Temperature monitoring system for steam turbine of thermal power generating unit

Abstract

The invention discloses a thermal power generating unit steam turbine temperature monitoring system which comprises a sensing layer, the sensing layer is connected with an edge calculation layer, the edge calculation layer is connected with a communication layer, the communication layer is connected with a data layer, the data layer is connected with an intelligent analysis layer, and the intelligent analysis layer is connected with a display and control layer. The system can improve the intelligent level and the overall performance of thermal power generating unit steam turbine temperature monitoring.

Description

Technical Field

[0001] The invention belongs to the technical field of temperature monitoring and relates to a temperature monitoring system for a steam turbine of a thermal power unit. Background Art

[0002] The steam turbine of a thermal power unit is a key equipment in a thermal power plant. It is used to convert the heat energy generated by combustion into mechanical energy, which in turn drives the generator to generate electricity. Its basic working principle is that the high-temperature and high-pressure steam generated by burning fuel (such as coal, natural gas or oil) passes through the blades of the steam turbine, causing it to rotate, thereby driving the generator to operate. The reason for temperature monitoring of the steam turbine of a thermal power unit is that the steam turbine will experience high temperature and high pressure conditions during operation. Temperature monitoring can ensure that the equipment operates within a safe range to avoid damage or failure caused by overheating. By monitoring the temperature, the operating parameters of the steam turbine can be optimized, thereby improving power generation efficiency and reducing fuel consumption. Abnormal temperature may be a precursor to equipment failure. Through real-time monitoring, problems can be discovered early, repaired or adjusted, and serious failures can be avoided. High temperature environment has an impact on various components of the steam turbine (such as blades, bearings, etc.). Temperature monitoring helps manage these impacts, thereby extending the service life of the equipment. The operation of thermal power units involves dangerous factors such as high temperature and high pressure. Ensuring that the temperature is within the specified range is an important measure to ensure the safety of operators. Therefore, temperature monitoring is an important part of the operation management of the steam turbine of a thermal power unit, which can help ensure the safety and economy of the equipment.

[0003] At present, the temperature monitoring of steam turbines in thermal power plants is mainly achieved by thermocouples and thermal resistors, infrared thermometers, distributed optical fiber temperature sensors and wireless temperature sensors. Among them, thermocouples and thermal resistors measure temperature based on the thermoelectric potential difference generated by two different metals when the temperature changes. The temperature is measured by using the property that the resistance of metal changes with temperature. Infrared thermometers use infrared radiation emitted by objects to measure temperature, and are mainly used for non-contact temperature measurement. Distributed optical fiber temperature sensors are based on the principle of optical fiber Brillouin scattering, and can perform distributed monitoring of temperature over a long distance. Wireless temperature sensors transmit temperature data to the central control system through wireless communication technology, which facilitates real-time monitoring and data collection. However, the single sensor type in these existing technologies usually relies on a single type of temperature sensor, such as thermocouples or thermal resistors, and the types of monitored data are limited, making it difficult to provide comprehensive equipment status information. In addition, the data collected by the sensor is usually transmitted directly to the central control system for processing, relying on centralized data processing methods, and there are data transmission delays and centralized processing bottlenecks. And it usually relies on traditional wired communications, with complex wiring and high costs, and it is difficult to adapt to dynamically changing industrial environments. Data security mainly relies on traditional encryption and firewall technologies, which may pose security risks in complex industrial environments. The intelligence level and overall performance of temperature monitoring need to be improved. Summary of the invention

[0004] The purpose of the present invention is to overcome the shortcomings of the above-mentioned prior art and provide a thermal power unit steam turbine temperature monitoring system, which can improve the intelligence level and overall performance of thermal power unit steam turbine temperature monitoring.

[0005] To achieve the above-mentioned purpose, the temperature monitoring system of a steam turbine of a thermal power unit described in the present invention includes a perception layer, the perception layer is connected to an edge computing layer, the edge computing layer is connected to a communication layer, the communication layer is connected to a data layer, the data layer is connected to an intelligent analysis layer, and the intelligent analysis layer is connected to a display and control layer.

[0006] Furthermore, the perception layer includes multimodal sensors, intelligent sensor nodes and self-powered sensors, and the multimodal sensors, intelligent sensor nodes and self-powered sensors are connected to the edge computing layer.

[0007] Furthermore, the edge computing layer includes edge computing nodes and a collaborative processing framework, and the edge computing nodes and the collaborative processing framework are connected to the communication layer and the perception layer.

[0008] Further, the communication layer includes a time-sensitive network and a hybrid network;

[0009] Furthermore, the data layer includes a distributed data lake system and a blockchain module;

[0010] The intelligent analysis layer includes self-learning AI models and multimodal data fusion;

[0011] The display and control layer includes a user interface and an online simulation and optimization module.

[0012] Furthermore, the multimodal sensor includes a thermocouple sensor, a thermal resistor sensor and an infrared sensor.

[0013] Furthermore, the intelligent sensor node is a sensor node in which a microprocessor is integrated, and the self-powered sensor is used to utilize heat energy and vibration energy in the environment to power the multi-modal sensor.

[0014] Furthermore, the edge computing node performs data processing and analysis through the edge computing device CPU, GPU or FPGA, uses the embedded computing device NVIDIA Jetson or Intel Movidius for calculation, the operating system adopts embedded Linux or real-time operating system, and the middleware uses EdgeX Foundry or KubeEdge open source middleware.

[0015] Furthermore, the collaborative processing framework is used to distribute data processing tasks to multiple edge computing nodes using distributed computing.

[0016] Furthermore, the distributed data lake system uses a combination of relational databases, NoSQL databases and time series databases.

[0017] The present invention has the following beneficial effects:

[0018] The thermal power unit steam turbine temperature monitoring system of the present invention adopts multiple types of sensors during specific operation to ensure the accuracy and wide coverage of temperature measurement, provide comprehensive equipment status information, overcome the limitations of a single sensor type, integrate a microprocessor in the sensor node, can perform preliminary processing and screening of data, improve data quality, have preliminary early warning capabilities, can issue alarms in a timely manner, use environmental energy, thermal energy, and vibration energy for power supply, reduce external power dependence, improve flexibility and adaptability, use edge computing nodes for preliminary data processing, reduce centralized processing pressure, reduce data transmission delays, and use distributed computing and collaborative processing frameworks to improve system processing efficiency and fault tolerance. Combine wired and wireless communication technologies to build a stable and flexible hybrid network, time Sensitive network technology ensures the certainty and real-time nature of data transmission, uses AES or TLS encryption algorithms to ensure data transmission security, implements strong authentication mechanisms and multi-level security measures to protect data privacy and security, blockchain technology ensures the immutability of data storage, provides reliable timestamps and data source verification, and smart contracts enable automated data management to ensure rule execution. Self-learning AI models and multimodal data fusion technology improve the accuracy of monitoring and prediction, collect and process new data in real time, and achieve intelligent decision-making and adaptive optimization. The system significantly improves the intelligence level and overall performance of temperature monitoring of steam turbines in thermal power units through the integration of multiple technical means such as multimodal sensing, edge computing, efficient communication, data security, intelligent analysis, display and control. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 It is a schematic diagram of the system framework flow of the present invention.

[0020] Among them, 1 is the perception layer, 2 is the edge computing layer, 3 is the communication layer, 4 is the data layer, 5 is the intelligent analysis layer, 6 is the display and control layer, 7 is the multimodal sensor, 8 is the intelligent sensor node, 9 is the self-powered sensor, 10 is the time-sensitive network, 11 is the hybrid network, 12 is the self-learning AI model, 13 is the multimodal data fusion, 14 is the online simulation and optimization module, 15 is the user interface, 16 is the blockchain module, 17 is the distributed data lake system, 18 is the collaborative processing framework, and 19 is the edge computing node. DETAILED DESCRIPTION

[0021] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiment of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiment of the present invention. Obviously, the described embodiment is only an embodiment of a part of the present invention, not all embodiments, and is not intended to limit the scope of the present invention. In addition, in the following description, the description of well-known structures and technologies is omitted to avoid unnecessary confusion of the concepts disclosed in the present invention. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work should fall within the scope of protection of the present invention.

[0022] The accompanying drawings show schematic diagrams of structures according to embodiments disclosed in the present invention. These figures are not drawn to scale, and some details are magnified and some details may be omitted for the purpose of clear expression. The shapes of various regions and layers shown in the figures and the relative sizes and positional relationships therebetween are only exemplary, and may deviate in practice due to manufacturing tolerances or technical limitations, and those skilled in the art may additionally design regions / layers with different shapes, sizes, and relative positions according to actual needs.

[0023] Embodiment 1

[0024] refer to Figure 1 The temperature monitoring system of a steam turbine of a thermal power unit described in the present invention includes a perception layer 1, the perception layer 1 is connected to an edge computing layer 2, the edge computing layer 2 is connected to a communication layer 3, the communication layer 3 is connected to a data layer 4, the data layer 4 is connected to an intelligent analysis layer 5, and the intelligent analysis layer 5 is connected to a display and control layer 6; the perception layer 1 includes a multimodal sensor 7, an intelligent sensor node 8 and a self-powered sensor 9, and the multimodal sensor 7, the intelligent sensor node 8 and the self-powered sensor 9 are connected to the edge computing layer 2; the edge computing layer 2 includes an edge computing node 19 and a collaborative processing framework 18, and the edge computing node 19 and the collaborative processing framework 18 are connected to the communication layer 3 and the perception layer 1.

[0025] Embodiment 2

[0026] refer to Figure 1 The temperature monitoring system for a steam turbine of a thermal power unit described in the present invention includes a perception layer 1, the perception layer 1 is connected to an edge computing layer 2, the edge computing layer 2 is connected to a communication layer 3, the communication layer 3 is connected to a data layer 4, the data layer 4 is connected to an intelligent analysis layer 5, and the intelligent analysis layer 5 is connected to a display and control layer 6.

[0027] As an implementation mode of the present invention, the perception layer 1 includes a multimodal sensor 7 , an intelligent sensor node 8 and a self-powered sensor 9 .

[0028] Multimodal sensor 7 uses multiple types of temperature sensors, such as thermocouples, thermal resistors, and infrared sensors, to ensure the accuracy and wide coverage of temperature measurement. Thermocouples are suitable for high temperature measurement, thermal resistors are suitable for medium and low temperature measurement, and infrared sensors can be used for non-contact temperature monitoring to provide comprehensive temperature data.

[0029] The intelligent sensor node 8 integrates a microprocessor in the sensor node, which can perform preliminary processing and screening on the collected raw data. Specifically, it uses filtering algorithms to remove noise in the sensor data and improve data quality. Some computing tasks are completed in the sensor node, such as simple trend analysis and anomaly detection. The intelligent node has preliminary early warning capabilities. When abnormal data (such as excessive temperature and abnormal vibration) is detected, it can issue an alarm in time and trigger the alarm through predefined rules. A simple model trained based on historical data is used to identify potential failure modes.

[0030] The self-powered sensor 9 uses energy in the environment (such as heat energy and vibration energy) to power the sensor, reducing the reliance on external power supplies and improving the flexibility of the sensor node. This self-powered design enables flexible deployment of sensors without additional wiring, adapting to changing industrial environments, and reducing maintenance costs.

[0031] As an implementation mode of the present invention, the edge computing layer 2 includes an edge computing node 19 and a collaborative processing framework 18 .

[0032] Among them, the edge computing node 19 selects the edge computing device CPU, GPU or FPGA for data processing and analysis, uses the embedded computing device NVIDIA Jetson or Intel Movidius for calculation, the operating system adopts embedded Linux or real-time operating system, and the middleware uses EdgeX Foundry or KubeEdge open source middleware to provide device management and data processing.

[0033] The edge computing node 19 performs preliminary processing on the data from the perception layer 1 to remove noise, fill in missing values, and perform standardization. Specifically, the filtering algorithm uses a low-pass filter or a Kalman filter algorithm to remove noise from the sensor data. Data correction is based on known sensor characteristics to correct the data to ensure data accuracy. Real-time data analysis uses the computing power of edge computing devices to perform real-time analysis on pre-processed data. Fast Fourier transform (FFT) is used to analyze the frequency domain of vibration data to detect abnormal vibration patterns. Through sliding windows and trend detection algorithms, temperature changes are monitored in real time to identify abnormal temperature rise.

[0034] The collaborative processing framework 18 uses distributed computing to distribute data processing tasks to multiple edge computing nodes 19 to improve processing efficiency and system fault tolerance. Use the distributed task scheduling system Kubernetes to manage the distribution and execution of computing tasks. Dynamically adjust task allocation according to the load of the computing node to ensure that the resource utilization of each node is maximized. Realize data and computing collaboration between multiple edge computing nodes 19 to improve the processing power of the overall system. Data sharing between edge nodes is achieved through high-speed local area networks to facilitate collaborative analysis and decision-making. Finally, based on the distributed AI reasoning framework TensorFlow Lite or ONNX Runtime, edge reasoning of complex AI models is realized.

[0035] Deploy pre-trained AI models on edge computing nodes 19 to achieve local intelligent decision-making. Based on the fault detection model trained with historical data, analyze sensor data in real time to identify potential faults. Through the prediction model, estimate the remaining life and maintenance requirements of the equipment to reduce unplanned downtime. Implement the early warning mechanism on the edge node, and trigger the early warning immediately when an abnormal situation is detected. Trigger alarms based on predefined thresholds, such as temperature exceeding the limit and vibration abnormality. Use pattern recognition algorithms to identify complex abnormal patterns and provide more accurate early warnings.

[0036] As an implementation mode of the present invention, the communication layer 3 includes a time-sensitive network 10 and a hybrid network 11 .

[0037] Combining wired and wireless communications, a stable and flexible hybrid network architecture is constructed, where a local area network is used to transmit data between edge computing nodes and sensor nodes, ensuring high bandwidth and low latency. A wide area network is used to transmit data to a central control system for remote monitoring and data analysis.

[0038] TSN technology is used in Time Sensitive Network 10 to ensure the determinism and real-time nature of data transmission in industrial environments, and TSN time synchronization and traffic scheduling are configured to ensure that the transmission delay of key data meets system requirements. Lightweight edge transmission protocols MQTT or CoAP are used to achieve efficient edge data transmission and processing.

[0039] Encryption is required during data transmission, using encryption algorithms such as AES or TLS to ensure the security and integrity of data during transmission. A strong authentication mechanism is implemented during encrypted transmission to ensure that only authorized devices and users can access the communication network and data. Digital certificates and two-factor authentication technology are used to improve system security. End-to-end encryption is performed on the transmitted data to prevent data theft and tampering during transmission.

[0040] Deploy a network management system to monitor the operation status of the network in real time, manage network equipment and communication links, realize network topology visualization, fault location and performance optimization, ensure network stability and efficient operation, use traffic monitoring tools such as NetFlow or sFlow to monitor network traffic in real time, analyze data transmission performance, configure alarm mechanisms, and issue alarms and take measures in a timely manner when abnormal situations occur in the network, such as sudden increase in traffic and transmission delay.

[0041] As an embodiment of the present invention, the data layer 4 includes a distributed data lake system 17 and a blockchain module 16.

[0042] Combine relational databases, NoSQL databases, and time series databases to build a data lake to store raw data from different sources, support a variety of data types, and use Apache Hadoop or Amazon S3 platforms to achieve centralized data storage and management. Maintain data metadata and record data sources, structures, and update time information through Apache Atlas to support data management and tracking. Implement automated metadata management. Use data quality tools such as Talend or Informatica to implement data quality monitoring and verification to ensure data integrity, consistency, and accuracy. Perform data cleaning and conversion.

[0043] Define data lifecycle policies and manage data creation, storage, archiving, and deletion processes. Use automated tools to manage data lifecycles and optimize storage resource utilization. Use Apache Hadoop or Spark platforms for big data batch processing. Deploy Apache Kafka or Apache Flink stream processing frameworks to achieve real-time data processing and analysis. Use data analysis tools Apache Hive or Presto for data query and analysis. Deploy data mining and machine learning platforms Apache Mahout or TensorFlow to support advanced data analysis and model training.

[0044] The blockchain module 16 is used to ensure data storage and immutability. The blockchain platform selects one or a combination of public chain, alliance chain, etc. The key monitoring data is written into the blockchain to ensure that the data cannot be tampered with once written, and reliable timestamps and data source verification are provided. Smart contracts are used to realize automated data management to ensure the execution of rules during data writing, updating and access. Data stored on the blockchain is encrypted to ensure that the data cannot be read without authorization. The TLS encryption protocol is used to protect the security of data during transmission.

[0045] Through smart contracts and multi-signature mechanisms, we control the access rights of different users and systems to data, ensuring that only authorized users can access and operate data. We use public key infrastructure (PKI) for identity authentication to ensure the authenticity and legitimacy of the visitor's identity. We desensitize data before writing it into the blockchain to hide sensitive information to protect privacy. Through anonymization technology, we prevent individuals or specific devices from being identified through data tracing, protecting privacy information.

[0046] As an embodiment of the present invention, the intelligent analysis layer 5 includes a self-learning AI model 12 and multimodal data fusion 13.

[0047] Collect and organize historical data, including temperature, pressure, and vibration sensor data, as the initial training data set. Perform data cleaning, preprocessing, and feature engineering to ensure data quality and feature relevance. Select machine learning algorithms such as deep neural networks and gradient boosting decision trees to train the initial model. Use cross-validation and validation sets to evaluate model performance and tune model parameters to achieve the best results. Collect and process new data in real time, including temperature, pressure, and vibration data under various operating conditions. Implement real-time processing and feature extraction of data streams to ensure that the model can obtain the latest information in a timely manner.

[0048] Use incremental learning algorithms such as online gradient descent or online K-means to update model weights and parameters. Adjust model structure or introduce new features based on the characteristics of new data and feedback to enhance the model's adaptability. Combine known labeled data with unlabeled data and use self-supervised learning methods, autoencoders or contrastive learning to further optimize the model. Automatically annotate unlabeled data to enrich training data sets. Continuously monitor the model's predictive performance, including accuracy, recall, and F1 score metrics. Implement anomaly detection mechanisms to detect outliers and confidence levels in model outputs and evaluate the reliability of the model.

[0049] Multimodal data fusion13 refers to the fusion of data from different types of sensors, such as temperature, pressure, vibration, etc., through reasonable algorithms and technologies to improve the accuracy of analysis and prediction. The following is a specific implementation plan: Unify the timestamps of different sensor data to ensure data synchronization. Deal with data loss and inconsistency issues, and complete and correct data through interpolation, smoothing and other technologies. Extract key features from different sensor data, such as the rate of change of temperature, the amplitude of pressure fluctuations, and the spectral characteristics of vibration. Use dimensionality reduction technology PCA to reduce feature dimensions and reduce model complexity.

[0050] Concatenate or weighted average the data features of different sensors to form a unified feature vector input model. Use multi-layer perceptron and neural network models to fuse multimodal features for learning. Independently train models for different modal data (such as temperature model, pressure model, vibration model), and then fuse the outputs of each model. Use weighted voting and confidence weighting methods to combine the prediction results of each model to make the final decision.

[0051] Use the deep learning model Long Short-Term Memory Network (LSTM) to process time series and spatial features simultaneously. Achieve the fusion of different modal data at different levels to capture complex nonlinear relationships and long-term and short-term dependencies. Use the cross-validation method to evaluate the performance of the multimodal fusion model to ensure the generalization ability of the model. Compare the performance of the single-modal model and the multimodal fusion model to verify the effectiveness of the fusion.

[0052] Self-learning AI models 12 and multimodal data fusion 13 are important components of the intelligent analysis layer 5. Through advanced learning and data fusion technologies, the system can more accurately monitor and predict equipment status, achieve adaptive optimization and intelligent decision-making. The implementation of these technologies will significantly improve the intelligence level and overall performance of the thermal power unit steam turbine temperature monitoring system.

[0053] As an implementation mode of the present invention, the display and control layer 6 includes a user interface 15 and an online simulation and optimization module 14 .

[0054] A user interface 15 is provided to provide visual monitoring, remote control and fault diagnosis of the system operation status. An online simulation and optimization module 14 is used in combination with digital twin technology to support online simulation and optimization, detect potential problems in advance and make optimization suggestions.

[0055] Data flow and communication mechanisms

[0056] Between the data layer 4 and the intelligent analysis layer 5, data flows and interacts through the edge computing nodes 19 and the communication network. After receiving the data from the perception layer, the edge computing nodes 19 perform preliminary processing and analysis, and then upload the data to the data layer 4 for storage. The intelligent analysis layer 5 obtains data from the data layer 4 for in-depth analysis and modeling, and the generated prediction results and decision information are then fed back to the display and control layer 6 through the communication layer for user reference.

[0057] Security and Privacy Measures

[0058] During data transmission and storage, the system adopts multi-level security and privacy protection measures. All data transmission is encrypted by AES or TLS to ensure the security and integrity of data during transmission. At the same time, the system also implements a strong authentication mechanism, using digital certificates and two-factor authentication technology to ensure that only authorized devices and users can access the communication network and data. In order to protect data privacy, data is desensitized before being written to the blockchain, and sensitive information is hidden through anonymization technology to prevent data tracing from identifying individuals or specific devices.

[0059] Real-time performance and fault tolerance

[0060] In order to ensure the real-time performance and high fault tolerance of the system, time-sensitive network 10 (TSN) technology is introduced to ensure the determinism and real-time performance of data transmission. When a network failure or node failure occurs, the system distributes data processing tasks to other edge computing nodes 19 through a distributed computing and collaborative processing framework 18 to ensure the continuous operation and efficient processing of the system.

[0061] Explanation of terms and abbreviations

[0062] To make it easier for non-specialist readers to understand the technical details, a glossary and notes will be provided in the document.

[0063] For example:

[0064] TSN: Time Sensitive Network10, FFT: Fast Fourier Transform, PCA: Principal Component Analysis, AES: Advanced Encryption Standard, TLS: Transport Layer Security, MQTT: Message Queuing Telemetry Transport Protocol, CoAP: Constrained Application Protocol, PKI: Public Key Infrastructure.

[0065] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A thermal power unit steam turbine temperature monitoring system, characterized in that: The invention comprises a perception layer (1), wherein the perception layer (1) is connected to an edge computing layer (2), wherein the edge computing layer (2) is connected to a communication layer (3), wherein the communication layer (3) is connected to a data layer (4), wherein the data layer (4) is connected to an intelligent analysis layer (5), and wherein the intelligent analysis layer (5) is connected to a display and control layer (6).

2. The thermal power unit steam turbine temperature monitoring system according to claim 1, characterized in that: The perception layer (1) comprises a multimodal sensor (7), an intelligent sensor node (8) and a self-powered sensor (9), and the multimodal sensor (7), the intelligent sensor node (8) and the self-powered sensor (9) are connected to the edge computing layer (2).

3. The thermal power unit steam turbine temperature monitoring system according to claim 1, characterized in that: The edge computing layer (2) includes edge computing nodes (19) and a collaborative processing framework (18), and the edge computing nodes (19) and the collaborative processing framework (18) are connected to the communication layer (3) and the perception layer (1).

4. The thermal power unit steam turbine temperature monitoring system according to claim 1, characterized in that: The communication layer (3) includes a time-sensitive network (10) and a hybrid network (11).

5. The thermal power unit steam turbine temperature monitoring system according to claim 1, characterized in that: The data layer (4) includes a distributed data lake system (17) and a blockchain module (16); The intelligent analysis layer (5) includes a self-learning AI model (12) and multimodal data fusion (13); The display and control layer (6) includes a user interface (15) and an online simulation and optimization module (14).

6. The thermal power unit steam turbine temperature monitoring system according to claim 2, characterized in that: The multimodal sensor (7) comprises a thermocouple sensor, a thermal resistance sensor and an infrared sensor.

7. The thermal power unit steam turbine temperature monitoring system according to claim 1, characterized in that: The intelligent sensor node (8) is a sensor node in which a microprocessor is integrated, and the self-powered sensor (9) is used to utilize heat energy and vibration energy in the environment to power the multi-modal sensor (7).

8. The thermal power unit steam turbine temperature monitoring system according to claim 1, characterized in that: The edge computing node (19) performs data processing and analysis through edge computing devices such as CPU, GPU or FPGA, performs calculations using embedded computing devices such as NVIDIA Jetson or Intel Movidius, uses an embedded Linux or real-time operating system as an operating system, and uses EdgeX Foundry or KubeEdge open source middleware as a middleware.

9. The thermal power unit steam turbine temperature monitoring system according to claim 1, characterized in that: The collaborative processing framework (18) is used to distribute data processing tasks to multiple edge computing nodes (19) using distributed computing; The distributed data lake system (17) is a combination of a relational database, a NoSQL database and a time series database.

10. A thermal power unit steam turbine temperature monitoring system, characterized in that: The system comprises a perception layer (1), wherein the perception layer (1) is connected to an edge computing layer (2), wherein the edge computing layer (2) is connected to a communication layer (3), wherein the communication layer (3) is connected to a data layer (4), wherein the data layer (4) is connected to an intelligent analysis layer (5), and wherein the intelligent analysis layer (5) is connected to a display and control layer (6); The perception layer (1) comprises a multimodal sensor (7), an intelligent sensor node (8) and a self-powered sensor (9), and the multimodal sensor (7), the intelligent sensor node (8) and the self-powered sensor (9) are connected to the edge computing layer (2); The edge computing layer (2) includes edge computing nodes (19) and a collaborative processing framework (18), and the edge computing nodes (19) and the collaborative processing framework (18) are connected to the communication layer (3) and the perception layer (1).

Images