Online monitoring system and method for operation state of hot water boiler

Abstract

The invention provides a hot water boiler operation state on-line monitoring system and method, and relates to the field of equipment intelligent monitoring, and the method comprises the steps: firstly, collecting time queue data of boiler operation parameters including a temperature value and a pressure value through a sensor assembly installed at a key part of a boiler; time queue data of the parameters are subjected to data encryption and compression to form a boiler operation parameter compressed data packet, the boiler operation parameter compressed data packet is sent to the central monitoring server through an MQTT protocol, the received compressed data packet is decompressed and decrypted at the central monitoring server end to be restored into original time queue data, and finally, the boiler operation parameter is obtained based on the data. And the central monitoring server analyzes and determines the operation state of the hot water boiler, and displays an abnormal early warning prompt once an abnormal state is found. In this way, abnormal conditions in the operation of the boiler can be found in time, and then safe and stable operation of the boiler can be ensured.

Description

Technical Field

[0001] The present application relates to the field of intelligent equipment monitoring, and more specifically, to an online monitoring system and method for the operation status of a hot water boiler. Background Art

[0002] Hot water boilers are key equipment in industrial production, heating systems and many other applications. Their stability and safety are directly related to the normal production and life. With the development of modern industry and the advancement of technology, higher requirements are placed on the safety, efficiency and maintenance of hot water boilers.

[0003] However, traditional hot water boiler monitoring methods mainly rely on threshold alarm systems based on preset rules. Specifically, the alarm mechanism based on fixed thresholds is too simple, difficult to adapt to complex operating environments, and prone to false alarms or missed alarms. In addition, this method cannot capture subtle and long-term trends in real time, such as gradually deteriorating equipment conditions or nonlinear performance degradation, which are often precursors to major failures. Moreover, due to the lack of in-depth data mining, traditional methods have difficulty identifying potential safety hazards, especially when faced with complex and changing industrial environments, which not only reduces the reliability of the system, but also increases operational risks and maintenance costs.

[0004] Therefore, there is an urgent need for an online monitoring solution for the operating status of a hot water boiler, which uses more intelligent and efficient monitoring technology to perform real-time monitoring to ensure the safe and stable operation of the hot water boiler. Summary of the invention

[0005] In order to solve the above technical problems, the present application is proposed. The embodiments of the present application provide a system and method for online monitoring of the operation status of a hot water boiler.

[0006] According to one aspect of the present application, a hot water boiler operating status online monitoring system is provided, which includes: a data acquisition module, used to collect boiler operating parameters collected by sensor components installed at key positions of the boiler to obtain a time queue of boiler operating parameters, wherein the boiler operating parameters include temperature values ​​and pressure values; a data transmission module, used to first encrypt and compress the time queue of the boiler operating parameters into a compressed data packet of boiler operating parameters, and transmit the compressed data packet of boiler operating parameters to a central monitoring server through the MQTT protocol; a data processing and analysis module, used to decompress and decrypt the compressed data packet of boiler operating parameters on the central monitoring server to obtain the time queue of the boiler operating parameters, and determine whether the operating status of the hot water boiler exists based on the time queue of the boiler operating parameters. Abnormality; a display and alarm module, which is used to display the abnormal status warning prompt; wherein, the data processing and analysis module includes: an operating parameter splitting unit, which is used to split the time queue of the boiler operating parameters to obtain the time queue of the temperature value and the time queue of the pressure value; an operating parameter encoding and strengthening unit, which is used to perform sequence encoding and self-attention strengthening on the time queue of the temperature value and the time queue of the pressure value respectively to obtain the temperature time series autocorrelation strengthening feature and the pressure time series autocorrelation strengthening feature; an operating parameter adjustment collaborative unit, which is used to perform external knowledge feature collaborative strengthening on the temperature time series autocorrelation strengthening feature and the pressure time series autocorrelation strengthening feature to obtain the boiler state time series collaborative strengthening representation, and judge whether the operating state of the hot water boiler is abnormal based on the boiler state time series collaborative strengthening representation.

[0007] According to another aspect of the present application, a method for online monitoring of the operating status of a hot water boiler is provided, and the method is used to execute the online monitoring system for the operating status of a hot water boiler as described above, which includes: performing data splitting on the time queue of the boiler operating parameters to obtain a time queue of temperature values ​​and a time queue of pressure values; performing sequence encoding and self-attention enhancement on the time queue of the temperature values ​​and the time queue of the pressure values ​​respectively to obtain temperature time series autocorrelation enhancement features and pressure time series autocorrelation enhancement features; performing external knowledge feature collaborative enhancement on the temperature time series autocorrelation enhancement features and the pressure time series autocorrelation enhancement features to obtain a boiler state time series collaborative enhancement representation, and judging whether the operating status of the hot water boiler is abnormal based on the boiler state time series collaborative enhancement representation.

[0008] Compared with the prior art, the online monitoring system and method for the operation status of a hot water boiler provided by the present application first collects the time queue data of the boiler operation parameters including the temperature value and the pressure value through the sensor components installed at the key parts of the boiler. Subsequently, the time queue data of these parameters are encrypted and compressed to form a compressed data packet of the boiler operation parameters, and sent to the central monitoring server through the MQTT protocol. On the central monitoring server side, the received compressed data packet is decompressed and decrypted to restore it to the original time queue data. Finally, based on these data, the central monitoring server analyzes and determines the operation status of the hot water boiler. Once an abnormal state is found, an abnormal warning prompt is displayed. In this way, it is helpful to timely discover abnormal conditions in the operation of the boiler, thereby ensuring the safe and stable operation of the boiler. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0010] In the attached picture: Figure 1 4 is a system block diagram of an online monitoring system for the operation status of a hot water boiler according to an embodiment of the present application.

[0011] Figure 2 The block diagram is a data processing and analysis module in the online monitoring system of the hot water boiler operation status according to an embodiment of the present application.

[0012] Figure 3 It is a block diagram of an operating parameter adjustment coordination unit in an online monitoring system for the operating status of a hot water boiler according to an embodiment of the present application.

[0013] Figure 4 The present invention is a flowchart of a method for online monitoring of the operating status of a hot water boiler according to an embodiment of the present application. DETAILED DESCRIPTION

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

[0015] As a core device in industrial production, heating systems and many other application scenarios, the stability and safety of hot water boiler operation are crucial to ensuring smooth production activities and normal daily life. With the vigorous development of modern industry and the continuous innovation of technology, hot water boilers are facing more stringent requirements in terms of safety, efficiency and maintenance.

[0016] However, in the past, the monitoring methods for hot water boilers mainly relied on the threshold alarm system based on preset rules. This traditional method has obvious limitations. Specifically, the alarm mechanism based on fixed thresholds is too simple and difficult to effectively cope with the complex and changing operating environment, so it is very easy to have false alarms or missed alarms. In addition, this method is unable to capture those subtle and long-term trends in real time, such as the gradual decline in equipment performance or nonlinear decay, which are often early warning signals before major failures occur. More importantly, due to the lack of in-depth mining and analysis of data, traditional methods are difficult to accurately identify potential safety hazards in complex and changing industrial environments. This not only weakens the overall reliability of the system, but also further increases the risk of operation and the cost of maintenance.

[0017] Based on this, the present application proposes an online monitoring system for the operation status of a hot water boiler. Figure 1 FIG. 1 is a system block diagram of an online monitoring system for the operation status of a hot water boiler according to an embodiment of the present application. Figure 1 As shown, in the hot water boiler operating status online monitoring system 100, it includes: a data acquisition module 110, which is used to collect boiler operating parameters collected by sensor components installed at key positions of the boiler to obtain a time queue of boiler operating parameters, wherein the boiler operating parameters include temperature values ​​and pressure values; a data transmission module 120, which is used to first encrypt and compress the time queue of the boiler operating parameters into a compressed data packet of boiler operating parameters, and transmit the compressed data packet of boiler operating parameters to a central monitoring server through the MQTT protocol; a data processing and analysis module 130, which is used to decompress and decrypt the compressed data packet of boiler operating parameters on the central monitoring server to obtain the time queue of the boiler operating parameters, and determine whether there is an abnormality in the operating status of the hot water boiler based on the time queue of the boiler operating parameters; a display and alarm module 140, which is used to display an abnormal status warning prompt.

[0018] In the embodiment of the present application, the data acquisition module 110 is used to collect the boiler operating parameters collected by the sensor assembly installed at the key part of the boiler to obtain the time queue of the boiler operating parameters, and the boiler operating parameters include temperature values ​​and pressure values. It should be understood that the temperature reflects the thermal energy state in the boiler and the efficiency of the heat exchange process. For example, in a normally operating hot water boiler, the water temperature needs to be maintained within a reasonable range to ensure that the heat can be effectively transferred to the required place. The pressure is closely related to the fluid mechanics state inside the boiler, which can reflect the capacity of the water in the boiler, the circulation state and the smoothness of the pipeline system. Too high or too low pressure may indicate potential problems, such as pipe blockage, leakage or water pump failure. Considering that the operation of the hot water boiler is a dynamic process, the temperature and pressure will change over time. In order to capture the operating state of the hot water boiler under different working conditions, it is necessary to collect the time queue data of the temperature value and the pressure value. For example, during the boiler startup phase, the temperature and pressure will gradually rise; during stable operation, they should remain relatively stable; and when a fault occurs or is subject to external interference, the temperature and pressure may show abnormal fluctuations or change trends. In general, by analyzing the time queue data of the collected temperature and pressure values, potential problems in the operation of the hot water boiler can be discovered in a timely manner.

[0019] In an embodiment of the present application, the data transmission module 120 is used to first encrypt and compress the time queue of the boiler operating parameters into a compressed data packet of the boiler operating parameters, and transmit the compressed data packet of the boiler operating parameters to the central monitoring server through the MQTT protocol. It should be understood that in order to prevent the operating parameter data of the hot water boiler from being accessed, stolen or tampered with by an unauthorized third party during the transmission process, it is necessary to encrypt the time queue of the boiler operating parameters. Even if the encrypted data is intercepted during the transmission process, the attacker cannot understand its content, which greatly improves the security of the data. Accordingly, considering that the time queue of the boiler operating parameters is a continuous data sequence. As time goes by and the frequency of acquisition increases, the amount of data will become very large. For example, if the temperature and pressure sensors collect data once every 10 seconds, a large number of data points will be generated in one hour. When transmitting these data, if they are not compressed, a large amount of network bandwidth will be occupied. In actual industrial environments, network resources are usually limited. There may be multiple devices that need to transmit data at the same time, and the network bandwidth will also be limited by factors such as cost and infrastructure. A large amount of uncompressed boiler operation data may cause network congestion, affect the data transmission of other devices, and may even cause data transmission failure. In order to reduce the volume of data and improve the efficiency of data transmission, it is necessary to compress the time queue of the boiler operating parameters to obtain a compressed data packet of the boiler operating parameters. By compression, the bandwidth required for data transmission can be significantly reduced. This enables faster data transmission under limited network resources. At the same time, a smaller amount of data means a shorter data transmission time, which is very important for real-time monitoring of the boiler operating status, because timely data transmission can ensure that the central monitoring server obtains the latest data for analysis faster and reduces the impact of data transmission delay on the timeliness of fault monitoring. MQTT is a lightweight message transmission protocol with a very small message header and low protocol overhead, which is suitable for use in resource-constrained IoT devices (such as sensors). Compared with other complex communication protocols, MQTT can more effectively utilize network resources when transmitting small data packets (such as compressed boiler operating parameter data packets) and reduce resource waste in the communication process. At the same time, due to the high efficiency of the MQTT protocol itself, it occupies relatively less bandwidth when transmitting data. The use of the MQTT protocol can ensure the reliable transmission of the compressed data packet of the boiler operating parameters without occupying too much bandwidth, which is conducive to real-time and continuous data transmission. Since the central monitoring server usually has more powerful computing resources and storage capabilities. Compared with sensors or other edge devices, it can run more complex data analysis algorithms, such as deep mining of time queue data, building complex models to determine boiler operation status, etc. In other words, by transmitting boiler operation data to the central monitoring server, it can make full use of the advantages of the server and more accurately determine whether there is an abnormality in the boiler.

[0020] The following is a detailed process of implementing "first encrypting and compressing the time queue of the boiler operating parameters into a compressed data packet of the boiler operating parameters, and transmitting the compressed data packet of the boiler operating parameters to the central monitoring server through the MQTT protocol": Data encryption is the primary link to ensure the secure transmission of boiler operating parameter data. In terms of encryption algorithm selection, multiple factors need to be considered comprehensively. Symmetric encryption algorithms (such as AES algorithm) have the advantage of fast encryption speed and are suitable for scenarios with high encryption speed requirements and relatively simple key management. For example, in the hot water boiler monitoring system of a small factory, the data volume is not large and the network environment is stable. The AES algorithm can quickly encrypt the collected temperature and pressure value time queue data. In the implementation process, it is necessary to use the key management system (KMS) to generate a unique key for each boiler or monitoring node, and store it securely in a local encrypted storage device (such as a hardware security module HSM). At the same time, the corresponding decryption key must also be properly stored on the central monitoring server. During data transmission, it is important to ensure that the key is transmitted in a secure channel to prevent key leakage. Asymmetric encryption algorithms (such as RSA algorithm) are suitable for multi-device interaction and complex key distribution management. Taking the boiler monitoring system of a large enterprise as an example, boilers in multiple departments or locations transmit data to the central monitoring server, and the communication security requirements are high. The use of the RSA algorithm can better manage the key. When the device is registered to the central monitoring server, the public key is transmitted to the server. The server uses the public key to encrypt the returned data or instructions, and the device uses the private key to decrypt, thereby ensuring the security and reliability of data transmission.

[0021] The data compression link is crucial to reducing the data storage space and transmission bandwidth requirements. Lossless compression algorithms are ideal for processing boiler operating parameter time queue data, among which the LZ77 algorithm is more commonly used. This algorithm achieves compression by identifying and utilizing repeated strings in the data. It can effectively reduce the amount of data when the temperature and pressure values ​​change little and there are more repeated data during the stable operation stage of the boiler. For example, at the data acquisition terminal at the boiler site, after a certain amount of time queue data (such as data collected every 100 data points or every 1 second) is collected, the LZ77 algorithm is immediately run for real-time compression, and the compressed data is temporarily stored in the local cache. At the same time, adaptive compression algorithms such as the Deflate algorithm that combines the LZ77 algorithm with Huffman coding are also quite advantageous. It can dynamically adjust the compression strategy according to the actual data situation, and can achieve good compression effects for boiler operating parameter data under different working conditions. In the compression process, in addition to selecting a suitable algorithm, it is also necessary to encapsulate the compressed data, add necessary header information such as data source (boiler number or location), collection time range, and compression algorithm identifier, and form a boiler operating parameter compressed data packet so that the data can be correctly parsed and decompressed at the transmission and receiving ends.

[0022] MQTT protocol transmission is the key to achieve efficient and reliable data transmission from the boiler to the central monitoring server. On the data acquisition device or edge computing node at the boiler end, the MQTT client software or library needs to be installed and relevant configurations need to be performed. Configuration parameters include server address (can be IP address or domain name), port number (default 1883, can also be customized as needed), user name and password (used for authentication and authorization to ensure data transmission security), etc. For example, in the factory's hot water boiler monitoring system, the data collector at the boiler end is configured as an MQTT client, connected to the central monitoring server in the enterprise's internal network, and the assigned user name and password are used for authentication. Only authorized devices can transmit data to the server. At the same time, MQTT server software (such as Mosquitto, EMQ X, etc.) must be installed on the central monitoring server and a series of configurations must be performed. This includes setting the listening port, creating user authentication and authorization rules, setting different access rights according to device type or region, and configuring message storage policies (such as whether to store historical messages, storage duration, etc.). For example, after the enterprise central monitoring server installs the Mosquitto server, it is configured to allow connection requests from various boiler equipment in the factory, and at the same time set up a message storage strategy to facilitate historical storage of important boiler operating parameter data for subsequent query and analysis.

[0023] During the data transmission process, the data acquisition device at the boiler end uses the compressed boiler operation parameter data packet as the payload of the MQTT message and publishes it to the specified topic (such as "boiler / parameters / [boiler number]", where [boiler number] is the unique identifier of each boiler). For example, the data compression package collected by boiler 1 is published to the topic "boiler / parameters / 001". The central monitoring server subscribes to the corresponding topic (such as "boiler / parameters / #" means receiving parameter messages of all boilers), and when receiving the message, it processes it according to the configuration rules. This includes decompressing and decrypting the message, extracting the temperature value and pressure value time queue data and storing it in the database, judging whether the boiler operation status is normal according to the preset analysis rules, and triggering the alarm mechanism to notify the operation and maintenance personnel if an abnormality is found. In addition, the server can also send control instructions or configuration information to the boiler end, and realize two-way communication by publishing to a specific topic and subscribing to the boiler end, such as adjusting the boiler operation parameters (temperature setting value, pressure threshold, etc.) or updating the data collection frequency.

[0024] Therefore, by carefully selecting encryption algorithms and properly managing keys, reasonably using compression algorithms and correctly encapsulating data packets, accurately configuring MQTT clients and servers, and implementing efficient data publishing and subscription mechanisms, it is possible to ensure that boiler operating parameter data is transmitted from the boiler end to the central monitoring server, while ensuring data security and improving transmission efficiency, which can provide a solid foundation for subsequent data analysis and boiler operating status monitoring.

[0025] In an embodiment of the present application, the data processing and analysis module 130 is used to decompress and decrypt the compressed data packet of the boiler operating parameters on the central monitoring server to obtain the time queue of the boiler operating parameters, and determine whether there is an abnormality in the operating status of the hot water boiler based on the time queue of the boiler operating parameters.

[0026] It should be understood that the boiler operation parameter data is compressed and encrypted before transmission. In order to restore the encrypted and compressed boiler operation parameter data packet to the original data form, it is necessary to decompress and decrypt the boiler operation parameter compressed data packet. The boiler operation parameter time queue obtained after decompression and decryption provides complete and real basic data for subsequent data processing and analysis. These data contain the temperature and pressure information at various time points during the operation of the boiler, which can reflect the dynamic process of the boiler operation. By analyzing these data, various situations such as boiler startup, stable operation, load changes, and possible abnormal fluctuations can be understood, and the boiler operation status can be judged more accurately.

[0027] Specifically, in determining whether there is an abnormality in the operating state of the hot water boiler based on the time queue of the boiler operating parameters, the technical concept of the present application is to use AI data analysis and processing algorithms to first split the boiler operating parameters, and then perform time series encoding and autocorrelation enhancement on the split time queue of temperature values ​​and the time queue of pressure values, respectively, so as to realize intelligent judgment of the operating state of the hot water boiler based on the collaborative representation between the temperature time series autocorrelation enhancement features and the pressure time series autocorrelation enhancement features. In this way, real-time monitoring and analysis can be performed to capture subtle change trends, rather than just relying on fixed threshold alarms, which helps to discover potential problems in advance, thereby greatly reducing the possibility of false alarms and missed alarms. In addition, it can adapt to the complexity and diversity of industrial environments and enhance the generalization ability of the model.

[0028] Specifically, Figure 2 FIG. 1 is a block diagram of a data processing and analysis module in an online monitoring system for hot water boiler operation status according to an embodiment of the present application. Figure 2 As shown, the data processing and analysis module 130 includes: an operating parameter splitting unit 131, which is used to split the time queue of the boiler operating parameters to obtain a time queue of temperature values ​​and a time queue of pressure values; an operating parameter encoding and strengthening unit 132, which is used to perform sequence encoding and self-attention strengthening on the time queue of temperature values ​​and the time queue of pressure values ​​respectively to obtain temperature time series autocorrelation strengthening features and pressure time series autocorrelation strengthening features; an operating parameter adjustment collaborative unit 133, which is used to perform external knowledge feature collaborative strengthening on the temperature time series autocorrelation strengthening features and the pressure time series autocorrelation strengthening features to obtain a boiler state time series collaborative strengthening representation, and judge whether the operating state of the hot water boiler is abnormal based on the boiler state time series collaborative strengthening representation.

[0029] In an embodiment of the present application, the operating parameter splitting unit 131 is used to split the time queue of the boiler operating parameters to obtain a time queue of temperature values ​​and a time queue of pressure values. Specifically, in an embodiment of the present application, the operating parameter splitting unit is used to: split the time queue of the boiler operating parameters according to the parameter sample dimension to obtain the time queue of the temperature value and the time queue of the pressure value. It should be understood that different physical quantities (such as temperature and pressure) in the boiler operating parameters may have different change rules and characteristics. Therefore, in order to be able to more finely understand and monitor the time series change characteristics of the operating state of the hot water boiler, in the technical solution of the present application, the time queue of the boiler operating parameters is split according to the parameter sample dimension to obtain the time queue of the temperature value and the time queue of the pressure value. In this way, the accuracy of the temperature and pressure time series feature analysis can be improved, and the relationship between temperature and pressure and its change trend over time can be captured and mined.

[0030] In an embodiment of the present application, the operating parameter encoding and strengthening unit 132 is used to perform sequence encoding and self-attention strengthening on the time queue of the temperature value and the time queue of the pressure value, respectively, to obtain the temperature time series autocorrelation strengthening feature and the pressure time series autocorrelation strengthening feature. Specifically, the operating parameter encoding and strengthening unit includes: a temperature-pressure time series data encoding subunit, which is used to input the time queue of the temperature value and the time queue of the pressure value into a sequence encoder based on an RNN model, respectively, to obtain a temperature time series implicit association feature vector and a pressure time series implicit association feature vector; a temperature-pressure time series feature strengthening subunit, which is used to input the temperature time series implicit association feature vector and the pressure time series implicit association feature vector into a time series self-attention model based on a converter structure to obtain a temperature time series autocorrelation strengthening feature vector as the temperature time series autocorrelation strengthening feature and a pressure time series autocorrelation strengthening feature vector as the pressure time series autocorrelation strengthening feature.

[0031] It should be understood that the time queue of the temperature values ​​and the time queue of the pressure values ​​both have time series implicit correlation features and time series change trends on different time scales, and the RNN model is particularly good at processing data with time series properties because they can maintain memory of previous information when processing sequence data, which makes the RNN model very suitable for analyzing the time dependency and dynamic change pattern in boiler operating parameters. Based on this, in the technical solution of the present application, the time queue of the temperature values ​​and the time queue of the pressure values ​​are respectively input into the sequence encoder based on the RNN model to effectively capture the change trend of temperature and pressure over time and the potential connection between these changes, and obtain the temperature time series implicit correlation feature vector and the pressure time series implicit correlation feature vector. In particular, the RNN model can automatically learn and extract implicit features in the time series, such as periodicity, trend, mutation point, etc. For hot water boilers, these features may be key clues for predicting faults or abnormal conditions.

[0032] The time queue of the temperature value and the time queue of the pressure value are respectively input into the sequence encoder based on the RNN model to obtain the temperature time series implicit correlation feature vector and the pressure time series implicit correlation feature vector. This process covers multiple key links such as data preprocessing, RNN model construction, training and optimization, and feature extraction. Each link is closely linked and influences each other to jointly realize the time series encoding processing of the input data. The following is a detailed description of its implementation process: First is the data preprocessing stage. Cleaning the collected temperature value and pressure value time queue data is a crucial first step. Due to factors such as sensor failure or interference, there may be obviously erroneous data points in the data, such as the maximum or minimum value of the temperature value beyond the normal operating range. Through the data cleaning technology based on statistical methods, a reasonable threshold range is set according to historical data and boiler operation characteristics, and the data beyond the range is regarded as an abnormal value for correction or deletion, thereby ensuring the accuracy and reliability of the data.

[0033] After data cleaning, the data needs to be standardized. Given that the range and magnitude of temperature and pressure values ​​may be different, such as temperature values ​​may be in the range of 0-100°C or higher, and pressure values ​​may be in the range of 0-10MPa or other ranges, in order to make different types of data comparable, a normalization method can be used to map the data to a specific interval (such as 0-1 or -1 to 1). The normalized data standardization process helps improve the training effect and convergence speed of the subsequent RNN model.

[0034] According to the input requirements of the RNN model, the data also needs to be grouped. If the RNN model is trained using supervised learning (for example, to predict future temperature or pressure trends), the time series data can be divided into certain time windows using methods such as the sliding window method. Assuming the time window size is w, the data of w consecutive time steps are used as a set of inputs, and the corresponding data of the next time step is used as the output label to construct a training data set to prepare for model training and optimization.

[0035] Then we move on to the RNN model building phase. When selecting the RNN unit type, we need to comprehensively consider the characteristics of the boiler operating parameter data and the analysis requirements. If there are long-term dependencies in the data and the gradient vanishing problem needs to be better handled, LSTM (Long Short-Term Memory Network) or GRU (Gated Recurrent Unit) may be more appropriate. Taking LSTM as an example, its unique internal structure includes a forget gate, an input gate, and an output gate, which can effectively control the flow and memory of information, and has obvious advantages in capturing long-term trends and periodic changes in temperature and pressure time series.

[0036] After determining the RNN unit type, it is necessary to design the architecture of the RNN model, including determining the model's number of layers, number of hidden units, and other hyperparameters. Initially, you can start with a simpler architecture, such as a single-layer RNN model. Taking a small hot water boiler monitoring system as an example, a single-layer LSTM model was initially tried with the number of hidden units set to 32, and then the performance of the model on the training set and validation set was closely observed. If the model is found to be underfitting, that is, the training error and validation error are large and convergence is slow, you can consider increasing the number of model layers or the number of hidden units, such as changing to a double-layer LSTM model with the number of hidden units in each layer increased to 64, to enhance the model's expressive power and its ability to learn complex data patterns.

[0037] After the model architecture design is completed, the parameters of the RNN model need to be initialized, including the weight matrix and bias vector. Common initialization methods include random initialization (such as normal distribution or uniform distribution initialization), but special attention should be paid to the appropriate range of initialization parameters to prevent gradient explosion or gradient disappearance problems. For example, for the weight matrix, normal distribution initialization can be used, the mean is set to 0, and the standard deviation is adjusted according to the number of hidden units to ensure that the initial parameter value is neither too large nor too small, which can create favorable conditions for stable training of the model.

[0038] At the same time, it is necessary to set the training parameters of the model, including the learning rate, number of iterations, batch size, etc. The learning rate determines the update step size of the model parameters in each iteration. The initial learning rate can be set to a smaller value (such as 0.001) and adjusted in time according to the training process (such as adopting a learning rate decay strategy to gradually reduce the learning rate as the number of training rounds increases). The number of iterations directly affects the number of rounds of model training. It is necessary to determine the appropriate value through experiments to avoid overfitting (too many training rounds cause the model to perform well on the training set but the performance on the test set declines) or underfitting (insufficient training rounds cause the model to fail to learn effective patterns in the data). The batch size is related to the efficiency of model training and memory usage. It is generally selected according to the size of the data set and hardware resources. Common ones are 32, 64, 128, etc.

[0039] Then we enter the model training and optimization phase. In this phase, we first need to select a suitable loss function to measure the difference between the model prediction result and the true value. During the model training process, we also need to calculate the loss function value based on the input temperature and pressure time queue data and the corresponding output label (if it is supervised learning). After each batch of data is forward propagated through the model to obtain the prediction result, the loss function is calculated and used as the basis for back propagation to update the model parameters.

[0040] Selecting a suitable optimization algorithm to update model parameters is also a key step. Common optimization algorithms include stochastic gradient descent (SGD) and its variants (such as Adagrad, Adadelta, Adam, etc.). The Adam optimization algorithm has performed well in practice. It combines the advantages of momentum method and adaptive learning rate. It can adaptively adjust the learning rate when different parameters are updated, accelerate model convergence and improve stability. During model training, the Adam optimization algorithm is used to calculate the gradient of each parameter based on the calculated loss function, and the parameters are updated according to the gradient and learning rate. During the training process, model verification and tuning are indispensable. Divide the data set into training set, validation set and test set (such as 70%, 15%, and 15%). After each round of training, use the validation set to evaluate the performance of the model (such as calculating the loss value or accuracy on the validation set). If it is found that the performance of the model on the validation set no longer improves or even decreases (signs of overfitting), corresponding measures can be taken, such as terminating training early, adding regularization terms (such as L1 or L2 regularization) to penalize model complexity, adjusting the model architecture or hyperparameters, etc., so as to improve the generalization ability of the model.

[0041] The last part is feature extraction. After the above series of training and optimization processes, when the model achieves relatively ideal performance (such as the validation set loss function value is stable and small), the trained RNN model is used to extract features from the temperature and pressure time queue data. The complete temperature and pressure time queue data are input into the RNN model respectively, and the temperature time series implicit correlation feature vector and the pressure time series implicit correlation feature vector are obtained through the hidden layer output of the model. The extracted feature vector will serve as the basis for subsequent in-depth analysis and processing.

[0042] Then, considering that the temporal importance between different time scales in the temperature time series implicit correlation feature vector and the pressure time series implicit correlation feature vector is different in the entire time range, therefore, in order to more accurately understand and analyze the importance of different time points, so as to highlight and strengthen the key time node features, in the technical solution of the present application, the temperature time series implicit correlation feature vector and the pressure time series implicit correlation feature vector are input into the temporal self-attention model based on the converter structure to assign different weights to each time point through the self-attention mechanism, and obtain the temperature time series autocorrelation enhancement feature vector and the pressure time series autocorrelation enhancement feature vector, so as to highlight those historical information that are more important for the prediction of the current time point. In particular, the multi-head self-attention mechanism in the converter structure allows multiple "perspectives" to work simultaneously, analyzing different parts of the same time series from different angles. This method helps to more comprehensively understand the complex temporal change patterns in the temperature and pressure time series, including but not limited to periodicity, trend and abnormal points, etc., to further enhance the feature representation.

[0043] In the embodiment of the present application, the operation parameter adjustment coordination unit 133 is used to coordinately enhance the temperature time series autocorrelation enhancement feature and the pressure time series autocorrelation enhancement feature with external knowledge features to obtain a boiler state time series coordinated enhancement representation, and determine whether the operation state of the hot water boiler is abnormal based on the boiler state time series coordinated enhancement representation. Specifically, Figure 3 FIG. 1 is a block diagram of an operation parameter adjustment coordination unit in an online monitoring system for operation status of a hot water boiler according to an embodiment of the present application. Figure 3 As shown, the operating parameter adjustment collaborative unit 133 includes: a temperature-pressure external knowledge feature collaborative subunit 1331, which is used to perform external knowledge feature collaborative enhancement on the temperature time series autocorrelation enhancement feature vector and the pressure time series autocorrelation enhancement feature vector to obtain a boiler state time series collaborative enhancement coding vector as the boiler state time series collaborative enhancement representation; a hot water boiler operating state identification subunit 1332, which is used to obtain an identification result based on the boiler state time series collaborative enhancement coding vector, and the identification result is used to indicate whether there is an abnormality in the operating state of the hot water boiler.

[0044] In an embodiment of the present application, the temperature-pressure external knowledge feature collaborative subunit 1331 is used to perform external knowledge feature collaborative enhancement on the temperature time series autocorrelation enhancement feature vector and the pressure time series autocorrelation enhancement feature vector to obtain a boiler state time series collaborative enhancement encoding vector as the boiler state time series collaborative enhancement representation. It should be understood that the temperature time series autocorrelation enhancement feature vector and the pressure time series autocorrelation enhancement feature vector have interactions and deep implicit connections at different time points. Therefore, in order to adopt a more sophisticated method to reveal the subtle and deep dynamic relationships hidden in the time series data, so as to more accurately understand and capture the complex interaction pattern between the two, in the technical solution of the present application, the temperature time series autocorrelation enhancement feature vector and the pressure time series autocorrelation enhancement feature vector are subjected to external knowledge feature collaborative enhancement to obtain a boiler state time series collaborative enhancement encoding vector. In particular, by introducing external knowledge (such as physical models, historical fault data, etc.) to modulate the characteristic fine-grained interaction network, the data from different sensors (temperature, pressure, etc.) can be more accurately integrated, which helps to capture complex patterns and relationships that are difficult to reveal from a single sensor, thereby obtaining a more comprehensive and detailed understanding of the operating status of the hot water boiler.

[0045] Specifically, in an embodiment of the present application, the temperature-pressure external knowledge feature collaborative subunit includes: a temperature-pressure fine-grained feature interaction secondary subunit, which is used to perform fine-grained feature interaction on the temperature time series autocorrelation enhancement feature vector and the pressure time series autocorrelation enhancement feature vector to obtain a temperature-pressure fine-grained time series interaction feature matrix; a temperature-pressure interaction feature external knowledge optimization secondary subunit, which is used to perform external knowledge attention optimization on the temperature-pressure fine-grained time series interaction feature matrix to obtain a temperature-pressure fine-grained time series external knowledge optimization interaction feature matrix; a temperature-pressure feature optimization secondary subunit, which is used to embed and optimize the temperature time series autocorrelation enhancement feature vector and the pressure time series autocorrelation enhancement feature vector with the temperature-pressure fine-grained time series external knowledge optimization interaction feature matrix as a common interaction feature to obtain an optimized temperature time series autocorrelation enhancement feature vector and an optimized pressure time series autocorrelation enhancement feature vector; a temperature-pressure feature response secondary subunit, which is used to calculate the response relationship between the optimized temperature time series autocorrelation enhancement feature vector and the optimized pressure time series autocorrelation enhancement feature vector to obtain the boiler state time series collaborative enhancement encoding vector.

[0046] Specifically, firstly, fine-grained feature interaction is performed on the temperature time series autocorrelation enhancement feature vector and the pressure time series autocorrelation enhancement feature vector to obtain a temperature-pressure fine-grained time series interaction feature matrix. The above process can be expressed as: ;in, is the temperature time series autocorrelation enhancement eigenvector, is the pressure time series autocorrelation enhanced eigenvector, for The transposed vector of is matrix multiplication, is the temperature-pressure fine-grained time series interaction feature matrix.

[0047] It should be understood that by performing fine-grained feature interaction on the temperature time series autocorrelation enhancement feature vector and the pressure time series autocorrelation enhancement feature vector, the fine-grained interaction information between the temperature time series autocorrelation enhancement feature vector and the pressure time series autocorrelation enhancement feature vector can be captured, which enables the model to understand the correlation between the temperature time series data and the pressure time series data from a microscopic level, thereby facilitating subsequent information fusion operations.

[0048] Next, the temperature-pressure fine-grained time series interaction feature matrix is ​​optimized by external knowledge attention to obtain the temperature-pressure fine-grained time series external knowledge optimized interaction feature matrix. The above process can be expressed as: ;in, is the temperature-pressure fine-grained time series interaction feature matrix, and They represent the learnable memory parameter matrix, for The transposed matrix of represents the normalization function, It is the temperature-pressure fine-grained timing external knowledge optimization interaction feature matrix.

[0049] It should be understood that data related to temperature and pressure are often in specific professional fields, behind which there are many professional knowledge mastered by experts in the field and a large amount of historical data accumulated in the past. By optimizing the external knowledge attention of the temperature-pressure fine-grained time series interaction feature matrix, this external knowledge can be integrated into the feature matrix processing process, so that the model can make full use of this additional information and better understand the changing laws and mutual relationships of temperature and pressure in the corresponding context.

[0050] Specifically, in the embodiment of the present application, the temperature and pressure feature optimization secondary subunit is used to: perform a linear transformation on the temperature time series autocorrelation enhancement feature vector to obtain a temperature query feature vector and a temperature value feature vector and use the temperature-pressure fine-grained time series external knowledge optimization interaction feature matrix as a key matrix, perform information optimization of the Transformer structure on the temperature query feature vector, the temperature value feature vector and the key matrix to obtain the optimized temperature time series autocorrelation enhancement feature vector. The process can be expressed as: ;in, is the temperature time series autocorrelation enhancement eigenvector, is matrix multiplication, and are the temperature query embedding matrix and the temperature first bias vector respectively, is the temperature query feature vector, and are the temperature value embedding matrix and the temperature second bias vector respectively, is the temperature value eigenvector, for The transposed matrix of yes Length, yes function, is the optimized temperature time series autocorrelation enhanced eigenvector.

[0051] The pressure time series autocorrelation enhanced feature vector is linearly transformed to obtain a pressure query feature vector and a pressure value feature vector, and the temperature-pressure fine-grained time series external knowledge optimization interaction feature matrix is ​​used as a key matrix. The pressure query feature vector, the pressure value feature vector and the key matrix are subjected to information optimization of the Transformer structure to obtain the optimized pressure time series autocorrelation enhanced feature vector. The process can be expressed as: ;in, is the pressure time series autocorrelation enhanced eigenvector, is matrix multiplication, yes function, and are the pressure query embedding matrix and the pressure first bias vector respectively, is the pressure query feature vector, and are respectively the pressure value embedding matrix and the pressure second bias vector, is the pressure value eigenvector, yes Length, is the optimized pressure time series autocorrelation enhanced eigenvector.

[0052] It should be understood that the temperature time series autocorrelation enhancement feature vector is linearly transformed to obtain the temperature query feature vector and the temperature value feature vector, and the temperature-pressure fine-grained time series external knowledge optimization interaction feature matrix is ​​used as the key matrix, and the temperature query feature vector, the temperature value feature vector and the key matrix are optimized for information of the Transformer structure, and the self-attention mechanism at the core of the Transformer structure can be used to realize the information exchange and integration between the internal and external knowledge of the temperature time series autocorrelation enhancement feature vector, so as to improve the quality of the temperature time series feature expression. Similarly, the pressure time series autocorrelation enhancement feature vector is linearly transformed to obtain the pressure query feature vector and the pressure value feature vector, and the temperature-pressure fine-grained time series external knowledge optimization interaction feature matrix is ​​used as the key matrix, and the pressure query feature vector, the pressure value feature vector and the key matrix are optimized for information of the Transformer structure. In this way, the pressure time series autocorrelation enhancement feature vector can also benefit from external knowledge, thereby improving the expression quality of the pressure time series feature. It should be particularly noted that when the Transformer structure is used to optimize the information of the temperature time series autocorrelation enhancement feature vector and the pressure time series autocorrelation enhancement feature vector, the temperature-pressure fine-grained time series external knowledge optimization interaction feature matrix is ​​a common key matrix space, which enables the optimized temperature time series autocorrelation enhancement feature vector and the pressure time series autocorrelation enhancement feature vector to have relatively better alignment characteristics in the high-dimensional feature space.

[0053] Specifically, in the embodiment of the present application, the temperature and pressure characteristic response secondary subunit is used to calculate the position point between the optimized temperature time series autocorrelation enhancement feature vector and the optimized pressure time series autocorrelation enhancement feature vector to obtain the boiler state time series collaborative enhancement encoding vector. The above process can be expressed as: ;in, is the optimized temperature time series autocorrelation enhanced eigenvector, is the optimized pressure time series autocorrelation enhanced eigenvector, is the boiler state time series collaborative reinforcement coding vector.

[0054] It should be understood that by calculating the response relationship between the optimized temperature time series autocorrelation enhancement feature vector and the optimized pressure time series autocorrelation enhancement feature vector by dividing by position, the interactive relationship between the two data of temperature and pressure can be better reflected, and then a boiler state time series collaborative enhancement coding vector that integrates the information of the temperature time series autocorrelation enhancement feature vector and the pressure time series autocorrelation enhancement feature vector can be generated.

[0055] In an embodiment of the present application, the hot water boiler operating state identification subunit 1332 is used to obtain the identification result according to the boiler state time series synergistic reinforcement coding vector, and the identification result is used to indicate whether the operating state of the hot water boiler is abnormal. Specifically, in an embodiment of the present application, the hot water boiler operating state identification subunit is used to: input the boiler state time series synergistic reinforcement coding vector into the classifier-based operating state identifier to obtain the identification result, and the identification result is used to indicate whether the operating state of the hot water boiler is abnormal. That is, the boiler state time series synergistic reinforcement coding vector obtained by external knowledge reinforcement using the temperature time series autocorrelation reinforcement feature vector and the pressure time series autocorrelation reinforcement feature vector is classified and processed to realize intelligent judgment of the operating state of the hot water boiler. Specifically, the classifier is a deep learning model whose main function is to divide data into different categories according to the input feature vector. The task of the classifier is to divide the operating state of the boiler into "normal" or "abnormal" categories according to the input boiler state time series synergistic reinforcement coding vector. It constructs classification rules by learning a large amount of sample data with labels (normal or abnormal). During the training phase, the classifier will be exposed to many time queue data of temperature and pressure under known normal operating states and known abnormal operating states, and learn how to distinguish these two states by analyzing the characteristic differences of these data. When the new boiler state time series synergistic reinforcement coding vector is input into the trained classifier, the classifier will judge whether the operating state of the hot water boiler is abnormal according to the learned pattern. In particular, in a specific embodiment of the present application, the boiler state time series synergistic reinforcement coding vector is input into the classifier-based operating state identifier to obtain the recognition result, and the recognition result is used to indicate whether the operating state of the hot water boiler is abnormal, including: using the fully connected layer of the classifier to fully connect the boiler state time series synergistic reinforcement coding vector to obtain the boiler state time series synergistic reinforcement fully connected coding feature vector; inputting the boiler state time series synergistic reinforcement fully connected coding feature vector into the Softmax classification function of the classifier to obtain the probability value of the boiler state time series synergistic reinforcement coding vector belonging to each classification label, and the classification label includes a classification label for indicating that the operating state of the hot water boiler is abnormal and a classification label for indicating that the operating state of the hot water boiler is not abnormal; the classification label corresponding to the largest of the probability values ​​is determined as the recognition result.

[0056] Here, considering that the temperature time series autocorrelation enhancement feature vector and the pressure time series autocorrelation enhancement feature vector represent the cyclic autocorrelation enhancement time series association characteristics of the temperature value and the pressure value respectively, when performing fine-grained feature interaction based on external knowledge modulation, the insufficient correspondence of the external knowledge modulation of the time series features of the autocorrelation enhancement under the time series of different data sources will lead to insufficient representation of the long-distance time series interaction features of the boiler state time series collaborative enhancement encoding vector, thereby affecting the accuracy of the recognition results obtained by the classifier-based operating state identifier.

[0057] Preferably, inputting the boiler state time series synergistic reinforcement coding vector into a classifier-based operating state identifier to obtain an identification result comprises: arranging the characteristic values ​​of the boiler state time series synergistic reinforcement coding vector in ascending order into a boiler state time series synergistic reinforcement sequence coding vector; The eigenvalue and The absolute value of the difference between the eigenvalues ​​is less than or equal to the distance difference hyperparameter , calculate the The eigenvalues ​​are similar to the The weighted sum between the eigenvalues ​​is the optimized Eigenvalue; Calculate the square root of the sum of squares of all eigenvalues ​​of the boiler state time series collaborative reinforcement encoding vector: ;in, The first one represents the boiler state time series synergistic reinforcement sequence encoding vector Eigenvalues, represents the length of the boiler state time series collaborative reinforcement coding vector, Represents the square root of the sum of squares of all eigenvalues ​​of the boiler state time series cooperative reinforcement encoding vector.

[0058] The square root of the sum of the squares of all eigenvalues ​​of the boiler state time series synergistic reinforcement coding vector is multiplied by 2 and then divided by the square of the length of the boiler state time series synergistic reinforcement coding vector to obtain the boiler state time series synergistic reinforcement space primitive value .

[0059] In response to the boiler state timing cooperative reinforcement sequence encoding vector The eigenvalue and The absolute value of the difference between the eigenvalues ​​is greater than the distance difference hyperparameter , multiply the boiler state time series collaborative reinforcement space primitive value by the first After the eigenvalue, calculate the product with the first The weighted reduction between the eigenvalues ​​is the optimized Eigenvalue.

[0060] On the basis of keeping the first eigenvalue of the boiler state time series synergistic reinforcement sequence coding vector unchanged, the second eigenvalue of the combined optimization The eigenvalues ​​are used to obtain the optimized boiler state timing cooperative reinforcement encoding vector.

[0061] The optimized boiler state time series cooperative reinforcement coding vector is input into a classifier-based operating state identifier to obtain an identification result.

[0062] Therefore, in order to address the problem of insufficient global fusion representation capability of the feature set of the boiler state time series collaborative reinforcement coding vector under a predetermined eigenvalue sequential distribution due to the long distance exceeding the predetermined local distribution interval threshold, a high-dimensional feature space primitive representation based on self-inner product fusion of the boiler state time series collaborative reinforcement coding vector is used to capture the complex structure of the global network interaction of its eigenvalues, thereby reconstructing the interactive query fusion relationship between the eigenvalues ​​of the boiler state time series collaborative reinforcement coding vector by simulating the scale-based high-dimensional feature space potential primitives, so as to realize the coding reconstruction of the real sequence distribution behavior of the boiler state time series collaborative reinforcement coding vector under long distance, improve the interactive query fusion expression effect of the boiler state time series collaborative reinforcement coding vector, and improve the accuracy of the recognition result obtained by the classifier-based operating state identifier.

[0063] In summary, it is clearly explained that whether there is an abnormality in the operating state of the hot water boiler based on the time queue of the boiler operating parameters, the data analysis and processing algorithm using AI is firstly used to split the boiler operating parameters, and then the time queue of the split temperature value and the time queue of the pressure value are respectively time-series encoded and autocorrelatedly enhanced, so as to realize the intelligent judgment of the operating state of the hot water boiler based on the collaborative representation between the temperature time series autocorrelation enhancement feature and the pressure time series autocorrelation enhancement feature. In this way, real-time monitoring and analysis can be carried out to capture subtle change trends, rather than just relying on fixed threshold alarms, which helps to discover potential problems in advance, thereby greatly reducing the possibility of false alarms and missed alarms. In addition, it can adapt to the complexity and diversity of industrial environments and enhance the generalization ability of the model.

[0064] In an embodiment of the present application, the display and alarm module 140 is used to display abnormal status warning prompts. It should be understood that once the hot water boiler is in an abnormal state, the operation and maintenance personnel need to be informed in time. Displaying abnormal status warning prompts can convey the abnormal information of the boiler to the staff responsible for monitoring and maintenance as soon as possible, ensuring that they can take measures quickly to avoid further deterioration of the problem. And by displaying abnormal warning prompts, complex boiler operation data can be refined into intuitive abnormal information. This method highlights the key problems and enables operation and maintenance personnel to quickly locate areas where failures may occur, which helps to improve the response speed of fault handling.

[0065] In summary, the online monitoring system 100 for the operation status of a hot water boiler based on the embodiment of the present application is explained. It first collects the time queue data of the boiler operation parameters including the temperature value and the pressure value through the sensor components installed at the key parts of the boiler. Then, the time queue data of these parameters are encrypted and compressed to form a compressed data packet of the boiler operation parameters, and sent to the central monitoring server through the MQTT protocol. At the central monitoring server end, the received compressed data packet is decompressed and decrypted to restore it to the original time queue data. Finally, based on these data, the central monitoring server analyzes and determines the operation status of the hot water boiler. Once an abnormal state is found, an abnormal warning prompt is displayed. In this way, it is helpful to timely discover abnormal conditions in the operation of the boiler, thereby ensuring the safe and stable operation of the boiler.

[0066] In an embodiment of the present application, a method for online monitoring of the operating status of a hot water boiler according to an embodiment of the present application is also provided, which is used to execute the online monitoring system for the operating status of a hot water boiler as described above. Figure 4 Flow chart of the method for online monitoring of the operation status of a hot water boiler according to an embodiment of the present application. Figure 4 As shown, in the online monitoring method for the operation status of a hot water boiler, the method includes: S110, performing data splitting on the time queue of the boiler operation parameters to obtain the time queue of the temperature value and the time queue of the pressure value; S120, performing sequence encoding and self-attention enhancement on the time queue of the temperature value and the time queue of the pressure value respectively to obtain the temperature time series autocorrelation enhancement feature and the pressure time series autocorrelation enhancement feature; S130, performing external knowledge feature collaborative enhancement on the temperature time series autocorrelation enhancement feature and the pressure time series autocorrelation enhancement feature to obtain the boiler state time series collaborative enhancement representation, and judging whether the operation status of the hot water boiler is abnormal based on the boiler state time series collaborative enhancement representation.

[0067] Here, those skilled in the art can understand that the specific operations of each step in the above-mentioned hot water boiler operating status online monitoring method have been referred to above. Figures 1 to 3 The description of the online monitoring system for the operation status of a hot water boiler has been described in detail, and therefore, its repeated description will be omitted.

[0068] In summary, the online monitoring method for the operation status of a hot water boiler based on the embodiment of the present application is explained. It first collects the time queue data of the boiler operation parameters including the temperature value and the pressure value through the sensor components installed at the key parts of the boiler. Then, the time queue data of these parameters are encrypted and compressed to form a compressed data packet of the boiler operation parameters, and sent to the central monitoring server through the MQTT protocol. At the central monitoring server, the received compressed data packet is decompressed and decrypted to restore it to the original time queue data. Finally, based on these data, the central monitoring server analyzes and determines the operation status of the hot water boiler. Once an abnormal state is found, an abnormal warning prompt is displayed. In this way, it is helpful to timely discover abnormal conditions in the operation of the boiler, thereby ensuring the safe and stable operation of the boiler.

Claims

1. An online monitoring system for the operation status of a hot water boiler, characterized in that: include: A data acquisition module, used to collect boiler operating parameters collected by sensor components installed at key parts of the boiler to obtain a time queue of boiler operating parameters, wherein the boiler operating parameters include temperature values ​​and pressure values; A data transmission module, used for first encrypting and compressing the time queue of the boiler operation parameter into a boiler operation parameter compressed data packet, and transmitting the boiler operation parameter compressed data packet to a central monitoring server through the MQTT protocol; A data processing and analysis module, configured to decompress and decrypt the compressed data packet of the boiler operating parameters on the central monitoring server to obtain a time queue of the boiler operating parameters, and determine whether the operating state of the hot water boiler is abnormal based on the time queue of the boiler operating parameters; A display and alarm module is used to display abnormal status warning prompts; wherein the data processing and analysis module includes: an operating parameter splitting unit, which is used to split the time queue of the boiler operating parameters to obtain a time queue of temperature values ​​and a time queue of pressure values; an operating parameter encoding and strengthening unit, which is used to perform sequence encoding and self-attention strengthening on the time queue of temperature values ​​and the time queue of pressure values ​​respectively to obtain temperature time series autocorrelation strengthening features and pressure time series autocorrelation strengthening features; an operating parameter adjustment collaborative unit, which is used to perform external knowledge feature collaborative strengthening on the temperature time series autocorrelation strengthening features and the pressure time series autocorrelation strengthening features to obtain a boiler state time series collaborative strengthening representation, and judge whether the operating state of the hot water boiler is abnormal based on the boiler state time series collaborative strengthening representation.

2. The hot water boiler operating status online monitoring system according to claim 1 is characterized in that: The operating parameter splitting unit is used to split the time queue of the boiler operating parameters according to the parameter sample dimension to obtain the time queue of the temperature value and the time queue of the pressure value.

3. The hot water boiler operating status online monitoring system according to claim 2 is characterized in that: The operating parameter encoding and strengthening unit includes: a temperature-pressure time series data encoding subunit, which is used to input the time queue of the temperature value and the time queue of the pressure value into a sequence encoder based on the RNN model respectively to obtain a temperature time series implicit correlation feature vector and a pressure time series implicit correlation feature vector; a temperature-pressure time series feature strengthening subunit, which is used to input the temperature time series implicit correlation feature vector and the pressure time series implicit correlation feature vector into a time series self-attention model based on a converter structure to obtain a temperature time series autocorrelation strengthening feature vector as the temperature time series autocorrelation strengthening feature and a pressure time series autocorrelation strengthening feature vector as the pressure time series autocorrelation strengthening feature.

4. The hot water boiler operating status online monitoring system according to claim 3 is characterized in that: The operating parameter adjustment collaborative unit includes: a temperature-pressure external knowledge feature collaborative sub-unit, which is used to perform external knowledge feature collaborative enhancement on the temperature time series autocorrelation enhancement feature vector and the pressure time series autocorrelation enhancement feature vector to obtain a boiler state time series collaborative enhancement coding vector as the boiler state time series collaborative enhancement representation; a hot water boiler operating state identification sub-unit, which is used to obtain an identification result based on the boiler state time series collaborative enhancement coding vector, and the identification result is used to indicate whether there is an abnormality in the operating state of the hot water boiler.

5. The hot water boiler operating status online monitoring system according to claim 4 is characterized in that: The temperature-pressure external knowledge feature collaborative subunit includes: a temperature-pressure fine-grained feature interaction secondary subunit, which is used to perform fine-grained feature interaction on the temperature time series autocorrelation enhancement feature vector and the pressure time series autocorrelation enhancement feature vector to obtain a temperature-pressure fine-grained time series interaction feature matrix; a temperature-pressure interaction feature external knowledge optimization secondary subunit, which is used to perform external knowledge attention optimization on the temperature-pressure fine-grained time series interaction feature matrix to obtain a temperature-pressure fine-grained time series external knowledge optimization interaction feature matrix; a temperature-pressure feature optimization secondary subunit, which is used to embed and optimize the temperature time series autocorrelation enhancement feature vector and the pressure time series autocorrelation enhancement feature vector with the temperature-pressure fine-grained time series external knowledge optimization interaction feature matrix as a common interaction feature to obtain an optimized temperature time series autocorrelation enhancement feature vector and an optimized pressure time series autocorrelation enhancement feature vector; a temperature-pressure feature response secondary subunit, which is used to calculate the response relationship between the optimized temperature time series autocorrelation enhancement feature vector and the optimized pressure time series autocorrelation enhancement feature vector to obtain the boiler state time series collaborative enhancement encoding vector.

6. The hot water boiler operating status online monitoring system according to claim 5, characterized in that: The temperature-pressure feature optimization secondary subunit is used to: perform linear transformation on the temperature time series autocorrelation enhancement feature vector to obtain a temperature query feature vector and a temperature value feature vector, and use the temperature-pressure fine-grained time series external knowledge optimization interactive feature matrix as a key matrix, perform Transformer structure information optimization on the temperature query feature vector, the temperature value feature vector and the key matrix to obtain the optimized temperature time series autocorrelation enhancement feature vector; perform linear transformation on the pressure time series autocorrelation enhancement feature vector to obtain a pressure query feature vector and a pressure value feature vector, and use the temperature-pressure fine-grained time series external knowledge optimization interactive feature matrix as a key matrix, perform Transformer structure information optimization on the pressure query feature vector, the pressure value feature vector and the key matrix to obtain the optimized pressure time series autocorrelation enhancement feature vector.

7. The hot water boiler operating status online monitoring system according to claim 6, characterized in that: The temperature and pressure characteristic response secondary subunit is used to calculate the boiler state time series collaborative enhancement coding vector by dividing the optimized temperature time series autocorrelation enhancement characteristic vector and the optimized pressure time series autocorrelation enhancement characteristic vector by a position point.

8. The hot water boiler operating status online monitoring system according to claim 7, characterized in that: The hot water boiler operating state identification subunit is used to: input the boiler state time series collaborative reinforcement coding vector into the classifier-based operating state identifier to obtain the identification result, and the identification result is used to indicate whether the operating state of the hot water boiler is abnormal.

9. A method for online monitoring of the operation status of a hot water boiler, used to implement the online monitoring system for the operation status of a hot water boiler according to claim 1, characterized in that: The online monitoring method for the operation status of a hot water boiler comprises: performing data splitting on a time queue of the boiler operation parameters to obtain a time queue of a temperature value and a time queue of a pressure value; performing sequence encoding and self-attention enhancement on the time queue of the temperature value and the time queue of the pressure value respectively to obtain a temperature time series autocorrelation enhancement feature and a pressure time series autocorrelation enhancement feature; performing external knowledge feature collaborative enhancement on the temperature time series autocorrelation enhancement feature and the pressure time series autocorrelation enhancement feature to obtain a boiler state time series collaborative enhancement representation, and judging whether the operation status of the hot water boiler is abnormal based on the boiler state time series collaborative enhancement representation.

10. The method for online monitoring of the operation status of a hot water boiler according to claim 9, characterized in that: The time queue of the boiler operating parameters is split into data to obtain a time queue of temperature values ​​and a time queue of pressure values, including: splitting the time queue of the boiler operating parameters according to the parameter sample dimension to obtain a time queue of the temperature values ​​and a time queue of the pressure values.

Images