressure time series characteristics can be revealed, but also periodic patterns and abnormal events in the data can be discovered.

[0049] Specifically, in an embodiment of the present application, the step S1534 includes: performing multiple gated feature modulation on the sequence of the pressure data temporal context semantic associations and the sequence of the pressure data temporal context semantic space jump values ​​to obtain a sequence of pressure data temporal context feature transfer modulation weights; using the sequence of the pressure data temporal context feature transfer modulation weights as the sequence of weights, calculating the position-weighted sum of the sequence of the pressure data temporal context enhanced feature heterogeneous modulation coding feature vectors to obtain the pressure data temporal context heterogeneous feature significant propagation representation vector; performing residual connection on the pressure data temporal context heterogeneous feature significant propagation representation vector and the pressure data temporal context enhanced feature heterogeneous propagation representation vector to obtain the pressure data temporal context heterogeneous feature significant propagation compensation representation vector; performing Euclidean space mapping on the pressure data temporal context heterogeneous feature significant propagation compensation representation vector to obtain the pressure data temporal context heterogeneous feature significant propagation aggregation representation vector. The processing process of the above step S1534 can be expressed as: ;in, is the first in the sequence of the temporal contextual reinforcement feature of the pressure data, the heterogeneous modulation coding feature vector The temporal contextual enhancement feature of the pressure data is modulated and coded by a feature vector. is the representation vector of the temporal context-enhanced feature propagation in different domains of pressure data, is a sequence of temporal context semantic associations of the pressure data, is a sequence of hop values ​​in the temporal context semantic space of the pressure data, and are the semantic association weight matrix and the semantic space weight matrix respectively. is matrix multiplication, is the bias vector, yes function, is the sequence of modulation weights transferred by the temporal context features of pressure data, yes The corresponding pressure data temporal context feature transfer modulation weight, is the number of vectors in the sequence of the pressure data temporal context enhanced feature heterogeneous modulation coding feature vectors, is the compensation representation vector of the temporal context heterogeneous features of the pressure data, and are the third weight matrix and the fourth weight matrix respectively, is the Euclidean space mapping operation, is a vector representing the significant propagation aggregation of the pressure data time series.

[0050] That is, by performing multiple gated feature modulation on the sequence of the semantic association of the pressure data temporal context and the sequence of the semantic space jump value of the pressure data temporal context to dynamically adjust the direction and intensity of the information flow, the model can adaptively and selectively strengthen or weaken the transmission of specific information according to the relationship between the pressure temporal features, thereby improving the accuracy and robustness of the pressure data feature representation. Subsequently, the sequence of the modulation weights of the pressure data temporal context feature transmission is used as the sequence of weights, and the position-weighted sum of the sequence of the heterogeneous modulation coding feature vectors of the pressure data temporal context reinforcement feature is calculated to obtain the pressure data temporal context heterogeneous feature significant propagation representation vector. Here, the importance and temporal changes of the pressure data temporal features are comprehensively considered by weighted summation, and a pressure data temporal context heterogeneous feature significant propagation representation vector that comprehensively reflects the temporal evolution characteristics of the pressure features is obtained. Specifically, the calculation method of the weighted sum by position not only considers the importance of the pressure data temporal features, but also combines the changing trend in the time dimension, so that the final feature vector can fully reflect the spatiotemporal characteristics of the original pressure data. Next, the pressure data temporal context heterogeneous feature significant propagation representation vector and the pressure data temporal context enhanced feature heterogeneous propagation representation vector are residually connected. By introducing residual connections, not only can the gradient vanishing problem that may occur during deep network training be solved, but also the direct transmission of pressure features can be promoted. Finally, the pressure data temporal context heterogeneous feature significant propagation compensation representation vector is Euclidean space mapped to obtain the pressure data temporal significant propagation aggregate representation vector. In this way, the complex processed pressure features can be converted back to the traditional Euclidean space, realizing effective projection from high-dimensional feature space to low-dimensional space, which is convenient for subsequent analysis.

[0051] In step S154, based on the pressure data temporal significant propagation aggregation representation vector, a recognition result is obtained, and the recognition result is used to indicate whether a pressure abnormality warning is triggered. Specifically, in an embodiment of the present application, the step S154 includes: inputting the pressure data temporal significant propagation aggregation representation vector into a pressure abnormality identifier based on a classifier to obtain the recognition result. That is, the pressure data temporal significant propagation aggregation representation vector obtained by multi-gated temporal aggregation using the sequence of the pressure data temporal context enhanced encoding feature vector is classified and processed to intelligently determine whether a pressure abnormality warning is triggered. A classifier is a machine learning model that can analyze and judge new input data according to the learned pattern after being trained with a large amount of labeled data, and then map it to different output categories. By inputting the pressure data temporal significant propagation aggregation representation vector into a pressure abnormality identifier based on a classifier, an automated and intelligent pressure abnormality warning can be achieved. In particular, in an embodiment of the present application, a possible implementation method for inputting the pressure data time series significant propagation aggregation representation vector into a pressure anomaly identifier based on a classifier to obtain a recognition result indicating whether a pressure anomaly warning is triggered may be: using the fully connected layer of the classifier to fully connect encode the pressure data time series significant propagation aggregation representation vector to obtain a fully connected encoded feature vector of the pressure data time series significant propagation aggregation representation; inputting the pressure data time series significant propagation aggregation representation fully connected encoded feature vector into the Softmax classification function of the classifier to obtain the probability value of the pressure data time series significant propagation aggregation representation vector belonging to each classification label, and the classification labels include those for indicating the triggering of a pressure anomaly warning and those for indicating the non-triggering of a pressure anomaly warning; and determining the classification label corresponding to the largest of the probability values ​​as the recognition result.

[0052] In particular, in one example, considering that each pressure data temporal context enhancement encoding feature vector in the sequence of the pressure data temporal context enhancement encoding feature vectors respectively represents the context-related temporal features of the pressure data in the local time domain, when performing feature dynamic propagation based on a cross-domain multiple gate structure, due to the dynamic differences in the cross-domain gate structure caused by the differences in the distribution of temporal features between local time domains, it is expected to improve the semantically consistent aggregation expression effect of the pressure data temporal significant propagation aggregation representation vector obtained by feature dynamic propagation.

[0053] Based on this, when the pressure data temporal significant propagation aggregation representation vector is input into the classifier-based pressure anomaly identifier to obtain the recognition result, the present application optimizes the pressure data temporal significant propagation aggregation representation vector, including: performing feature clustering on the feature set of the pressure data temporal significant propagation aggregation representation vector to obtain the pressure data temporal significant propagation aggregation class feature set. And the pressure data time series significant propagation aggregation out-of-class feature set .

[0054] Calculate the ratio of the number of eigenvalues ​​in the feature set of the pressure data time series significant propagation aggregation class to the number of eigenvalues ​​in the feature set of the pressure data time series significant propagation aggregation representation vector .

[0055] Calculate the sum of the absolute values ​​of all feature values ​​in the feature set of the pressure data time series significant propagation aggregation class The power of the pressure data time series significantly spreads the sum of the absolute values ​​of all eigenvalues ​​in the feature set of the aggregate representation vector The ratio of the powers is used to obtain the significant propagation aggregation weight of the first pressure data time series.

[0056] Calculate the sum of squares of all feature values ​​in the feature set of the pressure data time series significant propagation aggregation class The power of the pressure data time series significantly spreads the sum of the squares of all eigenvalues ​​in the feature set of the aggregate representation vector The ratio of the powers is used to obtain the significant propagation aggregation weight of the second pressure data time series.

[0057] That is, the first pressure data time series significantly propagates the aggregation weight and the second pressure data time series significantly spreads the aggregation weights It is expressed as: ;in, The pressure data time series significantly propagates each feature value in the feature set within the aggregation class, The respective eigenvalues ​​in the feature set of the feature aggregation representation vector are significantly propagated for the pressure data time series, is the number of eigenvalues ​​in the feature set within the significant propagation aggregation class of the pressure data time series, The number of eigenvalues ​​of the feature set of the significant propagation aggregation representation vector for the pressure data time series, express and The ratio between represents the significant propagation aggregation weight of the first pressure data time series, The second pressure data time series significantly propagates the aggregation weight.

[0058] Then, for each feature value in the feature set within the pressure data time series significant propagation aggregation class, the product of the feature value and the second pressure data time series significant propagation aggregation weight is calculated, and then the product of the feature value and the first pressure data time series significant propagation aggregation weight is added to obtain the optimized feature value; this process can be expressed as: in, The pressure data time series significantly propagates each feature value in the feature set within the aggregation class, represents the significant propagation aggregation weight of the first pressure data time series, represents the significant propagation aggregation weight of the second pressure data time series, The optimized feature values ​​of each feature set in the feature set within the pressure data time series are significantly propagated and aggregated.

[0059] For each feature value in the pressure data time series significant propagation aggregation out-of-class feature set, the product of the feature value and the first pressure data time series significant propagation aggregation weight is calculated to obtain the optimized feature value; this process can be expressed as: ;in, The pressure data time series significantly propagates each feature value in the aggregated out-of-class feature set, represents the significant propagation aggregation weight of the first pressure data time series, It represents the optimized eigenvalues ​​of each eigenvalue in the feature set outside the significant propagation aggregation class of the pressure data time series.

[0060] Finally, an optimized pressure data time series significant propagation aggregation representation vector is formed based on the optimized feature values ​​of the pressure data time series significant propagation aggregation class internal feature set and the pressure data time series significant propagation aggregation class external feature set.

[0061] That is, while the pressure data time series significant propagation aggregation representation vector is subjected to feature clustering, the interaction description of the key feature information of the pressure data time series significant propagation aggregation representation vector in the clustering process is carried out, and the geometric equivariant topology of the feature is constructed by low-rank harmonic modulation of the features on the equivariance basis of the clustering features of the pressure data time series significant propagation aggregation representation vector and the features as a whole, so as to obtain the graphical distribution translation and rotation symmetry of the clustering features of the pressure data time series significant propagation aggregation representation vector relative to the features as a whole, thereby introducing geometric message passing into the feature expression of the pressure data time series significant propagation aggregation representation vector and realizing the cluster mapping symmetry of the pressure data time series significant propagation aggregation representation vector through manipulation of irreducible low-rank order coefficients, thereby improving the consistency of the clustering-based feature representation of the pressure data time series significant propagation aggregation representation vector, thereby improving the accuracy of the recognition result obtained by inputting the pressure data time series significant propagation aggregation representation vector into the classifier-based pressure anomaly identifier.

[0062] In summary, step S150 is explained clearly, which uses data analysis and coding technology based on deep learning to perform local time domain correlation on pressure data, and then performs time series enhancement of spatial and time dimensions on the local time domain implicit correlation features of each pressure data, so as to intelligently determine whether to trigger a pressure anomaly warning based on the multi-gated time series dynamic aggregation representation between the time series context enhancement coding features of each pressure data. In this way, complex patterns and trends in pressure data can be learned and potential abnormal behaviors can be identified, which helps to discover potential problems at an early stage. And through continuous learning and optimization, its detection strategy is automatically adjusted to adapt to different working conditions and environmental changes, thereby significantly improving the accuracy and intelligence level of pressure anomaly warning in the fly ash melting processing system.

[0063] In step S160, in the central control system, if only the temperature abnormality warning is triggered, the cooling program is executed; if the temperature abnormality warning and the pressure abnormality warning are triggered at the same time, the emergency pressure relief program is executed. It should be understood that when only the temperature abnormality warning is triggered, it means that the abnormal conditions of the current system are mainly concentrated in the temperature. In order to intervene in the temperature abnormality in time, the cooling program needs to be executed. By executing the cooling program, the temperature can be adjusted back to a reasonable range to ensure that the process of fly ash melting is carried out normally, and the occurrence of potential chain reactions is prevented, and the overall operation state of the fly ash melting treatment system is maintained relatively stable, which can avoid the occurrence of more complex and more serious abnormal conditions in the future, and is conducive to reducing system failure shutdowns and reducing adverse effects on the environment. When the temperature abnormality warning and the pressure abnormality warning are triggered at the same time, it means that the situation faced by the fly ash melting treatment system is more complex and severe. In the case where the temperature abnormality causes the pressure abnormality, the internal pressure of the fly ash melting treatment system may exceed the safety limit that the equipment can bear, which will bring great safety threats to the system, such as furnace explosions, pipeline ruptures and other catastrophic consequences. By executing the emergency pressure relief procedure, the excessive pressure in the fly ash melting treatment system can be quickly released to a relatively safe level, eliminating the direct safety hazards caused by excessive pressure, ensuring that the system will not collapse due to excessive pressure in a short period of time, and ensuring the safety of on-site personnel and equipment. Here, the specific response measures under different abnormal combinations are clarified, so that when facing temperature abnormalities, temperature and pressure abnormalities at the same time, etc., the abnormal warning system can respond quickly and accurately without going through complex manual judgment and decision-making processes, saving precious time, improving the efficiency of the entire system in responding to abnormal situations, and reducing the duration of abnormal states, thereby reducing the degree of damage caused by abnormalities to the system. In addition, the mechanism of flexibly executing corresponding procedures according to different warning triggering situations makes the system more adaptable, able to cope with a variety of actual working conditions and abnormal scenarios, and can maintain a good operating state as much as possible under various conditions, which is conducive to ensuring the effect of fly ash melting treatment and the safety of the system.

[0064] In summary, the abnormal monitoring and early warning method of the fly ash melting treatment system based on the embodiment of the present application is explained. By adopting advanced sensor technology and central control system, the temperature and pressure in the fly ash melting process can be monitored in real time. When abnormal temperature and pressure are detected, the cooling and emergency pressure relief procedures are executed for processing. In this way, the abnormal situation of the fly ash melting treatment system can be accurately captured, and measures can be taken in time to deal with it, thereby improving the safety and reliability of the fly ash melting treatment system.

[0065] In an embodiment of the present application, there is also provided a fly ash melting processing system abnormality monitoring and early warning system for executing the fly ash melting processing system abnormality monitoring and early warning method as described above. Figure 4 FIG. 1 is a system block diagram of an abnormal monitoring and early warning system for a fly ash melting processing system according to an embodiment of the present application. Figure 4 As shown, according to the abnormal monitoring and early warning system of the fly ash melting processing system of the embodiment of the present application, it includes: a pressure data time series feature extraction module, which is used to perform local time domain implicit association on the time series of the pressure data to obtain a sequence of local time domain implicit association feature vectors of the pressure data; a pressure time series feature context association module, which is used to perform spatial attention enhancement and temporal attention enhancement on the sequence of local time domain implicit association feature vectors of the pressure data to obtain a sequence of pressure data time series context enhancement coding feature vectors; a pressure data time series significant propagation aggregation representation module, which is used to perform multi-gated time series feature dynamic propagation aggregation on the sequence of pressure data time series context enhancement coding feature vectors to obtain a pressure data time series significant propagation aggregation representation vector; a pressure abnormality early warning module, which is used to obtain a recognition result based on the pressure data time series significant propagation aggregation representation vector, and the recognition result is used to indicate whether a pressure abnormality warning is triggered.

[0066] Here, those skilled in the art will appreciate that the specific functions and operations of the various units and modules in the above-mentioned abnormal monitoring and early warning system for the fly ash melting treatment system have been described in detail above. Figures 1 to 3 The description of the abnormal monitoring and early warning method of the fly ash melting processing system has been introduced in detail, and therefore, its repeated description will be omitted.

[0067] In summary, the abnormal monitoring and early warning system of the fly ash melting treatment system based on the embodiment of the present application is explained. It uses advanced sensor technology and central control system to monitor the temperature and pressure in the fly ash melting process in real time. When abnormal temperature and pressure are detected, the temperature reduction and emergency pressure relief procedures are executed for processing. In this way, the abnormal situation of the fly ash melting treatment system can be accurately captured, and measures can be taken in time to deal with it, thereby improving the safety and reliability of the fly ash melting treatment system.

Claims

1. A method for abnormal monitoring and early warning of a fly ash melting processing system, characterized in that: include: The temperature data and the pressure data are collected at a predetermined sampling frequency by temperature sensors and pressure sensors installed at key positions, and the temperature data and the pressure data are sent to a central control system; In the central control system, the temperature data is compared with a temperature safety threshold to obtain a temperature comparison result; In the central control system, when the temperature comparison result exceeds a set range, a temperature abnormality warning is triggered; In the central control system, after the temperature abnormality warning is triggered, the received pressure data is arranged according to the time dimension to obtain a time series of the pressure data; In the central control system, determining whether to trigger a pressure abnormality warning based on the time series of the pressure data; In the central control system, if only the abnormal temperature warning is triggered, a cooling procedure is executed; if both the abnormal temperature warning and the abnormal pressure warning are triggered at the same time, an emergency pressure relief procedure is executed; Wherein, in the central control system, determining whether to trigger a pressure abnormality warning based on the time series of the pressure data includes: Performing local time-domain implicit correlation on the time series of the pressure data to obtain a sequence of local time-domain implicit correlation feature vectors of the pressure data; Performing spatial attention enhancement and temporal attention enhancement on the sequence of local time-domain implicitly associated feature vectors of the pressure data to obtain a sequence of temporal context enhanced encoding feature vectors of the pressure data; Performing multi-gated temporal feature dynamic propagation aggregation on the sequence of the pressure data temporal context enhanced encoding feature vectors to obtain a pressure data temporal significant propagation aggregation representation vector; Based on the pressure data time series significant propagation aggregation representation vector, a recognition result is obtained, and the recognition result is used to indicate whether a pressure abnormality warning is triggered.

2. The method for abnormal monitoring and early warning of a fly ash melting processing system according to claim 1, characterized in that: Performing local time-domain implicit association on the time series of the pressure data to obtain a sequence of local time-domain implicit association feature vectors of the pressure data, including: inputting the time series of the pressure data into a sequence encoder based on a 1D-CNN model to obtain a sequence of local time-domain implicit association feature vectors of the pressure data.

3. The method for abnormal monitoring and early warning of a fly ash melting treatment system according to claim 2, characterized in that: Performing spatial attention enhancement and temporal attention enhancement on the sequence of the local temporal implicit correlation feature vectors of the pressure data to obtain a sequence of temporal context enhanced encoding feature vectors of the pressure data, including: Inputting the sequence of local temporal implicit correlation feature vectors of the pressure data into a spatial attention module based on a converter structure to obtain a sequence of local temporal attention enhanced feature vectors of the pressure data; The sequence of the pressure data local temporal attention enhanced feature vectors is input into a temporal attention module based on a converter structure to obtain a sequence of the pressure data temporal context enhanced encoding feature vectors.

4. The abnormal monitoring and early warning method for the fly ash melting processing system according to claim 3 is characterized in that: Performing multi-gated temporal feature dynamic propagation aggregation on the sequence of the pressure data temporal context enhanced encoding feature vectors to obtain a pressure data temporal significant propagation aggregation representation vector, including: Mapping the sequence of the pressure data temporal context enhanced coding feature vectors to the Poincare space to obtain a sequence of the pressure data temporal context enhanced feature heterogeneous modulation coding feature vectors; Performing forward LSTM sequence encoding on the sequence of the pressure data temporal context enhancement feature heterogeneous modulation coding feature vectors to obtain a pressure data temporal context enhancement feature heterogeneous propagation representation vector; Based on the pressure data temporal context enhanced feature heterogeneous propagation representation vector, the semantic association and semantic space jump value of each pressure data temporal context enhanced feature heterogeneous modulation coding feature vector in the sequence of the pressure data temporal context enhanced feature heterogeneous modulation coding feature vector are calculated to obtain a sequence of pressure data temporal context semantic associations and a sequence of pressure data temporal context semantic space jump values; Based on the sequence of semantic associations of the pressure data temporal context and the sequence of semantic space jump values ​​of the pressure data temporal context, the sequence of heterogeneous propagation representation vectors of the pressure data temporal context enhancement features and the sequence of heterogeneous modulation coding feature vectors of the pressure data temporal context enhancement features are compensated dynamically aggregated to obtain the pressure data temporal significant propagation aggregated representation vector.

5. The abnormal monitoring and early warning method for the fly ash melting processing system according to claim 4 is characterized in that: Based on the pressure data temporal context enhanced feature foreign domain propagation representation vector, the semantic association and semantic space jump value of each pressure data temporal context enhanced feature foreign domain modulation coding feature vector in the sequence of the pressure data temporal context enhanced feature foreign domain modulation coding feature vector are calculated to obtain a sequence of pressure data temporal context semantic associations and a sequence of pressure data temporal context semantic space jump values, including: Calculating the semantic association between each pressure data temporal context enhanced feature foreign-domain modulation coding feature vector in the sequence of the pressure data temporal context enhanced feature foreign-domain modulation coding feature vector and the pressure data temporal context enhanced feature foreign-domain propagation representation vector to obtain a sequence of the pressure data temporal context semantic associations; Calculate the semantic space jump value of each pressure data timing context enhancement feature alien modulation coding feature vector in the sequence of the pressure data timing context enhancement feature alien modulation coding feature vector relative to the pressure data timing context enhancement feature alien propagation representation vector to obtain a sequence of the pressure data timing context semantic space jump values.

6. The method for abnormal monitoring and early warning of a fly ash melting processing system according to claim 5, characterized in that: Based on the sequence of the pressure data temporal context semantic associations and the sequence of the pressure data temporal context semantic space jump values, the pressure data temporal context enhanced feature heterogeneous propagation representation vector and the sequence of the pressure data temporal context enhanced feature heterogeneous modulation coding feature vector are compensated and dynamically aggregated to obtain the pressure data temporal significant propagation aggregated representation vector, including: Performing multiple gated feature modulation on the sequence of pressure data temporal context semantic associations and the sequence of pressure data temporal context semantic space jump values ​​to obtain a sequence of pressure data temporal context feature transfer modulation weights; Taking the sequence of modulation weights of the pressure data temporal context feature transfer as the sequence of weights, calculating the position-weighted sum of the sequence of the pressure data temporal context enhanced feature heterogeneous modulation coding feature vectors to obtain the pressure data temporal context heterogeneous feature significant propagation representation vector; Residual connection is performed on the pressure data temporal context heterogeneous feature significant propagation representation vector and the pressure data temporal context enhanced feature heterogeneous propagation representation vector to obtain the pressure data temporal context heterogeneous feature significant propagation compensation representation vector; The pressure data temporal context foreign feature significant propagation compensation representation vector is mapped into Euclidean space to obtain the pressure data temporal significant propagation aggregation representation vector.

7. The abnormal monitoring and early warning method for the fly ash melting processing system according to claim 6 is characterized in that: Based on the pressure data time series significant propagation aggregate representation vector, obtaining a recognition result includes: inputting the pressure data time series significant propagation aggregate representation vector into a classifier-based pressure anomaly identifier to obtain the recognition result.

8. A fly ash melting treatment system abnormality monitoring and early warning system, used to execute the fly ash melting treatment system abnormality monitoring and early warning method according to claim 1, characterized in that: The abnormal monitoring and early warning system of the fly ash melting treatment system comprises: A pressure data time series feature extraction module, used for performing local time domain implicit correlation on the time series of the pressure data to obtain a sequence of local time domain implicit correlation feature vectors of the pressure data; A pressure time series feature context association module, used for performing spatial attention enhancement and temporal attention enhancement on the sequence of local time domain implicit association feature vectors of the pressure data to obtain a sequence of pressure data time series context enhancement encoding feature vectors; A pressure data temporal significant propagation aggregation representation module is used to perform multi-gated temporal feature dynamic propagation aggregation on the sequence of the pressure data temporal context enhanced encoding feature vectors to obtain a pressure data temporal significant propagation aggregation representation vector; The pressure abnormality warning module is used to obtain a recognition result based on the pressure data time series significant propagation aggregation representation vector, and the recognition result is used to indicate whether the pressure abnormality warning is triggered.

9. The abnormal monitoring and early warning system for the fly ash melting processing system according to claim 8, characterized in that: The pressure data time series feature extraction module is used to: input the time series of the pressure data into a sequence encoder based on the 1D-CNN model to obtain a sequence of local time domain implicit correlation feature vectors of the pressure data.

Images