Intelligent thermal power plant energy consumption monitoring system and method based on PLC

Abstract

The embodiment of the invention relates to the field of intelligent monitoring, provides an intelligent thermal power plant energy consumption monitoring system and method based on a PLC, and relates to the field of intelligent monitoring. The method comprises the following steps: extracting a set of energy consumption data time sequences marked as normal energy consumption, and then adopting a data analysis and coding technology based on artificial intelligence; and intelligently judging whether the energy consumption of the target energy consumption equipment is abnormal or not according to the query response semantic representation between the query energy consumption data time sequence semantic feature obtained by encoding the energy consumption data and each normal energy consumption data time sequence semantic feature. In this way, the complex energy consumption mode in the time sequence data of the equipment can be carefully captured, and false alarms and missing alarms are remarkably reduced. And the judgment standard can be automatically adjusted according to the real-time energy consumption data, so that more flexible and accurate anomaly detection is provided.

Description

Technical Field

[0001] The present disclosure relates to the field of intelligent monitoring, and more specifically, to a PLC-based intelligent thermal power plant energy consumption monitoring system and method. Background Art

[0002] As global attention to energy efficiency continues to grow, energy management in the industrial sector is becoming increasingly important. Especially in large energy-consuming facilities such as thermal power plants, optimizing energy consumption not only helps reduce costs but also significantly reduces carbon emissions and other environmental impacts. As a key component of industrial automation, programmable logic controllers (PLCs) play a central role in the manufacturing and energy industries. PLCs collect and process data by controlling mechanical operations and integrating sensors and actuators. In thermal power plants, PLCs are used to monitor the energy consumption status of key equipment such as boilers, turbines, and generators to prevent equipment damage and increased energy consumption.

[0003] However, traditional thermal power plant energy consumption monitoring PLC systems primarily rely on preset fixed thresholds to trigger alarms. This approach is significantly inadequate when dealing with complex nonlinear changes or progressive faults. Because fixed thresholds cannot flexibly adapt to dynamically changing operating conditions, this can lead to false alarms or missed alarms, reducing the accuracy of anomaly detection, impacting maintenance efficiency and power plant safety. Furthermore, traditional PLC systems struggle to identify complex temporal patterns or abnormal behavior in energy consumption data, making it difficult to detect potential problems or provide early warning of faults.

[0004] Therefore, an optimized PLC-based smart thermal power plant energy consumption monitoring solution is desired. Summary of the Invention

[0005] The present disclosure aims to solve at least one of the problems existing in the prior art and provide a PLC-based smart thermal power plant energy consumption monitoring system and method.

[0006] One aspect of the present disclosure provides a PLC-based smart thermal power plant energy consumption monitoring method, comprising:

[0007] A time queue for collecting energy consumption data of target energy-consuming devices through energy consumption sensors;

[0008] Inputting the time queue of the energy consumption data into a PLC-based energy consumption monitoring processor;

[0009] The PLC-based energy consumption monitoring processor extracts a set of energy consumption data time series marked as normal energy consumption from a background database;

[0010] In the PLC-based energy consumption monitoring processor, sequence encoding is performed on the set of energy consumption data time series marked as normal energy consumption to obtain a set of normal energy consumption data time series semantic features;

[0011] In the PLC-based energy consumption monitoring processor, the time queue of the energy consumption data is sequence-encoded to obtain a time series semantic feature of the query energy consumption data;

[0012] In the PLC-based energy consumption monitoring processor, query response encoding is performed on the set of the query energy consumption data temporal semantic features and the normal energy consumption data temporal semantic features to obtain energy consumption data query response semantic features, including: calculating a fast semantic matching factor between each normal energy consumption data temporal semantic feature in the set of the query energy consumption data temporal semantic features and the normal energy consumption data temporal semantic features to obtain a set of fast semantic matching factors; based on the set of fast semantic matching factors, performing a fast semantic query on the set of the query energy consumption data temporal semantic features and the normal energy consumption data temporal semantic features to obtain the energy consumption data query response semantic features;

[0013] In the PLC-based energy consumption monitoring processor, based on the energy consumption data query response semantic features, a monitoring result indicating whether the target energy-consuming device has abnormal energy consumption is obtained.

[0014] Optionally, the set of energy consumption data time series marked as normal energy consumption is sequence-encoded to obtain a set of normal energy consumption data time series semantic features, including: inputting each energy consumption data time series marked as normal energy consumption in the set of energy consumption data time series marked as normal energy consumption into an energy consumption data sequence encoder based on an RNN-LSTM hybrid model to obtain a set of normal energy consumption data time series semantic coding vectors as the set of normal energy consumption data time series semantic features.

[0015] Optionally, the time queue of the energy consumption data is sequence-encoded to obtain the query energy consumption data temporal semantic features, including: inputting the time queue of the energy consumption data into the energy consumption data sequence encoder based on the RNN-LSTM hybrid model to obtain the query energy consumption data temporal semantic encoding vector as the query energy consumption data temporal semantic features.

[0016] Optionally, the fast semantic matching factors between the query energy consumption data temporal semantic features and each normal energy consumption data temporal semantic feature in the set of normal energy consumption data temporal semantic features are calculated to obtain a set of fast semantic matching factors, including: calculating the mutual information between the query energy consumption data temporal semantic coding vector and each normal energy consumption data temporal semantic coding vector in the set of normal energy consumption data temporal semantic coding vectors to obtain a set of query-normal energy consumption data semantic fast semantic matching factors as the set of fast semantic matching factors.

[0017] Optionally, calculating the mutual information between the query energy consumption data time series semantic coding vector and each normal energy consumption data time series semantic coding vector in the set of normal energy consumption data time series semantic coding vectors to obtain a set of query-normal energy consumption data semantic fast semantic matching factors includes:

[0018] Inputting the query energy consumption data time series semantic encoding vector into a sigmoid function to obtain a query energy consumption data time series semantic transformation feature vector;

[0019] Inputting each normal energy consumption data time series semantic coding vector in the set of normal energy consumption data time series semantic coding vectors into the sigmoid function to obtain a set of normal energy consumption data time series semantic transformation feature vectors;

[0020] The mutual information between the query energy consumption data temporal semantic transformation feature vector and each normal energy consumption data temporal semantic transformation feature vector in the set of normal energy consumption data temporal semantic transformation feature vectors is calculated to obtain a set of query-normal energy consumption data semantic fast semantic matching factors.

[0021] Optionally, based on the set of the fast semantic matching factors, performing a fast semantic query on the set of the query energy consumption data time series semantic features and the normal energy consumption data time series semantic features to obtain the energy consumption data query response semantic features, including:

[0022] Based on the set of query-normal energy consumption data semantic fast semantic matching factors, determining a subset of fast-matched normal energy consumption data time series semantic coding vectors from the set of normal energy consumption data time series semantic coding vectors;

[0023] The query energy consumption data time series semantic coding vector is linearly transformed to obtain an energy consumption data semantic query vector and an energy consumption data semantic value vector, and a subset of the normal energy consumption data time series semantic coding vector is used as a subset of the key vector. The energy consumption data semantic query vector, the energy consumption data semantic value vector and the subset of the key vector are subjected to Transformer cross-domain query encoding to obtain an energy consumption data query response semantic coding vector as the energy consumption data query response semantic feature.

[0024] Optionally, based on the set of query-normal energy consumption data semantic fast semantic matching factors, determining a subset of fast-matched normal energy consumption data time series semantic coding vectors from the set of normal energy consumption data time series semantic coding vectors includes:

[0025] Identify a first query-normal energy consumption data semantic maximum value and a second query-normal energy consumption data semantic maximum value from the set of query-normal energy consumption data semantic fast semantic matching factors;

[0026] Based on the positions of the first query-normal energy consumption data semantic maximum and the second query-normal energy consumption data semantic maximum in the set of query-normal energy consumption data semantic fast semantic matching factors, a subset of the normal energy consumption data temporal semantic coding vectors that are quickly matched is determined from the set of normal energy consumption data temporal semantic coding vectors.

[0027] Optionally, based on the semantic features of the energy consumption data query response, a monitoring result is obtained to indicate whether the target energy consumption device has abnormal energy consumption, including: inputting the energy consumption data query response semantic encoding vector into an energy consumption monitoring module based on the SVM model to obtain the monitoring result, and the monitoring result is used to indicate whether the target energy consumption device has abnormal energy consumption.

[0028] Another aspect of the present disclosure provides a PLC-based smart thermal power plant energy consumption monitoring system, comprising:

[0029] Energy consumption data time queue collection module, used to collect the time queue of energy consumption data of target energy consumption equipment through energy consumption sensors;

[0030] An energy consumption data time queue input module, used for inputting the time queue of the energy consumption data into the PLC-based energy consumption monitoring processor;

[0031] A normal energy consumption data time series extraction module is used to extract a set of energy consumption data time series marked as normal energy consumption from a background database in the PLC-based energy consumption monitoring processor;

[0032] a normal energy consumption data time series encoding module, configured to perform sequence encoding on the set of energy consumption data time series marked as normal energy consumption in the PLC-based energy consumption monitoring processor to obtain a set of normal energy consumption data time series semantic features;

[0033] An energy consumption data time queue sequence encoding module is used to perform sequence encoding on the time queue of the energy consumption data in the PLC-based energy consumption monitoring processor to obtain a time series semantic feature of the query energy consumption data;

[0034] A query response encoding module is used to perform query response encoding on the set of the query energy consumption data time series semantic feature and the normal energy consumption data time series semantic feature in the PLC-based energy consumption monitoring processor to obtain an energy consumption data query response semantic feature;

[0035] The target energy-consuming equipment monitoring result analysis module is used to obtain a monitoring result indicating whether the target energy-consuming equipment has abnormal energy consumption based on the semantic features of the energy consumption data query response in the PLC-based energy consumption monitoring processor.

[0036] Optionally, the query response encoding module is used to:

[0037] Calculating a fast semantic matching factor between the query energy consumption data time series semantic feature and each normal energy consumption data time series semantic feature in the set of normal energy consumption data time series semantic features to obtain a set of fast semantic matching factors;

[0038] Based on the set of fast semantic matching factors, a fast semantic query is performed on the set of the query energy consumption data time series semantic features and the normal energy consumption data time series semantic features to obtain the energy consumption data query response semantic features.

[0039] Compared with the prior art, the present disclosure collects the time queue of energy consumption data of the target energy-consuming equipment and inputs it into the PLC-based energy consumption monitoring processor to extract a set of energy consumption data time series marked as normal energy consumption, and then uses artificial intelligence-based data analysis and encoding technology to intelligently judge whether the target energy-consuming equipment has abnormal energy consumption based on the query energy consumption data time series semantic features obtained by encoding the energy consumption data and the query response semantic representation between the time series semantic features of each normal energy consumption data. In this way, the complex energy consumption patterns in the equipment time series data can be captured more carefully, significantly reducing false alarms and missed reports. And the judgment criteria can be automatically adjusted according to the real-time energy consumption data, thereby providing more flexible and accurate anomaly detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] One or more embodiments are exemplarily illustrated by pictures in the corresponding drawings, and these exemplifications do not constitute limitations on the embodiments. Elements with the same reference numerals in the drawings represent similar elements, and unless otherwise stated, the figures in the drawings do not constitute proportional limitations.

[0041] Figure 1 Flowchart of a PLC-based smart thermal power plant energy consumption monitoring method according to an embodiment of the present application;

[0042] Figure 2 Schematic diagram of data flow of a PLC-based smart thermal power plant energy consumption monitoring method according to an embodiment of the present application;

[0043] Figure 3 A flowchart of performing query response encoding on a set of the query energy consumption data temporal semantic features and the normal energy consumption data temporal semantic features in the PLC-based energy consumption monitoring processor in the PLC-based smart thermal power plant energy consumption monitoring method according to an embodiment of the present application to obtain energy consumption data query response semantic features;

[0044] Figure 4 A flowchart of calculating the mutual information between the query energy consumption data time series semantic coding vector and each normal energy consumption data time series semantic coding vector in the set of normal energy consumption data time series semantic coding vectors to obtain a set of query-normal energy consumption data semantic fast semantic matching factors in the PLC-based smart thermal power plant energy consumption monitoring method according to an embodiment of the present application;

[0045] Figure 5 A flowchart of performing a rapid semantic query on a set of the query energy consumption data temporal semantic features and the normal energy consumption data temporal semantic features based on the set of the rapid semantic matching factors in the PLC-based smart thermal power plant energy consumption monitoring method according to an embodiment of the present application to obtain the energy consumption data query response semantic features;

[0046] Figure 6 This is a system block diagram of a PLC-based smart thermal power plant energy consumption monitoring system according to an embodiment of the present application. DETAILED DESCRIPTION

[0047] In order to make the purpose, technical solutions and advantages of the embodiments of the present disclosure clearer, the embodiments of the present disclosure will be described in detail below with reference to the accompanying drawings. However, it will be understood by those skilled in the art that in each embodiment of the present disclosure, many technical details are provided to enable readers to better understand the present disclosure. However, even without these technical details and various changes and modifications based on the following embodiments, the technical solutions claimed in the present disclosure can be implemented. The division of the following embodiments is for the convenience of description and should not constitute any limitation on the specific implementation of the present disclosure. The various embodiments can be combined and referenced with each other under the premise that there is no contradiction.

[0048] As the world places increasing emphasis on energy efficiency, energy management is becoming increasingly important in the industrial sector, particularly in large energy-consuming facilities such as thermal power plants. Programmable logic controllers (PLCs) play a key role in these facilities, controlling equipment operations and integrating sensors and actuators to collect and process data. PLCs monitor the energy consumption status of equipment such as boilers, turbines, and generators in thermal power plants, helping to avoid equipment damage and reduce energy consumption. However, traditional PLC energy consumption monitoring systems typically rely on fixed thresholds to trigger alarms, which is ineffective when faced with complex nonlinear changes or gradual failures. Because this approach cannot flexibly adapt to dynamic changes in operating conditions, it can lead to false alarms or missed alarms, reducing the accuracy of fault detection, affecting the efficiency of equipment maintenance, and the safety of power plants.

[0049] In response to the above technical problems, this application proposes a PLC-based smart thermal power plant energy consumption monitoring method. Figure 1 This is a flow chart of a PLC-based smart thermal power plant energy consumption monitoring method according to an embodiment of the present application.

[0050] Figure 2 Schematic diagram of data flow of the PLC-based smart thermal power plant energy consumption monitoring method according to the embodiment of the present application. Figure 1 and Figure 2As shown, according to the embodiment of the present application, the energy consumption monitoring method of a smart thermal power plant based on PLC includes: S110, collecting the time queue of energy consumption data of the target energy-consuming equipment through the energy consumption sensor; S120, inputting the time queue of the energy consumption data into the energy consumption monitoring processor based on PLC; S130, in the energy consumption monitoring processor based on PLC, extracting the set of energy consumption data time series marked as normal energy consumption from the background database; S140, in the energy consumption monitoring processor based on PLC, performing sequence encoding on the set of energy consumption data time series marked as normal energy consumption to obtain normal energy consumption data. A set of time series semantic features; S150, in the PLC-based energy consumption monitoring processor, serially encoding the time queue of the energy consumption data to obtain the time series semantic features of the query energy consumption data; S160, in the PLC-based energy consumption monitoring processor, query response encoding the set of the time series semantic features of the query energy consumption data and the time series semantic features of the normal energy consumption data to obtain the energy consumption data query response semantic features; S170, in the PLC-based energy consumption monitoring processor, based on the energy consumption data query response semantic features, obtaining a monitoring result for indicating whether the energy consumption of the target energy consuming device is abnormal.

[0051] In response to the above technical problems, the technical concept of the present application is to collect the time queue of energy consumption data of the target energy-consuming equipment through energy consumption sensors and input it into the energy consumption monitoring processor based on PLC. In the energy consumption monitoring processor based on PLC, a set of energy consumption data time series marked as normal energy consumption is extracted from the background database, and artificial intelligence-based data analysis and encoding technology is used to perform time series semantic encoding on each energy consumption data and energy consumption data marked as normal energy consumption, so as to intelligently judge whether the target energy-consuming equipment has abnormal energy consumption based on the query energy consumption data time series semantic features obtained by energy consumption data encoding and the query response semantic representation between the time series semantic features of each normal energy consumption data. In this way, by integrating advanced artificial intelligence technology and the real-time control capability of PLC, it is possible to capture the complex energy consumption patterns in the equipment time series data in a more detailed manner, significantly reducing false alarms and missed alarms. And it can automatically adjust its judgment criteria according to real-time energy consumption data, rather than relying solely on preset fixed thresholds, thereby providing more flexible and accurate anomaly detection.

[0052] In the aforementioned PLC-based smart thermal power plant energy consumption monitoring method, step S110 involves collecting a time series of energy consumption data from target energy-consuming devices using energy consumption sensors. It should be understood that energy consumption sensors are typically installed at the power input of a device, monitoring parameters such as current, voltage, and power in real time, and then calculating energy consumption data. The sensors record this data in a time series format. The time series typically has precise timestamps, with each data point representing the energy consumption level at a specific moment. High-frequency sampling allows for the analysis of energy consumption fluctuations across devices over different time periods, helping to analyze energy consumption patterns and identify peak periods and energy-saving potential. Specifically, the time series of energy consumption data can be aggregated by integrated devices and transmitted to a cloud platform or local storage system for further analysis. During the analysis process, the data series not only reveals the energy consumption of devices over different time periods, but also enables trend prediction, anomaly detection, and load analysis. By deeply analyzing this time series data, potential efficiency bottlenecks and unreasonable energy consumption behaviors can be identified, enabling the development of effective energy-saving strategies.

[0053] In the above-mentioned PLC-based smart thermal power plant energy consumption monitoring method, step S120 inputs the time queue of the energy consumption data into the PLC-based energy consumption monitoring processor. It should be understood that PLC, as a commonly used automation equipment in the field of industrial control, has powerful data acquisition and control capabilities. By inputting time series energy consumption data into the PLC, the energy consumption data can be efficiently processed, stored and analyzed. Specifically, the time queue of energy consumption data usually contains power consumption data at each moment. This data can be collected in real time by devices such as electricity meters and sensors, usually in units of timestamps and corresponding energy consumption values (such as watt-hours, kilowatt-hours, etc.). Inputting this data into the PLC is mainly completed through the PLC input module. The specific steps are: 1) using appropriate sensors or instruments, the energy consumption data of the equipment or system is collected and converted into a transmittable digital signal; 2) the collected digital signal is transmitted to the input port of the PLC. 3) The program or logic control system built into the PLC can receive these signals and process the energy consumption data according to the preset time queue order. PLCs can be programmed to store data, perform statistical analysis, or perform real-time monitoring, even dynamically adjusting equipment operating status. PLCs can also be integrated with other data processing modules, such as SCADA systems or human-machine interfaces (HMIs), to further optimize energy consumption monitoring systems.

[0054] Specifically, to enable data transmission from sensors to the PLC, a reliable communication interface must be designed. This interface must support a variety of industry-standard protocols, such as Modbus, Profibus, and Ethernet / IP, to accommodate different types of sensors and network environments. Using these protocols, the PLC can read data from various sensors and integrate it into its processing flow. For modern factories, industrial Ethernet has become the primary communication method of choice due to its high speed, low latency, and excellent compatibility. Once sensors begin operating, they collect energy consumption data at a pre-set frequency and timestamp it, forming an ordered time queue. This data is initially temporarily stored in a local cache or buffer to prevent data loss due to network issues. Once the network connection is confirmed to be stable and reliable, the cached data is gradually uploaded to the PLC. This approach ensures that even brief network outages will not affect the continuity and integrity of the entire system.

[0055] Furthermore, to ensure that data from all sensors can be compared within the same timeframe, a unified time source is needed to synchronize the timestamps of each sensor. For example, a GPS-synchronized clock can be deployed as a global time reference. This allows data from any sensor to be directly compared with data from other sensors without requiring additional time adjustments. Furthermore, the PLC itself should maintain consistency with this global time source to ensure that all time-related operations involved in its internal processing are accurate.

[0056] Furthermore, before energy consumption data enters the PLC, it typically undergoes a round of preprocessing and preliminary filtering. This includes verifying that each data point contains necessary information, such as a valid timestamp and measurement value; and identifying data points that are clearly outside the normal range, which may be due to sensor failure or other reasons. For these suspicious data, you can choose to mark them as invalid or further investigate the cause. If necessary, the data is formatted to conform to the input format required by the PLC. This may involve operations such as unit conversion and numerical scaling to ensure that the data can be directly used for subsequent calculations and analysis.

[0057] As more and more energy consumption data is transmitted to PLCs, to efficiently manage and process this real-time data stream, PLCs are typically equipped with dedicated data buffers or queue mechanisms to temporarily store newly arriving data. These buffers not only absorb short-term data spikes but also allow the PLC to process each task one by one according to priority. For example, emergency alarm signals can be responded to immediately, while routine monitoring data can be analyzed later during idle periods. Furthermore, to improve data processing efficiency, PLCs can utilize multi-threading or multi-core processor technology to execute multiple tasks simultaneously. This means that while receiving new energy consumption data, they can also perform complex time series analysis on existing data without causing system overload or delays.

[0058] In the above-mentioned PLC-based smart thermal power plant energy consumption monitoring method, in step S130, the PLC-based energy consumption monitoring processor extracts a set of energy consumption data time series marked as normal energy consumption from the background database. It should be understood that the "set of energy consumption data time series marked as normal energy consumption" refers to a group of data records that have been verified or confirmed to show that the equipment exhibits normal and expected energy consumption levels under specific operating conditions. These data are usually screened from historical operation records and represent the energy consumption pattern of the equipment under optimal or standard working conditions. Therefore, in order to be able to more accurately evaluate the current energy consumption usage, in the technical solution of the present application, a set of energy consumption data time series marked as normal energy consumption is extracted from the background database. In this way, by comparing the actual energy consumption data with the known normal energy consumption pattern, the algorithm's understanding of normal energy consumption behavior can be enhanced, making it possible to more accurately distinguish abnormal situations, thereby reducing the occurrence of false alarms.

[0059] Specifically, in order to ensure that the required data can be extracted from the background database efficiently and accurately, a well-structured database system needs to be designed and maintained. This database must not only store a large amount of historical energy consumption data, but also have the ability to quickly query and retrieve. Typically, such databases use relational database management systems (RDBMS), such as MySQL, PostgreSQL, etc., because they provide powerful data management and query capabilities. For large-scale data sets, you can also consider using NoSQL databases, such as MongoDB, to meet unstructured or semi-structured data needs. Each record in the database represents the energy consumption status of a device at a specific moment, including but not limited to timestamps, measurement values, and annotation information. Among them, the annotation information is very important, which indicates whether the record is considered to be "normal energy consumption". These annotations can be completed through automatic classification by machine learning algorithms.

[0060] Furthermore, data labeling is an ongoing process, occurring not only during system initialization but also as new data continuously arrives. Labeling standards are typically developed based on specifications provided by equipment manufacturers, industry standards, and the factory's own operating experience. For example, if a boiler's fuel consumption remains within a specific range under certain load conditions, energy consumption during that period can be considered normal; conversely, if it exceeds or falls below this range, it may be an anomaly. When new energy consumption data enters the system, the PLC performs a preliminary analysis to determine whether there are any obvious anomalies. Data that appears normal is added to the "normal" dataset, and over time, a large time series of normal energy consumption data is formed. Data marked as abnormal is stored separately for further investigation and corrective action. This dynamic update mechanism ensures that the "normal" data in the database always reflects the latest actual conditions, improving the accuracy of subsequent analysis.

[0061] Once the database has a sufficiently rich set of normal energy consumption time series data, the next step is to effectively extract useful information from it. This requires the PLC to possess powerful data processing capabilities capable of executing complex queries. In modern industrial environments, PLCs are often equipped with high-performance processors and ample memory resources, more than sufficient for such tasks. Furthermore, to improve query efficiency, indexing technology can be introduced, allowing the PLC to quickly locate the required data for a specific time period or device. Specifically, when the PLC needs to perform energy consumption analysis, it sends a query request containing specific criteria to the backend database, such as "Please return all boiler energy consumption data marked as 'normal' for the past year." Upon receiving the request, the database management system quickly responds, filtering the data that meets the criteria according to pre-set rules and organizing it into an easily processable time series format, returning it to the PLC. This time series data contains information from multiple dimensions, such as various parameters measured by different sensors, corresponding time points, and associated operating conditions, providing rich material for subsequent in-depth analysis.

[0062] It is worth noting that during the extraction process, necessary data cleaning and verification should also be carried out to ensure the quality of the data obtained. This includes removing duplicates, filling in missing values, and correcting erroneous records. For example, if it is found that the data at certain time points have obvious unreasonable fluctuations, it may be due to sensor failure or other external interference. At this time, it is necessary to adjust or eliminate it appropriately. Only data that has been strictly verified can be truly used to build a model of "normal" energy consumption patterns, and then serve as a comparison benchmark to evaluate the energy consumption status of the current equipment. In this way, not only can we more accurately define what is "normal" energy consumption behavior, but we can also provide early warning of possible problems, helping factories achieve more intelligent energy consumption management.

[0063] In the above-mentioned PLC-based smart thermal power plant energy consumption monitoring method, in step S140, the PLC-based energy consumption monitoring processor performs sequence encoding on the set of energy consumption data time series marked as normal energy consumption to obtain a set of normal energy consumption data time series semantic features. It should be understood that energy consumption data is usually presented in the form of a time series, which contains the energy consumption status of the equipment or system at different time points. By performing sequence encoding on these data, the original numerical data can be converted into high-dimensional features that are easier to understand, analyze and process, thereby achieving more accurate energy consumption pattern recognition and anomaly detection. Among them, the energy consumption data marked as "normal" represents the energy consumption pattern of the equipment under normal working conditions, and these data reflect the typical operating characteristics of the equipment. By performing sequence encoding on these normal energy consumption data, the time series features therein can be extracted, such as key factors such as periodic fluctuations, trend changes, and peak periods.

[0064] In an embodiment of the present application, in step S140, the set of energy consumption data time series marked as having normal energy consumption is sequence-encoded to obtain a set of normal energy consumption data time series semantic features, including: inputting each energy consumption data time series marked as having normal energy consumption in the set of energy consumption data time series marked as having normal energy consumption into an energy consumption data sequence encoder based on an RNN-LSTM hybrid model to obtain a set of normal energy consumption data time series semantic encoding vectors as the set of normal energy consumption data time series semantic features. It should be understood that, considering that each energy consumption data time series marked as having normal energy consumption exhibits dynamic change characteristics in the time dimension, and at the same time, there are interconnected time series dependencies between these sequences in different time periods. Therefore, in order to effectively capture the implicit time dependencies and long-term trends in energy consumption data, in the technical solution of the present application, each energy consumption data time series marked as having normal energy consumption in the set of energy consumption data time series marked as having normal energy consumption is inputted into an energy consumption data sequence encoder based on an RNN-LSTM hybrid model to obtain a set of normal energy consumption data time series semantic encoding vectors. In particular, traditional RNNs are able to handle temporal dependencies in input sequences, that is, the output at the current moment depends not only on the input at the current moment, but also on the state at the previous moment. However, RNNs are prone to vanishing or exploding gradients when processing long time series, making it difficult to capture long-term dependencies. LSTM is a variant of RNN, specifically designed to solve the vanishing gradient problem in RNNs. It controls the flow of information by introducing special "gating" units - input gates, forget gates, and output gates, so that long-term dependencies can be preserved more effectively. Therefore, combining RNN and LSTM units to form an RNN-LSTM hybrid model is an effective method that can fully utilize the advantages of both, enabling it to not only capture local sequence features, but also handle long-term dependencies, thereby improving the model's understanding and prediction capabilities for complex time series data. Specifically, traditional RNN units are used at the front end of the network to capture local features and short-term dynamic changes in the time series of energy consumption data labeled as normal energy consumption. LSTM units are introduced in subsequent layers to focus on capturing deeper temporal dependencies and long-term trends. In this way, both short-term and long-term dependencies of normal energy consumption data can be considered at the same time, thereby providing a richer temporal representation of normal energy consumption data.

[0065] In the above-mentioned PLC-based smart thermal power plant energy consumption monitoring method, in step S150, the PLC-based energy consumption monitoring processor performs sequence encoding on the time queue of the energy consumption data to obtain the time series semantic features of the query energy consumption data. It should be understood that the time series data itself contains the energy consumption information of the equipment or system within a specific time period. However, it is often difficult to reveal potential patterns or anomalies by relying solely on the original numerical data for analysis. Through sequence encoding, these time series data can be converted into features with higher-level semantics, thereby supporting more efficient query, analysis and decision-making. Among them, the key to sequence encoding is to effectively extract features from the energy consumption data in the time queue and convert the original numerical data into a structured and abstract semantic feature set. These features generally include but are not limited to periodic patterns, trend changes, abnormal fluctuations, and dependencies between data. Through such an encoding process, the PLC system can better understand the time structure of the energy consumption data, thereby achieving efficient query and dynamic analysis. For example, through technologies such as Fourier transform, wavelet transform or autoencoding neural network, the frequency domain characteristics, time domain characteristics and even nonlinear characteristics of the data can be extracted, thereby obtaining a deep understanding of the energy consumption data.

[0066] In an embodiment of the present application, the step S150, which performs sequence encoding on the time queue of the energy consumption data to obtain the time series semantic features of the query energy consumption data, includes: inputting the time queue of the energy consumption data into the energy consumption data sequence encoder based on the RNN-LSTM hybrid model to obtain the time series semantic encoding vector of the query energy consumption data as the time series semantic features of the query energy consumption data. It should be understood that, similarly, considering that the energy consumption data contains rich time series features, such as small abnormal changes in energy consumption data in the short term and long-term energy consumption stability. Based on this, in the technical solution of the present application, the time queue of the energy consumption data is input into the energy consumption data sequence encoder based on the RNN-LSTM hybrid model to effectively capture the short-term fluctuations and long-term trends in the energy consumption data, and obtain the time series semantic encoding vector of the query energy consumption data. In this way, the RNN layer can quickly respond to immediate changes, while the LSTM layer can handle complex long-term dependencies, thereby ensuring that the model can both identify minor anomalies and understand the overall energy consumption pattern.

[0067] In the above-mentioned PLC-based smart thermal power plant energy consumption monitoring method, in step S160, the PLC-based energy consumption monitoring processor performs query response encoding on the set of the query energy consumption data time series semantic features and the normal energy consumption data time series semantic features to obtain energy consumption data query response semantic features. It should be understood that, considering that the query energy consumption data time series semantic features represent the specific energy consumption behavior of the device in the current or recent period, they reflect the energy consumption time series pattern of the device at a specific time point. The set of normal energy consumption data time series semantic features defines the typical energy consumption patterns of the device under various standard operating conditions. These patterns constitute a reference benchmark for determining whether new energy consumption data meets expectations. Therefore, in order to quickly filter out the information that is most semantically closely related to the query energy consumption data time series semantic features from the set of normal energy consumption data time series semantic features, in the technical solution of the present application, query response encoding is performed on the set of the query energy consumption data time series semantic features and the normal energy consumption data time series semantic features to obtain energy consumption data query response semantic features. In particular, the query response encoding mechanism can identify the most relevant features between the query vector and the normal vector in a short time through fast semantic matching, thereby providing more accurate anomaly detection.

[0068] Figure 3 The present invention provides a flow chart of the energy consumption monitoring processor based on PLC in the energy consumption monitoring method of a smart thermal power plant based on PLC according to an embodiment of the present application, performing query response encoding on the set of the query energy consumption data time series semantic features and the normal energy consumption data time series semantic features to obtain the energy consumption data query response semantic features. Figure 3 As shown, in an embodiment of the present application, the step S160 includes: S161, calculating the fast semantic matching factors between the query energy consumption data temporal semantic features and the set of normal energy consumption data temporal semantic features to obtain a set of fast semantic matching factors; S162, based on the set of fast semantic matching factors, performing a fast semantic query on the query energy consumption data temporal semantic features and the set of normal energy consumption data temporal semantic features to obtain the energy consumption data query response semantic features.

[0069] Specifically, the step S161 calculates the fast semantic matching factors between the query energy consumption data temporal semantic features and each normal energy consumption data temporal semantic feature in the set of normal energy consumption data temporal semantic features to obtain a set of fast semantic matching factors, including: calculating the mutual information between the query energy consumption data temporal semantic coding vector and each normal energy consumption data temporal semantic coding vector in the set of normal energy consumption data temporal semantic coding vector to obtain a set of query-normal energy consumption data semantic fast semantic matching factors as the set of fast semantic matching factors. It should be understood that in the technical solution of the present application, mutual information is used to evaluate the semantic similarity between different feature vectors. The larger the mutual information value, the more information overlap there is between the two feature vectors, that is, to some extent, the concepts or information represented by the two feature vectors are highly correlated.

[0070] Figure 4 The present invention provides a flowchart of calculating the mutual information between the query energy consumption data time series semantic coding vector and each normal energy consumption data time series semantic coding vector in the set of normal energy consumption data time series semantic coding vectors in the PLC-based smart thermal power plant energy consumption monitoring method according to the embodiment of the present application to obtain a set of query-normal energy consumption data semantic fast semantic matching factors. Figure 4 As shown, in an embodiment of the present application, the step S161, calculating the mutual information between the query energy consumption data temporal semantic coding vector and each normal energy consumption data temporal semantic coding vector in the set of normal energy consumption data temporal semantic coding vectors to obtain a set of query-normal energy consumption data semantic fast semantic matching factors, includes: S1611, inputting the query energy consumption data temporal semantic coding vector into the sigmoid function to obtain a query energy consumption data temporal semantic transformation feature vector; S1612, inputting each normal energy consumption data temporal semantic coding vector in the set of normal energy consumption data temporal semantic coding vectors into the sigmoid function to obtain a set of normal energy consumption data temporal semantic transformation feature vectors; S1613, calculating the mutual information between the query energy consumption data temporal semantic transformation feature vector and each normal energy consumption data temporal semantic transformation feature vector in the set of normal energy consumption data temporal semantic transformation feature vectors to obtain the set of query-normal energy consumption data semantic fast semantic matching factors.

[0071] More specifically, the process of step S161 can be expressed as follows:

[0072] V={v 21 ,v 22 ,...,v 2i ,...,v 2n}

[0073] x1=sigmoid(v1)

[0074] x 2i =sigmoid(v 2i )

[0075] MI i =H(x1)+H(x 2i )-H(x1,x 2i )

[0076] Wherein, V is the set of time series semantic coding vectors of normal energy consumption data, v 21 , v 22 , v 2i and v 2n are the first, second, i-th and n-th normal energy consumption data time series semantic coding vectors in the set of normal energy consumption data time series semantic coding vectors, v1 is the query energy consumption data time series semantic coding vector, sigmoid(·) is the sigmoid function, x1 is the query energy consumption data time series semantic transformation feature vector, x 2i It is v 2i The corresponding normal energy consumption data time series semantic transformation feature vector, H(x1) represents the information entropy of x1, H(x 2i ) represents x 2i The information entropy of H(x1,x 2i ) represents x1 and x 2i The joint entropy between MI i represents x1 and x 2i The query-normal energy consumption data semantic fast semantic matching factor is calculated. In this way, calculating the mutual information not only helps identify the normal energy consumption data time series semantic encoding vector that is most relevant to the query energy consumption data time series semantic encoding vector, but also provides an objective standard for subsequent fast matching in a quantitative manner.

[0077] Figure 5 This is a flowchart of performing a fast semantic query on the set of the query energy consumption data time series semantic features and the normal energy consumption data time series semantic features based on the set of the fast semantic matching factors in the PLC-based smart thermal power plant energy consumption monitoring method according to the embodiment of the present application to obtain the energy consumption data query response semantic features. Figure 5As shown, in an embodiment of the present application, the step S162, based on the set of the fast semantic matching factors, performs a fast semantic query on the set of the query energy consumption data temporal semantic features and the set of the normal energy consumption data temporal semantic features to obtain the energy consumption data query response semantic features, including: S1621, based on the set of the query-normal energy consumption data semantic fast semantic matching factors, determines a subset of the fast-matched normal energy consumption data temporal semantic coding vectors from the set of the normal energy consumption data temporal semantic coding vectors; S1622, performs a linear transformation on the query energy consumption data temporal semantic coding vector to obtain an energy consumption data semantic query vector and an energy consumption data semantic value vector, and uses the subset of the normal energy consumption data temporal semantic coding vector as a subset of the key vector, performs Transformer cross-domain query encoding on the energy consumption data semantic query vector, the energy consumption data semantic value vector and the subset of the key vector to obtain an energy consumption data query response semantic coding vector as the energy consumption data query response semantic feature.

[0078] In an embodiment of the present application, the step S1621, based on the set of query-normal energy consumption data semantic fast semantic matching factors, determines a subset of fast-matched normal energy consumption data temporal semantic coding vectors from the set of normal energy consumption data temporal semantic coding vectors, including: S1621-1, identifying the first query-normal energy consumption data semantic maximum and the second query-normal energy consumption data semantic maximum from the set of query-normal energy consumption data semantic fast semantic matching factors; S1621-2, based on the positions of the first query-normal energy consumption data semantic maximum and the second query-normal energy consumption data semantic maximum in the set of query-normal energy consumption data semantic fast semantic matching factors, determines a subset of fast-matched normal energy consumption data temporal semantic coding vectors from the set of normal energy consumption data temporal semantic coding vectors. It should be understood that identifying the first query-normal energy consumption data semantic maximum and the second query-normal energy consumption data semantic maximum from the set of query-normal energy consumption data semantic fast semantic matching factors is actually a process of finding the two normal energy consumption data temporal semantic encoding vectors that are most semantically relevant to the query energy consumption data temporal semantic encoding vector. Based on the positions of the first query-normal energy consumption data semantic maximum and the second query-normal energy consumption data semantic maximum in the set of query-normal energy consumption data semantic fast semantic matching factors, a subsequence of normal energy consumption data temporal semantic encoding vectors that is a fast match is determined from the sequence of normal energy consumption data temporal semantic encoding vectors. This process can be considered a preliminary feature selection or filtering.

[0079] More specifically, the process of step S1621 can be expressed as follows:

[0080]

[0081] {k sub}={v 2arg max1 ,...,v 2j ,...,v 2arg max2}

[0082] in, is a set of query-normal energy consumption data semantic fast semantic matching factors, n is the number of query-normal energy consumption data semantic fast semantic matching factors in the set of query-normal energy consumption data semantic fast semantic matching factors, max(·) is the maximum value in the set, MI max1 The first query is the maximum semantic value of normal energy consumption data, MI max2 is the second query - the semantic maximum value of normal energy consumption data, arg max(·) is the position of the maximum value in the set, arg max1 is the first determined fast matching position, arg max2 is the second determined fast matching position, {k sub} is a subset of the normal energy consumption data temporal semantic encoding vectors. In this way, the system can focus resources on processing those normal energy consumption data temporal semantic encoding vectors that are most likely to meaningfully interact with the query energy consumption data temporal semantic encoding vector, thereby improving processing efficiency and result quality. The selection of the subsequence of the normal energy consumption data temporal semantic encoding vectors for rapid matching also reflects the system's judgment of the importance of the input data, helping to reduce noise interference and improve the model's generalization ability.

[0083] Specifically, in step S1622, the query energy consumption data time series semantic encoding vector is linearly transformed to obtain an energy consumption data semantic query vector and an energy consumption data semantic value vector, and a subset of the normal energy consumption data time series semantic encoding vector is used as a subset of the key vector, and the energy consumption data semantic query vector, the energy consumption data semantic value vector and the subset of the key vector are subjected to Transformer cross-domain query encoding to obtain an energy consumption data query response semantic encoding vector as the energy consumption data query response semantic feature. It should be understood that the core of the transformer model lies in the self-attention mechanism, which allows the model to dynamically focus on different parts of the sequence when processing sequence data, without having to process each element sequentially like a traditional recurrent neural network. In the technical solution of the present application, the energy consumption data semantic query vector is responsible for finding the key vector that is most relevant to itself, while the energy consumption data semantic value vector carries the specific information of these related parts.

[0084] More specifically, the process of step S1622 can be expressed as follows:

[0085] v q =v1Wq +b q

[0086] v v =v1W v +b v

[0087]

[0088] Among them, W q and b q are the query embedding matrix and query bias vector, respectively, v q is the semantic query vector of energy consumption data, W v and b v are the value embedding matrix and value bias vector, v v is the semantic value vector of energy consumption data, v 2j T It is v 2j The transposed vector of v 2j The length of , softmax(·) is the softmax function, is matrix multiplication, v r is the energy consumption data query response semantic encoding vector. In this way, the converter model can effectively capture and utilize the long-range dependencies between the query energy consumption data time series semantic encoding vector and a subset of the normal energy consumption data time series semantic encoding vectors to generate an accurate energy consumption data query response semantic encoding vector.

[0089] In the aforementioned PLC-based smart thermal power plant energy consumption monitoring method, in step S170, the PLC-based energy consumption monitoring processor obtains a monitoring result indicating whether the target energy-consuming device has abnormal energy consumption based on the semantic features of the energy consumption data query response. It should be understood that the core goal of an energy consumption monitoring system is to monitor the energy consumption status of a device in real time, promptly detect potential anomalies and faults, ensure efficient operation of the device, and avoid excessive energy consumption. In the PLC-based energy consumption monitoring processor, obtaining a monitoring result indicating whether the target energy-consuming device has abnormal energy consumption based on the semantic features of the energy consumption data query response primarily involves in-depth analysis and comparison of the time series features of the energy consumption data to determine whether the device exhibits abnormal energy consumption behavior inconsistent with its normal operating mode. The PLC-based energy consumption monitoring processor utilizes the semantic features of the energy consumption data query response to compare the differences between the time series features of the device's current energy consumption data and historical normal data. Specifically, the system assesses whether the current energy consumption data deviates significantly based on the device's energy consumption pattern and the "normal" features of historical data. For example, if a device's energy consumption suddenly exceeds a predetermined normal range during a specific period, and this change clearly deviates from historical trends, the system will flag this energy consumption data as abnormal based on the query's semantic features. Then, using a classification model based on normal energy consumption characteristics, the system can automatically determine whether a device exhibits abnormal energy consumption behavior based on the changing trends in real-time data, and then issue an alert or recommend appropriate control measures. This semantic feature-based anomaly detection not only helps promptly identify equipment failures and energy waste, but also optimizes equipment operating efficiency and reduces energy consumption.

[0090] In an embodiment of the present application, step S170, based on the semantic features of the energy consumption data query response, obtains a monitoring result indicating whether the target energy-consuming device has abnormal energy consumption, including: inputting the semantic encoding vector of the energy consumption data query response into an energy consumption monitoring module based on an SVM model to obtain the monitoring result, wherein the monitoring result indicates whether the target energy-consuming device has abnormal energy consumption. That is, the semantic features of the energy consumption data query response obtained by performing a rapid query response using a set of the query energy consumption data time series semantic encoding vector and the normal energy consumption data time series semantic encoding vector are processed to intelligently determine whether the target energy-consuming device has abnormal energy consumption. In this way, by integrating advanced artificial intelligence technology and the real-time control capabilities of the PLC, it is possible to more meticulously capture complex energy consumption patterns in device time series data, significantly reducing false positives and missed negatives. Furthermore, the judgment criteria can be automatically adjusted based on real-time energy consumption data, rather than relying solely on preset fixed thresholds, thereby providing more flexible and accurate anomaly detection. The SVM (support vector machine) model is used to process the semantic features based on the energy consumption data query response, aiming to intelligently determine whether the target energy-consuming device has abnormal energy consumption. Specifically, SVM is a supervised learning algorithm commonly used for classification and regression analysis, and performs particularly well when processing data with high-dimensional features. In this scenario, the SVM model compares and analyzes the query response semantic encoding vector of the energy consumption data with the time series semantic encoding vector of the normal energy consumption data to make a judgment on energy consumption anomaly. It is worth mentioning that in the technical solution of the present application, the training process of the SVM model requires a large amount of labeled data as input, which includes the time series semantic encoding vectors of "normal" and "abnormal" energy consumption data. During the training phase, the SVM will find an optimal hyperplane (decision boundary) so that the normal and abnormal energy consumption features can be separated to the maximum extent. Specifically, the SVM will improve the accuracy of classification by maximizing the margin between the two types of data. During the training process, the model continuously adjusts the weights and biases according to the position and label (normal or abnormal) of the feature vector until the optimal hyperplane that can accurately divide the two types of data is found. Then, when the PLC system receives a new query request, the query response semantic encoding vector of the energy consumption data will be passed as input to the trained SVM model. The SVM model uses the optimal hyperplane obtained during training to classify newly input semantic features and output corresponding monitoring results, indicating whether the target energy-consuming device exhibits abnormal energy consumption. If the input query response features have a low similarity to "normal" features, the SVM model will determine that the device's energy consumption behavior is abnormal; otherwise, it will be considered normal. Using the SVM model, the PLC system can dynamically adjust its judgment criteria based on real-time energy consumption data, avoiding the false positives and false negatives that can arise from relying solely on fixed thresholds.

[0091] In summary, the PLC-based smart thermal power plant energy consumption monitoring method based on the embodiment of the present application is explained, which collects the time queue of the energy consumption data of the target energy-consuming equipment and inputs it into the PLC-based energy consumption monitoring processor to extract a set of energy consumption data time series marked as normal energy consumption, and then uses artificial intelligence-based data analysis and encoding technology to intelligently judge whether the target energy-consuming equipment has abnormal energy consumption based on the query energy consumption data time series semantic features obtained by encoding the energy consumption data and the query response semantic representation between the time series semantic features of each normal energy consumption data. In this way, the complex energy consumption patterns in the equipment time series data can be captured more carefully, significantly reducing false alarms and omissions. And its judgment criteria can be automatically adjusted according to the real-time energy consumption data, thereby providing more flexible and accurate anomaly detection.

[0092] Figure 6 FIG is a system block diagram of a PLC-based smart thermal power plant energy consumption monitoring system according to an embodiment of the present application. Figure 6 As shown, according to the PLC-based smart thermal power plant energy consumption monitoring system 100 of the embodiment of the present application, it includes: an energy consumption data time queue acquisition module 110, which is used to collect the time queue of energy consumption data of the target energy consumption equipment through the energy consumption sensor; an energy consumption data time queue input module 120, which is used to input the time queue of the energy consumption data into the PLC-based energy consumption monitoring processor; an energy consumption normal data time series extraction module 130, which is used to extract the set of energy consumption data time series marked as normal energy consumption from the background database in the PLC-based energy consumption monitoring processor; an energy consumption normal data time series encoding module 140, which is used to sequence the set of energy consumption data time series marked as normal energy consumption in the PLC-based energy consumption monitoring processor Encoding to obtain a set of normal energy consumption data temporal semantic features; an energy consumption data time queue sequence encoding module 150, used to perform sequence encoding on the time queue of the energy consumption data in the PLC-based energy consumption monitoring processor to obtain the query energy consumption data temporal semantic features; a query response encoding module 160, used to perform query response encoding on the query energy consumption data temporal semantic features and the set of normal energy consumption data temporal semantic features in the PLC-based energy consumption monitoring processor to obtain energy consumption data query response semantic features; a target energy consumption equipment monitoring result analysis module 170, used to obtain a monitoring result indicating whether the target energy consumption equipment has abnormal energy consumption based on the energy consumption data query response semantic features in the PLC-based energy consumption monitoring processor.

[0093] Here, those skilled in the art will understand that the specific operations of each step in the above-mentioned PLC-based smart thermal power plant energy consumption monitoring system have been referred to above. Figures 1 to 5 The method has been introduced in detail in the description of the PLC-based smart thermal power plant energy consumption monitoring method, and therefore, its repeated description will be omitted.

[0094] As described above, the PLC-based smart thermal power plant energy consumption monitoring system 100 according to the embodiment of the present application can be implemented in various terminal devices. In one example, the PLC-based smart thermal power plant energy consumption monitoring system 100 can be integrated into the terminal device as a software module and / or hardware module. For example, the PLC-based smart thermal power plant energy consumption monitoring system 100 can be a software module in the operating system of the terminal device, or can be an application developed for the terminal device; of course, the PLC-based smart thermal power plant energy consumption monitoring system 100 can also be one of the many hardware modules of the terminal device.

[0095] Alternatively, in another example, the PLC-based smart thermal power plant energy consumption monitoring system 100 and the terminal device may also be separate devices, and the PLC-based smart thermal power plant energy consumption monitoring system 100 may be connected to the terminal device through a wired and / or wireless network, and transmit interactive information in accordance with an agreed data format.

[0096] In summary, the PLC-based smart thermal power plant energy consumption monitoring system based on the embodiment of the present application is explained, which collects the time queue of the energy consumption data of the target energy-consuming equipment and inputs it into the PLC-based energy consumption monitoring processor to extract a set of energy consumption data time series marked as normal energy consumption, and then uses artificial intelligence-based data analysis and encoding technology to intelligently judge whether the target energy-consuming equipment has abnormal energy consumption based on the query energy consumption data time series semantic features obtained by encoding the energy consumption data and the query response semantic representation between the time series semantic features of each normal energy consumption data. In this way, the complex energy consumption patterns in the equipment time series data can be captured more carefully, significantly reducing false alarms and omissions. And its judgment criteria can be automatically adjusted according to the real-time energy consumption data, thereby providing more flexible and accurate anomaly detection.

[0097] Those skilled in the art will appreciate that the above-mentioned embodiments are specific embodiments for implementing the present disclosure, and that in actual applications, various changes may be made thereto in form and detail without departing from the spirit and scope of the present disclosure.

Claims

1. A PLC-based smart thermal power plant energy consumption monitoring method, characterized in that: include: A time queue for collecting energy consumption data of target energy-consuming devices through energy consumption sensors; Inputting the time queue of the energy consumption data into a PLC-based energy consumption monitoring processor; In the PLC-based energy consumption monitoring processor, a set of energy consumption data time series marked as normal energy consumption is extracted from a background database; In the PLC-based energy consumption monitoring processor, sequence encoding is performed on the set of energy consumption data time series marked as normal energy consumption to obtain a set of normal energy consumption data time series semantic features; In the PLC-based energy consumption monitoring processor, the time queue of the energy consumption data is sequence-encoded to obtain a time series semantic feature of the query energy consumption data; In the PLC-based energy consumption monitoring processor, query response encoding is performed on the set of the query energy consumption data temporal semantic features and the normal energy consumption data temporal semantic features to obtain energy consumption data query response semantic features, including: calculating a fast semantic matching factor between each normal energy consumption data temporal semantic feature in the set of the query energy consumption data temporal semantic features and the normal energy consumption data temporal semantic features to obtain a set of fast semantic matching factors; based on the set of fast semantic matching factors, performing a fast semantic query on the set of the query energy consumption data temporal semantic features and the normal energy consumption data temporal semantic features to obtain the energy consumption data query response semantic features; In the PLC-based energy consumption monitoring processor, based on the energy consumption data query response semantic features, a monitoring result indicating whether the target energy-consuming device has abnormal energy consumption is obtained.

2. The PLC-based smart thermal power plant energy consumption monitoring method according to claim 1 is characterized in that: The set of energy consumption data time series marked as normal energy consumption is sequence-encoded to obtain a set of normal energy consumption data time series semantic features, including: inputting each energy consumption data time series marked as normal energy consumption in the set of energy consumption data time series marked as normal energy consumption into an energy consumption data sequence encoder based on an RNN-LSTM hybrid model to obtain a set of normal energy consumption data time series semantic coding vectors as the set of normal energy consumption data time series semantic features.

3. The PLC-based smart thermal power plant energy consumption monitoring method according to claim 2 is characterized in that: The time queue of the energy consumption data is sequence-encoded to obtain the time series semantic features of the query energy consumption data, including: inputting the time queue of the energy consumption data into the energy consumption data sequence encoder based on the RNN-LSTM hybrid model to obtain the time series semantic encoding vector of the query energy consumption data as the time series semantic features of the query energy consumption data.

4. The PLC-based smart thermal power plant energy consumption monitoring method according to claim 3 is characterized in that: Calculate the fast semantic matching factors between the query energy consumption data temporal semantic features and each normal energy consumption data temporal semantic feature in the set of normal energy consumption data temporal semantic features to obtain a set of fast semantic matching factors, including: calculating the mutual information between the query energy consumption data temporal semantic coding vector and each normal energy consumption data temporal semantic coding vector in the set of normal energy consumption data temporal semantic coding vector to obtain a set of query-normal energy consumption data semantic fast semantic matching factors as the set of fast semantic matching factors.

5. The PLC-based smart thermal power plant energy consumption monitoring method according to claim 4 is characterized in that: Calculating the mutual information between the query energy consumption data time series semantic coding vector and each normal energy consumption data time series semantic coding vector in the set of normal energy consumption data time series semantic coding vectors to obtain a set of query-normal energy consumption data semantic fast semantic matching factors, including: Inputting the query energy consumption data time series semantic encoding vector into a sigmoid function to obtain a query energy consumption data time series semantic transformation feature vector; Inputting each normal energy consumption data time series semantic coding vector in the set of normal energy consumption data time series semantic coding vectors into the sigmoid function to obtain a set of normal energy consumption data time series semantic transformation feature vectors; The mutual information between the query energy consumption data temporal semantic transformation feature vector and each normal energy consumption data temporal semantic transformation feature vector in the set of normal energy consumption data temporal semantic transformation feature vectors is calculated to obtain a set of query-normal energy consumption data semantic fast semantic matching factors.

6. The PLC-based smart thermal power plant energy consumption monitoring method according to claim 5 is characterized in that: Based on the set of the fast semantic matching factors, a fast semantic query is performed on the set of the query energy consumption data time series semantic features and the normal energy consumption data time series semantic features to obtain the energy consumption data query response semantic features, including: Based on the set of query-normal energy consumption data semantic fast semantic matching factors, determining a subset of fast-matched normal energy consumption data time series semantic coding vectors from the set of normal energy consumption data time series semantic coding vectors; The query energy consumption data time series semantic coding vector is linearly transformed to obtain an energy consumption data semantic query vector and an energy consumption data semantic value vector, and a subset of the normal energy consumption data time series semantic coding vector is used as a subset of the key vector. The energy consumption data semantic query vector, the energy consumption data semantic value vector and the subset of the key vector are subjected to Transformer cross-domain query encoding to obtain an energy consumption data query response semantic coding vector as the energy consumption data query response semantic feature.

7. The PLC-based smart thermal power plant energy consumption monitoring method according to claim 6 is characterized in that: Based on the set of query-normal energy consumption data semantic fast semantic matching factors, determining a subset of fast-matched normal energy consumption data time series semantic coding vectors from the set of normal energy consumption data time series semantic coding vectors, including: Identify a first query-normal energy consumption data semantic maximum value and a second query-normal energy consumption data semantic maximum value from the set of query-normal energy consumption data semantic fast semantic matching factors; Based on the positions of the first query-normal energy consumption data semantic maximum and the second query-normal energy consumption data semantic maximum in the set of query-normal energy consumption data semantic fast semantic matching factors, a subset of the normal energy consumption data temporal semantic coding vectors that are quickly matched is determined from the set of normal energy consumption data temporal semantic coding vectors.

8. The PLC-based smart thermal power plant energy consumption monitoring method according to claim 7 is characterized in that: Based on the semantic features of the energy consumption data query response, a monitoring result is obtained to indicate whether the target energy consumption device has abnormal energy consumption, including: inputting the energy consumption data query response semantic encoding vector into an energy consumption monitoring module based on an SVM model to obtain the monitoring result, and the monitoring result is used to indicate whether the target energy consumption device has abnormal energy consumption.

9. A PLC-based smart thermal power plant energy consumption monitoring system, characterized in that: include: Energy consumption data time queue collection module, used to collect the time queue of energy consumption data of target energy consumption equipment through energy consumption sensors; An energy consumption data time queue input module, used for inputting the time queue of the energy consumption data into the PLC-based energy consumption monitoring processor; A normal energy consumption data time series extraction module is used to extract a set of energy consumption data time series marked as normal energy consumption from a background database in the PLC-based energy consumption monitoring processor; a normal energy consumption data time series encoding module, configured to perform sequence encoding on the set of energy consumption data time series marked as normal energy consumption in the PLC-based energy consumption monitoring processor to obtain a set of normal energy consumption data time series semantic features; An energy consumption data time queue sequence encoding module is used to perform sequence encoding on the time queue of the energy consumption data in the PLC-based energy consumption monitoring processor to obtain a time series semantic feature of the query energy consumption data; A query response encoding module is used to perform query response encoding on the set of the query energy consumption data time series semantic feature and the normal energy consumption data time series semantic feature in the PLC-based energy consumption monitoring processor to obtain an energy consumption data query response semantic feature; The target energy-consuming equipment monitoring result analysis module is used to obtain a monitoring result indicating whether the target energy-consuming equipment has abnormal energy consumption based on the semantic features of the energy consumption data query response in the PLC-based energy consumption monitoring processor.

10. The PLC-based smart thermal power plant energy consumption monitoring system according to claim 9 is characterized in that: The query response encoding module is used to: Calculating a fast semantic matching factor between the query energy consumption data time series semantic feature and each normal energy consumption data time series semantic feature in the set of normal energy consumption data time series semantic features to obtain a set of fast semantic matching factors; Based on the set of fast semantic matching factors, a fast semantic query is performed on the set of the query energy consumption data time series semantic features and the normal energy consumption data time series semantic features to obtain the energy consumption data query response semantic features.

Images