Abnormality diagnosis method and device for heat supply system, storage medium and computer equipment

Abstract

The invention relates to the technical field of heat supply operation, in particular to a heat supply system abnormity diagnosis method and device, a storage medium and computer equipment. The method comprises the following steps: constructing a model training set based on multi-source heterogeneous data of the heat supply system in a historical period; training a preset deep neural network by using the model training set to obtain a prediction model for predicting the reading of the heat supply sensor; based on the model training set and the prediction model, determining reading prediction error thresholds of the prediction model for the heat supply sensor under different operation conditions of the heat supply system; determining a real-time reading prediction error of the prediction model for the heat supply sensor based on the multi-source heterogeneous data of the heat supply system in the monitoring period and the prediction model; and comparing the real-time reading prediction error with a reading prediction error threshold value, and determining an abnormal sensor in the heat supply sensors. According to the method, high-precision and self-adaptive diagnosis of the abnormity of the heat supply sensor in the heat supply system is realized, and the fault identification accuracy and the operation and maintenance efficiency are remarkably improved.

Description

Technical Field

[0001] This application relates to the field of heating operation technology, and in particular to a method, device, storage medium and computer equipment for diagnosing abnormalities in a heating system. Background Technology

[0002] With the rapid development of smart heating systems, relying on massive amounts of sensor data for operation monitoring and optimization decisions has become the norm in the industry.

[0003] In related technologies, production and operation data, user-side data, business management data, and external environmental data of heating systems are typically stored in independent systems, making it impossible to conduct collaborative analysis and accurate insights from a holistic system perspective. Secondly, the assessment of sensor data quality largely relies on pre-set static threshold alarm mechanisms, which are extremely insensitive to complex anomaly patterns such as slowly occurring sensor drift and intermittent failures, resulting in severe false alarms and missed alarms. This leads to problems such as low data utilization, limited intelligence levels, high operation and maintenance costs, and significant decision-making risks. Summary of the Invention

[0004] In view of this, this application provides a method, device, storage medium and computer equipment for diagnosing anomalies in a heating system. By integrating multi-source data and utilizing predictive models and dynamic threshold technology, it achieves high-precision and adaptive diagnosis of anomalies in heating sensors within the heating system, significantly improving data quality, fault identification accuracy and system operation and maintenance efficiency.

[0005] According to one aspect of this application, a method for diagnosing abnormalities in a heating system is provided, comprising: A model training set was constructed based on multi-source heterogeneous data from the heating system over a historical period. The preset deep neural network is trained using the model training set to obtain a prediction model for predicting the readings of heating sensors deployed in the heating site. Based on the model training set and the prediction model, the prediction error threshold of the prediction model for the readings of the heating sensor under different operating conditions of the heating system is determined. Based on the multi-source heterogeneous data of the heating system during the monitoring period and the prediction model, the prediction error of the prediction model for the real-time reading of the heating sensor is determined. Based on the multi-source heterogeneous data of the heating system during the monitoring period, the prediction error of the real-time reading of the heating sensor by the prediction model is compared with the reading prediction error threshold to identify the abnormal sensor among the heating sensors.

[0006] According to another aspect of this application, a heating system anomaly diagnosis device is provided, comprising: The training module is used to construct a model training set based on multi-source heterogeneous data of the heating system in historical periods; and to train a preset deep neural network using the model training set to obtain a prediction model for predicting the readings of heating sensors deployed in the heating site. The determination module is configured to: determine, based on the model training set and the prediction model, a prediction error threshold for the readings of the heating sensor under different operating conditions of the heating system; determine, based on the multi-source heterogeneous data of the heating system during the monitoring period and the prediction model, a real-time reading prediction error for the heating sensor; and determine, based on the multi-source heterogeneous data of the heating system during the monitoring period, an abnormal sensor among the heating sensors by comparing the real-time reading prediction error of the prediction model for the heating sensor with the reading prediction error threshold.

[0007] According to another aspect of this application, a readable storage medium is provided having a program or instructions stored thereon, which, when executed by a processor, implement the steps of the above-described heating system anomaly diagnosis method.

[0008] According to another aspect of this application, a computer device is provided, including a storage medium, a processor, and a computer program stored on the storage medium and executable on the processor, wherein the processor executes the program to implement the steps of the above-described heating system abnormality diagnosis method.

[0009] By employing the above technical solutions, this application provides a method, apparatus, storage medium, and computer equipment for diagnosing anomalies in heating systems. It constructs a model training set by aggregating historical multi-source heterogeneous data from the heating system and trains a predictive model capable of predicting heating sensor readings based on this set. Furthermore, an operating condition sensing mechanism is introduced to determine the dynamic reading prediction error threshold of the predictive model under different operating conditions based on the historical multi-source heterogeneous data. Finally, in the application stage, the reading prediction error of real-time multi-source heterogeneous data is calculated and compared with the dynamic threshold under the corresponding operating condition, combined with business data verification, thereby accurately identifying abnormal sensors. Thus, this embodiment improves the availability and consistency of data across the entire heating system through multi-source data fusion and preprocessing, laying a high-quality data foundation for upper-level intelligent applications. Moreover, by fusing a recurrent neural network prediction model and a dynamic threshold algorithm, real-time, online, and intelligent evaluation of sensor data is achieved, which can not only keenly capture sudden anomalies but also effectively identify complex fault modes such as progressive drift, significantly improving the accuracy and timeliness of diagnosis. Furthermore, this embodiment realizes the transformation from relying on regular manual inspections to data-based predictive maintenance, which significantly reduces the cost of manual intervention and the decision-making risks caused by abnormal sensor data, providing a reliable guarantee for the safe, stable, efficient and economical operation of the smart heating system.

[0010] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application. Attached Figure Description

[0011] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments of this application and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings: Figure 1 A flowchart illustrating the abnormal diagnosis method for a heating system provided in an embodiment of this application is shown. Figure 2 A flowchart illustrating a heating system anomaly diagnosis method according to another embodiment of this application is shown; Figure 3 A structural block diagram of the heating system anomaly diagnosis device provided in an embodiment of this application is shown. Detailed Implementation

[0012] The present application will be described in detail below with reference to the accompanying drawings and embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in the embodiments of the present application can be combined with each other.

[0013] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain this application, and should not be construed as limiting this application.

[0014] Those skilled in the art will understand that, unless specifically stated otherwise, the singular forms “a,” “an,” “the,” and “the” used herein may also include the plural forms. It should be further understood that the term “comprising” as used in this application means the presence of the stated features, integers, steps, operations, elements, and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof. It should be understood that when we say an element is “connected” or “attached” to another element, it can be directly connected or attached to the other element, or there may be intermediate elements. Furthermore, “connected” or “attached” as used herein can include wireless connections or wireless interconnections. The term “and / or” as used herein includes all or any unit and all combinations of one or more associated listed items.

[0015] Exemplary embodiments according to this application will now be described in more detail with reference to the accompanying drawings. However, these exemplary embodiments may be implemented in many different forms and should not be construed as being limited to the embodiments set forth herein. It should be understood that these embodiments are provided so that the disclosure of this application is thorough and complete, and that the concept of these exemplary embodiments is fully conveyed to those skilled in the art.

[0016] This application provides a method for diagnosing abnormalities in a heating system, such as... Figure 1 As shown, the method includes: Step 101: Construct a model training set based on multi-source heterogeneous data of the heating system during historical periods.

[0017] Step 102: Train the preset deep neural network using the model training set to obtain a prediction model for predicting the readings of heating sensors deployed in the heating site.

[0018] Step 103: Based on the model training set and the prediction model, determine the prediction error threshold for the readings of the heating sensor under different operating conditions of the heating system.

[0019] Step 104: Based on the multi-source heterogeneous data and prediction model of the heating system during the monitoring period, determine the prediction error of the prediction model for the real-time readings of the heating sensor.

[0020] Step 105: Based on the multi-source heterogeneous data of the heating system during the monitoring period, compare the prediction error of the real-time reading of the heating sensor with the reading prediction error threshold of the prediction model to identify the abnormal sensors in the heating sensor.

[0021] The heating system anomaly diagnosis method provided in this application can be applied to a terminal, a server, or software running on either a terminal or a server. In some embodiments, the terminal can be a smartphone, tablet, laptop, desktop computer, etc.; the server can be configured as an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms; the software can be an application that implements the heating system anomaly diagnosis method, but is not limited to the above forms.

[0022] This embodiment first constructs a model training set by aggregating historical multi-source heterogeneous data from the heating system, and trains a predictive model capable of predicting heating sensor readings based on this set. Then, an operating condition sensing mechanism is introduced to determine the dynamic reading prediction error threshold of the predictive model under different operating conditions based on the historical multi-source heterogeneous data. Finally, in the application phase, the reading prediction error of real-time multi-source heterogeneous data is calculated and compared with the dynamic threshold under the corresponding operating condition, combined with business data verification, thereby accurately identifying abnormal sensors. Thus, this embodiment improves the availability and consistency of data across the entire heating system through multi-source data fusion and preprocessing, laying a high-quality data foundation for upper-level intelligent applications. Furthermore, by fusing a recurrent neural network prediction model and a dynamic threshold algorithm, real-time, online, and intelligent evaluation of sensor data is achieved, which can not only keenly capture sudden anomalies but also effectively identify complex fault modes such as progressive drift, significantly improving the accuracy and timeliness of diagnosis. Furthermore, this embodiment realizes the transformation from relying on regular manual inspections to data-based predictive maintenance, which significantly reduces the cost of manual intervention and the decision-making risks caused by abnormal sensor data, providing a reliable guarantee for the safe, stable, efficient and economical operation of the smart heating system.

[0023] Another embodiment of this application provides a method for diagnosing abnormalities in a heating system, such as... Figure 2 As shown, the method includes: Step 201: Obtain multi-source heterogeneous data of the heating system during historical periods.

[0024] It should be noted that a heating system is a comprehensive technical system with boiler room units, outdoor heating pipe networks and radiators as its main components, and heat energy is transmitted and distributed through the circulation of heat medium.

[0025] In this embodiment, multi-source heterogeneous data from the heating system can be acquired to aggregate all potentially related information and break down data silos. Specifically, the multi-source heterogeneous data includes the heating system's production and operation data, external status data, heating effect data, and event records. It should be noted that the heating system's production and operation data is collected in real time by heating sensors deployed throughout the heating system site. This production and operation data includes the readings from each heating sensor. Specifically, the heating sensors are those installed at key heating locations within the heating system.

[0026] Specifically, key heating locations can include heat sources such as boiler outlets, heating networks such as main pipelines, heating stations, and key user locations such as building entrances. Heating sensors can include measuring instruments such as temperature sensors, pressure transmitters, flow meters, and heat meters. Production and operation data can include heat source temperature, pressure, flow rate, and heat volume. External status data of the heating system can include ambient temperature, meteorological data, and holiday information. Heating system performance and event records can include user-side data, business data, and safety data. User-side data includes room temperature data and heat metering data; business data includes billing data, customer complaint data, and work order data; and safety data includes heating equipment status data, alarm records, and inspection records.

[0027] In practical applications, multi-source heterogeneous data is typically distributed across various data sources. A unified standard interface can be used to collect this multi-source heterogeneous data from key data sources associated with the heating system in a distributed manner. Specifically, key data sources may include heat metering devices, room temperature acquisition devices, station control systems, source station monitoring systems, and business management platforms.

[0028] It is worth mentioning that, in this embodiment, by using Industrial Internet identifier resolution technology, each data acquisition device used to collect multi-source heterogeneous data is pre-assigned a unique identifier, thus ensuring that each piece of data in the acquired multi-source heterogeneous data of the heating system carries a unique identifier. For example, the outlet temperature sensor of heat exchange station No. 1, and the room temperature data acquisition device of room XX in building XX of XX community. Therefore, this embodiment can acquire standardized multi-source data streams with unified spatiotemporal identifiers, and any data can be accurately located to a specific device or user in the heating system. It is understood that the data acquisition devices include heating sensors.

[0029] In this step, multi-source heterogeneous data of the heating system during historical periods are obtained.

[0030] Step 202: Based on the heating effect and event records of the heating system in historical periods, filter out the production operation data and external status data of the heating system under normal operation in historical periods; determine the training sub-samples based on the data belonging to the same historical moment in the production operation data and external status data of the heating system under normal operation in historical periods; determine the training samples based on the training sub-samples of historical moments in historical periods of the heating system in historical periods; use the production operation data of the predicted moment corresponding to the historical period to which the training sample belongs as the prediction label of the training sample; construct the model training set based on the training samples carrying prediction labels.

[0031] In this step, based on the heating effect and event records from the multi-source heterogeneous data of the heating system over a historical period, production operation data and external status data belonging to the normal operating state of the heating system are filtered out from the multi-source heterogeneous data of the heating system over a historical period. For example, the production operation data and external status data corresponding to the time periods in the heating effect and event records where the user room temperature was normal and there were no user complaints are regarded as the production operation data and external status data under the normal operating state. Next, the data belonging to the same historical moment in the production operation data and external status data of the heating system under the normal operating state over a historical period are formed into a training subsample. Furthermore, in each training subsample, based on the identity identifier carried by each data, all data in the training subsample are arranged in the same identity identifier order, so that data collected by the same acquisition device in the training subsample are placed in the same position. For a specific example, the training subsample is represented as: [timestamp, reading value of heating sensor 1, reading value of heating sensor 2, reading value of heating sensor 3, external ambient temperature collected by acquisition device 1, ...], such as [timestamp, water supply temperature of station 1, return water temperature of station 1, flow rate of station 1, outdoor temperature, ...].

[0032] Then, the historical period is divided into multiple adjacent historical time periods of equal duration. Training sub-samples corresponding to all historical moments within each historical time period are arranged chronologically to form training samples. This ensures that the training samples include not only multiple feature parameters such as temperature, pressure, and flow rate, but also serve as historical multivariate time series for subsequent model training. Simultaneously, the future moment after a preset prediction period is taken as the prediction moment corresponding to that historical time period. Therefore, the production and operation data corresponding to the prediction moment of the historical time period in the multi-source heterogeneous data of the heating system during the historical period are used as prediction labels for the training samples corresponding to that historical time period. This allows the preset deep neural network to predict the production and operation data of the heating system at the prediction moment corresponding to that historical time period based on the data from the historical time periods within the training samples. This results in predicted production and operation data that matches the actual situation at the prediction moment of the heating system at that historical time period, i.e., the predicted reading values ​​of each heating sensor at the prediction moment corresponding to that historical time period. Thus, a model training set is formed based on the training samples. The model training set is used to provide the normal patterns of how the production operation data collected by the heating sensors change in coordination under various external conditions and internal operating conditions of the heating system.

[0033] For specific examples, the historical period can be the previous year, the historical time period can be 10-30 minutes, and the preset prediction time period can be 5-10 minutes.

[0034] In this step, after obtaining the model training set, it needs to be preprocessed to update it. Specifically, for missing data in the training samples of the model training set, intelligent imputation can be performed using strongly correlated data. For example, if a certain training subsample in the model training set is missing the water supply temperature of Station 1, the return water temperature, flow rate, and other data of Station 1 that are strongly correlated with the historical time corresponding to that training subsample can be used to estimate and thus impute the missing data. Furthermore, wavelet transform can be used to filter out meaningless high-frequency spikes in the training samples caused by electromagnetic interference, etc. At the same time, the training samples can also be standardized to normalize data of different dimensions such as temperature and pressure to the [0,1] interval, eliminating the influence of numerical scale on model training.

[0035] This embodiment significantly improves the availability and consistency of data across the entire heating system through the model training set, constructing a high-quality data foundation. Preprocessing eliminates invalid data and biases caused by different units of measurement, enhancing data consistency and model training effectiveness. Thus, the high-quality preprocessed data effectively supports the reliability of subsequent data analysis and optimized control, reducing decision-making risks and manual processing costs caused by data anomalies.

[0036] Step 203: Use the model training set to train the preset deep neural network to obtain a prediction model for predicting the readings of heating sensors deployed in the heating site.

[0037] In this step, a pre-set deep neural network is trained using a model training set to obtain a prediction model, enabling the model to predict the true values ​​of heating sensor readings at future times. Specifically, when production operation data and external status data of the heating system for any historical period are input into the prediction model, the model can output the predicted production operation data of the heating system at the predicted time corresponding to that historical period, i.e., the predicted reading values ​​of each heating sensor at the predicted time corresponding to that historical period.

[0038] Here, the preset deep neural network can be a recurrent neural network (RNN), a long short-term memory network (LSTM), or a gated recurrent unit (GRU) to utilize the preset deep neural network to memorize the historical sequence over a period of time and predict the value at the next future moment.

[0039] This embodiment uses a recurrent neural network structure for model training. Through its powerful time-series modeling capabilities, it learns the dynamic coupling relationship and evolution law between multiple sensor parameters of the heating system under normal operating conditions, thereby establishing a predictive model that can accurately reflect the system characteristics.

[0040] Step 204: Determine the operating conditions of the heating system within a historical period, and group the training samples in the model training set belonging to the same operating conditions into the same operating condition group; input the training samples in the operating condition group into the prediction model to obtain the predicted production operation data of the heating system at the predicted time corresponding to the historical period to which the training samples in the operating condition group belong; the predicted production operation data of the heating system at the predicted time corresponding to the historical period to which the training samples in the operating condition group belong includes the predicted reading values ​​of the heating sensors at the predicted time corresponding to the historical period to which the training samples in the operating condition group belong; determine the prediction error of the prediction model for the readings of the heating sensors under the operating conditions corresponding to the operating conditions of the operating condition group based on the difference between the predicted reading values ​​and the actual reading values ​​of the heating sensors under the historical period to which the training samples in the operating condition group belong; determine the prediction error threshold of the prediction model for the readings of the heating sensors under the operating conditions corresponding to the operating condition group based on the statistical analysis value of the prediction error of the prediction model for the readings of the heating sensors under the operating conditions corresponding to the operating condition group.

[0041] In this step, each training sample from the model training set is input into the trained prediction model to obtain the predicted production and operation data of the heating system at the predicted time corresponding to the historical period to which the training sample belongs. Simultaneously, the actual production and operation data of the heating system at the predicted time corresponding to the historical period to which the training sample belongs is obtained. It can be understood that the actual production and operation data of the heating system at the predicted time corresponding to the historical period to which the training sample belongs includes the actual readings of each heating sensor at the predicted time corresponding to the historical period to which the training sample belongs.

[0042] Therefore, based on the difference between the predicted production operation data and the actual production operation data at the predicted time corresponding to the historical period to which the training sample belongs, that is, the difference between the predicted reading value and the actual reading value of each heating sensor at the predicted time corresponding to the historical period to which the training sample belongs, the prediction error of the prediction model for the reading of each heating sensor is obtained.

[0043] It is worth mentioning that the prediction error of the prediction model for each heating sensor reading is used to represent the normal, acceptable range of deviation of the heating sensor reading from the predicted reading of the prediction model under historical normal operating conditions.

[0044] For example, if the predicted reading of the temperature sensor in the heating sensor is 25℃ at the predicted time corresponding to the historical period of the training sample, while the actual reading of the temperature sensor at the predicted time corresponding to the historical period of the training sample is 25.5℃, then the prediction model has a prediction error of 0.5℃ for the reading of the temperature sensor.

[0045] Furthermore, the operating conditions of the heating system in historical periods are determined, and training samples belonging to the same operating condition are grouped into the same operating condition group. Specifically, different operating conditions can be determined based on different combinations of the total load rate, operating mode, and external ambient temperature range of the heating system. For example, the total load rate can be divided into high load, medium load, and low load, etc.; the operating mode can be divided into the initial heating period, the final heating period, and the severe cold period, etc.; and the external ambient temperature range can be divided into <-15℃, -15℃~-5℃, -5℃~5℃, and >5℃. Then, one operating condition can be the initial heating period and (-15~-5℃) with a high load. If a training sample belongs to the initial heating period in history, and the external ambient temperature in the training sample is all within (-15~-5℃), and the load rate calculated based on power is 92% (high load), then the training sample belongs to this operating condition.

[0046] Furthermore, for any operating condition group, the prediction error of the prediction model for each heating sensor reading, obtained based on the training samples in that operating condition group, is the prediction error of the prediction model for that heating sensor reading under that operating condition of the heating system. Therefore, for each heating sensor, the statistical analysis value of the prediction model's prediction error for that heating sensor reading under that operating condition is used as the prediction error threshold for that heating sensor reading under that operating condition. Specifically, the statistical analysis value can be the 99th percentile. The prediction error threshold for the heating sensor reading under that operating condition is expressed as follows: under that operating condition, the prediction error of the prediction model for that heating sensor reading in 99% of its historical data is less than this value. Thus, for any operating condition group, the prediction error threshold for each heating sensor reading under that operating condition can be obtained.

[0047] Step 205: Based on the production operation data and external status data of the heating system during the monitoring period, construct a real-time prediction sample; input the real-time prediction sample into the prediction model to obtain the predicted production operation data of the heating system at the prediction time corresponding to the monitoring period; based on the predicted production operation data and the actual production operation data of the heating system at the prediction time corresponding to the monitoring period, determine the prediction error of the prediction model for the real-time readings of the heating sensor.

[0048] In this step, multi-source heterogeneous data of the heating system during the monitoring period are acquired. Using the same method as generating training samples, the production operation data of the heating system during the monitoring period and the external status data are combined to form a real-time prediction sample in the form of a multivariate time series. The real-time prediction sample is preprocessed in the same way as the training sample to ensure that the sample format is consistent and to avoid misjudgment caused by inconsistent data processing. This embodiment will not be described in detail here.

[0049] Next, the real-time predicted samples of the heating system are input into the prediction model to obtain the predicted production and operation data of the heating system at the predicted time corresponding to the monitoring cycle, thereby obtaining a theoretical reference for the heating system under normal conditions. Then, based on the differences between the predicted production and operation data of the heating system at the predicted time corresponding to the monitoring cycle and the actual production and operation data, that is, the differences between the predicted readings and actual readings of each heating sensor at the predicted time corresponding to the monitoring cycle, the prediction error of the prediction model for the real-time readings of each heating sensor is obtained.

[0050] Step 206: Determine the target operating condition for real-time prediction sample matching; compare the prediction error of the prediction model for the real-time reading of the heating sensor with the prediction error threshold of the prediction model for the reading of the heating sensor under the target operating condition; based on the comparison results, determine the candidate abnormal sensors among the heating sensors; based on the heating effect and event records of the heating system during the monitoring period, verify the candidate abnormal sensors to obtain the abnormal sensors.

[0051] In this step, the real-time predicted samples are matched with the operating conditions obtained in the previous steps to obtain the target operating conditions matching the real-time predicted samples. Then, for any heating sensor, the prediction error of the real-time reading of the heating sensor by the prediction model is compared with the prediction error threshold of the reading of the heating sensor under the target operating conditions. Based on the heating effect and event records of the heating system within the monitoring period, abnormal sensors are identified. Here, the future time after the monitoring period has passed the preset prediction period is taken as the prediction time corresponding to that historical period.

[0052] Specifically, if the prediction error of the real-time reading of the heating sensor by the prediction model is greater than the prediction error threshold of the heating sensor's reading under the target operating conditions, the reading of the heating sensor is determined to deviate from the normal range, and the heating sensor is preliminarily determined to be abnormal, thus obtaining candidate abnormal sensors. If the prediction error of the real-time reading of the heating sensor by the prediction model is less than or equal to the prediction error threshold of the heating sensor's reading under the target operating conditions, the heating sensor is determined to be working normally, the data is reliable, no alarm is triggered, and unnecessary intervention is avoided. This not only improves the real-time performance of fault detection but also greatly reduces the false alarm rate.

[0053] At this point, by combining the heating effect and event records of the heating system during the monitoring period, if the heating effect and event records of the heating system during the monitoring period contain user complaints, maintenance work orders and abnormal room temperature related to the candidate abnormal sensor, then the candidate abnormal sensor is determined to be an abnormal sensor, so as to effectively identify various abnormal states such as heating sensor drift, blockage, and complete failure.

[0054] This embodiment can perform anomaly diagnosis and status identification on real-time sensor data, and adaptively determine the abnormal alarm thresholds of various sensors according to different operating conditions. It provides a scientific and flexible discrimination benchmark for the online diagnosis stage, significantly improves the accuracy of fault detection, and thus realizes online automatic verification and fault early warning of the working status of sensors in the heating system.

[0055] Step 207: Generate structured diagnostic results for the heating system based on the abnormal sensors, and visualize and push the diagnostic results in real time.

[0056] In this step, structured diagnostic results for the heating system are generated based on the abnormal sensors. These results are then visualized and pushed in real-time to the operation and maintenance monitoring screen or the staff's mobile app. This guides maintenance personnel to prioritize addressing truly problematic equipment, achieving a shift from routine inspections to predictive maintenance. Furthermore, normal data associated with normal sensors (excluding those with abnormal sensors) deployed on-site within the heating system are periodically added to the model training set, initiating incremental learning of the predictive model. This enables this embodiment to possess self-evolution capabilities, adapting to changes in the heating system over time.

[0057] For example, the diagnostic results could be [heating sensor identification, timestamp, diagnostic status (normal / abnormal), abnormality type (sudden / drift), confidence level, and recommended measures].

[0058] This embodiment performs cleaning, noise reduction, formatting, and normalization on the raw data, greatly reducing manual intervention and improving data preprocessing efficiency and quality, ensuring high availability of data used in subsequent analysis. Furthermore, by integrating a recurrent neural network prediction model and a dynamic threshold algorithm, it achieves real-time online evaluation of sensor data, not only capturing sudden and severe anomalies but also accurately identifying gradual data drift and complex fault modes, resulting in high diagnostic accuracy. Simultaneously, it can utilize continuously generated new data for online or offline updates and training, enabling the diagnostic model to adapt to changes in system operating conditions, dynamically adjust its prediction and diagnostic boundaries, and continuously optimize performance. In addition, this embodiment possesses excellent system compatibility and operating condition adaptability, allowing it to be embedded into existing heating management systems and play a crucial role, fundamentally avoiding decision-making errors caused by poor-quality data, and providing solid data quality assurance for the safe, stable, and efficient operation of smart heating systems.

[0059] It should be noted that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0060] Furthermore, such as Figure 3As shown, as a specific implementation of the above-mentioned heating system anomaly diagnosis method, this application embodiment provides a heating system anomaly diagnosis device 300, which includes: a training module 301, a determination module 302 and a diagnosis module 303.

[0061] Among them, the training module 301 is used to construct a model training set based on multi-source heterogeneous data of the heating system in historical periods; and to use the model training set to train a preset deep neural network to obtain a prediction model for predicting the readings of heating sensors deployed in the heating site. The determination module 302 is used to determine the prediction error threshold of the heating sensor readings under different operating conditions of the heating system based on the model training set and the prediction model; and to determine the prediction error of the real-time readings of the heating sensor based on the multi-source heterogeneous data of the heating system during the monitoring period and the prediction model. The diagnostic module 303 is used to identify abnormal sensors in the heating system by comparing the prediction error of the real-time reading of the heating sensor with the reading prediction error threshold based on the multi-source heterogeneous data of the heating system during the monitoring period.

[0062] Optionally, the training module 301 is specifically used to: filter out production operation data and external status data of the heating system under normal operation during the historical period based on the heating effect and event records of the heating system during the historical period; determine training sub-samples based on the data belonging to the same historical moment in the production operation data and external status data of the heating system under normal operation during the historical period; determine training samples based on the training sub-samples of historical moments within the historical period of the heating system; use the production operation data of the predicted moment corresponding to the historical period to which the training sample belongs as the prediction label of the training sample; and construct a model training set based on the training samples carrying prediction labels.

[0063] Optionally, module 302 is specifically used to determine the operating conditions of the heating system within a historical period, and to group training samples in the model training set belonging to the same operating conditions into the same operating condition group; input the training samples in the operating condition group into the prediction model to obtain the predicted production operation data of the heating system at the predicted time corresponding to the historical period to which the training samples in the operating condition group belong, the predicted production operation data of the heating system at the predicted time corresponding to the historical period to which the training samples in the operating condition group belong includes the predicted reading values ​​of the heating sensors at the predicted time corresponding to the historical period to which the training samples in the operating condition group belong; based on the predicted production operation data of the heating system at the predicted time corresponding to the historical period to which the training samples in the operating condition group belong and the actual production operation data, determine the prediction error threshold of the prediction model for the readings of the heating sensors under the operating conditions corresponding to the operating condition group.

[0064] Optionally, the determining module 302 is specifically used to determine the prediction error of the heating sensor reading under the corresponding operating conditions of the operating condition group based on the difference between the predicted reading value and the actual reading value at the predicted time corresponding to the historical time period of the training sample in the operating condition group; and to determine the prediction error threshold of the heating sensor reading under the corresponding operating conditions of the operating condition group based on the statistical analysis value of the prediction error of the heating sensor reading under the corresponding operating conditions of the operating condition group.

[0065] Optionally, the determining module 302 is specifically used to construct a real-time prediction sample based on the production operation data and external status data of the heating system during the monitoring period; input the real-time prediction sample into the prediction model to obtain the predicted production operation data of the heating system at the prediction time corresponding to the monitoring period; and determine the prediction error of the prediction model for the real-time reading of the heating sensor based on the predicted production operation data and the actual production operation data of the heating system at the prediction time corresponding to the monitoring period.

[0066] Optionally, the diagnostic module 303 is specifically used to construct a real-time prediction sample based on the production operation data and external status data of the heating system during the monitoring period, and determine the target operating condition matching the real-time prediction sample; compare the prediction error of the real-time reading of the heating sensor by the prediction model with the prediction error threshold of the reading of the heating sensor under the target operating condition by the prediction model; based on the comparison results, determine the candidate abnormal sensors among the heating sensors; and verify the candidate abnormal sensors based on the heating effect and event records of the heating system during the monitoring period to obtain the abnormal sensors.

[0067] The heating system anomaly diagnostic device 300 also includes: The early warning module is used to generate structured diagnostic results for the heating system based on abnormal sensors, and to visualize and push the diagnostic results in real time.

[0068] Specific limitations regarding the heating system anomaly diagnostic device can be found in the limitations of the heating system anomaly diagnostic method described above, and will not be repeated here. Each module in the aforementioned heating system anomaly diagnostic device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device in hardware form, or stored in the memory of a computer device in software form, so that the processor can call and execute the corresponding operations of each module.

[0069] Based on the above, Figures 1 to 2 Accordingly, embodiments of this application also provide a readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-described method. Figures 1 to 2 The method for diagnosing abnormalities in the heating system is shown.

[0070] Based on this understanding, the technical solution of this application can be embodied in the form of a software product. The software product can be stored in a non-volatile storage medium (such as a CD-ROM, USB flash drive, or portable hard drive), and includes several instructions to cause a computer device (such as a personal computer, server, or network device) to execute the methods described in the various implementation scenarios of this application.

[0071] Based on the above, Figures 1 to 2 The method shown, and Figure 3 To achieve the above objectives, the present application also provides a computer device, specifically a personal computer, server, network device, etc., as shown in the virtual device embodiment. This computer device includes a storage medium and a processor; the storage medium stores a computer program; the processor executes the computer program to achieve the above-described objectives. Figures 1 to 2 The method for diagnosing abnormalities in the heating system is shown.

[0072] Optionally, the computer device may also include a user interface, a network interface, a camera, radio frequency (RF) circuitry, sensors, audio circuitry, a Wi-Fi module, etc. The user interface may include a display screen, input units such as a keyboard, etc., and optional user interfaces may also include USB ports, card reader ports, etc. The network interface may optionally include standard wired interfaces, wireless interfaces (such as Bluetooth interfaces, Wi-Fi interfaces), etc.

[0073] Those skilled in the art will understand that the computer device structure provided in this embodiment does not constitute a limitation on the computer device, and may include more or fewer components, or combine certain components, or have different component arrangements.

[0074] The storage medium may also include an operating system and a network communication module. The operating system is a program that manages and stores the hardware and software resources of a computer device, supporting the operation of information processing programs and other software and / or programs. The network communication module is used to enable communication between the various components within the storage medium, as well as communication with other hardware and software within the physical device.

[0075] Through the above description of the embodiments, those skilled in the art can clearly understand that this application can be implemented by means of software plus necessary general-purpose hardware platform, or the embodiments of this application can be implemented by hardware.

[0076] Those skilled in the art will understand that the accompanying drawings are merely schematic diagrams of a preferred embodiment, and the modules or processes shown in the drawings are not necessarily essential for implementing this application. Those skilled in the art will understand that the modules in the apparatus of the embodiment can be distributed within the apparatus of the embodiment as described, or can be modified to be located in one or more apparatuses different from this embodiment. The modules of the above-described embodiment can be combined into one module, or further divided into multiple sub-modules.

[0077] The serial numbers in this application are for descriptive purposes only and do not represent the superiority or inferiority of any particular implementation scenario. The above disclosures are merely a few specific implementation scenarios of this application; however, this application is not limited thereto, and any variations conceived by those skilled in the art should fall within the protection scope of this application.

Claims

1. A method for diagnosing abnormalities in a heating system, characterized in that, The method includes: A model training set was constructed based on multi-source heterogeneous data from the heating system over a historical period. The preset deep neural network is trained using the model training set to obtain a prediction model for predicting the readings of heating sensors deployed in the heating site. Based on the model training set and the prediction model, the prediction error threshold of the prediction model for the readings of the heating sensor under different operating conditions of the heating system is determined. Based on the multi-source heterogeneous data of the heating system during the monitoring period and the prediction model, the prediction error of the prediction model for the real-time reading of the heating sensor is determined. Based on the multi-source heterogeneous data of the heating system during the monitoring period, the prediction error of the real-time reading of the heating sensor by the prediction model is compared with the reading prediction error threshold to identify the abnormal sensor among the heating sensors.

2. The method for diagnosing abnormalities in a heating system according to claim 1, characterized in that, The multi-source heterogeneous data includes production and operation data of the heating system, external status data, heating effect and event records. The construction of a model training set based on the multi-source heterogeneous data of the heating system over historical periods specifically includes: Based on the heating effect and event records of the heating system during the historical period, the production operation data and external status data of the heating system that were in normal operation during the historical period are filtered out. Training subsamples are determined based on the production operation data of the heating system under normal operating conditions during the historical period and the data of the external status data belonging to the same historical moment. The training samples are determined based on the training sub-samples of the heating system at the historical moment within the historical period of the historical time. The production operation data of the predicted time corresponding to the historical time period to which the training sample belongs is used as the prediction label of the training sample, and the predicted time corresponding to the historical time period is the future time after the historical time period has passed a preset prediction time period. The model training set is constructed based on the training samples carrying the predicted labels.

3. The method for diagnosing abnormalities in a heating system according to claim 1, characterized in that, The step of determining the prediction error threshold of the heating sensor readings under different operating conditions of the heating system based on the model training set and the prediction model specifically includes: Determine the operating conditions of the heating system during the historical period, and divide the training samples in the model training set that belong to the same operating conditions into the same operating condition group; The training samples in the operating condition group are input into the prediction model to obtain the predicted production operation data of the heating system at the predicted time corresponding to the historical period to which the training samples in the operating condition group belong. The predicted production operation data of the heating system at the predicted time corresponding to the historical period to which the training samples in the operating condition group belong includes the predicted reading value of the heating sensor at the predicted time corresponding to the historical period to which the training samples in the operating condition group belong. Based on the predicted production operation data and the actual production operation data of the training samples of the heating system in the operating condition group, the prediction error threshold of the prediction model for the reading of the heating sensor under the operating condition corresponding to the operating condition in the operating condition group is determined.

4. The method for diagnosing abnormalities in a heating system according to claim 3, characterized in that, The actual production operation data of the heating system at the predicted time corresponding to the historical period to which the training samples of the heating system belong in the operating condition group include the actual reading values ​​of the heating sensor at the predicted time corresponding to the historical period to which the training samples of the heating system belong in the operating condition group. The step of determining the prediction error threshold for the readings of the heating sensor under the corresponding operating condition in the operating condition group based on the predicted production operation data and the actual production operation data of the heating system at the predicted time corresponding to the historical period to which the training samples of the heating system belong in the operating condition group specifically includes: Based on the difference between the predicted reading value and the actual reading value of the heating sensor at the predicted time corresponding to the historical time period to which the training sample of the heating sensor belongs in the operating condition group, the prediction error of the prediction model for the reading of the heating sensor under the operating condition corresponding to the operating condition group is determined. Based on the statistical analysis of the prediction error of the heating sensor reading under the corresponding operating condition of the operating condition group, the threshold value of the prediction error of the heating sensor reading under the corresponding operating condition group is determined.

5. The method for diagnosing abnormalities in a heating system according to claim 1, characterized in that, The multi-source heterogeneous data includes production and operation data of the heating system, external status data, heating effect and event records. The determination of the prediction error of the prediction model for the real-time readings of the heating sensor based on the multi-source heterogeneous data of the heating system within the monitoring period and the prediction model specifically includes: Based on the production operation data and external status data of the heating system during the monitoring period, a real-time prediction sample is constructed. The real-time prediction sample is input into the prediction model to obtain the predicted production and operation data of the heating system at the prediction time corresponding to the monitoring cycle. Based on the predicted production operation data and actual production operation data of the heating system at the predicted time corresponding to the monitoring cycle, the prediction error of the prediction model for the real-time reading of the heating sensor is determined.

6. The method for diagnosing abnormalities in a heating system according to claim 1, characterized in that, The multi-source heterogeneous data includes production and operation data of the heating system, external status data, heating effect and event records. Based on the multi-source heterogeneous data of the heating system within the monitoring period, the prediction model's prediction error for the real-time readings of the heating sensors is compared with the prediction error threshold to identify abnormal sensors among the heating sensors. Specifically, this includes: Based on the production operation data and external status data of the heating system during the monitoring period, a real-time prediction sample is constructed, and the target operating condition matching the real-time prediction sample is determined. Compare the prediction error of the real-time reading of the heating sensor by the prediction model with the prediction error threshold of the reading of the heating sensor under the target operating condition by the prediction model; Based on the comparison results, candidate abnormal sensors among the heating sensors were identified; Based on the heating effect and event records of the heating system during the monitoring period, the candidate abnormal sensors are verified to obtain the abnormal sensors.

7. The method for diagnosing abnormalities in a heating system according to claim 1, characterized in that, The method further includes: The system generates structured diagnostic results for the heating system based on the abnormal sensors, and then visualizes and pushes these diagnostic results in real time.

8. A diagnostic device for abnormalities in a heating system, characterized in that, The device includes: The training module is used to construct a model training set based on multi-source heterogeneous data of the heating system in historical periods; and to train a preset deep neural network using the model training set to obtain a prediction model for predicting the readings of heating sensors deployed in the heating site. The determination module is used to determine, based on the model training set and the prediction model, the prediction error threshold of the prediction model for the readings of the heating sensor under different operating conditions of the heating system; and, based on the multi-source heterogeneous data of the heating system during the monitoring period and the prediction model, the prediction error of the prediction model for the real-time readings of the heating sensor. The diagnostic module is used to identify abnormal sensors among the heating sensors by comparing the prediction error of the real-time readings of the heating sensors with the reading prediction error threshold based on the multi-source heterogeneous data of the heating system during the monitoring period.

9. A readable storage medium having a program or instructions stored thereon, characterized in that, When the program or instructions are executed by the processor, they implement the steps of the heating system anomaly diagnosis method as described in any one of claims 1 to 7.

10. A computer device, comprising a storage medium, a processor, and a computer program stored on the storage medium and executable on the processor, characterized in that, When the processor executes the program, it implements the heating system abnormality diagnosis method as described in any one of claims 1 to 7.

Images