Method, device and equipment for identifying abnormal pollution discharge based on multi-source data and medium

Abstract

The invention relates to a method, a device and equipment for identifying abnormal pollution discharge based on multi-source data, and a medium, and belongs to the technical field of environment monitoring. The method comprises the following steps: collecting working condition electricity consumption historical data and pollutant discharge historical data; extracting at least two key variables from the working condition electricity consumption historical data and the pollutant emission historical data, taking the key variables as endogenous variables, constructing a vector autoregression model according to the endogenous variables, and determining a lag order of the vector autoregression model by using an information criterion method; processing the working condition electricity consumption historical data and the pollutant emission historical data to obtain a training set, and training a vector autoregression model by using the training set; obtaining input data according to a to-be-predicted moment and the lag order, inputting the input data into the trained vector autoregression model, outputting a predicted value of a key variable at the to-be-predicted moment, and obtaining a corresponding measured value; and judging whether pollution discharge is abnormal or not according to the predicted value and the measured value of the key variable, and analyzing an abnormal reason.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of environmental monitoring, and specifically relates to a method and device for identifying abnormal pollution discharge based on multi-source data, equipment and a medium. BACKGROUND

[0002] Nowadays, society pays more and more attention to environmental problems, and identifying abnormal discharge of pollutants is an important prerequisite for targeted management. In the current field of environmental monitoring, CEMS systems have been widely used in atmospheric pollution source emission monitoring. They can collect parameters such as the concentrations of gaseous pollutants such as SO2 and NOx, the concentration of particulate matter, and the temperature and flow rate of flue gas in real time. Work condition power monitoring reflects the operating state of equipment by collecting power data (such as voltage, current, power factor, etc.) of production equipment and pollution control equipment in enterprises.

[0003] Industrial production is a complex system, and the operating state of production equipment directly affects the amount of pollutants generated, while the operating state of pollution control equipment determines the final amount of pollutants discharged. However, in the prior art, only the pollutant emission data detected by the CEMS system or the work condition power data are compared with a simple threshold to determine whether there is abnormal pollution discharge, without considering the internal relationship between production, control and pollutant concentration. This results in low identification accuracy and the inability to trace the source. SUMMARY

[0005] In view of the above analysis, the embodiments of the present application aim to provide a method, device, equipment and medium for identifying abnormal pollution discharge based on multi-source data, in order to solve the technical problems of low identification accuracy and inability to trace the source in the prior art.

[0006] The purpose of the present application is achieved as follows: The first aspect of the present application provides a method for identifying abnormal pollution discharge based on multi-source data, comprising: collecting work condition power historical data and pollutant emission historical data; extracting at least two key variables from the work condition power historical data and the pollutant emission historical data, taking the key variables as endogenous variables, constructing a vector autoregressive model according to the endogenous variables, and determining the lag order of the vector autoregressive model using an information criterion method; processing the work condition power historical data and the pollutant emission historical data to obtain a training set, and training the vector autoregressive model using the training set; obtaining input data according to a to-be-predicted time and a lag order, inputting the input data into the trained vector autoregressive model, outputting a predicted value of the key variable at the to-be-predicted time, and obtaining a corresponding measured value; determining whether there is abnormal pollution discharge according to the predicted value and the measured value of the key variable, and analyzing the cause of the abnormality.

[0007] Further, the operating condition electricity consumption history data at least includes power data of production equipment and pollution control equipment, and the pollutant emission history data at least includes VOCs concentration data, and the operating condition electricity consumption history data and the pollutant emission history data need to be time-aligned and processed for removing outliers.

[0008] Further, the vector autoregressive model is constructed according to the endogenous variables, and is represented as: wherein, represents t a column vector composed of endogenous variables at the moment, p represents a lag order, represents a lag coefficient matrix, represents t a random error vector at the moment.

[0009] Further, the information criterion method is used to determine the lag order of the vector autoregressive model, including: fitting the vector autoregressive model using different lag orders respectively, calculating the information criterion value at each fitting, and screening out the lag order corresponding to the minimum information criterion value.

[0010] Further, the operating condition electricity consumption history data and the pollutant emission history data are processed to obtain a training set, including: screening data corresponding to the endogenous variables in the operating condition electricity consumption history data and the pollutant emission history data, cutting and splicing the screened data according to the lag order, obtaining training sample pairs meeting the input and output requirements of the vector autoregressive model, and all training sample pairs forming the training set.

[0011] Further, whether the emission is abnormal in the to-be-predicted period is determined according to the predicted data and the measured data of the key variables, and an abnormal reason is obtained, including: calculating the residual between the predicted value and the measured value of each key variable, and forming a joint residual vector; calculating the Mahalanobis distance according to the joint residual vector and the residual covariance matrix during training of the vector autoregressive model, and if the Mahalanobis distance exceeds a preset threshold, determining that the emission is abnormal; and if it is determined that the emission is abnormal, performing dimension-by-dimension analysis on the joint residual vector, and identifying the abnormal reason in combination with the coupling relationship between the key variables.

[0012] Further, the method further comprises: performing model testing and evaluation on the trained vector autoregressive model, specifically: performing white noise and stability testing on the trained vector autoregressive model; calculating evaluation indexes of the model based on the predicted value and the measured value of the key variable, the evaluation indexes at least including root mean square error and mean absolute error; and adjusting parameters of the abnormal pollution model or retraining the model according to the testing and evaluation results.

[0013] The second aspect embodiment of the present application provides a device for identifying abnormal pollution based on multi-source data, comprising: a data collection module configured to collect historical working condition power consumption data and historical pollutant emission data; a model construction module configured to extract at least two key variables from the historical working condition power consumption data and the historical pollutant emission data, use the key variables as endogenous variables, and construct a vector autoregressive model according to the endogenous variables and determine a lag order of the vector autoregressive model using an information criterion method; a model training module configured to process the historical working condition power consumption data and the historical pollutant emission data to obtain a training set, and train the vector autoregressive model using the training set; a prediction module configured to obtain input data according to a to-be-predicted time and the lag order, input the input data into the trained vector autoregressive model, output a predicted value of the key variable at the to-be-predicted time, and obtain a corresponding measured value; an abnormality judgment module configured to judge whether the pollution is abnormal according to the predicted value and the measured value of the key variable, and analyze the abnormality reason.

[0014] The third aspect embodiment of the present application provides an electronic device comprising a memory and a processor, wherein the memory stores a computer program, and the computer program is executed by the processor to implement the method for identifying abnormal pollution based on multi-source data according to any one of the embodiments.

[0015] The fourth aspect embodiment of the present application provides a computer readable storage medium having a computer program stored thereon, and the computer program is executed by a processor to implement the method for identifying abnormal pollution based on multi-source data according to any one of the embodiments.

[0016] Compared with the prior art, the present application can at least achieve one of the following beneficial effects: The method for identifying abnormal pollution based on multi-source data provided by the application can reveal the dynamic coupling relationship between endogenous variables by constructing a vector autoregressive model with key variables in power consumption data and pollutant emission data as endogenous variables and enabling the model to generate multi-dimensional joint residuals in the prediction stage after training, and can realize accurate identification of abnormal pollution through statistical discrimination and dimension-by-dimension analysis. The method can comprehensively consider the internal relationship among production, treatment and emission, avoid false positives and false negatives, and significantly improve the identification accuracy. Meanwhile, when determining abnormality, the residual characteristics can be combined to locate the abnormal source, and good traceability is achieved. BRIEF DESCRIPTION OF DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or prior art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments described in the present application, and other drawings can also be obtained by those skilled in the art based on these drawings.

[0018] Figure 1 The flowchart of the method for identifying abnormal pollution based on multi-source data provided by the application embodiment 1 is shown in the figure. Figure 2 The schematic diagram of the device for identifying abnormal pollution based on multi-source data provided by the application embodiment 2 is shown in the figure. Figure 3 The schematic diagram of the electronic device architecture provided by the application embodiment 3 is shown in the figure. DETAILED DESCRIPTION

[0019] In order to make the purpose, technical scheme and advantages of the embodiments of the present application more clear, the technical scheme in the embodiments of the present application will be described clearly and completely in the following with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all the embodiments. It should be noted that the embodiments and features in the embodiments can be combined, separated, interchanged and / or rearranged without conflict. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0020] Embodiment 1 One specific embodiment of the present application, as shown in the figure, discloses a method for identifying abnormal pollution based on multi-source data, comprising the following steps: Figure 1 S1, collecting working condition power consumption history data and pollutant emission history data. S1, collecting working condition power consumption history data and pollutant emission history data.

[0021] In this embodiment, the operating condition electricity consumption history data at least includes power data of production equipment and pollution control equipment, and the pollutant emission history data at least includes VOCs concentration data. The operating condition electricity consumption history data and the pollutant emission history data need to be subjected to time alignment and abnormal value cleaning processing.

[0022] Specifically, the electricity consumption data can be obtained by installing intelligent electric meters, power collection modules or industrial Internet of Things collection terminals on the production equipment and pollution control equipment. The production equipment power and pollution control equipment power represent the production intensity and pollution control level to some extent, and they have a causal relationship and dynamic coupling with the emission amount, can reflect the whole process of “production, emission and control”, and the pollutant emission data is collected by the CEMS (continuous emission monitoring system) installed at the enterprise end, at least including VOCs concentration. VOCs is a typical pollutant index of processes such as automotive repair paint spraying and paint baking, and the concentration level can directly reflect whether the emission is over standard.

[0023] For example, the production equipment and pollution control equipment in the paint baking workshop of a certain automotive repair enterprise from March 1, 2024 to December 31, 2024 are monitored to collect current, voltage and power data every hour; the VOCs concentration and emission rate of waste gas emitted by the paint baking room during the same period are collected by the CEMS system every hour; the abnormal values in the electricity consumption data and the pollutant emission data can be identified and corrected by using the 3σ principle, and the missing values can be filled by using the linear interpolation method.

[0024] S2, at least two key variables are extracted from the operating condition electricity consumption history data and the pollutant emission history data, the key variables are used as endogenous variables, a vector autoregressive model is constructed according to the endogenous variables, and the lag order is determined by using the information criterion method.

[0025] Specifically, the key variable is a data variable selected from the original multi-source data and capable of directly representing the main influencing factor of pollution emission, such as production equipment power, pollution control equipment power and VOCs concentration. The endogenous variable refers to a variable that simultaneously serves as an input and an output in the vector autoregressive model and is determined by the historical state of the model itself. The key variable, as an endogenous variable of the VAR model, participates in modeling, and its current value depends not only on its past changes but also on the historical values of other key variables, thereby reflecting the dynamic relationship between multiple variables. After the key variables are extracted, the subsequent processing includes: (1) a vector autoregressive model is constructed according to the endogenous variables, which is represented as: wherein, represents t a column vector composed of endogenous variables at time t, p represents the lag order, represents the lag coefficient matrix, denotes t a random error vector of time.

[0026] (2) Use different lag orders to fit the vector autoregressive model respectively, calculate the information criterion value at each fitting, and select the lag order corresponding to the minimum information criterion value.

[0027] For example, the key variables are the power of production equipment, the power of pollution control equipment, and the concentration of VOCs. Assuming that the selected lag order that minimizes the information criterion value is 2, then the vector autoregressive model is: wherein, , , denote t the power of production equipment, the power of pollution control equipment, and the concentration of VOCs at time t, and the information criterion value can be AIC or BIC, denoted as: . Wherein, , m denotes the number of endogenous variables, ( n = total sample number- p ) denotes the effective sample number, RSS denotes the sum of the prediction residual squares of the vector autoregressive model for all endogenous variables.

[0028] When p is small (such as p =1), the model only uses the information of the first time to predict the current value, and may not be able to capture the dynamic change rule of the data, and the fitting effect is poor, RSS is large, the information criterion value is high; as p increases, the model incorporates more historical information and can better capture the lag correlation between variables, and the fitting effect improves, RSS will gradually decrease; when p is too large, the model may overfit the training data, and the reduction amplitude of RSS will slow down and tend to be stable, but the information criterion value will also increase.

[0029] S3, processing the historical power consumption data and the historical pollutant emission data of the working condition to obtain a training set, and training the vector autoregressive model using the training set.

[0030] Specifically, the data corresponding to the endogenous variables in the historical power consumption data and the historical pollutant emission data of the working condition are selected, and the selected data are cut and spliced according to the lag order to obtain training sample pairs that meet the input and output requirements of the vector autoregressive model. All training sample pairs form a training set.

[0031] Exemplarily, according to the endogenous variable and the lag order determined in step S2, data belonging to the endogenous variable is screened out from the working condition electricity consumption historical data and the pollutant emission historical data to form new time series data, and then a training set is constructed; taking the production equipment power, the pollution control equipment power and the VOCs concentration as the key variables and the lag order as 2 for example, the training set is constructed through sample rolling, specifically: When t = 3, the input is two column vectors composed of the production equipment power, the pollution control equipment power and the VOCs concentration at the time points of t = 1 and t = 2, and the output is a column vector composed of the production equipment power, the pollution control equipment power and the VOCs concentration at the time point of t = 3, that is, sample pair; until , the sample pair is , and thus a training set including -3 training sample pairs is obtained, and in the training process, the least square method is adopted to estimate the parameters of the model, the lag coefficient matrix is determined through iterative calculation to make the residual sum of squares minimum, so that the vector autoregressive model can predict the values of the key variables in the to-be-predicted period according to the historical data of the lag order.

[0032] In some embodiments, the method further comprises: model testing and evaluation on the trained vector autoregressive model, specifically: white noise and stability testing on the trained vector autoregressive model; calculating evaluation indexes of the model based on the predicted data and the measured data of the key variables, the evaluation indexes at least including root mean square error and mean absolute error; adjusting the parameters of the abnormal pollutant emission model or retraining the model according to the testing and evaluation results.

[0033] S4, acquiring input data according to the to-be-predicted time point and the lag order, inputting the input data into the trained vector autoregressive model, outputting the predicted value of the key variable at the to-be-predicted time point, and acquiring the corresponding measured value.

[0034] Specifically, taking the auto repair enterprise in S1 as an example, assuming that it is required to identify whether the pollutant emission of the enterprise from January 1, 2025 to January 15, 2025 is abnormal, then the to-be-predicted time point is each hour within the 15 days, and when the lag order is 2, the input data is the measured value of the key variable at the previous two hours of the to-be-predicted time point, the input data is input into the trained vector autoregressive model, the predicted value of the key variable at the to-be-predicted time point is output, and the measured value of the key variable at the to-be-predicted time point is acquired, thereby providing a basis for subsequent identification of abnormality.

[0035] S5. Based on the predicted and measured data of the key variables, determine whether the sewage discharge is abnormal during the predicted period and find out the cause of the abnormality.

[0036] In this embodiment, step S5 specifically includes: (1) Calculate the residuals between the predicted and measured values ​​of each key variable and form a joint residual vector; For example, the joint residual vector composed of the residuals between the predicted and actual values ​​of key variables—production equipment power, pollution control equipment power, and VOCs concentration—is represented as follows: , , , They represent t Real-time measured values ​​of production equipment power, pollution control equipment power, and VOCs concentration. , , They represent t Predicted values ​​of real-time production equipment power, pollution control equipment power, and VOCs concentration. These represent the residuals of the power of the production equipment, the power of the pollution control equipment, and the VOCs concentration at time t, respectively.

[0037] (2) Calculate the Mahalanobis distance based on the joint residual vector and the residual covariance matrix during the training of the vector autoregression model. If the Mahalanobis distance exceeds a preset threshold, it is determined to be an abnormal discharge. Specifically, the Mahalanobis distance is expressed as: ,in, This represents the residual covariance matrix. The preset threshold can be set and adjusted according to the actual situation.

[0038] (3) If it is determined to be abnormal discharge, perform a dimension-by-dimensional analysis on the joint residual vector, and identify the cause of the abnormality by combining the coupling relationship between the key variables.

[0039] Specifically, when determining by Mahalanobis distance t After an abnormal discharge, the joint residual vector is further analyzed dimension by dimension to identify the cause of the anomaly, which includes the following situations: (1) If the residual power of the production equipment >0 or ≈0, residual power of pollution control equipment <0, and the residual of VOCs concentration A value >0 indicates that production has increased, but pollution control efficiency has not been improved simultaneously, resulting in abnormal emissions and increased VOCs concentration. (2) If the residual power of the production equipment ≈0 or <0, residual of pollution control equipment power ≈0, residual of VOCs concentration >0, indicating that both production and pollution control are normal, and VOCs concentration may be elevated due to sensor abnormalities or non-working condition emissions.

[0040] Compared with the prior art, the method for identifying abnormal pollution based on multi-source data provided by the embodiment can construct a vector autoregressive model by taking key variables in power consumption data and pollutant emission data as endogenous variables, enable the model to reveal dynamic coupling relationships between the endogenous variables after training, generate multi-dimensional joint residuals in the prediction stage, and achieve accurate identification of abnormal pollution through statistical discrimination and dimension-by-dimension analysis. The method can comprehensively consider the internal relationship between production, control and emission, avoid false positives and false negatives, and significantly improve identification accuracy. At the same time, when an abnormality is determined, the residual characteristics can be combined to locate the abnormal source, and good traceability is achieved.

[0041] Embodiment 2 The embodiment provides an apparatus for identifying abnormal pollution based on multi-source data, as shown in Figure 2 , which comprises: a data collection module configured to collect working condition power consumption historical data and pollutant emission historical data; a model construction module configured to extract at least two key variables from the working condition power consumption historical data and the pollutant emission historical data, take the key variables as endogenous variables, construct a vector autoregressive model according to the endogenous variables, and determine a lag order of the vector autoregressive model using an information criterion method; a model training module configured to process the working condition power consumption historical data and the pollutant emission historical data to obtain a training set, and train the vector autoregressive model using the training set; a prediction module configured to obtain input data according to a to-be-predicted time and the lag order, input the input data into the trained vector autoregressive model, output predicted values of the key variables at the to-be-predicted time, and obtain corresponding measured values; an abnormality judgment module configured to judge whether pollution is abnormal according to the predicted values and the measured values of the key variables, and analyze abnormal reasons.

[0042] Embodiment 3 The embodiment provides an electronic device, as shown in Figure 3 , which comprises a memory and a processor, and the memory stores a computer program, and the computer program is executed by the processor to implement the method for identifying abnormal pollution based on multi-source data according to any one of the above embodiments.

[0043] Embodiment 4 The embodiment provides a computer readable storage medium, which stores a computer program, and the program is executed by a processor to realize the method for identifying abnormal pollution based on multi-source data according to any one of the above embodiments.

[0044] Computer readable storage media includes permanent and non-permanent, removable and non-removable media, which can be realized by any method or technology to store information. The information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassette, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transmission medium that can be used to store information accessible by a computing device. According to the definition herein, computer readable media does not include transitory computer readable media, such as modulated data signals and carriers.

[0045] The skilled person should further appreciate that the units and algorithm steps of each example described in connection with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. In order to clearly illustrate the interchangeability of hardware and software, the components and steps of each example have been described in general terms in the above description. Whether the functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. The skilled person can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0046] The steps of the methods or algorithms described in connection with the embodiments disclosed herein can be implemented in hardware, software modules executed by a processor, or a combination of both. The software modules can be placed in random access memory (RAM), memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disks, removable disks, CD-ROMs, or any other form of storage medium known in the art.

[0047] The above specific embodiments further illustrate the purpose, technical solutions and beneficial effects of the present application. It should be understood that the above description is only a specific embodiment of the present application and is not used to limit the protection scope of the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the present application should be included in the protection scope of the present application.

Claims

1. A method for identifying abnormal pollution based on multi-source data, characterized in that, The method comprises the following steps: collecting historical data of working condition electricity consumption and historical data of pollutant emission; extracting at least two key variables from the historical data of working condition electricity consumption and the historical data of pollutant emission, taking the key variables as endogenous variables, constructing a vector autoregressive model according to the endogenous variables, and determining the lag order of the vector autoregressive model by using an information criterion method; processing the historical data of working condition electricity consumption and the historical data of pollutant emission to obtain a training set, and training the vector autoregressive model by using the training set; obtaining input data according to a to-be-predicted time and the lag order, inputting the input data into the trained vector autoregressive model, outputting a predicted value of the key variable at the to-be-predicted time, and obtaining a corresponding measured value; judging whether the pollutant emission is abnormal according to the predicted value and the measured value of the key variable, and analyzing the abnormal reason.

2. The method for identifying abnormal pollution discharge based on multi-source data according to claim 1, characterized in that, The historical data of working condition electricity consumption at least comprises power data of production equipment and pollution control equipment, the historical data of pollutant emission at least comprises VOCs concentration data, and the historical data of working condition electricity consumption and the historical data of pollutant emission need to be processed by time alignment and abnormal value cleaning.

3. The method for identifying abnormal pollution discharge based on multi-source data according to claim 1, characterized in that, The vector autoregressive model is constructed according to the endogenous variables, and is represented as: wherein, denotes t a column vector of endogenous variables at time t, p denotes the lag order, denotes the lag coefficient matrix, denotes t a random error vector at time t.

4. The method for identifying abnormal pollution discharge based on multi-source data according to claim 1, characterized in that, The lag order of the vector autoregressive model is determined by using the information criterion method, which comprises the following steps: fitting the vector autoregressive model by using different lag orders respectively, calculating the information criterion value at each fitting time, and screening the lag order corresponding to the minimum information criterion value.

5. The method for identifying abnormal pollution discharge based on multi-source data according to claim 1, characterized in that, The historical data of working condition electricity consumption and the historical data of pollutant emission are processed to obtain a training set, which comprises the following steps: screening data corresponding to the endogenous variables in the historical data of working condition electricity consumption and the historical data of pollutant emission, cutting and splicing the screened data according to the lag order to obtain training sample pairs meeting the input and output requirements of the vector autoregressive model, and all the training sample pairs form a training set.

6. The method for identifying abnormal pollution discharge based on multi-source data according to claim 1, characterized in that, The predicted value and the measured value of the key variable are used to judge whether the pollutant emission is abnormal, and the abnormal reason is analyzed, which comprises the following steps: calculating the residual error between the predicted value and the measured value of each key variable, and forming a joint residual error vector; calculating the Mahalanobis distance according to the joint residual error vector and the residual error covariance matrix during the training of the vector autoregressive model, and determining that the pollutant emission is abnormal if the Mahalanobis distance exceeds a preset threshold; if it is determined that the pollutant emission is abnormal, performing dimension-by-dimension analysis on the joint residual error vector, and identifying the abnormal reason in combination with the coupling relationship between the key variables.

7. The method for identifying abnormal pollution discharge based on multi-source data according to any one of claims 1-6, characterized in that, The method further comprises the following steps of model testing and evaluation of the trained vector autoregressive model: performing white noise and stability testing on the trained vector autoregressive model; calculating evaluation indexes of the model based on the predicted value and the measured value of the key variable, wherein the evaluation indexes at least comprise a root mean square error and a mean absolute error; adjusting parameters of the abnormal pollutant emission model or retraining the model according to the testing and evaluation results.

8. An apparatus for identifying abnormal pollution based on multi-source data, characterized in that, The device comprises: a data collection module configured to collect historical data of working condition electricity consumption and historical data of pollutant emission; A model construction module is configured to extract at least two key variables from the working condition power consumption history data and the pollutant emission history data, use the key variables as endogenous variables, construct a vector autoregressive model according to the endogenous variables, and determine a lag order of the vector autoregressive model by using an information criterion method; A model training module is configured to process the working condition power consumption history data and the pollutant emission history data to obtain a training set, and train the vector autoregressive model by using the training set; A prediction module is configured to obtain input data according to a to-be-predicted time and the lag order, input the input data into the trained vector autoregressive model, output a predicted value of the key variable at the to-be-predicted time, and obtain a corresponding measured value; An abnormality judgment module is configured to judge whether the pollutant emission is abnormal according to the predicted value and the measured value of the key variable, and analyze an abnormality cause.

9. An electronic device, comprising: A computer program is stored in a memory and executed by a processor, and the computer program implements the method for identifying abnormal pollutant emission based on multi-source data according to any one of claims 1-7.

10. A storage medium, characterized by A computer program is stored in a memory and executed by a processor, and the computer program implements the method for identifying abnormal pollutant emission based on multi-source data according to any one of claims 1-7.

Images