Intelligent remote supervision and early warning method and system based on Internet of Things

Abstract

The present invention discloses a method and system for intelligent remote supervision and early warning based on the Internet of Things, and belongs to the field of intelligent Internet of Things technology. The present invention obtains the state information of the target special equipment system and its associated systems in different fault warning task processes, and performs correlation analysis and state attribute identification, thereby refining state attribute knowledge that is crucial to fault early warning. Furthermore, by combining this knowledge with the fault early warning algorithm, the present invention can accurately reveal the potential fault hidden danger state of the target system in a specific operating action link, and provide strong support for the preventive maintenance and safety management of the equipment. In this way, the problem of insufficient accuracy and timeliness of traditional fault early warning methods can be solved, providing new guarantees for the safe operation of special equipment systems.

Description

Technical Field

[0001] The present invention belongs to the technical field of smart Internet of Things, and specifically relates to a smart remote supervision and early warning method and system based on the Internet of Things. Background Art

[0002] In the operation and maintenance process of special equipment systems, timely detection and treatment of potential fault hazards is the key to ensuring safe and stable operation of equipment. However, traditional fault warning methods often rely only on the status information of a single system, ignoring the correlation between similar systems in different task processes, resulting in limited accuracy and timeliness of warnings. At the same time, these methods usually lack in-depth analysis of the characteristics of equipment status changes and operation response time characteristics, making it difficult to fully assess the potential failure risks of equipment. Summary of the invention

[0003] The present invention provides an Internet of Things-based intelligent remote supervision and early warning method and system, which can solve or partially solve the technical problems involved in the above-mentioned background technology.

[0004] The present invention provides an Internet of Things-based intelligent remote supervision and early warning method, which is applied to an Internet of Things early warning system. The method comprises: obtaining first Internet of Things sensor state information and first device response state information of a target special equipment system in a first fault early warning task process, and obtaining second Internet of Things sensor state information and second device response state information of an associated special equipment system in a second fault early warning task process, wherein the associated special equipment system and the target special equipment system are special equipment systems of the same category, and the first fault early warning task process is different from the second fault early warning task process; after associating the first Internet of Things sensor state information and the second Internet of Things sensor state information, state attribute recognition is performed to obtain first state attribute knowledge, and the first device response state information and the second device response state information are associated with each other. After the status information is associated, the status attribute is identified to obtain the second status attribute knowledge; the first status attribute knowledge and the first equipment response status information are integrated and introduced into the first fault warning algorithm for forward-looking decision analysis to obtain the target sensor state change characteristics of the target special equipment system in the target operation action link label; the second status attribute knowledge and the first IoT sensor state information are integrated and introduced into the second fault warning algorithm for forward-looking decision analysis, and the operation response timeliness characteristics of the target special equipment system in the target operation action link label are determined according to the decision analysis viewpoint of the second fault warning algorithm; the potential fault hidden danger state of the target special equipment system in the target operation action link label is determined according to the target sensor state change characteristics and the operation response timeliness characteristics.

[0005] The present invention provides an Internet of Things early warning system, comprising at least one processor and a memory; the memory stores computer-executable instructions; the at least one processor executes the computer-executable instructions stored in the memory, so that the at least one processor executes the above method.

[0006] The present invention provides a readable storage medium, on which a program or instruction is stored, and when the program or instruction is executed by a processor, the steps of the above method are implemented.

[0007] The present invention obtains the state information of the target special equipment system and its associated systems in different fault warning task processes, and performs correlation analysis and state attribute identification, thereby refining state attribute knowledge that is crucial to fault warning. Furthermore, by combining this knowledge with the fault warning algorithm, the present invention can accurately reveal the potential fault hidden danger state of the target system in a specific operating action link, providing strong support for the preventive maintenance and safety management of the equipment. In this way, the problem of insufficient accuracy and timeliness of traditional fault warning methods can be solved, providing new guarantees for the safe operation of special equipment systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] Figure 1 This is a flow chart of a method for intelligent remote supervision and early warning based on the Internet of Things provided by the present invention.

[0009] Figure 2 This is a structural schematic diagram of an Internet of Things early warning system provided by the present invention. DETAILED DESCRIPTION

[0010] The technical solution of the present invention will be described clearly and completely below in conjunction with the accompanying drawings of the present invention. Obviously, the described embodiments are 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 creative work are within the scope of protection of the present invention.

[0011] The terms "first", "second", etc. in the present invention are used to distinguish similar objects, and are not used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances, so that the embodiments of the present invention can be implemented in an order other than those illustrated or described herein, and the objects distinguished by "first", "second", etc. are generally of the same type, and the number of objects is not limited. For example, the first object can be one or more. In addition, in the present invention, "and / or" means at least one of the connected objects, and the character " / " generally indicates that the objects associated with each other are in an "or" relationship.

[0012] Figure 1A method for early warning based on intelligent remote supervision of the Internet of Things is shown, which is applied to an early warning system of the Internet of Things. The method includes the following steps S110-S150.

[0013] With the rapid development of industrial technology, the safety management of special equipment has become an important link in industrial production. In particular, the safety and stability of quick-opening pressure vessels during use are directly related to production efficiency and personnel safety. In order to effectively prevent and reduce safety accidents caused by equipment failure, the present invention provides strong technical support for the safety interlock of quick-opening pressure vessels through real-time monitoring and intelligent analysis.

[0014] S110. The Internet of Things early warning system obtains the first Internet of Things sensor status information and the first device response status information of the target special equipment system in the first fault early warning task process, and obtains the second Internet of Things sensor status information and the second device response status information of the associated special equipment system in the second fault early warning task process, wherein the associated special equipment system and the target special equipment system are special equipment systems of the same category, and there is a difference between the first fault early warning task process and the second fault early warning task process.

[0015] Among them, S110 is the information acquisition stage of the IoT early warning system. In the application scenario of the quick-opening pressure vessel, the IoT early warning system first bears the heavy responsibility of information acquisition. The IoT early warning system captures the IoT sensor status information and equipment response status information of the equipment in the first fault warning task process in real time through various sensors deployed on the target special equipment system (i.e., a specific quick-opening pressure vessel). This information includes but is not limited to the pressure value, temperature value, switch status of the pressure vessel, and various mechanical or electronic signals related to safety interlocks.

[0016] At the same time, in order to enhance the accuracy and comprehensiveness of the early warning, the IoT early warning system will also obtain the IoT sensor status information and equipment response status information of the associated special equipment system (i.e., a quick-opening pressure vessel of the same type as the target equipment but in a different fault early warning task process) in the second fault early warning task process. "Association" in the present invention means that although these devices have different specific tasks, they belong to the same category, and their operating status and failure modes have certain similarities and comparability. By monitoring these two types of information at the same time, the IoT early warning system can build a richer and more diverse data foundation to provide support for subsequent analysis and decision-making.

[0017] S120. The IoT early warning system associates the first IoT sensor status information with the second IoT sensor status information, and then performs status attribute identification to obtain first status attribute knowledge; and associates the first device response status information with the second device response status information, and then performs status attribute identification to obtain second status attribute knowledge.

[0018] Among them, S120 is the state attribute identification stage of the IoT early warning system. After obtaining the original state information, the IoT early warning system enters the state attribute identification stage. The IoT early warning system first associates the IoT sensor state information of the target device and the associated device in the first and second fault warning task processes, and identifies the state attribute knowledge contained in these state information through advanced data processing algorithms to form the first state attribute knowledge. This process aims to extract valuable information for equipment status evaluation from a large amount of sensor data, such as pressure fluctuation patterns, temperature change trends, etc.

[0019] Similarly, the IoT early warning system will also associate the device response status information of the target device and the associated device to identify the second state attribute knowledge. This includes the response time and response method of the device when it receives a specific instruction or encounters a specific working condition. This information is crucial for evaluating the operating performance and safety of the device.

[0020] S130, the Internet of Things early warning system integrates the first state attribute knowledge and the first device response state information and imports them into the first fault early warning algorithm for forward-looking decision analysis to obtain the target sensor state change characteristics of the target special equipment system in the target operation action link label.

[0021] Among them, S130 is the stage of forward-looking decision analysis and target sensor state change feature extraction. After completing the state attribute recognition, the IoT early warning system enters the forward-looking decision analysis stage. The system first integrates the first state attribute knowledge with the first device response state information to form a comprehensive device state data set. Subsequently, this data set is imported into the specially designed first fault early warning algorithm for in-depth analysis and prediction.

[0022] The first fault warning algorithm is based on advanced technologies such as machine learning and deep learning. It can discover the correlation between equipment status and potential faults through learning and pattern recognition of historical data. In the present invention, the algorithm focuses on analyzing the sensor state change characteristics of the target special equipment system in specific operating action links (such as opening, closing, pressurizing, and depressurizing). These characteristics can reflect whether the equipment has an abnormality or a trend of deviating from the normal state in this link, thereby providing a basis for subsequent fault warnings.

[0023] S140. The IoT early warning system integrates the second state attribute knowledge and the first IoT sensor state information and imports them into the second fault early warning algorithm for forward-looking decision analysis, and determines the operational response time characteristics of the target special equipment system in the target operation action link label based on the decision analysis viewpoint of the second fault early warning algorithm.

[0024] Among them, S140 is the stage of determining the operation response time characteristics. In order to more comprehensively evaluate the operating status of the equipment, the IoT early warning system will also conduct a second round of forward-looking decision analysis. The IoT early warning system integrates the second state attribute knowledge with the first IoT sensor state information and imports it into the second fault warning algorithm. Unlike the first fault warning algorithm, the second algorithm focuses more on analyzing the operation response time characteristics of the equipment in a specific operation action link.

[0025] The operational response time efficiency characteristics refer to the time required for the equipment to respond from the beginning to the completion of the response after receiving the operation instruction, as well as the stability and accuracy of the response process. These characteristics are of great significance for judging the control performance of the equipment, the reliability of the actuator, and the effectiveness of the safety interlock. Through the analysis of the second fault warning algorithm, the IoT early warning system can accurately identify the operational response time efficiency characteristics of the target special equipment system in each operation link, providing an important basis for the subsequent fault hidden danger assessment.

[0026] S150. The Internet of Things early warning system determines the potential fault hazard status of the target special equipment system in the target operation action link label based on the target sensor state change characteristics and the operation response time characteristics.

[0027] Among them, S150 is the stage of determining the potential fault hidden danger state. After completing two rounds of forward-looking decision analysis, the IoT early warning system enters the final stage of determining the potential fault hidden danger state. The IoT early warning system comprehensively evaluates the potential fault hidden danger state of the target special equipment system in a specific operation link based on the target sensor state change characteristics obtained from the first fault warning algorithm and the operation response time efficiency characteristics obtained from the second fault warning algorithm.

[0028] The evaluation process combines the real-time status information of the equipment, historical fault data and expert knowledge base. Through multi-dimensional and multi-level analysis and comparison, it can accurately identify the potential fault points, fault modes and possible consequences of the equipment. Based on these evaluation results, the IoT early warning system can issue early warning signals in a timely manner to remind operators to take appropriate measures for intervention and prevention, thereby effectively avoiding the occurrence of faults and ensuring the safe operation of equipment.

[0029] It should be noted that in order to more specifically illustrate the application of the present invention in the safety interlock of a quick-opening door pressure vessel, the above content will be further refined below to ensure that those skilled in the art can implement it clearly and completely.

[0030] In the actual application scenario of fast-opening pressure vessels, the IoT early warning system captures the following data in real time through sensors deployed on the target equipment: pressure value: for example, recorded once per second, ranging from 0-30MPa, and the current value is 15MPa; temperature value: also recorded every second, ranging from -20℃-200℃, and the current value is 100℃; switch status: 0 means closed, 1 means open, and the current status is 1; safety interlock signal: including the status of various mechanical or electronic safety devices, such as locking devices, pressure relief valves, etc., expressed in binary form (0 or 1), and the current status of all safety devices is 1 (normal). At the same time, the IoT early warning system also obtains the corresponding data of associated equipment (fast-opening pressure vessels of the same type but in different task processes). For example, the pressure value of the associated equipment is currently 12MPa, the temperature value is 90℃, the switch status is 0, and the safety interlock signals are all normal.

[0031] The IoT early warning system associates the sensor status information of the target device and the associated device, and identifies the state attribute knowledge through the following algorithms: Pressure fluctuation algorithm: Calculate the pressure fluctuation range of the target device and the associated device in the past minute to obtain the first state attribute knowledge. For example, the pressure fluctuation range of the target device is ±0.5MPa, and that of the associated device is ±0.3MPa; Temperature trend algorithm: Analyze the temperature change trend of the target device and the associated device in the past ten minutes, and obtain the rate of temperature rise or fall as part of the second state attribute knowledge. For example, the temperature rise rate of the target device is 2℃ / min, and that of the associated device is 1.5℃ / min.

[0032] For device response status information, the IoT early warning system uses the following algorithms for identification: Response time algorithm: Calculates the time from when the device receives the operation instruction to when it starts to respond, and the time from when the response starts to when it is completed. For example, the target device's door opening response time is 2s, and the door closing response time is 3s; the corresponding time of the associated device is slightly different; Response stability algorithm: Evaluates the stability of the device during the response process, such as the fluctuation of pressure and temperature, and the accuracy of the actuator. For example, the target device's pressure fluctuation during the door opening process is less than 0.1MPa, and the temperature remains stable; the performance of the associated device in the same process is also similar.

[0033] Next, the IoT early warning system integrates the first state attribute knowledge with the first device response state information and imports it into the first fault early warning algorithm. The algorithm is based on the random forest model in machine learning and performs forward-looking decision analysis through the following steps:

[0034] (1) Data preprocessing: cleaning and normalizing the integrated data;

[0035] (2) Feature selection: Select the most valuable features for fault warning from the data, such as pressure fluctuation range, temperature rise rate, response time, etc.

[0036] (3) Model training: Use historical failure data to train the random forest model so that it can identify characteristic patterns associated with potential failures;

[0037] (4) Prediction and analysis: Use the trained model to predict the current data and obtain the sensor state change characteristics of the target special equipment system in a specific operation link (such as opening the door). For example, the prediction results show that the pressure fluctuation may increase abnormally during the door opening process and exceed the threshold of the normal range.

[0038] Furthermore, the IoT early warning system integrates the second state attribute knowledge with the first IoT sensor state information and imports it into the second fault early warning algorithm. This algorithm uses the long short-term memory network (LSTM) in deep learning and focuses on analyzing the operational response time characteristics of the equipment in a specific operation action link. The specific steps are as follows:

[0039] Data serialization: Arrange the integrated data in chronological order to form time series data;

[0040] Feature extraction: Use LSTM networks to automatically extract features from time series data, such as the tendency of response time extension, stability changes during the response process, etc.

[0041] Model training and prediction: Use historical operation response data to train the LSTM model and make predictions to obtain the operation response time characteristics of the target special equipment system in each operation action link. For example, the prediction results show that the response time may be prolonged and the stability may decrease during the door closing process.

[0042] Finally, the IoT early warning system comprehensively evaluates the potential fault status of the target special equipment system based on the target sensor state change characteristics and operation response time characteristics obtained from the first and second fault warning algorithms. The specific evaluation process is as follows:

[0043] Feature fusion: The features obtained by the two algorithms are combined to form a comprehensive feature set;

[0044] Multi-dimensional analysis: Analyze and compare feature sets from multiple dimensions such as real-time status information, historical fault data, and expert knowledge base;

[0045] Fault point identification: Through comparative analysis, potential fault points of the equipment can be identified, such as pressure sensor failure, actuator jamming, etc.

[0046] Failure mode and effect analysis: Based on the identified fault points, analyze the possible failure modes and consequences, such as explosion risk caused by abnormal pressure increase, switch failure caused by actuator jamming, etc.

[0047] Warning signal issuance: Based on the analysis results, the system promptly issues a warning signal to remind operators to take appropriate measures for intervention and prevention.

[0048] In this way, the application of the present invention in the safety interlock of the quick-opening door pressure vessel (that is, the real-time monitoring and intelligent analysis method based on the Internet of Things early warning system) can effectively improve the safety and stability of the safety interlock of the quick-opening door pressure vessel and ensure the efficient progress of industrial production.

[0049] It is worth mentioning that the S120 link is the core technical innovation of the present invention, which involves how the IoT early warning system uses advanced algorithms to conduct in-depth analysis and processing of large amounts of data collected from the target special equipment system and related special equipment systems, thereby extracting status attribute knowledge that is crucial to equipment status assessment.

[0050] At this stage, the IoT early warning system will first associate the first IoT sensor state information captured from the target special equipment system (i.e., the quick-opening pressure vessel of primary concern) in the first fault early warning task process with the second IoT sensor state information obtained from the associated special equipment system of the same category but in a different task process. The IoT sensor state information in the present invention covers key parameters such as the real-time pressure value, temperature value, and switch state of the pressure vessel. For example, the current pressure value of the target device may be 15MPa, while the pressure value of the associated device is 12MPa; the temperature of the target device is maintained at 100°C, while the temperature of the associated device is 95°C. By correlating and analyzing these data, the system can begin to build a more comprehensive and dynamic picture of the equipment status.

[0051] Next, the IoT early warning system will use complex data processing algorithms to conduct in-depth mining of these associated sensor state information to identify the implicit first-state attribute knowledge. This process is not just a simple data aggregation, but the algorithm is used to reveal the regularity and trends behind the data, such as the change pattern of pressure values ​​over time, the fluctuation range of temperature under different operating conditions, etc. For example, the system may find that under certain working conditions, the pressure fluctuation range of the target device is always maintained within ±0.5MPa, while the fluctuation range of the associated device is slightly smaller, at ±0.3MPa. These findings constitute the first-state attribute knowledge, which provides an important reference for subsequent fault warnings.

[0052] At the same time, the IoT early warning system will also perform the same correlation analysis on the device response status information of the target device and the associated devices. This includes key indicators such as the device's response time after receiving the operation instruction, the response method, and the stability during the response process. For example, the system may record that the target device's response time in the door opening operation is 2 seconds, the door closing operation is 3 seconds, and the pressure fluctuation during the response process is less than 0.1MPa; while the corresponding time of the associated device may be slightly different. By correlating and analyzing these response status information, the system can extract the second state attribute knowledge, which is also crucial for evaluating the operating performance and safety of the device.

[0053] It can be seen that through the intelligent analysis of the IoT early warning system, S120 not only effectively associates the sensor status information and response status information of the target device with the associated devices, but also identifies the first state attribute knowledge and second state attribute knowledge that are of great value to the equipment status assessment. This information provides a rich data foundation for the subsequent fault warning algorithm, enabling the early warning system to identify potential fault hazards more accurately and timely, thereby ensuring the safe operation of the quick-opening pressure vessel.

[0054] It is worth mentioning that the S150 link, as another core technical innovation of the present invention, demonstrates how the Internet of Things early warning system can accurately lock the potential fault hidden danger status of the target special equipment system in a specific operating action link by relying on the target sensor state change characteristics and operation response time characteristics.

[0055] In this process, the IoT early warning system first focuses on the target sensor state change characteristics. These characteristics are obtained by continuously monitoring and analyzing the sensor data of the target special equipment system, and they reveal the subtle changes in the equipment status over time. Taking the fast-opening pressure vessel as an example, the target sensor state change characteristics may include the fluctuation of the pressure value before and after the door is opened. For example, at the moment of opening the door, the pressure drops rapidly from 15MPa to 10MPa and stabilizes within a few seconds. The pattern and speed of this pressure change are important sensor state change characteristics.

[0056] Next, the IoT early warning system considers the operational response time efficiency feature. This feature focuses on the response speed and stability of the device when performing a specific operation. For example, in the door closing action, the IoT early warning system may record that the time required from issuing the door closing command to the door being completely closed is 2.5 seconds, and the pressure fluctuation during the door closing process remains within ±0.1MPa. Such response time and stability performance constitute the key part of the operational response time efficiency feature.

[0057] With the above two features as the basis, the IoT early warning system further combines these features with preset target operation action link labels. The target operation action link labels are clear identifications of the key stages in the equipment operation process, such as "opening the door", "closing the door", "pressurizing" and "releasing pressure". By matching the features with the labels, the IoT early warning system can more accurately understand the performance of the equipment in each link.

[0058] Finally, the IoT early warning system comprehensively considers the target sensor state change characteristics, operation response time characteristics, and target operation action link labels to evaluate the potential failure status of the equipment in each operation action link. This evaluation may reveal the high-risk status of the equipment in a specific link. For example, in the door opening link, the system may detect an abnormal pressure drop rate. Combined with historical data, this may indicate that the door sealing mechanism is worn and there is a potential risk of leakage. Such an evaluation result is the potential failure status, which provides valuable advance information for preventive maintenance of the equipment.

[0059] It can be seen that through the comprehensive analysis of the Internet of Things early warning system, the S150 link not only links the target sensor state change characteristics and operation response time characteristics with specific operation action links, but also accurately identifies the potential fault hazards of the equipment in these links, providing strong support for the safe management and maintenance of the equipment.

[0060] In summary, the present invention realizes efficient early warning of potential fault hazards of special equipment systems by innovatively integrating the status information of the target special equipment system and the associated special equipment system. Specifically, by respectively obtaining the IoT sensor status information and equipment response status information of the target system and the associated system in different fault warning task processes, and correlating these information to perform state attribute identification, the present invention can extract the first state attribute knowledge and the second state attribute knowledge that are crucial to fault warning.

[0061] Subsequently, these state attribute knowledge are integrated with the equipment response state information or the IoT sensor state information, and imported into a special fault warning algorithm for forward-looking decision analysis, thereby accurately revealing the sensor state change characteristics and operation response time characteristics of the target special equipment system in a specific operation link. Ultimately, based on these characteristics, the present invention can accurately determine the potential fault hidden danger status of the target system, providing strong support for the preventive maintenance and safety management of the equipment.

[0062] In this way, not only the accuracy and timeliness of fault warning are improved, but also deeper fault potential patterns are discovered by correlating and analyzing the status information in different task processes. In addition, by combining the state attribute knowledge with the fault warning algorithm, the present invention realizes a comprehensive and dynamic evaluation of the system status of special equipment, providing a strong guarantee for the stable operation of the equipment.

[0063] In some optional embodiments, the method of associating the first IoT sensor state information and the second IoT sensor state information and then performing state attribute identification to obtain first state attribute knowledge includes: associating the first IoT sensor state information and the second IoT sensor state information to obtain a first sensor state information network; performing feature mining on the first sensor state information network through different feature mining modules to obtain a number of first sensor state features with consistent feature sizes; combining feature members with the same distribution labels in each of the first sensor state features to obtain a number of first sensor state feature chains, and performing state attribute identification on the several first sensor state feature chains to obtain first state attribute knowledge.

[0064] It is understandable that it is crucial to achieve efficient and accurate state attribute recognition for the complex and changeable sensor data in the IoT environment. This process requires not only the ability to effectively integrate state information from different sensors, but also the ability to extract state attribute knowledge with practical value from this information. To this end, this embodiment aims to associate the first IoT sensor state information with the second IoT sensor state information, and perform in-depth state attribute recognition on this basis, and finally obtain the first state attribute knowledge.

[0065] First of all, the core of this embodiment is to build a comprehensive sensor state information network, which is the basis for state attribute identification. In order to achieve this goal, the system needs to associate the first IoT sensor state information with the second IoT sensor state information. In this process, the first IoT sensor state information and the second IoT sensor state information respectively represent the state data captured by two sensors of different sources or types in the IoT environment. These data may cover a variety of physical quantities such as temperature, humidity, pressure, position, speed, etc., and each is transmitted in different formats and frequencies. Therefore, the first step in associating this information is to perform data preprocessing, including data cleaning, format unification, time synchronization, etc., to ensure the accuracy and effectiveness of subsequent analysis.

[0066] After completing data preprocessing, the system uses advanced association algorithms, such as time series-based association analysis and event-driven association rule mining, to closely link the first IoT sensor state information with the second IoT sensor state information. The key to this step is to identify the inherent connection and mutual influence between the two. For example, the temperature change detected by one sensor may be directly related to the humidity fluctuation recorded by another sensor. Through this association, the system is able to build a first sensor state information network containing multi-dimensional state information, which not only reflects the state changes of a single sensor, but also reveals the dynamic relationship between sensors.

[0067] Next, in order to extract useful features from the first sensor state information network, the system deploys multiple feature mining modules. These modules use different algorithms and techniques, such as feature selection in machine learning, convolutional neural networks (CNN) or recurrent neural networks (RNN) in deep learning, and principal component analysis (PCA) in statistical learning, to ensure that the features of state information can be fully captured from multiple angles and levels. Importantly, these feature mining modules are designed to output first sensor state features with consistent feature sizes, in order to ensure the smooth progress of subsequent feature combination and state attribute identification. The consistency of feature size means that each feature vector has the same dimension, which facilitates comparison, combination and analysis.

[0068] After obtaining a series of first sensor state features, the system further organizes and optimizes these features. Specifically, according to the distribution labels of the features, the feature members with the same distribution labels in each first sensor state feature are combined to form several first sensor state feature chains. In the present invention, the distribution label can be understood as the position or category of the feature in the state space, which reflects the state mode or trend to which the feature belongs. By combining feature members with the same distribution labels, the system can construct more representative and explanatory feature chains, which not only simplify the subsequent analysis process, but also help to improve the accuracy of state attribute recognition.

[0069] Subsequently, the system identifies the state attributes of the constructed first sensor state feature chain. This step is a key link in the technical solution, which directly determines the quality and value of the first state attribute knowledge finally obtained. State attribute identification can adopt a variety of methods, including but not limited to rule-based reasoning, machine learning classifiers (such as support vector machines, random forests), deep learning models (such as deep neural networks, convolutional neural networks), etc. When selecting the identification method, it is necessary to comprehensively consider factors such as the characteristics of the data, the requirements for recognition accuracy, and the limitations of computing resources. Through state attribute identification, the system can extract key state attributes from the first sensor state feature chain, such as the operating status of the device, abnormal changes in the environment, and user behavior patterns. These attribute knowledge is of great significance for decision support, fault warning, and user experience optimization of IoT applications.

[0070] It is worth noting that the state attribute recognition in this embodiment is not just a one-time static process, but a dynamic process of continuous iteration and optimization. With the continuous generation of new data and the continuous updating of old data in the IoT environment, the system needs to update the first sensor state information network regularly or in real time, and re-execute the steps of feature mining, feature combination and state attribute recognition. This dynamicity ensures that the system can adapt to the rapid changes in the IoT environment and capture new state attributes and trends in a timely manner, thereby providing users with more accurate and timely services.

[0071] In addition, in order to improve the efficiency and accuracy of state attribute identification, this embodiment can also combine domain knowledge and expert experience for auxiliary analysis. Domain knowledge can include the working principle of IoT devices, the mutual influence of environmental factors, the rules of user behavior, etc., while expert experience comes from the in-depth understanding and long-term practice of IoT applications. By integrating domain knowledge and expert experience into the process of state attribute identification, the system can more accurately interpret and predict the changes in state attributes, thereby improving the overall analysis effect and decision support capabilities.

[0072] In summary, this embodiment associates the first IoT sensor state information with the second IoT sensor state information to construct a first sensor state information network, and finally obtains the first state attribute knowledge through steps such as feature mining, feature combination, and state attribute recognition. In this way, not only the effective integration and analysis of complex state information in the IoT environment is achieved, but also accurate state attribute knowledge is provided, which provides strong support for the intelligentization and efficiency of IoT applications.

[0073] In the next step, the state attribute identification of several first sensor state feature chains is performed to obtain first state attribute knowledge, including: sequentially importing several first sensor state feature chains into the first long-term and short-term monitoring algorithm for processing, and respectively obtaining the first state trend description features corresponding to each of the first sensor state feature chains; performing feature focusing operation on each of the first state trend description features to obtain the key state trend coefficient corresponding to each of the first state trend description features, and integrating each of the first state trend description features according to the key state trend coefficient to obtain first state attribute knowledge.

[0074] In actual application, the process of identifying the state attributes of several first sensing state feature chains to obtain the first state attribute knowledge constitutes a complex and sophisticated technical link. This process requires not only efficient processing and analysis of a large amount of feature chain data, but also accurate capture and extraction of the most representative state attribute information. To this end, this embodiment achieves in-depth analysis of the first sensing state feature chain and accurate identification of the state attributes by combining long-term and short-term monitoring algorithms with feature focus operations.

[0075] First, considering that the first sensing state feature chain contains rich state change information and potential trend laws, this embodiment uses the first long-term monitoring algorithm (Long Short-Term Memory, LSTM) as the core processing tool. LSTM is a recurrent neural network (RNN) that is particularly suitable for processing and predicting time series data. It can effectively capture long-term dependencies in the data while avoiding the gradient disappearance or explosion problems that may occur in traditional RNNs. Importing several first sensing state feature chains into the LSTM algorithm in sequence means that the system can analyze each feature chain one by one according to the time sequence and the intrinsic connection between the feature chains, so as to obtain the first state trend description features corresponding to each feature chain.

[0076] In this process, the LSTM algorithm, through its unique memory unit structure, can learn and memorize the key state change points and trend turning points in the feature chain, and then generate high-dimensional feature vectors that describe these state trends. These first state trend description features not only reflect the overall trend of state changes in the feature chain, but also contain detailed information about state changes, such as the rate, amplitude, and periodicity of change. Therefore, they are an important basis for subsequent state attribute identification.

[0077] Next, in order to further extract the most representative state attribute information from the first state trend description feature, this embodiment introduces a feature focus operation. The core idea of ​​this operation is to determine the feature elements that contribute most to the state trend description, i.e., the key state trend coefficients, by calculating and comparing the element values ​​in each feature vector. These coefficients represent the most important information points in the feature vector, and their changes are often directly related to the key changes in the state attributes.

[0078] When performing feature focusing, the system first traverses each first state trend description feature and calculates the ratio of the value of each element to the average value or standard deviation of the entire feature vector as the importance index of the element. Then, according to the preset threshold or sorting rules, the system selects several elements with the highest importance index as key state trend coefficients. The determination of these coefficients is not only based on their numerical values, but also takes into account their position in the feature vector and their correlation with other elements, thereby ensuring the accuracy and comprehensiveness of the screening results.

[0079] After obtaining the key state trend coefficients corresponding to each first state trend description feature, this embodiment further integrates these coefficients to form the final first state attribute knowledge. The integration process includes weighted summation of key state trend coefficients, normalization processing, and necessary dimensionality reduction operations. Through these operations, the system can bring together the key information scattered in each feature vector to form a unified and compact state attribute description. This description not only contains the main trends and key nodes of state changes, but also reflects the inherent connections and mutual influences between state attributes, thereby providing a strong basis for decision support, state monitoring, and anomaly detection in IoT applications.

[0080] It is worth noting that the state attribute identification process in this embodiment is not a one-time static analysis, but a dynamic process of continuous iteration and optimization. With the continuous generation of new data and the continuous updating of old data in the IoT environment, the system needs to reprocess and analyze the first sensor state feature chain regularly or in real time. This dynamicity ensures that the system can adapt to the rapid changes in the IoT environment and capture new state attributes and trends in a timely manner, thereby providing users with more accurate and timely services.

[0081] In addition, in order to improve the accuracy and efficiency of state attribute recognition, this embodiment can also be combined with other advanced technical means for auxiliary analysis. For example, the results of multiple LSTM models or feature focus operations can be integrated using the ensemble learning method in machine learning to obtain more robust and reliable state attribute knowledge. At the same time, domain knowledge and expert experience can also be introduced to interpret and verify the output results of the LSTM algorithm, thereby ensuring the accuracy and credibility of the recognition results.

[0082] In addition, considering the diversity and complexity of data in the IoT environment, this embodiment also provides flexible configuration options and parameter adjustment mechanisms. Users can select appropriate LSTM model parameters, feature focus operation thresholds and sorting rules, etc. according to specific application scenarios and data characteristics to achieve the best state attribute recognition effect. This flexibility and configurability enables this embodiment to adapt to the application needs of the IoT in different fields and provide users with more customized and personalized services.

[0083] In some examples, there are several associated special equipment systems, and the second fault warning task processes corresponding to each of the associated special equipment systems are different. The first Internet of Things sensor status information and the second Internet of Things sensor status information are associated to obtain a first sensor status information network, including: associating the second Internet of Things sensor status information corresponding to each of the associated special equipment systems according to the security priority of each of the second fault warning task processes to obtain an original sensor status information network; determining an addition area for the first Internet of Things sensor status information in the original sensor status information network according to the security priority of the first fault warning task process and each of the second fault warning task processes; and adding the first Internet of Things sensor status information to the original sensor status information network according to the added area to obtain a first sensor status information network.

[0084] In order to more specifically explain the technical solution of this embodiment, relevant examples will be introduced below to further illustrate how to associate the first IoT sensor state information with the second IoT sensor state information and construct a first sensor state information network.

[0085] 1. Building a raw sensor state information network

[0086] 1.1. In the safety interlock system of quick-opening pressure vessels, there are three related special equipment systems (system A, system B, system C), which are responsible for monitoring pressure, temperature and door lock status respectively. Each system corresponds to a unique second fault warning task process, and the safety priorities of these processes are different.

[0087] System A (pressure monitoring): has the highest safety priority, with a weight of 0.5.

[0088] System B (temperature monitoring): medium security priority, weight 0.3.

[0089] System C (door lock status monitoring): has the lowest security priority and a weight of 0.2.

[0090] 1.2 Algorithm Example

[0091] Weight allocation: Based on the evaluation of security priority, weights are allocated to the second IoT sensor status information of each system.

[0092] Ordered arrangement: Arrange the second IoT sensor state information in order from high to low weight (or from low to high according to specific needs).

[0093] Association construction: Use graph theory algorithms, such as the Dijkstra algorithm or the Floyd-Warshall algorithm, to construct the original sensor state information network based on weights and arrangement orders. In this network, nodes represent individual sensor state information, edges represent the associations between them, and the weights of edges reflect the security priority.

[0094] 2. Determine the area where the first IoT sensor status information is added

[0095] 2.1. The first IoT sensor status information comes from the core monitoring system, which mainly monitors the overall operating status of the container. Its security priority is between system A and system B, with a weight of 0.4.

[0096] 2.2 Algorithm Example:

[0097] Correlation analysis: Calculate the correlation coefficient between the first IoT sensor state information and each node in the original sensor state information network. For example, the Pearson correlation coefficient or the Spearman rank correlation coefficient may be used for calculation.

[0098] Key node identification: Based on the correlation analysis results, identify the node with the strongest correlation with the first IoT sensor state information as the adding point. Set the node with the strongest correlation with a node of system A.

[0099] Determine the added area: In the original sensor state information network, find the area adjacent to the identified key node as the added area of ​​the first IoT sensor state information.

[0100] 3. Add the first IoT sensor status information to the original sensor status information network

[0101] 3.1、Solution of technical problems:

[0102] Data format mismatch: Use data conversion tools (such as JSON, XML, etc.) to convert the data format of the First IoT sensor status information into a format that is compatible with the original sensor status information network.

[0103] Incompatible communication protocols: Use protocol adaptation technology (such as MQTT, CoAP, etc.) to ensure that the first IoT sensor status information can be smoothly transmitted to the original sensor status information network.

[0104] 3.2 Algorithm Example:

[0105] Data fusion: Use data fusion algorithms (such as Kalman filtering, particle filtering, etc.) to fuse the first IoT sensor state information with the data in the original sensor state information network to improve the accuracy and reliability of the data.

[0106] Network optimization: Use complex network analysis algorithms (such as community discovery algorithms, network robustness analysis, etc.) to optimize the structure and performance of the added network to ensure its stability and reliability.

[0107] 3.3 Dynamic iteration and optimization

[0108] In order to keep the first sensor status information network updated and improved, the following steps can be performed regularly:

[0109] Data collection and analysis: Continuously collect operating data and fault records of special equipment systems, and use data analysis tools (such as Python's Pandas, NumPy, etc.) for processing and analysis.

[0110] Safety hazard identification: Use machine learning algorithms (such as decision trees, random forests, etc.) to train and learn from the collected data to identify potential safety hazards and failure modes.

[0111] Network adjustment and optimization: Based on the identified security risks and failure modes, certain nodes or connections in the network are adjusted and optimized to improve the security and reliability of the network.

[0112] 3.4 Technical Effects

[0113] Safety monitoring accuracy: By comparing the failure rate and safety hazard detection rate of special equipment systems before and after implementation, the degree of improvement in safety monitoring accuracy is evaluated.

[0114] Fault warning timeliness: The timeliness of fault warning is evaluated by measuring the time interval from the abnormality of sensor status information to the issuance of fault warning.

[0115] System stability and reliability: Evaluate the stability and reliability of the system by monitoring the network's operating status and performance indicators (such as node connection rate, data transmission rate, etc.).

[0116] In summary, not only the safety monitoring accuracy and fault warning timeliness of the quick-opening door pressure vessel safety interlock system are improved, but also the stability and reliability of the system are enhanced.

[0117] In an optional design idea, the first state attribute knowledge and the first device response state information are integrated and imported into the first fault warning algorithm for forward-looking decision analysis to obtain the target sensor state change characteristics of the target special equipment system in the target operation action link label, including:

[0118] (1) Obtaining a first category of quality inspection indicators corresponding to the target special equipment system; performing quality inspection element identification on the first category of quality inspection indicators to obtain a first quality inspection element knowledge vector;

[0119] (2) The first quality inspection factor knowledge vector, the first state attribute knowledge, and the first equipment response state information are integrated and imported into a first fault warning algorithm for forward-looking decision analysis to obtain target sensor state change characteristics of the target special equipment system in the target operation action link label.

[0120] The goal of the above design idea is to obtain the target sensor state change characteristics of the target special equipment system in a specific operating action link (specified by the target operating action link label) by integrating multiple information and using the first fault warning algorithm to perform forward-looking decision analysis.

[0121] Phase 1: Obtain the first category of quality inspection indicators and identify quality inspection elements

[0122] 1.1. Obtain the first category of quality inspection indicators:

[0123] This step involves extracting quality inspection indicators related to the target special equipment system category from the relevant documents, specifications or databases of the target special equipment system. These indicators are usually used to evaluate the performance, safety and compliance of the system.

[0124] 1.2. Identification of quality inspection elements:

[0125] Conduct in-depth analysis on the first category of quality inspection indicators obtained to identify the key quality inspection elements that constitute these indicators. These elements may include but are not limited to size, material, process, functional testing, etc.

[0126] Through expert knowledge, historical data or machine learning algorithms, these quality inspection factors are transformed into quantifiable knowledge vectors, namely the first quality inspection factor knowledge vectors. This vector will be used in subsequent decision analysis.

[0127] Phase 2: Integrate information and conduct forward-looking decision analysis

[0128] 2.1 Information Integration

[0129] The first quality inspection element knowledge vector, the first state attribute knowledge (information describing the current state of the system) and the first device response state information (information on the response of the system to a specific stimulus or operation) are integrated.

[0130] The purpose of this step is to fuse different sources and types of information together to provide a comprehensive perspective to assess the status of the target special equipment system.

[0131] 2.2. Import the first fault warning algorithm:

[0132] The integrated information is imported into the first fault warning algorithm. This algorithm may be a model based on machine learning, deep learning or rule reasoning, which can process and analyze the input information to predict the future state changes of the system.

[0133] 2.3. Forward-looking decision analysis:

[0134] In the first fault warning algorithm, the integrated information is used for forward-looking decision analysis, which includes evaluating the characteristics of sensor state changes that may occur in different operating action links of the system.

[0135] The algorithm considers multiple factors such as quality inspection elements, current status and device response, as well as their interactions and potential impacts, to generate predictions about the changing characteristics of the target sensor state.

[0136] 2.4. Output target sensor state change characteristics:

[0137] Finally, the algorithm will output the target sensor state change characteristics predicted by the target special equipment system under the target operation action link label. These characteristics may include the change trend and abnormal pattern of sensor parameters such as temperature, pressure, and vibration.

[0138] Through the above steps, the design idea integrates and applies multiple information (quality inspection indicators, state attributes, equipment response) to the first fault warning algorithm for forward-looking decision analysis. This method helps to identify and predict the sensor state change characteristics that may occur in the target special equipment system under specific operating links in advance, thereby supporting more timely and effective fault warning and maintenance decisions.

[0139] Furthermore, the following is an example of the above content combined with the application scenario of the safety interlock of a quick-opening pressure vessel.

[0140] 1. Application Background

[0141] Quick-opening pressure vessels, such as autoclaves and vulcanizers, are widely used in the chemical, building materials, food and other industries. Due to their special structure and frequent operation requirements, they have extremely high requirements for safety performance. As a key device to ensure the safe operation of pressure vessels, the role of safety interlock devices cannot be ignored.

[0142] 2. Numerical Examples

[0143] Pressure threshold: The safe working pressure of the autoclave is set to 1.0MPa, and the safety interlock device is set to allow the door to be opened when the pressure is lower than 0.1MPa. This is to ensure that the pressure in the container has been completely released before opening the door, avoiding danger caused by residual pressure when opening the door.

[0144] Temperature threshold: For some pressure vessels that need to control temperature, such as vulcanization tanks, the safety interlock device may also set a temperature threshold. For example, when the temperature drops below 50°C, the door is allowed to open.

[0145] Time delay: In some cases, in order to prevent false operation caused by too fast operation, the safety interlock device may also be set with a time delay function. For example, after the pressure drops to the safety threshold, it is necessary to wait for 3 minutes before allowing the door to open.

[0146] 3. Algorithm Example

[0147] plaintext

[0148] initialization:

[0149] Set the pressure threshold P_threshold = 0.1MPa

[0150] Set the temperature threshold T_threshold = 50℃

[0151] Set time delay T_delay = 3 minutes

[0152] Cycle detection:

[0153] Read the current pressure P_current and temperature T_current

[0154] ifP_current<=P_thresholdandT_current<=T_threshold:

[0155] Start timing T_start = current time

[0156] if current time - T_start>=T_delay:

[0157] Allow door opening operation

[0158] else:

[0159] Door opening is prohibited and a warning signal is issued

[0160] Handling door opening request:

[0161] When receiving a door opening request:

[0162] Check whether the door opening conditions are met (pressure, temperature, time delay)

[0163] If satisfied, perform door opening operation

[0164] Otherwise, the door opening request is rejected and an error message is fed back.

[0165] 4. The specific role of the safety interlock device

[0166] Real-time monitoring: Use sensors to monitor key parameters such as pressure and temperature of pressure vessels in real time.

[0167] Logical judgment: Based on preset thresholds and logical rules, determine whether the current state allows specific operations (such as opening a door).

[0168] Control execution: When safety conditions are met, the control actuator (such as solenoid valve, pneumatic lock, etc.) performs the door opening operation; when the conditions are not met, the operation is prohibited and a warning signal is issued.

[0169] Recording and feedback: Record the status and results of each operation to provide a basis for subsequent troubleshooting and maintenance; at the same time, provide real-time feedback to operators through indicator lights, display screens, etc.

[0170] Such a design effectively improves the safety performance of the pressure vessel and reduces the risk of accidents.

[0171] In other optional embodiments, the second state attribute knowledge and the first IoT sensor state information are integrated and imported into a second fault warning algorithm for forward-looking decision analysis, and the operational response time efficiency characteristics of the target special equipment system in the target operation action link label are determined based on the decision analysis viewpoint of the second fault warning algorithm, including: obtaining the second category quality inspection indicators corresponding to the target special equipment system; performing quality inspection element identification on the second category quality inspection indicators to obtain a second quality inspection element knowledge vector; integrating the second quality inspection element knowledge vector, the second state attribute knowledge and the first IoT sensor state information and importing them into a second fault warning algorithm for forward-looking decision analysis, and determining the operational response time efficiency characteristics of the target special equipment system in the target operation action link label based on the decision analysis viewpoint of the second fault warning algorithm.

[0172] In the present invention, the core characteristics of the target special equipment system: the quick-opening door pressure vessel and its operating environment need to be given priority consideration. Quick-opening door pressure vessels are usually designed to withstand the storage and transportation of high pressure, high temperature or corrosive media. Their door opening and closing operations are frequent and have extremely high requirements for the safety interlock mechanism. The safety interlock system must not only ensure that the quick-opening door cannot be opened before the internal pressure of the container is completely released, but also respond quickly in an emergency to avoid accidents. Therefore, comprehensive status monitoring and fault warning of the target special equipment system is the basis for ensuring its safe and efficient operation.

[0173] To this end, this embodiment aims to integrate the second state attribute knowledge, the first IoT sensor state information and the knowledge vector of specific quality inspection indicators, and use the second fault warning algorithm to perform forward-looking decision analysis, so as to accurately evaluate the operational response time characteristics of the target special equipment system in specific operating action links (such as door opening and closing operations).

[0174] Specifically, the technical solution first focuses on obtaining the second category quality inspection indicators corresponding to the target special equipment system. These quality inspection indicators are formulated based on industry standards, equipment manufacturer recommendations and historical failure data analysis, and are intended to comprehensively cover the key factors that affect the safety performance of quick-opening pressure vessels. For example, they may include the material strength of the container, sealing performance, reliability of the interlocking device, accuracy and response time of the pressure sensor, etc. The selection of these indicators is intended to provide solid data support for subsequent forward-looking decision-making analysis.

[0175] Next, the quality inspection elements of the acquired second category quality inspection indicators are deeply identified. This process involves breaking down each quality inspection indicator into specific quality inspection elements, and converting them into the second quality inspection element knowledge vector through quantitative processing. For example, for the material strength of the container, it may be necessary to examine specific parameters such as the tensile strength, yield strength, and corrosion resistance of the material, and after standardizing these parameters, form a knowledge vector that can be processed by the algorithm. The completion of this step not only helps to improve the accuracy of data analysis, but also lays a solid foundation for subsequent algorithm integration and decision analysis.

[0176] After completing the construction of the second quality inspection element knowledge vector, the technical solution enters the core link: integrating this knowledge vector, the second state attribute knowledge and the first IoT sensor state information, and importing them into the second fault warning algorithm for forward-looking decision analysis. Among them, the second state attribute knowledge covers the basic configuration information of the equipment, historical maintenance records, operating environment parameters, etc., which are crucial for understanding the current state of the equipment and predicting future trends. The first IoT sensor state information comes directly from various sensors deployed on the equipment, such as pressure sensors, temperature sensors, displacement sensors, etc. The data they collect in real time can directly reflect the immediate state of the equipment.

[0177] The second fault warning algorithm, as the core component of the technical solution, its design needs to fully consider the particularity of the fast-opening pressure vessel and its operating environment. The algorithm should be able to process high-dimensional input data, including the above-mentioned integrated various types of state information and quality inspection element knowledge vectors, and through advanced machine learning or deep learning models, such as neural networks, support vector machines, random forests, etc., to mine the complex associations between data and achieve early identification and warning of equipment failures. More importantly, the algorithm should have good generalization capabilities and be able to accurately predict the operational response time characteristics of the equipment under different working conditions, such as the delay time of door opening and closing operations, the response time of the interlocking device, etc., so as to provide timely and accurate decision support for operators.

[0178] Based on the decision analysis viewpoint of the second fault warning algorithm, the operational response time characteristics of the target special equipment system under the target operation action link label can be further determined. This process not only involves the interpretation of the algorithm output results, but also includes comparing the analysis results with the preset safety threshold to determine whether the current state of the equipment meets the safety requirements. If the operational response time characteristics predicted by the algorithm exceed the safety range, the early warning mechanism should be triggered immediately to notify the operator to take corresponding measures, such as suspending operation, performing maintenance inspections, etc., to prevent potential accidents.

[0179] In addition, the technical solution also emphasizes the continuous monitoring and feedback optimization of decision analysis results. By collecting feedback data from actual operation, such as the actual response time of the equipment and the occurrence of faults, the second fault warning algorithm can be iteratively optimized to continuously improve its prediction accuracy and decision-making efficiency. This closed-loop management method helps to form a continuously improved safety management mechanism to ensure the long-term safe operation of quick-opening pressure vessels.

[0180] Of course, in order to more specifically explain the implementation of the above technical solution in the application scenario of the safety interlock of a quick-opening pressure vessel, an example will be introduced below to illustrate the actual operation and effect of the solution.

[0181] 1. Numerical Examples

[0182] The target special equipment system is a quick-opening pressure vessel with a designed working pressure of 2.0 MPa and a working temperature not exceeding 150°C. To ensure safety, its door opening and closing operation and interlocking mechanism need to be closely monitored.

[0183] The second category of quality inspection indicators: The following items were selected as key quality inspection indicators: container wall thickness (standard value is 10mm, allowable deviation is ±0.5mm), sealing surface flatness (standard deviation does not exceed 0.1mm), interlock device response time (no more than 2 seconds), pressure sensor accuracy (±0.05MPa).

[0184] The second quality inspection element knowledge vector:

[0185] Container wall thickness: [9.5mm, 10.0mm, 10.5mm] (represents the measured values ​​at three different positions);

[0186] Sealing surface flatness: [0.08mm, 0.05mm, 0.12mm] (flatness deviation at three different sealing points);

[0187] Interlock device response time test: [1.8s, 1.9s, 2.0s] (average of three tests);

[0188] Pressure sensor accuracy calibration: [+0.03MPa, -0.02MPa, +0.04MPa] (results of three calibrations).

[0189] First IoT sensor status information:

[0190] Current pressure: 1.2MPa;

[0191] Current temperature: 120℃;

[0192] Interlocking device status: closed;

[0193] Door status: Closed.

[0194] 2. Algorithm Example

[0195] In order to realize the forward-looking decision analysis of fast-opening pressure vessels, a second fault warning algorithm based on support vector machine (SVM) is designed. The algorithm can process multi-dimensional input data and output the prediction of the operation response time characteristics of the equipment in a specific operation link (such as door opening and closing operation).

[0196] Algorithm input: second quality inspection factor knowledge vector; second state attribute knowledge (such as equipment model, service life, last maintenance time, etc.); first IoT sensor state information.

[0197] Algorithm output: prediction of door opening and closing operation response time; interlocking device reliability assessment; potential fault warning signal.

[0198] Algorithm implementation steps: Data preprocessing: Standardize the input state information and quality inspection element knowledge vectors to eliminate the impact of different dimensional data on algorithm performance; Feature extraction: Extract features useful for fault warning from the preprocessed data. For example, the average value and deviation of the container wall thickness, the maximum value and average value of the sealing surface flatness, and the stability index of the interlocking device response time can be calculated; Model training: Use historical fault data and normal operation data to train the SVM model. During the training process, the prediction performance of the model is optimized by adjusting the model parameters (such as kernel function type, penalty factor, etc.); Online prediction: Input the real-time collected state information and quality inspection element knowledge vectors into the trained SVM model for online prediction. The model will output the predicted value of the door opening and closing operation response time of the equipment in the current state, the reliability evaluation result of the interlocking device, and the potential fault warning signal; Decision support: Provide decision support for operators based on the prediction results of the algorithm. For example, if the predicted door opening and closing operation response time exceeds the safety threshold, or the reliability evaluation result of the interlocking device is poor, the algorithm will trigger a warning signal to prompt the operator to check or repair.

[0199] Through the specific implementation of the above technical solution, it is possible to achieve forward-looking decision analysis of the quick-opening pressure vessel during the door opening and closing operation. The algorithm can accurately predict the operational response time characteristics of the equipment and provide timely and accurate decision support for operators. This helps to improve the safety performance of the equipment and reduce the possibility of failure, thereby ensuring the smooth progress of the production process.

[0200] For example, in a certain actual operation, the algorithm predicted that the response time of the interlock device exceeded the safety threshold. The operator promptly checked according to the early warning signal and found that the interlock device was stuck. Through timely maintenance, potential safety hazards were avoided and the normal operation of the equipment was ensured.

[0201] In a preferred embodiment, the second quality inspection element knowledge vector, the second state attribute knowledge and the first IoT sensor state information are integrated and imported into a second fault warning algorithm for forward-looking decision analysis, and the operation response time characteristics of the target special equipment system in the target operation action link label are determined based on the decision analysis viewpoint of the second fault warning algorithm, including: combining the second quality inspection element knowledge vector, the second state attribute knowledge and the first IoT sensor state information to obtain linkage warning information to be analyzed; importing the linkage warning information to be analyzed into the second fault warning algorithm, and performing discrimination processing on the linkage warning information to be analyzed according to the discrimination branch of the second fault warning algorithm to obtain a discrimination signal. Thermal characteristic diagram, wherein the discrimination thermal characteristic diagram is used to indicate the possibility weights of several preliminary screening equipment action analysis modules; according to the discrimination thermal characteristic diagram, a target equipment action analysis module is determined among several said preliminary screening equipment action analysis modules, and the first operation action link label of the first fault warning task process, the target operation action link label and each operation action link label between the first operation action link label and the target operation action link label are respectively imported into the target equipment action analysis module to obtain several target action response continuity evaluations; based on several said target action response continuity evaluations, the operation response time efficiency characteristics of the target special equipment system at the target operation action link label are determined.

[0202] When further discussing the forward-looking decision analysis of the safety interlock system of the quick-opening pressure vessel, a specific example will be used below to further improve the introduction of the above embodiment, aiming to provide a more specific and operational framework so that those skilled in the art can better understand how this technical solution works in practical applications.

[0203] 1. Numerical Examples

[0204] The design parameters and working conditions of an actual quick-opening pressure vessel are as follows:

[0205] Design working pressure: 2.5 MPa

[0206] Designed operating temperature: 180 ℃

[0207] Standard value of container wall thickness: 12 mm, allowable deviation ±0.8 mm

[0208] Sealing surface flatness standard deviation: no more than 0.15 mm

[0209] Interlock device response time: no more than 1.5 seconds

[0210] Pressure sensor accuracy: ±0.05 MPa.

[0211] Based on this, the following second quality inspection factor knowledge vector, second state attribute knowledge and first IoT sensor state information are collected:

[0212] The second quality inspection element knowledge vector: container wall thickness measurement value: [11.9 mm, 12.2 mm, 12.1 mm]; sealing surface flatness deviation: [0.10 mm, 0.12 mm, 0.14 mm]; interlock device response time test: [1.45 s, 1.48 s, 1.49s]; pressure sensor accuracy calibration: [+0.03 MPa, -0.04 MPa, +0.02 MPa].

[0213] Second state attribute knowledge: Equipment model: XYZ-1234; Service life: 5 years; Last maintenance time: 6 months ago; Operating environment: Indoor, temperature control range 20-25 ℃.

[0214] First IoT sensor status information: current pressure: 1.8 MPa; current temperature: 160 ℃; interlock device status: closed; door status: closed.

[0215] 2. Algorithm Implementation

[0216] In order to perform forward-looking decision analysis, a second fault warning algorithm based on machine learning is used, such as support vector machine (SVM) or random forest. The following is a simplified algorithm implementation process:

[0217] Data preprocessing: Standardize the second quality inspection factor knowledge vector, the second state attribute knowledge, and the first IoT sensor state information to ensure data consistency and comparability. For example, convert the container wall thickness measurement value into a percentage deviation from the standard value, and use the sealing surface flatness deviation, interlock device response time, and pressure sensor accuracy verification results directly as input features.

[0218] Feature engineering: Extract features useful for fault warning from preprocessed data. For example, calculate the average deviation of container wall thickness, the maximum deviation of sealing surface flatness, the mean and standard deviation of interlock device response time, and the mean absolute error of pressure sensor accuracy.

[0219] Model training: Use historical fault data and normal operation data to train the second fault warning algorithm. During the training process, cross-validation and parameter tuning are used to optimize the model's prediction performance.

[0220] Online prediction: Input the real-time collected status information (i.e., the first IoT sensor status information) and the latest quality inspection results (i.e., the second quality inspection element knowledge vector) into the trained model. The model outputs a discriminant thermal feature map, indicating the probability weight of each preliminary screening equipment action analysis module.

[0221] Decision analysis: According to the discriminant thermal characteristic diagram, the equipment action analysis module with the highest probability weight is selected as the target. The first operation action link label of the first fault warning task process (such as "start pressurization"), the target operation action link label (such as "quick door opening"), and the operation action link labels between them (such as "pressure stability", "temperature control", etc.) are imported into the target equipment action analysis module. The module outputs several target action response continuity evaluations, such as the response time and stability score of each link.

[0222] Determination of operation response time efficiency characteristics: Based on the evaluation of the target action response continuity, the operation response time efficiency characteristics of the target special equipment system in the target operation action link label are comprehensively evaluated. For example, if the response time of the "quick door opening" link exceeds the preset threshold or the stability score is low, a fault warning signal is triggered.

[0223] The following is an example of a random forest algorithm that illustrates how to provide fault warnings based on input features:

[0224] Python

[0225] from sklearn.ensemble import RandomForestClassifier

[0226] from sklearn.model_selection import train_test_split

[0227] from sklearn.metrics import accuracy_score, classification_report

[0228] # Assume that there is already a data set containing historical fault data and normal operation data

[0229] # The dataset contains features (such as container wall thickness deviation, sealing surface flatness deviation, etc.) and labels (such as "normal", "fault")

[0230] # Data preprocessing and feature engineering

[0231] # ...

[0232] # Divide into training set and test set

[0233] X_train, X_test, y_train, y_test = train_test_split(features, labels, test_size=0.2, random_state=42)

[0234] # Train the random forest model

[0235] model = RandomForestClassifier(n_estimators=100, random_state=42)

[0236] model.fit(X_train, y_train)

[0237] # Online prediction (the input is set to real-time collected status information and the latest quality inspection results)

[0238] real_time_features = [...] # Real-time feature values, fill in according to actual conditions

[0239] predicted_label = model.predict([real_time_features])

[0240] # Make decisions based on the prediction results

[0241] # ...

[0242] # If the prediction result is "failure", the fault warning signal is triggered

[0243] if predicted_label == 'failure':

[0244] print("Fault warning signal has been triggered!")

[0245] else:

[0246] print("The device is running normally.")

[0247] Such a design not only helps to improve the safety performance and production efficiency of the equipment, but also provides new ideas and methods for the safety management of special equipment.

[0248] In some optional embodiments, the first fault warning algorithm is debugged based on the following steps: obtaining the third IoT sensor status information and the third device response status information of the first historical special equipment system in the third fault warning task process, obtaining the fourth IoT sensor status information of the second historical special equipment system in the fourth fault warning task process, wherein the second historical special equipment system and the first historical special equipment system are special equipment systems of the same category, there is a difference between the third fault warning task process and the fourth fault warning task process, the monitoring period of the third IoT sensor status information is equal to the step period between the target operation action link label and the first operation action link label of the first fault warning task process, the monitoring period of the third device response status information and the monitoring period of the fourth IoT sensor status information are both equal to the monitoring period of the first IoT sensor status information; debug the first fault warning algorithm based on the third IoT sensor status information, the third device response status information and the fourth IoT sensor status information.

[0249] It can be understood that for the application scenario of safety interlock of fast-opening pressure vessel, the training and debugging process of the first fault warning algorithm is a complex process that comprehensively considers multi-source information, cross-system data and different task process characteristics. The process aims to build an efficient, accurate and adaptable warning model to achieve early identification and warning of potential faults in special equipment systems.

[0250] First, in order to fully capture the characteristics of special equipment systems under different operating conditions, it is necessary to obtain the third IoT sensor state information and the third device response state information of the first historical special equipment system in the third fault warning task process. This information constitutes the basic data set for algorithm training. Among them, the third IoT sensor state information covers various physical parameters of the pressure vessel in a specific task process, such as pressure, temperature, vibration, etc. These parameters are important indicators reflecting the operating status of the system. At the same time, the third device response state information records the response of the system when it receives a warning signal or performs a specific operation, such as the action time of the interlocking device, the opening and closing status of the door, etc. This information is crucial for evaluating the response performance of the system.

[0251] It is worth noting that the monitoring period of the third IoT sensor state information is set to be equal to the step period between the target operation action link label and the first operation action link label of the first fault warning task process. This setting ensures that the algorithm can fully consider the dynamic changes of the system from the initial state to the target state during the training process, so as to more accurately capture the early signs of potential faults.

[0252] Secondly, in order to enhance the generalization ability and adaptability of the algorithm, it is also necessary to obtain the fourth IoT sensor state information of the second historical special equipment system in the fourth fault warning task process. In the present invention, the second historical special equipment system and the first historical special equipment system belong to the same category, that is, both are fast-opening pressure vessel systems, but they run in different task processes, and the fourth fault warning task process is different from the third fault warning task process. By introducing data from different task processes, the algorithm can learn more diverse fault modes and improve the ability to identify unknown faults.

[0253] After obtaining the above data, the next step is to debug the first fault warning algorithm based on this information. The core of the debugging process is how to effectively use this data to train the algorithm so that it can accurately identify potential fault modes. To this end, supervised learning methods in machine learning can be used, such as support vector machines, neural networks, or random forests. These methods can build a prediction model based on the relationship between the input features (i.e., IoT sensor status information and device response status information) and the output label (i.e., whether there is a fault).

[0254] During the debugging process, the data must first be preprocessed, including data cleaning, feature extraction, and normalization. Data cleaning is to remove noise and outliers to ensure the accuracy and reliability of the data; feature extraction is to extract useful information for fault prediction from the original data, such as the rate of change of pressure, the range of temperature fluctuation, etc.; normalization is to convert the data to a unified dimension to facilitate algorithm processing and comparison.

[0255] Next, the preprocessed data is divided into a training set and a test set. The training set is used to train the algorithm, by constantly adjusting the parameters and structure of the model so that it can best fit the data; the test set is used to verify the performance of the algorithm and evaluate its accuracy and generalization ability in practical applications. During the training process, the cross-validation method can be used, that is, the data is divided into different training sets and test sets for training multiple times to avoid overfitting and underfitting problems.

[0256] In the process of algorithm debugging, special attention should be paid to the optimization of hyperparameters. Hyperparameters are parameters that need to be manually set in the algorithm, such as learning rate, regularization coefficient, etc. They have an important impact on the performance of the algorithm. Through methods such as grid search, random search or Bayesian optimization, the optimal hyperparameter combination can be found to further improve the performance of the algorithm.

[0257] In addition, in order to enhance the algorithm's ability to adapt to dynamic changes in special equipment systems, it is also possible to consider introducing a dynamic adjustment mechanism. For example, the threshold or weight of the algorithm can be dynamically adjusted according to the real-time operating status and historical data of the system to achieve a more accurate early warning of potential failures.

[0258] After debugging and optimization of the above steps, the first fault warning algorithm was finally determined. This algorithm can comprehensively consider the IoT sensor status information and equipment response status information of the special equipment system, as well as the differences between different task processes and historical data, to achieve accurate early warning of potential faults. In practical applications, this algorithm can be embedded in the monitoring system of the special equipment system to monitor the operating status of the system in real time. Once a potential fault is found, it will automatically trigger a warning signal and promptly notify relevant personnel to handle it, thereby effectively avoiding accidents and ensuring the safety and stability of production.

[0259] In this way, by acquiring multi-source information, cross-system data, and data from different task processes, machine learning methods are used to train and optimize the model, and finally an efficient, accurate, and adaptable early warning model is constructed. The application of this model will greatly improve the safety and stability of special equipment systems and provide strong protection for the safe production of enterprises.

[0260] In addition, the following points should be noted during implementation:

[0261] Data quality and integrity: Ensure that the acquired data is accurate, complete and free of bias, which is the basis for algorithm training. For missing or abnormal data, reasonable methods need to be used to fill or process them to avoid affecting the performance of the algorithm;

[0262] Interpretability of the algorithm: While pursuing algorithm performance, it is also necessary to consider its interpretability. That is, the output of the algorithm should be able to intuitively reflect the operating status of the system and the cause of potential failures, so that relevant personnel can understand and handle it;

[0263] Real-time performance and response speed: For special equipment systems such as fast-opening pressure vessels, real-time performance and response speed are crucial. Therefore, in the process of algorithm design and implementation, its computational efficiency and response time need to be fully considered to ensure that accurate early warning signals can be given in the shortest possible time;

[0264] Continuous update and optimization: With the continuous upgrading and transformation of special equipment systems and changes in the operating environment, algorithms also need to be continuously updated and optimized. By continuously introducing new data and feedback information, the algorithm is iteratively updated to adapt to new operating conditions and failure modes.

[0265] In other optional embodiments, the obtaining of the third IoT sensor status information and the third device response status information of the first historical special equipment system in the third fault warning task process includes: obtaining the initial screening IoT sensor status information and the initial screening device response status information of the first historical special equipment system in the initial screening fault warning task process, wherein the monitoring period of the initial screening IoT sensor status information is greater than the step period between the target operation action link label and the first operation action link label, and the monitoring period of the initial screening device response status information is greater than the monitoring period of the first IoT sensor status information; performing a roaming scan in the initial screening IoT sensor status information according to the first information scanning window to obtain the third IoT sensor status information, wherein the coverage size of the first information scanning window is smaller than the number of information units of the initial screening IoT sensor status information; performing a roaming scan in the initial screening device response status information according to the second information scanning window to obtain the third device response status information, wherein the coverage size of the second information scanning window is smaller than the number of information units of the initial screening device response status information.

[0266] It is understandable that for the application scenario of safety interlock of fast-opening pressure vessel, the specific implementation method of obtaining the third IoT sensor state information and the third equipment response state information of the first historical special equipment system in the third fault warning task process involves a series of sophisticated data processing and information screening steps. This process aims to extract the most critical information for fault warning from massive historical data, laying a solid foundation for the training of subsequent algorithms and the construction of warning models.

[0267] In order to fully capture the operating status of the special equipment system, its IoT sensor status information and equipment response status information under various task processes are usually continuously monitored and recorded. However, not all data are of equal value for fault warning. Therefore, in an embodiment, the first thing to obtain is the initial screening IoT sensor status information and the initial screening equipment response status information of the first historical special equipment system in the initial screening fault warning task process. The "initial screening" here means that these data are original and have not been finely screened, but they provide a basis for subsequent fine analysis.

[0268] It is worth noting that the monitoring period of the initial screening of the IoT sensor state information is set to be greater than the step period between the target operation action link label and the first operation action link label. This setting is to ensure that the entire change process from the initial state to the target state of the system can be covered, so as not to miss any possible signs of failure. At the same time, the monitoring period of the initial screening device response state information is also greater than the monitoring period of the first IoT sensor state information, in order to more comprehensively understand the response characteristics of the system at different time scales.

[0269] Next, in order to extract truly useful information for fault warning from the initial screening data, it is necessary to use the information scanning window method to perform wandering scanning. This is an efficient data processing strategy that allows for flexible selection and analysis of specific parts of the data without changing the original data structure.

[0270] The third IoT sensor state information can be obtained by performing a roaming scan in the primary screening IoT sensor state information according to the first information scanning window. In the present invention, the coverage size of the first information scanning window is carefully designed to be smaller than the number of information units of the primary screening IoT sensor state information. This means that the scanning window will not cover all the data at once, but will gradually move in the data set, processing only a part of the data each time. Such a design helps to analyze the changing trends and characteristics of the data in more detail, thereby more accurately capturing the early signs of potential faults.

[0271] Similarly, the third device response status information can be obtained by performing a roaming scan in the primary screening device response status information according to the second information scanning window. The coverage size of the second information scanning window is also smaller than the number of information units of the primary screening device response status information to ensure a detailed analysis of the device response status. Through this step, we can gain an in-depth understanding of the detailed response of the system when it receives an early warning signal or performs a specific operation, such as the action time of the interlocking device, the opening and closing status of the door, etc. This information is crucial for evaluating the system's response performance and identifying potential faults.

[0272] During the wandering scan, several key points need to be noted. First, the moving step size of the scanning window should be reasonably set according to the characteristics of the data and the warning requirements. Too small a step size may lead to data redundancy and low computational efficiency, while too large a step size may miss important information. Secondly, the coverage size of the scanning window also needs to be adjusted according to the distribution of the data and the fault characteristics. In some cases, it may be necessary to perform multiple scans with scanning windows of different sizes to obtain more comprehensive information.

[0273] In addition, in order to improve the efficiency and accuracy of data processing, advanced technologies such as parallel processing, data compression and feature extraction can also be used. Parallel processing allows multiple data blocks to be processed simultaneously, thereby speeding up processing; data compression can reduce data storage space and transmission time, improving data processing efficiency; feature extraction is to extract useful information for fault warning from the original data, such as the rate of change of pressure, the range of temperature fluctuation, etc., which can be used as input features for subsequent algorithm training.

[0274] After the above steps, the third IoT sensor state information and the third device response state information are finally obtained. These information are the result of fine screening and processing, and they accurately reflect the operating status and response characteristics of the special equipment system under the third fault warning task process. This information will serve as an important input for subsequent algorithm training to build an accurate and efficient fault warning model.

[0275] This implementation is of great significance in the application scenario of safety interlock of fast-opening pressure vessels. Due to the complex and changeable working environment of pressure vessels, the occurrence of potential failures is often sudden and unpredictable. Therefore, it is crucial to improve the accuracy and reliability of the early warning model by carefully screening and processing historical data to extract truly useful information for fault warning. This implementation can not only help to better understand the operating status and response characteristics of the system, but also provide strong support for subsequent fault warning and maintenance decisions.

[0276] In this way, by using the information scanning window method to perform wandering scanning, combined with parallel processing, data compression and feature extraction technologies, the most critical information for fault warning can be extracted from massive historical data. This information will serve as an important input for subsequent algorithm training to build an accurate and efficient fault warning model, providing a strong guarantee for the safe operation of fast-opening pressure vessels.

[0277] In some independent embodiments, before obtaining the third IoT sensor status information and the third equipment response status information of the first historical special equipment system in the third fault warning task process, the IoT-based intelligent remote supervision and warning method also includes: determining a historical category from a number of preliminary screening categories, and determining the special equipment system corresponding to the historical category as the first historical special equipment system; or, determining the special equipment systems corresponding to a number of preliminary screening categories as preliminary screening special equipment systems corresponding to each of the preliminary screening categories, screening each of the preliminary screening special equipment systems, and determining the first historical special equipment system according to the screening results.

[0278] For the application scenario of safety interlock of fast-opening pressure vessel, the IoT-based intelligent remote supervision and early warning method also includes a series of crucial pre-steps before obtaining the third IoT sensor state information and the third equipment response state information of the first historical special equipment system in the third fault early warning task process. These steps are designed to ensure that the selected historical special equipment system is representative and can provide an accurate and valuable data basis for the subsequent construction of the fault early warning model.

[0279] First, it is necessary to determine the historical categories from several preliminary screening categories. This process is a key link in the preliminary classification and screening of large amounts of historical data. In the application scenario of the quick-opening pressure vessel safety interlock, the preliminary screening categories may include but are not limited to the model of the pressure vessel, age, working environment, historical maintenance records and other dimensions. Through a detailed analysis of these categories, it is possible to preliminarily identify which categories of pressure vessels are more likely to have specific failure modes, thereby identifying these categories as historical categories. The completion of this step provides direction for subsequent focus on specific categories of special equipment systems.

[0280] After the historical category is determined, the next step is to determine the special equipment system derived from the historical category as the first historical special equipment system. This step is the process of directly applying the screening results to the actual system. In the scenario of quick-opening pressure vessels, this means selecting one or more representative systems from all pressure vessels belonging to the selected historical category as the first historical special equipment system. These systems usually have complete operation records, sensor data, and equipment response logs, which are ideal data sources for building fault warning models.

[0281] Alternatively, as an alternative, the special equipment systems corresponding to several preliminary screening categories can be respectively determined as the preliminary screening special equipment systems corresponding to each preliminary screening category. The advantage of this approach is that it can cover all possible failure conditions more comprehensively and avoid deviations caused by the selection of a single category. When screening each preliminary screening special equipment system, the suitability of each system will be evaluated based on preset screening criteria, such as data integrity, system stability, and frequency of failures. The completion of this step provides a basis for the subsequent selection of the optimal first historical special equipment system from multiple systems.

[0282] During the screening process, special attention should be paid to the availability and quality of data. For special equipment systems such as quick-opening pressure vessels, the accuracy and real-time nature of sensor data are crucial. Therefore, when screening the first historical special equipment system, it is necessary to ensure that the selected system can provide high-quality, high-frequency sensor data, as well as detailed equipment response records. These data will serve as the basis for subsequent fault warning model training, which directly affects the accuracy and reliability of the model.

[0283] After determining the first historical special equipment system, we can enter the stage of obtaining the third IoT sensor status information and the third device response status information. This process is achieved through IoT technology, which specifically involves deploying various sensors and actuators on the special equipment system to monitor the operating status and response of the system in real time. In the scenario of fast-opening door pressure vessels, these sensors may include pressure sensors, temperature sensors, displacement sensors, etc., which are used to capture information such as pressure changes, temperature changes, and the opening and closing status of the door inside the container. At the same time, the actuator is responsible for performing corresponding interlocking actions based on the sensor data and control logic to ensure the safe operation of the container.

[0284] When acquiring the third-party IoT sensor status information and the third-party device response status information, special attention should be paid to the synchronization and consistency of the data. Since the operation of the special equipment system is a dynamic process, the data of each sensor and actuator must be accurately aligned on the time axis to accurately reflect the actual state of the system. Therefore, in the process of data acquisition and processing, advanced time synchronization technology and data fusion algorithms need to be used to ensure the accuracy and reliability of the data.

[0285] In addition, in order to improve the accuracy and real-time performance of fault warning, the acquired data can be further analyzed and processed in combination with machine learning algorithms. By training the fault warning model, real-time monitoring and early warning of the operating status of special equipment systems can be achieved. When the system is abnormal or has potential faults, the model can issue early warning signals in a timely manner, providing operators with sufficient reaction time to avoid accidents.

[0286] In the application scenario of the safety interlock of the quick-opening pressure vessel, this series of implementation examples not only improves the accuracy and reliability of fault warning, but also realizes the remote supervision and intelligent management of special equipment systems. Through the application of Internet of Things technology and the support of machine learning algorithms, real-time monitoring, data analysis and fault warning of the operating status of pressure vessels can be realized, providing a strong guarantee for the safe production of enterprises.

[0287] Specifically, firstly, by determining the historical category from several preliminary screening categories, and then determining the first historical special equipment system, an accurate and valuable data basis is provided for the subsequent construction of the fault warning model. Secondly, when obtaining the third IoT sensor status information and the third device response status information, advanced time synchronization technology and data fusion algorithms are used to ensure the accuracy and reliability of the data. Finally, the acquired data is further analyzed and processed in combination with machine learning algorithms, and an efficient fault warning model is constructed, realizing remote supervision and intelligent management of special equipment systems.

[0288] The application of this series of embodiments not only improves the safety performance of quick-opening pressure vessels, but also brings significant economic and social benefits to enterprises. Through real-time monitoring and early warning, potential faults can be discovered and handled in a timely manner to avoid accidents, thereby reducing the economic losses and reputation damage of enterprises. At the same time, the intelligent management method also improves the management efficiency and production efficiency of enterprises, laying a solid foundation for the sustainable development of enterprises.

[0289] Based on the above examples, the representativeness of the selected historical special equipment system and the accuracy of the data are ensured, providing a solid foundation for the subsequent construction of the fault warning model. Through the application of Internet of Things technology and the support of machine learning algorithms, remote supervision and intelligent management of special equipment systems are realized, providing a strong guarantee for the safe production of enterprises.

[0290] In some other independent embodiments, the second fault warning algorithm is debugged based on the following steps: obtaining the third device response status information and the fifth Internet of Things sensor status information of the first historical special equipment system in the third fault warning task process, and obtaining the fourth device response status information of the second historical special equipment system in the fourth fault warning task process, wherein the second historical special equipment system and the first historical special equipment system are special equipment systems of the same category, there is a difference between the third fault warning task process and the fourth fault warning task process, the monitoring period of the third device response status information is equal to the step period between the target operation action link label and the first operation action link label of the first fault warning task process, the monitoring period of the fifth Internet of Things sensor status information and the monitoring period of the fourth device response status information are both equal to the monitoring period of the first Internet of Things sensor status information; debug the second fault warning algorithm based on the third device response status information, the fifth Internet of Things sensor status information and the fourth device response status information.

[0291] For the application scenario of safety interlock of fast-opening pressure vessel, the debugging process of the second fault warning algorithm is not only a comprehensive consideration of functionality and concept, but also a practical process involving specific numerical calculations and specific algorithm applications. The following will combine specific numerical examples and related algorithm examples to explain this process in more detail.

[0292] First, in order to debug the second fault warning algorithm, it is necessary to obtain the third equipment response status information and the fifth IoT sensor status information in the third fault warning task process from the first historical special equipment system. For example, the first historical special equipment system is a fast-opening pressure vessel with a working pressure range of 0-10MPa and a working temperature range of 0-100°C. In the third fault warning task process, the equipment response status of the system in 10 consecutive working cycles is recorded, including the number of switches of the pressure vessel, the duration of each switch, the number of actions of the safety interlock device, and the action time. At the same time, IoT sensor data in these cycles, such as pressure sensor readings, temperature sensor readings, and displacement sensor readings, are also obtained, and these data are collected at a frequency of once per second.

[0293] Next, the fourth equipment response status information in the fourth fault warning task process is obtained from the second historical special equipment system. For example, the second historical special equipment system has the same structure and working principle as the first historical special equipment system, but their working environment and use conditions are different. In the fourth fault warning task process, the equipment response status of the system in 10 consecutive working cycles is also recorded and compared with the data of the first historical special equipment system. Through comparison, it is found that there are obvious differences in the equipment response status when the two systems perform different tasks, which provides strong data support for the subsequent debugging algorithm.

[0294] After obtaining these data, we can start debugging the second fault warning algorithm. For example, a fault warning algorithm based on support vector machine (SVM) is used. First, the data needs to be preprocessed, including data cleaning, normalization, and feature selection. In this process, according to the characteristics of the data and the requirements of the algorithm, key features such as pressure, temperature, displacement, and number of switches are selected as the input of the algorithm.

[0295] Then, the third device response state information and the fifth IoT sensor state information are used to train the SVM model. During the training process, the data is divided into a training set and a test set, and the optimal model parameters are selected by cross-validation. At the same time, the fourth device response state information is used as a control to evaluate the generalization ability of the model in different task processes. Through repeated adjustments and optimizations, an SVM model with good prediction performance and generalization ability is finally obtained.

[0296] To illustrate this process more specifically, for example, during the training process, the radial basis function (RBF) was selected as the kernel function of the SVM, and the optimal values ​​of the penalty parameter C and the kernel parameter γ were determined by cross-validation. Then, the performance of the model was evaluated using the test set, and the following results were obtained: the accuracy of the model reached 95%, the recall rate reached 90%, and the F1 score reached 92.5%. These results show that the model has good predictive performance and generalization ability, and can be applied to actual fast-opening pressure vessel systems.

[0297] In addition, during the debugging process, special attention was paid to the stability and robustness of the algorithm. Since the working environment of the quick-opening pressure vessel may have large fluctuations and uncertainties, a variety of methods were used to improve the stability and robustness of the algorithm. For example, data enhancement technology was introduced to increase the diversity of training data; an ensemble learning method was used to combine the prediction results of multiple weak classifiers to improve the performance of the overall model; and the algorithm was also subjected to robustness analysis to evaluate its performance under different noise levels.

[0298] Finally, after the second fault warning algorithm has been fully debugged and optimized, it can be applied to the actual quick-opening pressure vessel system. In actual applications, the algorithm needs to be updated and maintained regularly to ensure that it can continue to accurately predict and diagnose system failures. At the same time, it is necessary to collect more operating data and failure cases to continuously enrich and improve the algorithm's knowledge base and model library, and improve its accuracy and reliability in actual applications.

[0299] Further, Figure 2 FIG. 2 is a schematic diagram of the structure of an Internet of Things early warning system 200 provided by the present invention. Figure 2 The IoT early warning system 200 shown includes a processor 210, which can call and run a computer program from a memory to implement the method of the present invention.

[0300] Alternatively, if Figure 2 As shown, the IoT early warning system 200 may further include a memory 230. The processor 210 may call and run a computer program from the memory 230 to implement the method of the present invention.

[0301] The memory 230 may be a separate device independent of the processor 210 , or may be integrated into the processor 210 .

[0302] Alternatively, if Figure 2 As shown, the IoT early warning system 200 may further include a transceiver 220, and the processor 210 may control the transceiver 220 to interact with other devices, specifically, may send information or data to other devices, or receive information or data sent by other devices.

[0303] Optionally, the IoT early warning system 200 may implement the corresponding processes corresponding to the storage engine or components in the storage engine (such as a processing module) or a device deployed with a storage engine in each method of the present invention, which will not be described in detail here for the sake of brevity.

[0304] It should be understood that the processor of the present invention may be an integrated circuit chip having signal processing capabilities.

[0305] It is understood that the memory in the present invention can be a volatile memory or a non-volatile memory, or can include both volatile and non-volatile memory. It should be noted that the memory of the systems and methods described herein is intended to include but is not limited to suitable types of memory.

[0306] Based on the above, a readable storage medium is provided, on which a program or instruction is stored, and when the program or instruction is executed by a processor, the steps of the above method are implemented.

[0307] It should be noted that, in this article, the terms "comprise", "include" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the presence of other identical elements in the process, method, article or device including the element. In addition, it should be pointed out that the scope of the method and device in the embodiment of the present invention is not limited to performing functions in the order shown or discussed, and may also include performing functions in a substantially simultaneous manner or in reverse order according to the functions involved, for example, the described method may be performed in an order different from that described, and various steps may also be added, omitted, or combined. In addition, the features described with reference to certain examples may be combined in other examples.

[0308] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a computer software product, which is stored in a storage medium (such as ROM / RAM, a magnetic disk, or an optical disk), and includes a number of instructions for enabling a terminal (which can be a mobile phone, a computer, a server, or a network device, etc.) to execute the methods described in each embodiment of the present invention.

[0309] The embodiments of the present invention are described above in conjunction with the accompanying drawings, but the present invention is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of the present invention, ordinary technicians in this field can also make many forms without departing from the purpose of the present invention and the scope of protection of the present invention, all of which are within the protection of the present invention.

Claims

1. A method for early warning of remote supervision based on the Internet of Things, characterized in that: The method is applied to an Internet of Things early warning system, and the method comprises: Acquire the first IoT sensor state information and the first device response state information of the target special equipment system in the first fault warning task process, and acquire the second IoT sensor state information and the second device response state information of the associated special equipment system in the second fault warning task process, wherein the associated special equipment system and the target special equipment system are special equipment systems of the same category, and there is a difference between the first fault warning task process and the second fault warning task process; Associating the first IoT sensor state information with the second IoT sensor state information, performing state attribute identification to obtain first state attribute knowledge, and associating the first device response state information with the second device response state information, performing state attribute identification to obtain second state attribute knowledge; Integrate the first state attribute knowledge and the first device response state information and import them into the first fault warning algorithm for forward-looking decision analysis to obtain the target sensor state change characteristics of the target special equipment system in the target operation action link label; The second state attribute knowledge and the first IoT sensor state information are integrated and imported into the second fault warning algorithm for forward-looking decision analysis, and the operation response time efficiency characteristics of the target special equipment system in the target operation action link label are determined according to the decision analysis viewpoint of the second fault warning algorithm; Determine the potential fault hidden danger state of the target special equipment system in the target operation action link label according to the target sensor state change characteristics and the operation response time characteristics; The step of associating the first IoT sensor state information with the second IoT sensor state information and then performing state attribute identification to obtain first state attribute knowledge includes: Associating the first IoT sensor state information with the second IoT sensor state information to obtain a first sensor state information network; Performing feature mining on the first sensor state information network respectively through different feature mining modules to obtain a plurality of first sensor state features with consistent feature sizes; Respectively combining feature members with the same distribution label in each of the first sensor state features to obtain a plurality of first sensor state feature chains, and performing state attribute recognition on the plurality of first sensor state feature chains to obtain first state attribute knowledge; The step of performing state attribute recognition on a plurality of the first sensing state feature chains to obtain first state attribute knowledge includes: Importing a plurality of the first sensor state feature chains into a first long-term and short-term monitoring algorithm in sequence for processing, and obtaining first state trend description features corresponding to each of the first sensor state feature chains; Perform a feature focusing operation on each of the first state trend description features to obtain a key state trend coefficient corresponding to each of the first state trend description features, and integrate each of the first state trend description features according to the key state trend coefficient to obtain first state attribute knowledge.

2. The method for early warning based on the intelligent remote supervision of the Internet of Things as claimed in claim 1 is characterized in that: The number of the associated special equipment systems is several, and the second fault warning task processes corresponding to the associated special equipment systems are different. The first IoT sensor state information and the second IoT sensor state information are associated to obtain a first sensor state information network, including: Associating the second IoT sensor state information corresponding to each of the associated special equipment systems according to the security priority of each of the second fault warning task processes to obtain an original sensor state information network; Determine, according to the security priorities of the first fault warning task process and each of the second fault warning task processes, an adding area of ​​the first IoT sensor state information in the original sensor state information network; According to the adding area, the first IoT sensor state information is added to the original sensor state information network to obtain a first sensor state information network.

3. The method for early warning based on the intelligent remote supervision of the Internet of Things as claimed in claim 1 is characterized in that: The first state attribute knowledge and the first device response state information are integrated and imported into the first fault warning algorithm for forward-looking decision analysis to obtain the target sensor state change characteristics of the target special equipment system in the target operation action link label, including: Obtaining the first category quality inspection indicators corresponding to the target special equipment system; Performing quality inspection element identification on the first category of quality inspection indicators to obtain a first quality inspection element knowledge vector; The first quality inspection element knowledge vector, the first state attribute knowledge and the first equipment response state information are integrated and imported into the first fault warning algorithm for forward-looking decision analysis to obtain the target sensor state change characteristics of the target special equipment system in the target operation action link label.

4. The method for early warning based on the intelligent remote supervision of the Internet of Things according to claim 1 is characterized in that: The second state attribute knowledge and the first IoT sensor state information are integrated and imported into the second fault warning algorithm for forward-looking decision analysis, and the operation response time efficiency characteristics of the target special equipment system in the target operation action link label are determined according to the decision analysis viewpoint of the second fault warning algorithm, including: Obtaining the second category quality inspection indicators corresponding to the target special equipment system; Performing quality inspection element identification on the second category quality inspection indicators to obtain a second quality inspection element knowledge vector; The second quality inspection factor knowledge vector, the second state attribute knowledge and the first IoT sensor state information are integrated and imported into the second fault warning algorithm for forward-looking decision analysis, and the operational response time characteristics of the target special equipment system in the target operation action link label are determined based on the decision analysis viewpoint of the second fault warning algorithm.

5. The method for early warning based on the intelligent remote supervision of the Internet of Things as claimed in claim 4 is characterized in that: The second quality inspection factor knowledge vector, the second state attribute knowledge and the first IoT sensor state information are integrated and imported into the second fault warning algorithm for forward-looking decision analysis, and the operation response time efficiency characteristics of the target special equipment system in the target operation action link label are determined according to the decision analysis viewpoint of the second fault warning algorithm, including: Combining the second quality inspection factor knowledge vector, the second state attribute knowledge, and the first IoT sensor state information to obtain linkage warning information to be analyzed; Importing the linkage warning information to be analyzed into the second fault warning algorithm, performing discrimination processing on the linkage warning information to be analyzed according to the discrimination branch of the second fault warning algorithm to obtain a discrimination thermodynamic characteristic diagram, wherein the discrimination thermodynamic characteristic diagram is used to indicate the possibility weights of several preliminary screening equipment action analysis modules; Determine a target device action analysis module among the plurality of the preliminary screening device action analysis modules according to the discrimination thermodynamic characteristic diagram, import the first operation action link label of the first fault warning task process, the target operation action link label, and each operation action link label between the first operation action link label and the target operation action link label into the target device action analysis module, and obtain a plurality of target action response continuity evaluations; Based on a plurality of target action response continuity evaluations, the operational response time efficiency characteristics of the target special equipment system in the target operation action link label are determined.

6. The method for early warning based on the intelligent remote supervision of the Internet of Things according to claim 1 is characterized in that: The first fault warning algorithm is debugged based on the following steps: Obtain the third IoT sensor state information and the third device response state information of the first historical special equipment system in the third fault warning task process, and obtain the fourth IoT sensor state information of the second historical special equipment system in the fourth fault warning task process, wherein the second historical special equipment system and the first historical special equipment system are special equipment systems of the same category, the third fault warning task process is different from the fourth fault warning task process, the monitoring period of the third IoT sensor state information is equal to the step period between the target operation action link label and the first operation action link label of the first fault warning task process, and the monitoring period of the third device response state information and the monitoring period of the fourth IoT sensor state information are both equal to the monitoring period of the first IoT sensor state information; The first fault warning algorithm is debugged according to the third IoT sensor state information, the third device response state information, and the fourth IoT sensor state information.

7. The method for early warning based on the intelligent remote supervision of the Internet of Things according to claim 6 is characterized in that: The obtaining of the third IoT sensor state information and the third device response state information of the first historical special equipment system in the third fault warning task process includes: Obtaining the initial screening IoT sensor state information and the initial screening equipment response state information of the first historical special equipment system in the initial screening fault warning task process, wherein the monitoring period of the initial screening IoT sensor state information is greater than the step period between the target operation action link label and the first operation action link label, and the monitoring period of the initial screening equipment response state information is greater than the monitoring period of the first IoT sensor state information; Performing a roaming scan in the primary screening IoT sensor state information according to the first information scanning window to obtain the third IoT sensor state information, wherein the coverage size of the first information scanning window is smaller than the number of information units of the primary screening IoT sensor state information; The third device response status information is obtained by performing a roaming scan in the primary screening device response status information according to the second information scanning window, wherein the coverage size of the second information scanning window is smaller than the number of information units of the primary screening device response status information.

8. An Internet of Things early warning system, characterized in that: The method comprises at least one processor and a memory; the memory stores computer-executable instructions; the at least one processor executes the computer-executable instructions stored in the memory, so that the at least one processor executes the method according to any one of claims 1 to 7.

Images