CEMS (Continuous Emission Monitoring System) data anomaly detection and completion method, device and equipment based on data driving and storage medium

Abstract

The invention discloses a data-driven CEMS (Continuous Emission Monitoring System) data anomaly detection and supplementation method, device and equipment and a storage medium. The method comprises the following steps: performing second-level monitoring on a plurality of parameters required by calculation of carbon dioxide emission in the CEMS system to obtain monitoring data of the plurality of parameters; performing anomaly detection on the monitoring data and marking a quality inspection code; screening normal data from the acquired monitoring data according to the quality inspection code to calculate the carbon dioxide emission; converting the calculated second-level carbon dioxide emission into minute-level carbon dioxide emission; and supplementing the missing carbon dioxide emission at the minute level by adopting a machine learning method. According to the method, the minute-level carbon dioxide emission data is directly supplemented by adopting a machine learning method, the defect that in the prior art, parameters need to be supplemented separately is overcome, the accuracy and efficiency of data supplementing are improved, and more reliable data support is provided for carbon emission accounting work and carbon market construction.

Description

Technical Field

[0001] The present disclosure relates to the technical field of continuous monitoring of flue gas emissions, and in particular to a method, device, equipment and storage medium for detecting and completing CEMS data anomaly based on data-driven. Background Art

[0002] In the actual operation of CEMS (Continuous Emission Monitoring System), data loss and data anomalies are inevitable due to the influence of various factors such as working environment, operation and maintenance and calibration, signal interruption, equipment failure, etc. The emergence of these situations will affect the integrity, accuracy and precision of the carbon dioxide CEMS measured data, thereby affecting the carbon emission accounting work and carbon market construction.

[0003] At present, there have been some studies on data anomaly detection and data completion in the field of continuous monitoring. In terms of data anomaly detection, the existing technology mainly relies on a single outlier coefficient comparison method to identify abnormal conditions of the monitored object. For example, a study disclosed a method of comparing the outlier coefficient of the monitored object at the current moment with the local outlier coefficient of historical data to determine the possible abnormal phenomena of the monitored object at the current moment. However, this method cannot cope with complex abnormal patterns.

[0004] In terms of data completion, the current data completion of the missing data in the CEMS carbon emission system is mainly aimed at filling in the various parameters required for carbon emission calculations (such as concentration and flow). Existing filling technologies mainly use linear or nonlinear models based on load or output to fit a single parameter. However, there is currently no method to directly fill in carbon dioxide emissions. When all relevant data in the CEMS system are missing at the same time due to multiple factors such as communication failure, server downtime, power outage, network failure or bad weather, the single parameter filling method is difficult to play a role. Moreover, traditional filling technologies rely on simple data relationship models based on load or output, which cannot fully consider the complex working conditions in the production process, and it is difficult to adapt to changes in production conditions.

[0005] In view of this, it is necessary to provide a higher quality CEMS data anomaly detection and completion technology to better adapt to complex production conditions and needs. Summary of the invention

[0006] The present disclosure proposes a data-driven CEMS data anomaly detection and completion technology, which uses machine learning methods to directly complete minute-level carbon dioxide emissions data. It can process complex data patterns more accurately and can adaptively respond to new data changes, thereby achieving higher quality data completion.

[0007] According to an embodiment of the present disclosure, a data-driven CEMS data anomaly detection and completion method is proposed, comprising:

[0008] Monitor multiple parameters required for calculating carbon dioxide emissions in the CEMS system in seconds and obtain monitoring data of multiple parameters;

[0009] Perform anomaly detection on monitoring data and mark quality inspection codes;

[0010] Calculate the carbon dioxide emissions by screening normal data from the acquired monitoring data according to the quality inspection code;

[0011] Convert the calculated second-level carbon dioxide emissions into minute-level carbon dioxide emissions;

[0012] Machine learning methods are used to fill in the missing CO2 emissions at the minute level.

[0013] In some embodiments, the plurality of parameters include carbon dioxide concentration, flue gas flow rate, flue gas temperature, pressure and humidity.

[0014] In some embodiments, the performing of anomaly detection on the monitoring data and marking of the quality inspection code includes performing the following missing status check:

[0015] Check the missing status of monitoring data for each parameter;

[0016] If it is found during the inspection that the monitoring data of any parameter is missing, the monitoring data of each parameter corresponding to the moment when the monitoring data is missing are all identified as abnormal data and are all marked with the first abnormal quality inspection code.

[0017] In some embodiments, the performing of anomaly detection on the monitoring data and marking of quality inspection codes includes performing the following quality inspections:

[0018] Receive quality status information obtained by monitoring the monitoring data of each parameter through quality monitoring by monitoring instruments;

[0019] If the quality status information indicates that any monitoring data does not meet expectations, the monitoring data of all parameters at the time corresponding to the monitoring data indicated to not meet expectations are identified as abnormal data and are marked with a second abnormal quality inspection code.

[0020] In some embodiments, the performing of anomaly detection on the monitoring data and marking of the quality inspection code comprises performing the following threshold checks:

[0021] Perform threshold checks on the monitoring data for each parameter;

[0022] If any monitoring data of any parameter exceeds the preset threshold range of the parameter, the monitoring data of all parameters at the time corresponding to the monitoring data exceeding the preset threshold range are identified as abnormal data and are all marked with a third abnormal quality inspection code.

[0023] In some embodiments, the performing of anomaly detection on the monitoring data and marking of the quality inspection code comprises performing the following continuity check:

[0024] Conduct continuity checks on monitoring data for each parameter;

[0025] If the absolute value of the difference between any monitoring data of any parameter and the monitoring data of the parameter at the previous monitoring moment is greater than the preset difference threshold, it is judged that the monitoring data has mutated, and it is judged that the monitoring data with a value of 0 or exceeding the preset upper limit within the preset range before and after the mutated monitoring data are all discontinuous data, and the monitoring data of all parameters at the moment corresponding to the discontinuous data are all identified as abnormal data and are all marked with the fourth abnormal quality inspection code.

[0026] In some implementations, the method further includes marking an abnormal cause and a data index for the detected abnormal data.

[0027] In some embodiments, the machine learning method employs a SAITS model.

[0028] In some embodiments, the loss function of the SAITS model includes:

[0029] The reconstruction loss is used to calculate the mean absolute error between the actual CO2 emissions and the CO2 emissions predicted by the SAITS model;

[0030] The interpolation loss is calculated as the mean absolute error between the actual CO2 emissions at the artificially missing locations and the CO2 emissions predicted by the SAITS model.

[0031] According to an embodiment of the present disclosure, a data-driven CEMS data anomaly detection and completion device is proposed, comprising:

[0032] Second-level parameter monitoring unit, used to monitor multiple parameters required for calculating carbon dioxide emissions in the CEMS system in seconds and obtain monitoring data of multiple parameters;

[0033] Anomaly detection and marking unit, used to detect anomalies in monitoring data and mark quality inspection codes;

[0034] A carbon dioxide emission calculation unit, used to calculate carbon dioxide emissions by screening normal data from the acquired monitoring data according to the quality inspection code;

[0035] A minute-level data conversion unit, used to convert the calculated second-level carbon dioxide emissions into minute-level carbon dioxide emissions;

[0036] The minute-level carbon dioxide emissions completion unit is used to use machine learning methods to complete the missing carbon dioxide emissions at the minute level.

[0037] According to one embodiment of the present disclosure, an electronic device is provided, the device comprising a memory and a processor, the memory being used to store computer instructions executable on the processor, the processor being used to implement any of the above methods when executing the computer instructions.

[0038] According to an embodiment of the present disclosure, a computer-readable storage medium is provided, on which a computer program is stored. When the program is executed by a processor, any of the above methods is implemented.

[0039] The technical solution proposed in the present disclosure can comprehensively identify and mark data anomalies and record the corresponding anomaly reasons by performing multi-dimensional anomaly detection on CEMS data, including missing status check, quality status check, threshold check and continuity check, so as to ensure the reliability and traceability of data quality. In terms of data completion, the present disclosure proposes a method for directly completing carbon dioxide emissions without the need to complete each parameter involved in the calculation separately. This method can not only more accurately reflect the actual carbon emissions, but also still play a role when all relevant data are missing at the same time due to communication failures, server downtime and other reasons in the CEMS system. The present disclosure can also effectively reduce the complexity of data processing by calculating second-level data as minute-level data before completing it, and introduces the use of machine learning methods for data completion, which can more accurately process complex data patterns and adaptively respond to changes in production conditions, thereby providing a higher quality data completion solution and providing more reliable data support for carbon emission accounting and carbon market construction.

[0040] Other features and advantages of the present disclosure are described in detail below. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the specification and, together with the description, serve to explain the principles of the specification.

[0042] Figure 1 A flow chart of a data-driven CEMS data anomaly detection and completion method according to an embodiment of the present disclosure is shown.

[0043] Figure 2 It is a schematic diagram of the structure of an electronic device shown in at least one embodiment of the present disclosure. DETAILED DESCRIPTION

[0044] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present disclosure. Instead, they are merely examples of devices and methods consistent with some aspects of the present disclosure as detailed in the appended claims.

[0045] The disclosed embodiments may be applied to a computer system / server that may operate with numerous other general or special computing system environments or configurations. Examples of well-known computing systems, environments, and / or configurations suitable for use with a computer system / server include, but are not limited to, personal computer systems, server computer systems, thin clients, thick clients, handheld or laptop devices, microprocessor-based systems, set-top boxes, programmable consumer electronics, networked personal computers, minicomputer systems, mainframe computer systems, and distributed cloud computing technology environments including any of the above systems, and the like.

[0046] Computer systems / servers may be described in the general context of computer system executable instructions (such as program modules) executed by a computer system. In general, program modules may include routines, programs, object programs, components, logic, data structures, etc., which perform specific tasks or implement specific abstract data types. Computer systems / servers may be implemented in a distributed cloud computing environment where tasks are performed by remote processing devices linked through a communication network. In a distributed cloud computing environment, program modules may be located on local or remote computing system storage media including storage devices.

[0047] Figure 1 A flow chart of a data-driven CEMS data anomaly detection and completion method according to an embodiment of the present disclosure is shown. As shown in the figure, the method includes steps 11 to 15.

[0048] Step 11, monitor multiple parameters required for calculating carbon dioxide emissions in the CEMS system in seconds to obtain monitoring data of the multiple parameters.

[0049] In some embodiments, the multiple parameters include carbon dioxide concentration, flue gas flow, flue gas temperature, pressure and humidity. Carbon dioxide concentration can reflect the content of carbon dioxide in flue gas, flue gas flow can reflect the volume of flue gas discharged, and flue gas temperature, pressure and humidity can be used to correct the flue gas flow to ensure the accuracy of the calculation result. Based on these parameters, the final carbon dioxide emissions can be accurately calculated.

[0050] According to the present disclosure, the above parameters can be monitored at the second level. It should be noted that second-level monitoring refers to monitoring granularity of several seconds. For example, it can be considered that when the monitoring granularity is below 5 seconds, it can be called second-level monitoring. For example, in one example, the monitoring granularity is 1 second, then a complete set of monitoring data can be obtained every second, including the carbon dioxide concentration value, flue gas flow value, flue gas temperature value, pressure value and humidity value at that moment; in another example, the monitoring granularity is 5 seconds, then a complete set of monitoring data can be obtained every five seconds, including the carbon dioxide concentration value, flue gas flow value, flue gas temperature value, pressure value and humidity value at that moment. Through second-level monitoring, changes in emission conditions can be reflected in a timely and accurate manner, providing basic data support for subsequent abnormality detection and data completion.

[0051] The monitoring of each parameter is completed by the corresponding monitoring instrument. In some embodiments, in addition to outputting monitoring data, each monitoring instrument can also output data status information, such as status information indicating that the quality has reached the expected state (e.g., Quality Good, Quality OK, Quality Valid, etc.) and status information indicating that the quality has not reached the expected state (e.g., Quality Error, Quality Bad, Quality Low, etc.), which will be used for subsequent data quality checks. The range of the monitoring instrument is set according to the actual working conditions to provide a judgment basis for subsequent continuity checks.

[0052] Step 12: perform anomaly detection on the monitoring data and mark the quality inspection code.

[0053] The anomaly monitoring may include missing state checking, quality checking, threshold checking, and continuity checking.

[0054] In some embodiments, the monitoring data of each parameter may be checked for missing status. If the check finds that any parameter is missing at a certain moment (e.g., swallow seconds), the monitoring data of all parameters corresponding to this moment are identified as abnormal data and are marked with a first abnormal quality inspection code. According to this embodiment, the integrity of the monitoring data can be ensured, and the erroneous calculation caused by the missing of some data can be avoided.

[0055] In some embodiments, quality status information obtained by monitoring the quality of monitoring data of various parameters by a monitoring instrument can be received. If the quality status information given by any monitoring instrument indicates that any monitoring data does not meet expectations, for example, the monitoring instrument gives a low quality status indication such as Quality Error, Quality Bad or Quality Low, then the monitoring data of all parameters at the time corresponding to the monitoring data indicated to have not met expectations are identified as abnormal data and are marked with a second abnormal quality inspection code. According to this embodiment, data quality abnormalities can be discovered in a timely manner through the quality status indication of the monitoring instrument.

[0056] In some embodiments, a threshold check may be performed on the monitoring data of each parameter. If any monitoring data of any parameter exceeds the preset threshold range of the parameter, the monitoring data of all parameters at that moment are identified as abnormal data and are marked with a third abnormal quality inspection code. This embodiment can identify abnormal data that is significantly deviated from the normal range through the threshold check.

[0057] In some embodiments, the monitoring data of each parameter can be checked for continuity. If the absolute value of the difference between the current monitoring data of a parameter and the monitoring data at the previous monitoring moment is greater than a preset difference threshold, the monitoring data is judged to have a mutation. At the same time, within a preset range before and after the monitoring data that has mutated, the monitoring data with a value of 0 or exceeding a preset upper limit are all identified as discontinuous data. For these discontinuous data, the monitoring data of all parameters at the corresponding moment are identified as abnormal data, and are all marked with a fourth abnormal quality inspection code. This embodiment is mainly used to handle data discontinuities generated during instrument calibration.

[0058] The continuity test is illustrated by taking the flue gas flow parameter as an example. For example, the range of the flue gas flow meter is 0 to 1,000,000 (unit: m 3 / h). When the flue gas flow meter is calibrated, the following method can be adopted: first, the reading is reduced to a value close to 0 and lasts for about 2 minutes, and then it suddenly increases to about 700,000 and also lasts for about 2 minutes. The preset difference threshold and the preset upper limit can be set to 690,000. First, the monitoring data whose absolute value of the difference with the flue gas flow monitoring data at the previous monitoring moment is greater than 690,000 can be found, and it can be regarded as a mutation point caused by potential calibration. Then, for the mutation point found, the data within 2 minutes before and after it are checked respectively. If there are monitoring data with a value close to 0 or greater than 690,000 within this range, these data are all judged as discontinuous data. The monitoring data of all parameters (carbon dioxide concentration, flue gas flow, flue gas temperature, pressure and humidity) at the time corresponding to all the monitoring data judged to be discontinuous are identified as abnormal data and are marked with the fourth abnormal quality inspection code.

[0059] In some embodiments, for monitoring data that is not detected to be abnormal after the above four checks, a normal quality inspection code may be marked therefor, indicating that the data can be used for subsequent emission calculations.

[0060] In some implementations, when marking the quality inspection code, the detected abnormal data is also marked with the corresponding abnormal cause and index, such as a time index or a record index in a database. According to this implementation, by marking the abnormal cause and index, data tracing and problem diagnosis are facilitated, which is beneficial for subsequent operation and maintenance personnel to perform equipment maintenance and provides support and basis for subsequent system optimization.

[0061] Step 13, filter normal data from the acquired monitoring data according to the quality inspection code to calculate the carbon dioxide emissions.

[0062] For data without abnormality detection, such as monitoring data marked with a normal quality inspection code, it can be used as valid data for calculating carbon dioxide emissions. In some embodiments, the data marked with a normal quality inspection code at that moment meets the following conditions:

[0063] None of the parameters were missing;

[0064] The quality status information of all parameters output by the monitoring instrument shows that the data quality meets the standards (such as QualityGood, QualityOK, etc.)

[0065] All parameters were within their respective preset thresholds;

[0066] There were no discontinuities due to calibration in any of the parameters.

[0067] Based on the screened valid data, the carbon dioxide emissions at the corresponding moment are calculated using parameters such as carbon dioxide concentration, flue gas flow, flue gas temperature, pressure and humidity in accordance with relevant technical specifications, thereby ensuring that the calculated second-level carbon dioxide emissions are extremely accurate and reliable.

[0068] Step 14, converting the calculated second-level carbon dioxide emissions into minute-level carbon dioxide emissions.

[0069] Based on the second-level carbon dioxide emissions data calculated in the previous step, several consecutive second-level carbon dioxide emissions are converted into one minute-level carbon dioxide emissions. Minute-level data means that each data can be used to represent the carbon dioxide emissions level within several minutes. For example, it can be considered that data representing the carbon dioxide emissions level within 15 minutes can be called minute-level data. In some examples, multiple second-level carbon dioxide emissions within each minute can be converted into one minute-level carbon dioxide emissions; in other examples, multiple second-level carbon dioxide emissions within every 8 minutes can be converted into one minute-level carbon dioxide emissions; in other examples, multiple second-level carbon dioxide emissions within every 15 minutes can be converted into one minute-level carbon dioxide emissions.

[0070] Since the aforementioned anomaly detection may cause some second-level data to be marked as abnormal and the carbon dioxide emissions corresponding to that moment are not calculated, the conversion can be performed based on the effective second-level carbon dioxide emissions that can be used for conversion within the time range corresponding to the minute-level data.

[0071] When the amount of valid second-level data within the time range corresponding to a certain minute-level data meets the specified requirements (for example, the proportion of valid second-level data is not less than a threshold, such as not less than 75%), the minute-level data can be converted. In this case, those skilled in the art can use any appropriate conversion method (such as arithmetic mean) to calculate the minute-level carbon dioxide emissions.

[0072] When the amount of valid second-level data within the time range corresponding to a certain minute-level data does not meet the specified requirements (for example, the proportion of valid second-level data is lower than the threshold, such as lower than 75%), the minute-level data can be marked as missing and left for subsequent completion processing.

[0073] By converting second-level data into minute-level data, the complexity of subsequent data completion can be significantly reduced while ensuring the reliability of the data.

[0074] Step 15, use machine learning methods to fill in the missing carbon dioxide emissions at the minute level.

[0075] In some embodiments, the machine learning method employs a SAITS model.

[0076] After in-depth research, the inventor believes that the SAITS model is based on the Transformer architecture and can effectively capture long-term dependencies and feature correlations in time series by designing a diagonal mask self-attention mechanism, and has better parallel computing performance. Compared with other deep learning models, SAITS avoids the shortcomings of slow training speed and memory limitation of RNN-type models, and also avoids the problem that the loss function of generative models such as GAN may not converge and training is difficult. It is particularly suitable for CEMS data completion.

[0077] The carbon dioxide emission time series data to be completed is input into the SAITS model. In some embodiments, the SAITS model can complete the data by only inputting the minute-level carbon dioxide emission data, without the assistance of other feature data. The input data may include known minute-level carbon dioxide emission data and the location information of the missing data.

[0078] In some embodiments, the training of the SAITS model may adopt a joint optimization method, including two training tasks: a reconstruction task and an interpolation task.

[0079] In the reconstruction task, the mean absolute error between the actual CO2 emissions and the CO2 emissions predicted by the SAITS model can be calculated as the reconstruction loss. In the final output of the model, the original data is used for the non-missing positions, and the reconstruction loss can be calculated based on the intermediate output of the model.

[0080] In the interpolation task, before the training data is input, some known data are artificially missing, and the mean absolute error between the actual CO2 emissions at the artificially missing positions and the CO2 emissions predicted by the SAITS model is calculated as the interpolation loss.

[0081] The final loss function of the model can be obtained by combining the reconstruction loss and the interpolation loss through weight coefficients, and the model training is completed by minimizing the loss function.

[0082] According to this embodiment, the trained SAITS model is used to fill in the missing carbon dioxide emissions data. The self-attention mechanism of the SAITS model gives it a global vision, can efficiently capture the long-term dependencies of time series, and more accurately understand contextual information. By introducing the improvement of the diagonal mask module to the multi-head attention mechanism, the model's ability to capture the temporal dependencies and feature correlations between time steps is further enhanced.

[0083] This implementation method introduces the SAITS model to complete the data, which has the following advantages: it can effectively handle the long-term dependencies of time series; it has good parallel computing performance; the training is stable and the risk of non-convergence is avoided; and the carbon dioxide emissions are directly completed without the assistance of other feature data.

[0084] As described above, the data-driven CEMS data anomaly detection and completion method proposed in this embodiment can fully identify and mark data anomalies and record the corresponding anomaly reasons by performing multi-dimensional anomaly detection on CEMS data, including missing state check, quality state check, threshold check and continuity check, so as to ensure the reliability and traceability of data quality. In terms of data completion, this embodiment directly completes the carbon dioxide emissions without completing the various parameters involved in the calculation separately. This method can not only more accurately reflect the actual carbon emissions, but also still play a role when all relevant data are missing at the same time due to communication failures, server downtime and other reasons in the CEMS system. This embodiment also effectively reduces the complexity of data processing by calculating the second-level data as minute-level data before completing it, and introduces the use of machine learning methods for data completion, which can more accurately process complex data patterns and adaptively respond to changes in production conditions, thereby providing a higher quality data completion solution and providing more reliable data support for carbon emission accounting and carbon market construction.

[0085] The following is an example of anomaly detection and completion of flue gas monitoring data in a thermal power plant.

[0086] 1. Monitor the five parameters involved in the calculation of carbon dioxide emissions from thermal power plants in seconds. These five parameters include carbon dioxide concentration, flue gas flow, flue gas temperature, pressure and humidity.

[0087] 2. Perform anomaly detection on the acquired monitoring data and mark the quality inspection code:

[0088] a) Check the missing status of the monitoring data of the five parameters one by one according to the monitoring time granularity. If one or more parameters are detected to be missing (such as swallow seconds), the monitoring data of the five parameters corresponding to this time will be marked with the quality inspection code 1;

[0089] b) Check the quality status of the monitoring data of the five parameters. If the monitoring instrument of one or more parameters is detected to give a low quality status such as Quality Error, Quality Bad or Quality Low, the monitoring data of the five parameters corresponding to this time will be marked with the quality inspection code 2;

[0090] c) Perform threshold check on the monitoring data of the five parameters. If the monitoring data of one or more parameters are found to be beyond the preset threshold range, the monitoring data of the five parameters corresponding to this time will be marked with the quality inspection code 3;

[0091] d) Check the continuity of the five parameters. Take the flue gas flow rate as an example. The flow meter range is 0 to 1,000,000m 3 / h, when the absolute value of the difference between the monitoring data of two consecutive seconds is detected to be greater than 690,000m 3 / h, identify it as a potential mutation point; check the monitoring data within 2 minutes before and after the mutation point; if there is a value close to 0m 3 / h or greater than 690,000m 3 / h monitoring data, it is determined to be discontinuous data in the calibration process; the monitoring data of the five parameters at the corresponding moments of these discontinuous data are all marked with the quality inspection code 4;

[0092] e) For monitoring data with no abnormalities detected, the quality inspection code is marked as 0.

[0093] 3. Filter out all monitoring data with a quality inspection code of 0, and calculate the second-level carbon dioxide emissions at the corresponding time based on these valid data.

[0094] 4. Convert the second-level data to minute-level data. If the proportion of valid second-level carbon dioxide emissions data within a number of consecutive minutes is not less than a threshold (e.g. 75%), the average carbon dioxide emissions within the consecutive minutes can be calculated as the corresponding minute-level carbon dioxide emissions; otherwise, the corresponding minute-level carbon dioxide emissions are marked as missing;

[0095] 5. Use the SAITS model to complete the missing CO2 emissions data at the minute level. Specifically, the minute-level CO2 emissions time series is input into the trained SAITS model to complete the missing data.

[0096] According to another embodiment of the present disclosure, a data-driven CEMS data anomaly detection and completion device is also proposed, including:

[0097] Second-level parameter monitoring unit, used to monitor multiple parameters required for calculating carbon dioxide emissions in the CEMS system in seconds and obtain monitoring data of multiple parameters;

[0098] Anomaly detection and marking unit, used to detect anomalies in monitoring data and mark quality inspection codes;

[0099] A carbon dioxide emission calculation unit, used to calculate carbon dioxide emissions by screening normal data from the acquired monitoring data according to the quality inspection code;

[0100] A minute-level data conversion unit, used to convert the calculated second-level carbon dioxide emissions into minute-level carbon dioxide emissions;

[0101] The minute-level carbon dioxide emissions completion unit is used to use machine learning methods to complete the missing carbon dioxide emissions at the minute level.

[0102] Figure 2An electronic device provided for at least one embodiment of the present disclosure includes a memory and a processor, wherein the memory is used to store computer instructions executable on the processor, and the processor is used to implement the data-driven CEMS data anomaly detection and completion method described in any embodiment or implementation of the present disclosure when executing the computer instructions.

[0103] At least one embodiment of the present disclosure further provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the data-driven CEMS data anomaly detection and completion method described in any embodiment or implementation of the present disclosure.

[0104] It should be understood by those skilled in the art that one or more embodiments of the present specification may be provided as a method, system or computer program product. Therefore, one or more embodiments of the present specification may take the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware. Moreover, one or more embodiments of the present specification may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0105] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the data processing device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0106] The above is a description of a specific embodiment of the specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the embodiments and still achieve the desired results. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0107] The embodiments of the subject matter and functional operations described in this specification may be implemented in the following: digital electronic circuits, tangibly embodied computer software or firmware, computer hardware including the structures disclosed in this specification and their structural equivalents, or a combination of one or more of them. The embodiments of the subject matter described in this specification may be implemented as one or more computer programs, i.e., one or more modules in computer program instructions encoded on a tangible non-temporary program carrier to be executed by a data processing device or to control the operation of the data processing device. Alternatively or additionally, the program instructions may be encoded on an artificially generated propagation signal, such as a machine-generated electrical, optical or electromagnetic signal, which is generated to encode information and transmit it to a suitable receiver device for execution by a data processing device. The computer storage medium may be a machine-readable storage device, a machine-readable storage substrate, a random or serial access memory device, or a combination of one or more of them.

[0108] The processes and logic flows described in this specification can be performed by one or more programmable computers executing one or more computer programs to perform corresponding functions by operating on input data and generating output. The processes and logic flows can also be performed by special purpose logic circuits, such as FPGAs (field programmable gate arrays) or ASICs (application-specific integrated circuits), and the apparatus can also be implemented as special purpose logic circuits.

[0109] Computers suitable for executing computer programs include, for example, general and / or special microprocessors, or any other type of central processing unit. Typically, the central processing unit will receive instructions and data from a read-only memory and / or a random access memory. The basic components of a computer include a central processing unit for implementing or executing instructions and one or more memory devices for storing instructions and data. Typically, the computer will also include one or more large-capacity storage devices for storing data, such as magnetic disks, magneto-optical disks, or optical disks, or the computer will be operably coupled to this large-capacity storage device to receive data from it or to transmit data to it, or both. However, the computer does not necessarily have such a device. In addition, the computer can be embedded in another device, such as a mobile phone, a personal digital assistant (PDA), a mobile audio or video player, a game console, a global positioning system (GPS) receiver, or a portable storage device such as a universal serial bus (USB) flash drive, to name a few.

[0110] Computer readable media suitable for storing computer program instructions and data include all forms of non-volatile memory, media and memory devices, including, for example, semiconductor memory devices (e.g., EPROM, EEPROM and flash memory devices), magnetic disks (e.g., internal hard disks or removable disks), magneto-optical disks, and CD ROM and DVD-ROM disks. The processor and memory may be supplemented by, or incorporated in, special purpose logic circuitry.

[0111] Although this specification includes many specific implementation details, these should not be interpreted as limiting the scope of any invention or the scope of protection claimed, but are mainly used to describe the features of the specific embodiments of specific inventions. Certain features described in multiple embodiments in this specification may also be implemented in combination in a single embodiment. On the other hand, the various features described in a single embodiment may also be implemented separately in multiple embodiments or in any suitable sub-combination. In addition, although features may work in certain combinations as described above and even initially claim protection, one or more features from the claimed combination may be removed from the combination in some cases, and the claimed combination may point to a sub-combination or a variation of a sub-combination.

[0112] Similarly, although operations are depicted in a particular order in the accompanying drawings, this should not be understood as requiring that these operations be performed in the particular order shown or performed sequentially, or requiring that all illustrated operations be performed to achieve the desired results. In some cases, multitasking and parallel processing may be advantageous. In addition, the separation of various system modules and components in the above-described embodiments should not be understood as requiring such separation in all embodiments, and it should be understood that the described program components and systems can generally be integrated together in a single software product, or packaged into multiple software products.

[0113] Thus, specific embodiments of the subject matter have been described. Other embodiments are within the scope of the appended claims. In some cases, the actions recited in the claims can be performed in a different order and still achieve the desired results. In addition, the processes depicted in the drawings do not necessarily require the particular order or sequential order shown to achieve the desired results. In some implementations, multitasking and parallel processing may be advantageous.

[0114] The above description is merely a preferred embodiment of one or more embodiments of the present specification and is not intended to limit one or more embodiments of the present specification. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of one or more embodiments of the present specification shall be included in the scope of protection of one or more embodiments of the present specification.

Claims

1. A data-driven CEMS data anomaly detection and completion method, characterized in that: include: Monitor multiple parameters required for calculating carbon dioxide emissions in the CEMS system in seconds and obtain monitoring data of multiple parameters; Perform anomaly detection on monitoring data and mark quality inspection codes; Calculate the carbon dioxide emissions by screening normal data from the acquired monitoring data according to the quality inspection code; Convert the calculated second-level carbon dioxide emissions into minute-level carbon dioxide emissions; Machine learning methods are used to fill in the missing CO2 emissions at the minute level.

2. The method according to claim 1, characterized in that The plurality of parameters include carbon dioxide concentration, flue gas flow, flue gas temperature, pressure and humidity.

3. The method according to claim 1, characterized in that The abnormality detection and quality inspection code marking of the monitoring data includes the following missing status check: Check the missing status of monitoring data for each parameter; If it is found during the inspection that the monitoring data of any parameter is missing, the monitoring data of each parameter corresponding to the moment when the monitoring data is missing are all identified as abnormal data and are all marked with the first abnormal quality inspection code.

4. The method according to claim 1, characterized in that The abnormality detection and quality inspection code marking of the monitoring data includes the following quality inspections: Receive quality status information obtained by monitoring the monitoring data of each parameter through quality monitoring by monitoring instruments; If the quality status information indicates that any monitoring data does not meet expectations, the monitoring data of all parameters at the time corresponding to the monitoring data indicated to not meet expectations are identified as abnormal data and are marked with a second abnormal quality inspection code.

5. The method according to claim 1, characterized in that: The abnormality detection and quality inspection code marking of the monitoring data includes the following threshold checks: Perform threshold checks on the monitoring data for each parameter; If any monitoring data of any parameter exceeds the preset threshold range of the parameter, the monitoring data of all parameters at the time corresponding to the monitoring data exceeding the preset threshold range are identified as abnormal data and are all marked with a third abnormal quality inspection code.

6. The method according to claim 1, characterized in that The abnormality detection and quality inspection code marking of the monitoring data includes the following continuity checks: Conduct continuity checks on monitoring data for each parameter; If the absolute value of the difference between any monitoring data of any parameter and the monitoring data of the parameter at the previous monitoring moment is greater than the preset difference threshold, it is judged that the monitoring data has mutated, and it is judged that the monitoring data with a value of 0 or exceeding the preset upper limit within the preset range before and after the mutated monitoring data are all discontinuous data, and the monitoring data of all parameters at the moment corresponding to the discontinuous data are all identified as abnormal data and are all marked with the fourth abnormal quality inspection code.

7. The method according to claim 1, characterized in that The method further comprises marking an abnormal reason and a data index for the detected abnormal data.

8. The method according to claim 1, characterized in that: The machine learning method adopts the SAITS model.

9. The method according to claim 8, characterized in that The loss function of the SAITS model includes: The reconstruction loss is used to calculate the mean absolute error between the actual CO2 emissions and the CO2 emissions predicted by the SAITS model; The interpolation loss is calculated as the mean absolute error between the actual CO2 emissions at the artificially missing locations and the CO2 emissions predicted by the SAITS model.

10. A data-driven CEMS data anomaly detection and completion device, characterized in that: include: Second-level parameter monitoring unit, used to monitor multiple parameters required for calculating carbon dioxide emissions in the CEMS system in seconds and obtain monitoring data of multiple parameters; Anomaly detection and marking unit, used to detect anomalies in monitoring data and mark quality inspection codes; A carbon dioxide emission calculation unit, used to calculate carbon dioxide emissions by screening normal data from the acquired monitoring data according to the quality inspection code; A minute-level data conversion unit, used to convert the calculated second-level carbon dioxide emissions into minute-level carbon dioxide emissions; The minute-level carbon dioxide emissions completion unit is used to use machine learning methods to complete the missing carbon dioxide emissions at the minute level.

11. An electronic device, characterized in that: The device comprises a memory and a processor, wherein the memory is used to store computer instructions executable on the processor, and the processor is used to implement the method according to any one of claims 1 to 9 when executing the computer instructions.

12. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method according to any one of claims 1 to 9 is implemented.

Images