Test Method and Device for Early Warning Model of Supercritical Units Based on Big Data Platform
By using a supercritical unit early warning model based on a big data platform, measuring point data is acquired and verified, early warning results and trigger times are determined, and the mathematical model is processed by parameter adjustment. This solves the problem of the single-target nature of early warning methods in thermal power plants and achieves more accurate and timely fault early warning.
Patent Information
- Application Number
- CN202310071732.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2023-01-18
- Publication Date
- 2026-11-17
- Estimated Expiration
- 2043-01-18
AI Technical Summary
Existing technologies in thermal power plants use threshold and operating condition classifications for individual equipment to provide fault warnings. This results in a limited and unreliable warning method that cannot identify potential faults in complex systems in a timely and accurate manner.
A supercritical unit early warning model based on a big data platform is adopted. By acquiring measurement point data from multiple mathematical models, the measurement points are verified and monitoring values are input to obtain early warning results and trigger times. It is then determined whether preset conditions are met, and parameter adjustments are made to improve the accuracy and timeliness of the early warning.
It has improved the accuracy and timeliness of early warning for supercritical units in multiple professional fields and enhanced the ability to identify faults in complex systems.
Smart Images

Figure CN116302957B_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of big data platform technology for supercritical units, and in particular to a testing method and apparatus for early warning models of supercritical units based on a big data platform. Background Technology
[0002] In related technologies, thermal power plants are characterized by complex systems, numerous equipment, and sophisticated and safety-critical processes. This necessitates meticulous and cautious operation by staff both on-site and remotely from the control room to maintain the plant's basic normal operation. When systems and equipment malfunction, failure to promptly shut down faulty equipment and activate appropriate protective devices from the control room can escalate the accident, making it more dangerous and severe. Currently, thermal power plants address these issues by classifying the boundary conditions and information of each system and piece of equipment for mechanistic-level judgment and early warning. This method typically uses factory parameters provided by equipment manufacturers, expert reference values developed by experienced personnel, and reasonable values from the past safe operation of similar units as the basis for mechanistic analysis. This allows for different types of fault warnings and classifications for various operating conditions, systems, and equipment within the thermal power plant. However, thermal power plants consist of complex systems and equipment, and the aforementioned early warning methods often classify individual equipment based on thresholds and operating conditions, resulting in limited specificity and adaptability. Summary of the Invention
[0003] Therefore, this application provides a testing method and apparatus for early warning models of supercritical units based on a big data platform. The technical solution of this application is as follows: According to a first aspect of the embodiments of this application, a method for testing early warning models of supercritical units based on a big data platform is provided, the method comprising: Acquire measurement point data for each of multiple mathematical models; wherein, the multiple mathematical models are machine learning models pre-trained based on historical operating data of supercritical units; the measurement point data includes measurement point name, measurement point identifier, and monitoring value; For each mathematical model, the multiple measurement points are verified based on their respective measurement point names. In response to the fact that all the multiple measuring points have passed the verification, the monitoring value is input into the mathematical model to obtain the early warning result output by the mathematical model and the trigger time of the mathematical model; Determine whether the operating data of the supercritical unit at the trigger time meets the preset triggering conditions of the mathematical model, obtain a first intermediate result, and determine whether the first intermediate result meets the preset requirements; Determine whether the triggering time meets the preset requirements; Determine whether the warning result meets the preset requirements; If any one or more of the first intermediate result, the trigger time, and the warning result fail to meet the preset requirements, the mathematical model is subjected to parameter tuning.
[0004] According to one embodiment of this application, determining whether the operating data of the supercritical unit at the trigger time meets the preset triggering conditions of the mathematical model, obtaining a first intermediate result, and determining whether the first intermediate result meets the preset requirements includes: Based on the mathematical model, at least one preset triggering condition for the mathematical model is determined; Obtain the operating data of the supercritical unit at the trigger time; Based on the operational data, obtain operational information corresponding to each of the at least one preset triggering condition; Determine whether each piece of runtime information meets its corresponding preset triggering conditions; In response to each running information satisfying its corresponding preset triggering condition, it is determined that the first intermediate result meets the preset requirements; In response to at least one running information failing to meet its corresponding preset trigger condition, it is determined that the first intermediate result does not meet the preset requirements.
[0005] According to one embodiment of this application, determining whether the trigger time meets the preset requirements includes: Obtain the preset triggering period of the mathematical model; The preset trigger period and the trigger time are compared to obtain the first comparison result; In response to the first comparison result indicating that the trigger time does not fall within the preset trigger period, it is determined that the trigger time does not meet the preset requirements.
[0006] According to one embodiment of this application, determining whether the warning result meets the preset requirements includes: Obtain the actual early warning results corresponding to the measurement point data; The warning result is compared with the actual warning result to obtain a second comparison result; In response to the second comparison result indicating that the warning result is not consistent with the actual warning result, it is determined that the warning result does not meet the preset requirements.
[0007] According to one embodiment of this application, the verification process for the multiple measurement points based on the mathematical model, using their respective measurement point names, includes: Based on the measurement point names and measurement point identifiers of multiple measurement points in the mathematical model, search the historical data of the multiple measurement point identifiers of the mathematical model that are pre-stored in the database. In response to the absence of historical data for at least one measurement point in the database, related measurement points for the at least one measurement point are determined from among the plurality of measurement points based on the measurement point name of the at least one measurement point. Search the database for historical data of the relevant measuring points that are stored in advance; In response to the failure to find historical data of the relevant measurement points pre-stored in the database, it is determined that at least one measurement point has failed verification. The mathematical model is then subjected to parameter tuning, and the step of searching for historical data of each of the multiple measurement point identifiers of the mathematical model pre-stored in the database is repeated.
[0008] According to one embodiment of this application, the historical operating data of the supercritical unit includes any one or more of the following: historical operating data of steam turbine, historical operating data of boiler, historical operating data of electrical equipment, historical operating data of thermal control equipment, and historical operating data of chemical equipment; wherein, each type of historical operating data includes shutdown data, startup process data, and operating data; the shutdown data, startup process data, and operating data all include normal data and abnormal data.
[0009] According to a second aspect of the embodiments of this application, a test device for an early warning model of a supercritical unit based on a big data platform is provided, the device comprising: The acquisition module is used to acquire the measurement point data of each of the multiple mathematical models; wherein, the multiple mathematical models are all machine learning models pre-trained based on the historical operating data of the supercritical unit; the measurement point data includes the measurement point name, measurement point identifier and monitoring value; The verification module is used to verify the multiple measurement points for each mathematical model based on the measurement point names of each measurement point in the mathematical model. The input module is used to input the monitoring value into the mathematical model in response to the multiple measuring points passing the verification, and to obtain the early warning result output by the mathematical model and the trigger time of the mathematical model. The first determining module is used to determine whether the operating data of the supercritical unit at the trigger time meets the preset triggering conditions of the mathematical model, obtain a first intermediate result, and determine whether the first intermediate result meets the preset requirements. The second determining module is used to determine whether the triggering time meets the preset requirements; The third determining module is used to determine whether the warning result meets the preset requirements; The parameter tuning module is used to perform parameter tuning on the mathematical model in response to any one or more of the first intermediate result, the trigger time, and the warning result failing to meet preset requirements.
[0010] According to a third aspect of the embodiments of this application, an electronic device is provided, including: a processor, and a memory communicatively connected to the processor; The memory stores computer-executed instructions; The processor executes computer execution instructions stored in the memory to implement the method as described in any one of the first aspects.
[0011] According to a fourth aspect of the embodiments of this application, a computer-readable storage medium is provided, wherein computer-executable instructions are stored therein, which, when executed by a processor, are used to implement the method as described in any one of the first aspects.
[0012] According to a fifth aspect of the embodiments of this application, a computer program product is provided, characterized in that it includes a computer program that, when executed by a processor, implements the method described in any one of the first aspects.
[0013] The technical solutions provided by the embodiments of this application bring at least the following beneficial effects: By acquiring measurement point data from multiple mathematical models; for each mathematical model, verifying multiple measurement points based on their respective measurement point names; in response to multiple measurement points passing verification, inputting monitoring values into the mathematical model, obtaining the early warning result and trigger time of the mathematical model; determining whether the supercritical unit's operating data at the trigger time meets the preset trigger conditions of the mathematical model, obtaining the first intermediate result, and determining whether the first intermediate result meets the preset requirements; determining whether the trigger time meets the preset requirements; determining whether the early warning result meets the preset requirements; in response to any one or more of the first intermediate result, trigger time, and early warning result failing to meet the preset requirements, adjusting the parameters of the mathematical model, thereby improving the accuracy and timeliness of early warning for supercritical units in multiple professional directions through mathematical models.
[0014] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and do not limit this application. Attached Figure Description
[0015] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application, and do not constitute an undue limitation of this application.
[0016] Figure 1This is a flowchart illustrating a test method for an early warning model of a supercritical unit based on a big data platform, as described in an embodiment of this application. Figure 2 This is a structural block diagram of a supercritical unit early warning model testing device based on a big data platform, as described in an embodiment of this application. Figure 3 This is a block diagram of an electronic device according to an embodiment of this application. Detailed Implementation
[0017] To enable those skilled in the art to better understand the technical solutions of this application, the technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings.
[0018] It should be noted that the terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.
[0019] It should be noted that, due to the complexity of thermal power plants, the large number of equipment, and the sophisticated and safety-critical processes, personnel must exercise extreme caution and meticulousness both during on-site operations and remote operations from the control room to maintain the basic normal operation of the power plant. When a system or equipment malfunctions, if operators cannot immediately shut down the faulty equipment and activate the corresponding protective devices remotely from the control room, the accident could potentially escalate, become more dangerous, and more severe. Currently, thermal power plants address these issues by classifying the boundary conditions and information of each system and piece of equipment to determine the underlying mechanisms and provide early warnings. This method typically uses factory parameters provided by equipment manufacturers, expert reference values developed by experienced personnel, and reasonable values from the past safe operation of similar units as the basis for mechanism analysis. This allows for different types of fault warnings and classifications for various operating conditions, systems, and equipment within the thermal power plant. However, thermal power plants consist of complex systems and equipment, and the above-mentioned early warning methods often classify individual equipment based on thresholds and operating conditions, resulting in limited specificity and adaptability.
[0020] To address the aforementioned issues, this application proposes a testing method and apparatus for early warning models of supercritical units based on a big data platform. This method involves acquiring measurement point data from multiple mathematical models; verifying multiple measurement points for each mathematical model based on their respective measurement point names; inputting the monitored values into the mathematical model in response to all verifications, and obtaining the early warning result and trigger time of the mathematical model; determining whether the supercritical unit's operating data at the trigger time meets the preset trigger conditions of the mathematical model, obtaining a first intermediate result, and determining whether the first intermediate result meets preset requirements; determining whether the trigger time meets preset requirements; determining whether the early warning result meets preset requirements; and adjusting the parameters of the mathematical model in response to any one or more of the first intermediate result, trigger time, and early warning result failing to meet preset requirements. This improves the accuracy and timeliness of early warnings for supercritical units across multiple professional areas using mathematical models.
[0021] Figure 1 This is a flowchart of a test method for an early warning model of a supercritical unit based on a big data platform, as described in an embodiment of this application.
[0022] like Figure 1 As shown, the testing method for the early warning model of supercritical units based on a big data platform includes: Step 101: Obtain the measurement point data for each of the multiple mathematical models.
[0023] In this embodiment of the application, multiple mathematical models are machine learning models pre-trained based on historical operating data of supercritical units.
[0024] In this embodiment of the application, the measurement point data includes the measurement point name, measurement point identifier, and monitoring value.
[0025] In some embodiments of this application, the historical operating data of the supercritical unit includes any one or more of the following: historical operating data of steam turbine, historical operating data of boiler, historical operating data of electrical equipment, historical operating data of thermal control equipment, and historical operating data of chemical equipment; wherein, each type of historical operating data includes shutdown data, startup process data, and operating data; shutdown data, startup process data, and operating data all include normal data and abnormal data.
[0026] As a possible implementation example, considering the characteristics of multiple systems in power plants, the modeling directions are divided into five major categories, including: mathematical modeling of steam turbine, mathematical modeling of boiler, mathematical modeling of electrical engineering, mathematical modeling of thermal control, and mathematical modeling of chemistry.
[0027] Extract operational data from the system across five modeling directions, including three main categories: shutdown data, startup process data, and operational data. The shutdown data includes: shutdown data during short-term unit maintenance, shutdown data during long-term unit maintenance, shutdown data during short-term unit accident conditions, shutdown data during long-term unit accident conditions, short-term normal unit shutdown data, and long-term normal unit shutdown data. The startup process data includes: unit cold start process data, unit warm start process data, unit hot start process data, unit extremely hot start process data, unit cold start process data under accident conditions, unit warm start process data under accident conditions, unit hot start process data under accident conditions, and unit extremely hot start process data under accident conditions; The operating data includes: normal operating data at low load (0MW-105MW), normal operating data at high load (105.1MW-297.5MW), normal operating data at full load (297.6MW-350MW), operating data at low load (0MW-105MW) under unit accident conditions, operating data at high load (105.1MW-297.5MW) under unit accident conditions, and operating data at full load (297.6MW-350MW) under unit accident conditions.
[0028] Optionally, normal data and data under fault conditions can be extracted from shutdown data, startup process data, and operation data: Normal data includes: unit shutdown data during short-term maintenance, unit shutdown data during long-term maintenance, unit short-term normal shutdown data, unit long-term normal shutdown data, unit cold start process data, unit warm start process data, unit hot start process data, unit extremely hot start process data, unit low load (0MW-105MW) normal operation data, unit high load (105.1MW-297.5MW) normal operation data, and unit full load (297.6MW-350MW) normal operation data. Data under accident conditions: shutdown data during short-term accident conditions, shutdown data during long-term accident conditions, data during cold start-up of the unit under accident conditions, data during warm start-up of the unit under accident conditions, data during hot start-up of the unit under accident conditions, data during extremely hot start-up of the unit under accident conditions, low-load (0MW-105MW) operating data of the unit under accident conditions, high-load (105.1MW-297.5MW) operating data of the unit under accident conditions, and full-load (297.6MW-350MW) operating data of the unit under accident conditions.
[0029] Optionally, one or more of the following algorithms can be used to train and learn the model based on normal and accident data: linear regression, support vector machine, nearest neighbor / k-nearest neighbor, logistic regression, decision tree, k-means, random forest, naive Bayes, dimensionality reduction, and gradient enhancement. This enables the mathematical model to learn to recognize the normal and accident states of the unit and complete the establishment of the mathematical model.
[0030] Optionally, after the mathematical model is established, a segment of data under accident conditions can be randomly selected as a validation set to test the functionality of the mathematical model. The test success condition is that the model issues an early warning when an accident is about to occur. If the warning is not timely, or if there are any omissions, false alarms, excessive warnings, or insufficient warnings, the mathematical model test is considered unsuccessful.
[0031] Step 102: For each mathematical model, the multiple measurement points are verified based on their respective measurement point names.
[0032] In some embodiments of this application, step 102 includes: Step a1: Based on the measurement point names and identifiers of multiple measurement points in the mathematical model, search the database for the historical data of the multiple measurement point identifiers of the mathematical model that are stored in advance.
[0033] Optionally, the measuring point identifier can be the power plant identification system code of the measuring point.
[0034] Step a2: In response to the absence of historical data for at least one measurement point in the database, determine the relevant measurement point of at least one measurement point among multiple measurement points based on the measurement point name of at least one measurement point.
[0035] As a possible implementation example, the historical data of the measurement points are verified within a big data platform using the power plant identification system codes of the corresponding mathematical model measurement points to determine the correctness, accuracy, and authenticity of the measurement points. Based on the measurement point names and identifiers of multiple measurement points in the mathematical model, the database is searched to see if historical data for each of the multiple measurement point identifiers of the mathematical model is stored. If no historical data for at least one measurement point is found in the database, it indicates that the measurement point does not meet the accuracy requirements. Based on the measurement point name of at least one measurement point, the related measurement points of at least one measurement point are determined from among the multiple measurement points.
[0036] Optionally, based on the measurement point name, keywords, and synonyms of the keywords can be determined. Related measurement points can then be found based on these keywords and synonyms, provided their detection value attributes match those of the aforementioned measurement points. It is understandable that data transmission may be incomplete or distorted due to network instability or other reasons. Therefore, it is necessary to verify the power plant identification system code of the measurement points. Multiple measurement points may monitor the same value, only with slight differences in their names; thus, the integrity of the data can be verified through related measurement points.
[0037] In this embodiment of the application, in response to finding historical data of at least one measurement point in the database, it is determined that at least one measurement point has passed the verification.
[0038] Step a3: Search the database for historical data of relevant measurement points that have been stored in advance.
[0039] Step a4: In response to the failure to find historical data of relevant measurement points pre-stored in the database, it is determined that at least one measurement point has failed the verification. Parameter tuning is performed on at least one measurement point, and the step of searching for historical data of each measurement point identifier of the mathematical model pre-stored in the database is repeated.
[0040] As an example of possible implementation, in response to the failure to find historical data of relevant measurement points pre-stored in the database, it is determined that at least one measurement point has failed verification. Parameter tuning is then performed on at least one measurement point, and the steps of searching for historical data of multiple measurement point identifiers of the mathematical model pre-stored in the database are repeated.
[0041] Step 103: In response to the successful verification of multiple measurement points, the monitoring values are input into the mathematical model to obtain the early warning results output by the mathematical model and the trigger time of the mathematical model.
[0042] Step 104: Determine whether the operating data of the supercritical unit at the trigger time meets the preset triggering conditions of the mathematical model, obtain the first intermediate result, and determine whether the first intermediate result meets the preset requirements.
[0043] In some embodiments of this application, step 104 includes: Step b1: Based on the mathematical model, determine at least one preset triggering condition for the mathematical model.
[0044] Understandably, since multiple mathematical models belong to different professional directions in power plants, it is necessary to formulate corresponding model triggering rules for each model so that the models can be used more flexibly, agilely, and conveniently.
[0045] For example, the triggering rules for the pulverizer body blockage model and the pulverizer outlet pipe blockage model in the pulverizing system can be pulverizer operation, coal feeder operation, and normal coal feed rate; the triggering rules for the lubricating oil tank leakage model and the high-pressure fire-resistant oil tank leakage model can be oil pump operation, normal oil tank level, and normal oil circuit main pipe pressure; the triggering rules for the high-pressure heater leakage model in the high-pressure heater system can be high-pressure heater liquid level, normal main pipe pressure, and normal flow rate; the triggering rules for the first-stage reverse osmosis fouling model and the second-stage reverse osmosis fouling model in the reverse osmosis system can be total water production, first-stage reverse osmosis differential pressure, and second-stage reverse osmosis differential pressure all being normal.
[0046] Step b2: Obtain the operating data of the supercritical unit at the trigger time.
[0047] Step b3: Based on the running data, obtain the running information corresponding to each of at least one preset triggering condition.
[0048] Step b4: Determine whether each piece of running information meets its corresponding preset triggering conditions.
[0049] It is understandable that the process involves acquiring the operating data of the supercritical unit at the trigger time, obtaining the operating information corresponding to each of the at least one preset trigger condition based on the operating data, and determining whether the current operating state of the supercritical unit meets the trigger conditions of the mathematical model based on the operating information.
[0050] Step b5: In response to each running information satisfying its corresponding preset triggering condition, determine that the first intermediate result meets the preset requirements.
[0051] Step b6: In response to at least one running information failing to meet its corresponding preset trigger condition, determine that the first intermediate result does not meet the preset requirements.
[0052] Step 105: Determine whether the trigger time meets the preset requirements.
[0053] In some embodiments of this application, step 105 includes: Step c1: Obtain the preset triggering period of the mathematical model.
[0054] Step c2: Compare the preset trigger cycle and trigger time to obtain the first comparison result.
[0055] Step c3: In response to the first comparison result indicating that the trigger time does not fall within the preset trigger period, it is determined that the trigger time does not meet the preset requirements.
[0056] As a possible implementation example, since different mathematical models have different judgment periods for different accidents, the triggering period and triggering rules can be divided into different categories.
[0057] For example, the preset triggering cycle of a mathematical model may include: Short-term, rapid-trigger accident warning, with a set time period of within 10 seconds; A gradual increase in the short term triggers an accident warning, with a time frame set between 10 seconds and 10 minutes. A mid-term uniform speed change triggers an accident warning, with a set time period of 10 to 60 minutes. Long-term, slow changes trigger accident warnings, with a time period set between 60 and 600 minutes.
[0058] For example, blockage of the coal mill body in the pulverizing system and blockage of the coal mill outlet pipe in the pulverizing system are both short-term, rapid-triggering accident warnings. The accident situation occurs very quickly and a short time needs to be set within the model judgment period. For example, leaks in the lubricating oil system tank and leaks in the high-pressure fire-resistant oil system tank are both short-term, gradually increasing accident warnings. The accident situation will gradually escalate in a short period of time, so a short time needs to be set within the model judgment period. For example, leakage of the high-pressure heater in the high-pressure heater system is an accident warning triggered by a medium-term uniform change. The accident situation will gradually increase over a period of time, and a certain period needs to be set within the model judgment period. For example, fouling in the first stage and the second stage of a reverse osmosis system are both slow, long-term changes that trigger accident warnings. The severity of the accident will gradually increase over a long period of time, so a long period needs to be set within the model's judgment cycle.
[0059] Step 106: Determine whether the warning result meets the preset requirements.
[0060] In some embodiments of this application, step 106 includes: Step d1: Obtain the actual early warning results corresponding to the measurement point data.
[0061] Step d2 involves comparing the warning results with the actual warning results to obtain the second comparison result.
[0062] Step d3: In response to the second comparison result being inconsistent with the actual warning result, it is determined that the warning result does not meet the preset requirements.
[0063] Step 107: In response to any one or more of the first intermediate result, trigger time, and warning result failing to meet the preset requirements, the mathematical model is subjected to parameter tuning.
[0064] For example, taking the high-pressure fire-resistant oil system leakage early warning system in steam turbine engineering as an example, the high-pressure fire-resistant oil tank level, main pipe pressure, and oil pump operation data are used as training data. The training data is divided into three stages: shutdown data, startup process data, and operation data. These three stages of data serve as the training set and validation set data. A convolutional neural network algorithm is used to build the model, and finally, the model is validated. Next, the high-pressure fire-resistant oil system leakage early warning model operator block is deployed to a big data platform. The input measurement points on the platform, including the high-pressure fire-resistant oil tank level, main pipe pressure, and oil pump operation, need to be checked. The judgment period is set to within 10 seconds to 10 minutes, and the model triggering rules are set as oil pump operation, normal oil tank level, and normal main pipe pressure. Finally, the mathematical model operator block is tested and verified using fault data.
[0065] According to the supercritical unit early warning model testing method based on a big data platform according to the embodiments of this application, the method acquires measurement point data of multiple mathematical models; for each mathematical model, multiple measurement points are verified based on their respective measurement point names; in response to multiple measurement points passing verification, the monitoring values are input into the mathematical model to obtain the early warning result and trigger time of the mathematical model; it determines whether the supercritical unit's operating data at the trigger time meets the preset trigger conditions of the mathematical model, obtaining a first intermediate result; it then determines whether the first intermediate result meets preset requirements; it further determines whether the trigger time meets preset requirements; and finally, it determines whether the early warning result meets preset requirements. In response to any one or more of the first intermediate result, trigger time, and early warning result failing to meet preset requirements, the mathematical model is adjusted. This improves the accuracy and timeliness of early warning for supercritical units in multiple professional directions using mathematical models.
[0066] Figure 2 This is a flowchart of a supercritical unit early warning model testing device based on a big data platform, as described in an embodiment of this application.
[0067] like Figure 2 As shown, the test device for the early warning model of a supercritical unit based on a big data platform includes: The acquisition module 201 is used to acquire the measurement point data of each of the multiple mathematical models; wherein, the multiple mathematical models are machine learning models pre-trained based on the historical operating data of the supercritical unit; the measurement point data includes the measurement point name, measurement point identifier and monitoring value; The verification module 202 is used to verify multiple measurement points for each mathematical model based on the measurement point names of each measurement point in the mathematical model. The input module 203 is used to input the monitoring values into the mathematical model in response to the fact that multiple measuring points have passed the verification, and to obtain the early warning result output by the mathematical model and the trigger time of the mathematical model. The first determining module 204 is used to determine whether the operating data of the supercritical unit at the trigger time meets the preset triggering conditions of the mathematical model, obtain the first intermediate result, and determine whether the first intermediate result meets the preset requirements. The second determining module 205 is used to determine whether the triggering time meets the preset requirements; The third determining module 206 is used to determine whether the warning result meets the preset requirements; The parameter tuning module 207 is used to perform parameter tuning on the mathematical model in response to any one or more of the first intermediate result, trigger time, and early warning result failing to meet preset requirements.
[0068] According to the embodiments of this application, the supercritical unit early warning model testing device based on a big data platform acquires measurement point data for multiple mathematical models. For each mathematical model, multiple measurement points are verified based on their respective measurement point names. In response to multiple measurement points passing verification, the monitored values are input into the mathematical model to obtain the early warning result and trigger time of the mathematical model. It determines whether the supercritical unit's operating data at the trigger time meets the preset trigger conditions of the mathematical model, obtaining a first intermediate result. It then determines whether the first intermediate result meets preset requirements, whether the trigger time meets preset requirements, and whether the early warning result meets preset requirements. In response to any one or more of the first intermediate result, trigger time, and early warning result failing to meet preset requirements, the mathematical model is adjusted. This improves the accuracy and timeliness of early warning for supercritical units in multiple professional directions using mathematical models.
[0069] Figure 3 This is a block diagram of an electronic device according to an embodiment of this application. For example... Figure 3 As shown, the electronic device may include: a transceiver 31, a processor 32, and a memory 33.
[0070] Processor 32 executes computer execution instructions stored in memory, causing processor 32 to perform the scheme in the above embodiments. Processor 32 can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.
[0071] The memory 33 is connected to the processor 32 via the system bus and completes communication between them. The memory 33 is used to store computer program instructions.
[0072] Transceiver 31 can be used to obtain the task to be run and its configuration information.
[0073] The system bus can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. The system bus can be divided into address bus, data bus, control bus, etc. For ease of representation, only one thick line is used in the diagram, but this does not indicate that there is only one bus or one type of bus. Transceivers are used to enable communication between database access devices and other computers (e.g., clients, read-write libraries, and read-only libraries). Memory may include random access memory (RAM) and may also include non-volatile memory.
[0074] The electronic device provided in this application embodiment can be the terminal device described in the above embodiments.
[0075] This application also provides a chip for executing instructions, which is used to execute the message processing method described in the above embodiments.
[0076] This application also provides a computer-readable storage medium storing computer instructions that, when executed on a computer, cause the computer to perform the message processing method described in the above embodiments.
[0077] This application also provides a computer program product, which includes a computer program stored in a computer-readable storage medium. At least one processor can read the computer program from the computer-readable storage medium, and when the at least one processor executes the computer program, it can implement the technical solution of the message processing method in the above embodiments.
[0078] Other embodiments of this application will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of this application that follow the general principles of this application and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of this application are indicated by the following claims.
[0079] It should be understood that this application is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this application is limited only by the appended claims.
Claims
1. A method for testing a supercritical unit early warning model based on a big data platform, characterized in that, The method includes: Acquire measurement point data for each of multiple mathematical models; wherein, the multiple mathematical models are all machine learning models pre-trained based on historical operating data of supercritical units; the measurement point data includes measurement point name, measurement point identifier, and monitoring value, wherein, the measurement point identifier is the power plant identification system code of the measurement point; For each mathematical model, the multiple measurement points are verified based on their respective measurement point names. In response to the verification of the multiple measuring points, the monitoring values are input into the mathematical model to obtain the early warning result output by the mathematical model and the trigger time of the mathematical model; Determine whether the operating data of the supercritical unit at the trigger time meets the preset triggering conditions of the mathematical model, obtain a first intermediate result, and determine whether the first intermediate result meets the preset requirements; Determine whether the triggering time meets the preset requirements; Determine whether the warning result meets the preset requirements; If any one or more of the first intermediate result, the trigger time, and the warning result fail to meet the preset requirements, the mathematical model is subjected to parameter tuning. The step of determining whether the operating data of the supercritical unit at the trigger time meets the preset triggering conditions of the mathematical model, obtaining a first intermediate result, and determining whether the first intermediate result meets the preset requirements includes: Based on the mathematical model, at least one preset triggering condition for the mathematical model is determined; Obtain the operating data of the supercritical unit at the trigger time; Based on the operational data, obtain operational information corresponding to each of the at least one preset triggering condition; Determine whether each piece of runtime information meets its corresponding preset triggering conditions; In response to each running information satisfying its corresponding preset triggering condition, it is determined that the first intermediate result meets the preset requirements; In response to the fact that at least one running information does not meet its corresponding preset trigger condition, it is determined that the first intermediate result does not meet the preset requirements; Determining whether the trigger time meets the preset requirements includes: Obtain the preset triggering period of the mathematical model; The preset trigger period and the trigger time are compared to obtain the first comparison result; In response to the first comparison result indicating that the trigger time does not fall within the preset trigger period, it is determined that the trigger time does not meet the preset requirements; Determining whether the warning result meets the preset requirements includes: Obtain the actual early warning results corresponding to the measurement point data; The warning result is compared with the actual warning result to obtain a second comparison result; In response to the second comparison result indicating that the warning result is not consistent with the actual warning result, it is determined that the warning result does not meet the preset requirements; The verification process for the multiple measurement points based on the mathematical model involves verifying each measurement point using its own measurement point name, including: Based on the measurement point names and measurement point identifiers of multiple measurement points in the mathematical model, search the historical data of the multiple measurement point identifiers of the mathematical model that are pre-stored in the database. In response to the absence of historical data for at least one measurement point in the database, based on the measurement point name of the at least one measurement point, related measurement points of the at least one measurement point are determined among the plurality of measurement points. The determination of related measurement points includes: based on the measurement point name of the measurement point, determining keywords, key words and synonyms of key words in the measurement point name, and searching for related measurement points of the measurement point based on the keywords, key words and synonyms of key words. Search the database for historical data of the relevant measuring points that are stored in advance; In response to the failure to find historical data of the relevant measurement points pre-stored in the database, it is determined that at least one measurement point has failed verification. The mathematical model is then subjected to parameter tuning, and the step of searching for historical data of each of the multiple measurement point identifiers of the mathematical model pre-stored in the database is repeated.
2. A supercritical unit early warning model testing device based on a big data platform, characterized in that, The apparatus implements the method as described in claim 1, the apparatus comprising: The acquisition module is used to acquire the measurement point data of each of the multiple mathematical models; wherein, the multiple mathematical models are all machine learning models pre-trained based on the historical operating data of the supercritical unit; the measurement point data includes the measurement point name, measurement point identifier and monitoring value; The verification module is used to verify the multiple measurement points for each mathematical model based on the measurement point names of each measurement point in the mathematical model. The input module is used to input the monitoring value into the mathematical model in response to the multiple measuring points passing the verification, and to obtain the early warning result output by the mathematical model and the trigger time of the mathematical model. The first determining module is used to determine whether the operating data of the supercritical unit at the trigger time meets the preset triggering conditions of the mathematical model, obtain a first intermediate result, and determine whether the first intermediate result meets the preset requirements. The second determining module is used to determine whether the triggering time meets the preset requirements; The third determining module is used to determine whether the warning result meets the preset requirements; The parameter tuning module is used to perform parameter tuning on the mathematical model in response to any one or more of the first intermediate result, the trigger time, and the warning result failing to meet preset requirements.
3. An electronic device, comprising: include: A processor, and a memory communicatively connected to the processor; The memory stores computer-executed instructions; The processor executes computer execution instructions stored in the memory to implement the method as described in claim 1.
4. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, which, when executed by a processor, are used to implement the method as described in claim 1.
5. A computer program product, characterised in that, Includes a computer program that, when executed by a processor, implements the method of claim 1.
Citation Information
Patent Citations
Power plant thermal performance analysis and optimization system
CN112200433A
