Cement and coking enterprise atmospheric pollutant emission monitoring method and device based on electric power data

Abstract

The invention discloses a cement and coking enterprise atmospheric pollutant emission monitoring method and device based on electric power data. The cement and coking enterprise atmospheric pollutant emission monitoring method based on electric power data comprises the following steps: obtaining the production load of an enterprise; obtaining annual productivity data of different processes of the enterprise; obtaining the daily total output of each process of the enterprise according to the production load of the enterprise and the annual productivity data of different processes of the enterprise; and obtaining one or more of the daily atmospheric pollutant discharge amount of each process of the enterprise, the annual atmospheric pollutant discharge amount of each process of the enterprise and the annual atmospheric pollutant discharge amount of each process of the enterprise. According to the method, the electric power big data is used as a driving factor of the atmospheric pollutant emission amount, so that long-term continuous online monitoring is not needed, additional purchasing of monitoring equipment is not needed, on the other hand, pollutants are obtained through the electric power big data, and due to the fact that the electric power data has the advantages of being high in real-time performance and fast in synchronization, the real-time performance of the pollutants is improved. Therefore, quicker response can be achieved in the aspect of timeliness.

Description

Technical Field

[0001] The present application relates to the technical field of calculating the emission of atmospheric pollutants from enterprises, and in particular to a method for monitoring the emission of atmospheric pollutants from cement and coking enterprises based on electric power data, and a device for monitoring the emission of atmospheric pollutants from cement and coking enterprises based on electric power data. Background Art

[0002] By monitoring the real-time emissions of pollution sources, we can comprehensively monitor the amount of pollutant emissions in the study area, which plays an irreplaceable role in controlling the total amount of pollutants, combating environmental violations, and serving the security of major events.

[0003] There are mainly the following methods for monitoring pollution sources: sampling monitoring method, which determines the concentration of pollutants by directly collecting samples in the environment and conducting subsequent laboratory analysis; remote sensing monitoring method, which uses remote sensing technology to obtain the characteristics of the earth's surface and then infer the distribution of pollutants; online monitoring method, which sets up online monitoring equipment at or near the pollution source, collects environmental parameters and pollutant concentration data in real time, and conducts real-time analysis and alarm; model-based monitoring method, which uses mathematical models to simulate and predict the diffusion and transformation process of pollutants in the environment.

[0004] The above methods have their own advantages and disadvantages. The sampling monitoring method and remote sensing monitoring method are costly and have low sampling frequency, with high data value and low data density. The online monitoring method is costly and is currently an important means of ecological monitoring. It has high sampling frequency and high data value and is often used in ecological law enforcement applications. However, due to excessive environmental and cost pressures, only a small part of the pollution source enterprises in the huge industrial system have installed online monitoring systems. Other enterprises still rely on traditional manual law enforcement and supervisory law enforcement, and have failed to fully cover the pollution sources of industrial enterprises. Summary of the invention

[0005] The purpose of the present invention is to provide a method for monitoring atmospheric pollutant emissions from cement and coking enterprises based on power data to solve at least one of the above-mentioned technical problems.

[0006] One aspect of the present invention provides a method for monitoring atmospheric pollutant emissions from cement and coking enterprises based on power data, wherein the method for calculating the atmospheric pollutant emissions from the enterprise comprises:

[0007] The cement and coking enterprise atmospheric pollutant emission monitoring method based on power data includes:

[0008] Obtain the production load of the enterprise;

[0009] Obtain annual production capacity data for different processes of the enterprise;

[0010] Obtain the total daily output of each process of the enterprise based on the production load of the enterprise and the annual production capacity data of different processes of the enterprise;

[0011] According to the daily total output of each process of the enterprise, obtain one or more of the daily atmospheric pollutant emissions of each process of the enterprise, the annual atmospheric pollutant emissions of each process of the enterprise, and the annual atmospheric pollutant emissions of each process of the enterprise.

[0012] Optionally, obtaining the production load of the enterprise includes:

[0013] Obtain the enterprise's daily electricity consumption information; obtain the trained load forecasting model;

[0014] The enterprise's daily electricity consumption information and electricity consumption status are input into the trained load forecasting model to obtain the enterprise's production load.

[0015] Optionally, before obtaining the trained load forecasting model, obtaining the production load of the enterprise further includes:

[0016] The load forecasting model is trained, wherein the training of the load forecasting model comprises:

[0017] Acquire a power consumption database, wherein the power consumption database includes daily power consumption data of multiple enterprises;

[0018] Performing cluster analysis on the daily electricity consumption data of each enterprise, thereby obtaining each electricity consumption clustering interval of each enterprise and the enterprise electricity consumption data within each electricity consumption clustering interval, wherein one electricity consumption clustering interval corresponds to one electricity consumption state;

[0019] The daily electricity consumption data, production capacity data, and output data of each enterprise after clustering are divided to form a training set and a test set;

[0020] Obtain load forecasting model;

[0021] The load forecasting model is trained by the training set and the trained load forecasting model is tested by the test set.

[0022] Optionally, the total daily output of each process of the enterprise is obtained by the following formula:

[0023] in,

[0024] i, j, and k represent the enterprise, process, and date, respectively;

[0025] C represents the output accounting result, in tons;

[0026] G represents the annual production capacity data of different processes of the enterprise, in tons;

[0027] F represents production load (%).

[0028] Optionally, the annual total output of each process of the enterprise is obtained according to the daily total output of each process of the enterprise using the following formula:

[0029] C i,j =∑ k C i,j,k ;in,

[0030] i, j, and k represent the enterprise, process, and date, respectively;

[0031] C represents the output accounting result (t).

[0032] Optionally, the daily atmospheric pollutant emissions of each process of the enterprise are obtained according to the daily total output of each process of the enterprise by the following formula:

[0033] E i,j,k =∑(C i,j,k ×EF×(1-η)); where

[0034] E is the daily atmospheric pollutant emission of each process of the enterprise, EF is the emission factor, C is the output accounting result, η is the pollutant control and emission reduction efficiency, i, j, k represent the enterprise, process and date respectively.

[0035] Optionally, obtaining a power consumption database includes:

[0036] Obtain industry power data;

[0037] Calculate the missing rate data set of the included enterprises and calculate the industry missing boundary L, find the missing rate data of all enterprises in the industry, and sort them from small to large to obtain the data set C miss ={C 1 , C 2 , …, C n}, where n is the number of enterprises, find the quartile Q 1 =C (n+1) / 4 , Q 2 =C (n+1) / 2 , Q 3 =C 3(n+1) / 4 , Q 4 =C n , and calculate the upper limit of the missing rate L = Q 3 +1.5(Q 3 -Q 1 ), divide each enterprise data into complete enterprise data, enterprise data to be cleaned and discarded enterprise data according to the data missing rate Q;

[0038] Eliminate outliers in complete enterprise data and enterprise data to be supplemented, and limit the data range using the Laida criterion Will Assign a null value and remove it; at the same time, combine the moving window W={x n-2 , x n-1 , x n , x n+1 , x n+2} to identify the magnitude of data jumps, such as x n >>x n-2 , x n-1 , x n , x n+1 , x n+2 , assign null values ​​to outliers and remove x n Outliers < 0, in the above data classification process, if the data of a certain enterprise is complete and the outliers are eliminated, they will be included in the enterprise data set to be supplemented;

[0039] To supplement the enterprise data to be supplemented, input the enterprise data to be supplemented in sequence, calculate the maximum consecutive missing days (MMD) index, and use different methods to fill in according to the maximum consecutive missing days (MMD) as follows:

[0040] When MMD = 1, considering that the power consumption will not change suddenly in a short time, the adjacent value supplement method is used to fill x n =x n -1;

[0041] When 2≤MMD≤7, the moving average method is used to supplement the missing data. Specifically, at the place where the missing data occurs, the maximum number of consecutive missing days i∈(2,7) is calculated and filled

[0042] When 8≤MMD≤30, the historical similarity value method is used to supplement the data, extract the data of the past year, and use the similar day data in the historical data to supplement the data. The data is constructed into a data set S = {A, B, C, D, E}, T = {T 1 ,T 2 ,T 3 ,...,T n}', T∈A,B,C,D,E, column A in the data set represents the power consumption data, and column B is obtained based on whether the power consumption is missing. Column B is the missing label (0 represents missing, 1 represents missing). Based on the assumption that the power consumption in the corresponding time period of the missing power data has a relatively similar periodic fluctuation, the day label is calibrated according to the holiday, date and week attributes. Column C is the holiday label (0 represents a holiday, 1 represents a working day), column D is the date label (1:31), and column E is the week data (0:6). The data set is divided into missing data set S' and complete data set S' according to the different values ​​of column B. The single missing data f in S' is m ,b m ,c m ,d m ,e m}, and calculate similar days with the complete data set S' Right i Arrange from large to small to get d j ', the number of similar days selected has a significant impact on data completion. A single similar day is accidental, and too many similar days will increase data uncertainty. According to the 95% confidence interval test method, the number of non-similar days is set to 95% of the total number of days, and the similar day distance threshold d limit =d m ', where m is j*95% and then rounded, and d is selected dynamically i <d limit N similar days, fill in a j is d j For the corresponding daily electricity consumption, data f is moved to S' and repeated until S' is empty. The complete data set S' is used to repair the enterprise data as needed. If the electricity consumption data for more than one month is missing, the enterprise data is considered unreliable and the corresponding enterprise data is discarded.

[0043] The present application also provides a cement and coking enterprise atmospheric pollutant emission monitoring device based on power data, and the atmospheric pollutant emission calculation device for cement and coking enterprises includes:

[0044] A production load acquisition module, wherein the production load acquisition module is used to acquire the production load of the enterprise;

[0045] An enterprise's different process annual capacity data acquisition module, the enterprise's different process annual capacity data acquisition module is used to acquire the enterprise's different process annual capacity data;

[0046] A daily total output acquisition module for each process of the enterprise, wherein the daily total output acquisition module for each process of the enterprise is used to acquire the daily total output of each process of the enterprise according to the production load of the enterprise and the annual production capacity data of different processes of the enterprise;

[0047] An enterprise's atmospheric pollutant emission acquisition module is used to obtain one or more of the daily atmospheric pollutant emissions of each process of the enterprise, the annual atmospheric pollutant emissions of each process of the enterprise, and the annual atmospheric pollutant emissions of each process of the enterprise based on the total daily output of each process of the enterprise.

[0048] Beneficial Effects

[0049] The method for calculating enterprise atmospheric pollutant emissions in the present application uses electric power big data as the driving factor for atmospheric pollutant emissions. First of all, the method of the present application does not require long-term continuous online monitoring and does not require the purchase of additional monitoring equipment. On the other hand, the present application obtains pollutants through electric power big data. Since electric power data has the characteristics of strong real-time and fast synchronization, it can achieve faster response in terms of timeliness, solve the problem of low time and spatial resolution of traditional pollutant emission acquisition, and has the advantages of fast calculation and intelligent identification, etc., and realizes dynamic monitoring of atmospheric pollutant emissions of typical polluting enterprises at different time scales of year / month / day based on electric power big data. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 It is a flow chart of a method for monitoring atmospheric pollutant emissions from cement and coking enterprises based on electricity data according to an embodiment of the present application.

[0051] Figure 2 It is a schematic diagram of an electronic device for implementing a method for monitoring atmospheric pollutant emissions from cement and coking enterprises based on power data according to an embodiment of the present application.

[0052] Figure 3 This is a schematic diagram of the tiered clustering of a cement enterprise.

[0053] Figure 4 This is a schematic diagram of the grading and clustering of a coking enterprise.

[0054] Figure 5 This is a diagram showing the effect of testing the load forecasting model in the cement industry.

[0055] Figure 6 This is the effect diagram of the load forecasting model tested in the coking industry.

[0056] Figure 7 This is an example diagram of the data supplement effect.

[0057] Figure 8 It is a schematic diagram comparing the atmospheric pollutant emissions obtained by Enterprise A through the method of this application and the atmospheric pollutant emissions obtained by other means.

[0058] Fig. 9It is a schematic diagram comparing the atmospheric pollutant emissions obtained by Enterprise B through the method of this application and the atmospheric pollutant emissions obtained by other methods.

[0059] Fig.10 It is a schematic diagram comparing the atmospheric pollutant emissions obtained by Enterprise C through the method of this application and the atmospheric pollutant emissions obtained by other methods.

[0060] Fig.11 This is an error analysis chart of the results of atmospheric pollutant emissions obtained by enterprises A, B, and C using the method of this application. DETAILED DESCRIPTION

[0061] In order to make the purpose, technical scheme and advantages of the implementation of this application clearer, the technical scheme in the embodiment of this application will be described in more detail below in conjunction with the drawings in the embodiment of this application. In the drawings, the same or similar reference numerals throughout represent the same or similar elements or elements with the same or similar functions. The described embodiments are part of the embodiments of this application, not all of them. The embodiments described below with reference to the drawings are exemplary and are intended to be used to explain this application, and should not be construed as limitations on this application. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application. The embodiments of this application are described in detail below in conjunction with the drawings.

[0062] like Figure 1 The air pollutant emission monitoring methods for cement and coking enterprises based on electricity data include:

[0063] Step 1: Obtain the production load of the enterprise;

[0064] Step 2: Obtain the annual production capacity data of different processes of the enterprise;

[0065] Step 3: Obtain the total daily output of each process of the enterprise based on the production load of the enterprise and the annual production capacity data of different processes of the enterprise;

[0066] Step 4: Based on the total daily output of each process of the enterprise, obtain one or more of the daily air pollutant emissions of each process of the enterprise, the annual air pollutant emissions of each process of the enterprise, and the annual air pollutant emissions of each process of the enterprise.

[0067] The method for calculating enterprise atmospheric pollutant emissions in the present application uses electric power big data as the driving factor for atmospheric pollutant emissions. First of all, the method of the present application does not require long-term continuous online monitoring and does not require the purchase of additional monitoring equipment. On the other hand, the present application obtains pollutants through electric power big data. Since electric power data has the characteristics of strong real-time and fast synchronization, it can achieve faster response in terms of timeliness, solve the problem of low time and spatial resolution of traditional pollutant emission acquisition, and has the advantages of fast calculation and intelligent identification, etc., and realizes dynamic monitoring of atmospheric pollutant emissions of typical polluting enterprises at different time scales of year / month / day based on electric power big data.

[0068] In this embodiment, obtaining the production load of the enterprise includes:

[0069] Obtain the enterprise's daily electricity consumption information; obtain the trained load forecasting model;

[0070] The enterprise's daily electricity consumption information and electricity consumption status are input into the trained load forecasting model to obtain the enterprise's production load.

[0071] Before obtaining the trained load forecasting model, obtaining the production load of the enterprise further includes:

[0072] The load forecasting model is trained, wherein the training of the load forecasting model comprises:

[0073] Obtain an electricity consumption database, wherein the electricity consumption database includes daily electricity consumption data, production capacity data, and output data of multiple enterprises; it is understandable that other data may also be included, such as total industrial output data and energy consumption data, wherein if the total industrial output value is included, then the "total industrial output value" can be used as a constant parameter of the model, and if energy consumption is included, then the "energy consumption" can be used as a parameter, thereby improving the accuracy of the model.

[0074] Performing cluster analysis on the daily electricity consumption data of each enterprise, thereby obtaining each electricity consumption clustering interval of each enterprise and the enterprise electricity consumption data within each electricity consumption clustering interval, wherein one electricity consumption clustering interval corresponds to one electricity consumption state;

[0075] The daily electricity consumption data, production capacity data, and output data of each enterprise after clustering are divided to form a training set and a test set;

[0076] Obtain load forecasting model;

[0077] The load forecasting model is trained by the training set and the trained load forecasting model is tested by the test set.

[0078] In this embodiment, obtaining the power consumption database includes:

[0079] Obtain industry power data;

[0080] Calculate the missing rate data set of the included enterprises and calculate the industry missing boundary L, find the missing rate data of all enterprises in the industry, and sort them from small to large to obtain the data set C miss ={C 1 , C 2 , …, C n}, where n is the number of enterprises, find the quartile Q 1 =C (n+1) / 4 , Q 2 =C (n+1) / 2 , Q 3 =C 3(n+1) / 4 , Q 4 =C n , and calculate the upper limit of the missing rate L = Q 3 +1.5(Q 3 -Q 1 ), divide each enterprise data into complete enterprise data, enterprise data to be cleaned and discarded enterprise data according to the data missing rate Q;

[0081] Eliminate outliers in complete enterprise data and enterprise data to be supplemented, and limit the data range using the Laida criterion Will Assign a null value and remove it; at the same time, combine the moving window W of length 5 = {x n-2 , x n-1 , x n , x n+1 , x n+2} to identify the magnitude of data jumps, such as x n >>x n-2 , x n-1 , x n , x n+1 , x n+2 , assign null values ​​to outliers and remove x n Outliers < 0, in the above data classification process, if the data of a certain enterprise is complete and the outliers are eliminated, they will be included in the enterprise data set to be supplemented;

[0082] To supplement the enterprise data to be supplemented, input the enterprise data to be supplemented in sequence, calculate the maximum consecutive missing days (MMD) index, and use different methods to fill in according to the maximum consecutive missing days (MMD) as follows:

[0083] When MMD = 1, considering that the power consumption will not change suddenly in a short time, the adjacent value supplement method is used to fill x n =x n -1;

[0084] When 2≤MMD≤7, the moving average method is used to supplement the missing data. Specifically, at the place where the missing data occurs, the maximum number of consecutive missing days i∈(2,7) is calculated and filled

[0085] When 8≤MMD≤30, the historical similarity value method is used to supplement the data, extract the data of the past year, and use the similar day data in the historical data to supplement the data. The data is constructed into a data set S = {A, B, C, D, E}, T = {T 1 ,T 2 ,T 3 ,...,T n}', T∈A,B,C,D,E, column A in the data set represents the power consumption data, and column B is obtained based on whether the power consumption is missing. Column B is the missing label (0 represents missing, 1 represents missing). Based on the assumption that the power consumption in the corresponding time period of the missing power data has a relatively similar periodic fluctuation, the day label is calibrated according to the holiday, date and week attributes. Column C is the holiday label (0 represents a holiday, 1 represents a working day), column D is the date label (1:31), and column E is the week data (0:6). The data set is divided into missing data set S' and complete data set S' according to the different values ​​of column B. The single missing data f in S' is m ,b m ,c m ,d m ,e m}, and calculate similar days with the complete data set S' Right i Arrange from large to small to get d j ', the number of similar days selected has a significant impact on data completion. A single similar day is accidental, and too many similar days will increase data uncertainty. According to the 95% confidence interval test method, the number of non-similar days is set to 95% of the total number of days, and the similar day distance threshold d limit =d m ', where m is j*95% and then rounded, and d is selected dynamically i <d limit N similar days, fill in a j is d j For the corresponding daily electricity consumption, data f is moved to S' and repeated until S' is empty. The complete data set S' is used to repair the enterprise data as needed. If the electricity consumption data for more than one month is missing, the enterprise data is considered unreliable and the corresponding enterprise data is discarded.

[0086] In this embodiment, the total daily output of each process of the enterprise is obtained by the following formula:

[0087] in,

[0088] i, j, and k represent the enterprise, process, and date, respectively;

[0089] C represents the output accounting result, in tons (t);

[0090] G represents the annual production capacity data of different processes of the enterprise, and the unit is tons (t);

[0091] F represents production load (%).

[0092] In this embodiment, the annual total output of each process of the enterprise is obtained according to the daily total output of each process of the enterprise using the following formula:

[0093] C i,j =∑ k C i,j,k ;in,

[0094] i, j, and k represent the enterprise, process, and date, respectively;

[0095] C represents the output accounting result (t).

[0096] In this embodiment, the daily atmospheric pollutant emissions of each process of the enterprise are obtained according to the daily total output of each process of the enterprise by the following formula:

[0097] E i,j,k =∑(C i,j,k ×EF×(1-η)); where

[0098] E is the daily emission of atmospheric pollutants for each process of the enterprise, EF is the emission factor. In the present embodiment, different types of pollutants have different emission factors. Therefore, E obtained by the emission factor corresponding to a certain type of pollutant is the daily emission of atmospheric pollutants for each process of the enterprise corresponding to the type of pollutant. C is the output accounting result, η is the pollutant control and emission reduction efficiency, and i, j, k represent the enterprise, process and date respectively.

[0099] The present application is further described in detail below by way of examples. It should be understood that the examples do not constitute any limitation to the present application.

[0100] The mode adopted in this application is:

[0101] First, obtain the company's production load, then obtain the output through the production load, and estimate the company's emissions of atmospheric pollutants through the output. In simple terms, it can be considered an "electricity-production-pollution" solution.

[0102] This model is adopted because electricity consumption is strongly correlated with pollutant emissions.

[0103] Specifically, this application determines the strong correlation between electricity consumption and pollutant emissions through the following method:

[0104] Electricity data runs through all aspects of industrial production, and electricity consumption information can reflect the production status of enterprises to a certain extent. However, the use of fossil energy in some industries will affect the linear relationship between electricity consumption, production load and pollution emissions to a certain extent. In order to further explore the feasibility of building an industry emission model with electricity consumption information, it is necessary to examine the correlation between industry emissions and electricity consumption. Taking into account the consistency of data sources and the characteristics of industry emissions, this study combines the annual electricity consumption data of the emergency emission reduction inventory of pilot cities in 2019 with the particulate matter emission data for fitting, and then conducts Pearson correlation analysis (as shown in Equation 1 below).

[0105]

[0106] In the formula and are the sample means of variables X and Y, respectively.

[0107] The correlation between particulate matter emissions and electricity consumption in the coking industry, brick and tile industry, cement industry, glass industry, steel industry, and ceramic industry is calculated to be 0.97, 0.65, 0.72, 0.69, 0.94, and 0.63, respectively. The correlation criteria are shown in Table 1. It can be seen that the correlation between electricity consumption and pollutant emissions in the six industries is strong correlation, correlation, strong correlation, correlation, strong correlation, and correlation, respectively. Therefore, it is feasible to use electricity consumption to build a pollutant emission model.

[0108] Table 1. Determination basis of correlation

[0109]

[0110] This study increases the sample size of correlation analysis, and conducts Pearson correlation analysis on the daily electricity consumption of key polluting industries (cement and coking) in pilot cities and the daily emissions calculated based on CEMS, to examine the feasibility of building an "electricity-pollution" model based on power big data and CEMS data. The results show (see Table 4) that there is a difference in the linear correlation between the two. The reason is that corporate pollutant emissions include not only organized CEMS emissions, but also unorganized emissions and organized emissions without CEMS installed. Directly building a relationship between electricity consumption data and CEMS data may miss some of the company's emissions, and it is currently impossible to obtain more accurate unorganized daily emissions for model construction.

[0111] We further conducted a Pearson correlation analysis on the monthly production data and monthly electricity consumption data of several companies. The results in Table 2 show that there is a correlation between the two and it is significantly stronger than the correlation between electricity consumption data and CEMS data.

[0112] Table 2 Correlation determination results

[0113]

[0114] Therefore, it is considered to build an "electricity-production-pollution" solution based on electricity data and production data to ensure the accuracy of data transmission and the comprehensiveness of the accounting process.

[0115] See also Figure 1 In this embodiment, the “electricity-production-pollution” scheme is as follows Figure 1 As shown in the figure. In the part of calculating production based on electricity, the production load is relatively stable in different power intervals. Based on cluster analysis, the electricity consumption data of different enterprises are processed in different grades to obtain the range of each electricity consumption interval; then the model is trained with the graded electricity data, total industrial output value, and enterprise production capacity as characteristic values ​​to obtain the load analysis model; finally, the output is calculated based on the production load. In the part of calculating pollution based on production, the daily pollutant emissions of the enterprise are obtained by multiplying the output data by the latest pollutant emission factor.

[0116] In this embodiment, cluster analysis adopts the following analysis method:

[0117] The actual production process of each enterprise can be roughly divided into several production states: such as stable production period, this period corresponds to high production load, high power consumption, and small fluctuations in load and power; controlled production period, load level decreases, load and power theoretically fluctuate greatly, especially for some industries that include process enterprises that can start and stop at any time; shutdown stage, low load and low power consumption, long-term shutdown enterprises have stable electricity consumption, and the load and power curves are stable during the shutdown period; in addition, there are some special states, such as the overload production state in the cement industry.

[0118] Due to the differences in processes and control measures among industries, enterprises, it is impossible to assign corresponding production status classification to each enterprise. The workload is large and the theoretical basis for status classification is not high. Therefore, considering the differences in electricity consumption among enterprises, the cluster analysis method is used to classify electricity consumption. That is, the python cluster analysis program K-means in the machine learning method is used to adaptively determine the number of electricity consumption clusters for each enterprise, and distinguish the electricity consumption status of different enterprises. According to the classification results, it is found that the electricity consumption data of cement production and coking industries can generally be divided into four levels ( Figure 3 and Figure 4 ). It is worth noting that in order to fully reflect the advantages of power big data, the study assumes that the power and load function curves between different tiers follow the changing characteristics of power consumption.

[0119] In this embodiment, the load forecasting model used in this application adopts the random forest algorithm. Specifically, it is assumed that the maximum power consumption of each enterprise in the normal production stage corresponds to 100% production load, so as to convert the upper and lower limits of each enterprise's power consumption into load. The power consumption and load are fitted in each grading interval to determine the type of function curve. There are no significant differences between processes between different enterprises in the cement industry and the coking industry. Therefore, according to the production capacity, output and total industrial output value of different enterprises in the "Emergency Emission Reduction List", a high-accuracy industry load forecasting model based on power big data can be developed using the random forest algorithm.

[0120] The specific process of load forecasting model construction is to train a reliable random forest model based on a set of characteristic values ​​(daily electricity consumption and enterprise production load), and use this model to realize the measurement of a series of production loads. In this algorithm, 70% of the data is randomly used as the training set and 30% as the validation set. The random forest algorithm in this study is based on the sklearn data package in the scikit-learn (version 1.0.1) and scipy (version 1.8.1) data packages. Due to the random forest version problem, the output results need to be parameter corrected. The data is input into the above random forest for continuous training, and the various parameters are adjusted to make the calculation results closer to the input results in the data.

[0121] In this embodiment, the total daily output of each process of the enterprise can be obtained according to the production load of the enterprise and the annual production capacity data of different processes of the enterprise by the following method:

[0122] Since different cement companies have different processes (only a single process of grinding or clinker, or a dual process of both), different processes of each company must be calculated separately. The annual production capacity of the company is evenly distributed to 365 days to obtain the theoretical value of daily production capacity; by multiplying the daily production capacity of each process with the production load of the corresponding date and the corresponding process, the daily output calculation value can be obtained.

[0123] The output of an enterprise's products in each process can be calculated by the following formula:

[0124]

[0125] C i,j =∑ k C i,j,k (3)

[0126] Where:

[0127] i, j, and k represent the enterprise, process, and date, respectively;

[0128] C represents the output accounting result (t);

[0129] G represents the annual production capacity data of different processes of enterprises in the Emission Reduction Inventory (t);

[0130] F represents production load (%).

[0131] In this embodiment, the atmospheric pollutant emissions of the enterprise are obtained according to the daily total output of each process of the enterprise as follows:

[0132] There is a clear correlation between the production load and electricity consumption of typical polluting industries, and based on the observation of electricity data, it can be seen that the load is relatively stable in different ranges. Therefore, the study uses the emission factor method to construct a pollutant emission accounting model for typical polluting industries; the production load data used in the accounting model is the output result of the model of the relationship between electricity consumption and production load of enterprises in various industries.

[0133] Specifically, the formula is as follows:

[0134] E=∑(C i ×EF×(1-η)) (4)

[0135] Wherein E is the atmospheric pollutant emission of an enterprise in a typical atmospheric pollution industry, EF is the emission factor. In this embodiment, different types of pollutants have different emission factors. Therefore, E obtained by the emission factor corresponding to a certain type of pollutant is the daily atmospheric pollutant emission of each process of the enterprise corresponding to the type of pollutant, C is the product output of an enterprise in a typical atmospheric pollution industry, and η is the pollutant control and emission reduction efficiency.

[0136] In this embodiment, cement enterprises can be divided into three types of enterprises according to the process: grinding process, clinker process, and dual process (grinding, clinker). The model needs to be constructed in combination with the proportion of different processes. Based on the literature, the dual process enterprise analyzes the electricity consumption by converting the electricity consumption of clinker: grinding = 0.684211; other industries are considered as a single production line. It is understandable that in actual use, the process can also be divided according to one's own needs.

[0137] In this embodiment, the installation status of pollutant control measures of each enterprise comes from enterprise surveys and other environmental public reports. The pollution control efficiency data η of each industry comes from the "Handbook for the Preparation of Urban Air Pollutant Emission Inventories", see Table 3 below. The emission factor EF is based on the 9 technical guidelines for the preparation of pollution source inventories issued by the Ministry of Ecology and Environment, based on the results of research projects such as the second pollution spectrum, national atmospheric special projects, and governance research, combined with current production and pollution control processes, and accumulated updates and expansions to obtain the latest industry emission factors. The emission factors of key industries such as cement, coking, and steel can be refined to the process. It is more comprehensive, more detailed, and more able to reflect the current production and emission of pollutants than the emission factors announced by the state. Tsinghua University took the lead in jointly conducting research with 7 universities and research institutions, and released it in the form of a group standard, see Table 4 below.

[0138] Table 3 Control efficiency of various pollutants in common pollution control measures in key polluting industries (unit: %)

[0139]

[0140] Table 4 Emission factors of typical polluting industries (g / t product)

[0141]

[0142] In order to verify the effect of the method of this application, the following verification is performed:

[0143] First, this application verifies the load forecasting model as follows:

[0144] Based on the established load forecasting model, daily electricity consumption data is extracted from the verification set data to calculate the enterprise load and compared with the corresponding production load value in the verification set. The verification indicators are mainly standard mean deviation NMB, absolute mean deviation NME, mean deviation MB, root mean square error RMSE, and determination coefficient R2. The prediction results of the coking industry and cement industry are as follows: Figure 5 , Figure 6 and as shown in Table 5. From the perspective of the determination coefficient, the random forest model has a good fitting effect; from the model verification results, the model can realize the enterprise production load forecast based on electricity consumption data.

[0145] Table 5 Reclassification data verification indicators for typical polluting industries

[0146]

[0147] In this embodiment, the above-mentioned verification set data and the training set data for training the load prediction model can be obtained by the following method:

[0148] The electricity data used in this application comes from the Big Data Center of the State Grid Corporation of China. The daily electricity data of the coking industry and the cement industry in 2019 were selected from the enterprise data set of the pilot cities for data quality assessment. Among them, there are 2,570 data from 11 enterprises in the coking industry, and 192,567 data from 564 enterprises in the cement industry. Since there are cases where one enterprise is equipped with multiple meters, the number of data items is not simply equal to the number of enterprises multiplied by the number of days in the year. The study further compares and analyzes the actual number of electricity data with the theoretical number of items to obtain the missing distribution of industry electricity data. At the same time, the data in the data of a single enterprise that is less than 0 and greater than 3 times the data mean is set as an outlier, and the data outliers are counted.

[0149] As shown in Table 6, there are missing data in all industries and the missing rate is above 30%, among which the missing rate of coking industry data is as high as 63.2%. On the contrary, although there are outliers in the data, the data anomaly rate is low, all below 3%. The direct use value of the data is low, and all data should be preprocessed to remove anomalies and supplement missing values. In view of the low data anomaly rate, the study optimizes the data by directly removing abnormal data after data anomaly detection. For missing values, the study uses different methods to supplement missing data according to the different missing states of different enterprises, and only optimizes data for one enterprise at a time.

[0150] Table 6 Industry power data overview

[0151]

[0152] This application pays special attention to the continuity characteristics of power data. The reduction in the amount of available coherent data will bring huge deviations to the next step of using algorithms and models that rely on data training and data-driven, and may even lead to conclusive misjudgments. The outliers of power big data are characterized by isolated and non-repeating characteristics; the missing situation of power big data is relatively complex, which can be divided into the following situations: random isolated missing of single-point data, continuous missing of multiple-point data, complete missing, etc. In order to meet the data quality requirements of the research, a power big data optimization method with high universality, easy sharing and high practicality is proposed, including three steps: industry data extraction, data classification and data cleaning. The use of this method to optimize power data can effectively deal with the differences in missing and abnormal situations of power data in different industries, reflect the overall situation of a specific industry, and improve the continuity and availability of power data in different industries.

[0153] First, extract the industry power data and divide it into different types according to the company name. Considering the large number of companies and complex processes in a specific industry, the company data can be divided into specific types according to the distribution of the missing rate of each company data. Calculate the missing rate data set of the included companies and calculate the industry missing boundary L, find the missing rate data of all companies in the industry, and sort them from small to large to obtain the data set C miss ={C 1 , C 2 , …, C n}, where n is the number of enterprises, find the quartile Q 1 =C (n+1) / 4 , Q 2 =C (n+1) / 2 , Q 3 =C 3(n+1) / 4 , Q 4 =C n , and calculate the upper limit of the missing rate L = Q 3 +1.5(Q 3 -Q 1), according to the data missing rate Q, the enterprise data are divided into complete enterprise data, enterprise data to be cleaned and discarded enterprise data. The classification basis is shown in Table 7.

[0154] Table 7 Enterprise data classification table (by data missing characteristics)

[0155]

[0156]

[0157] Then, outliers in the complete enterprise data and the enterprise data to be supplemented are eliminated. This method uses the Laida criterion to limit the data range Will Assign a null value and remove it; at the same time, combine the moving window W of length 5 = {x n-2 , x n-1 , x n , x n+1 , x n+2} to identify the magnitude of data jumps, such as x n >>x n-2 , x n-1 , x n , x n+1 , x n+2 (far greater than the hundredfold change of the previous day), assign null values ​​to the outliers and remove x n In the above data classification process, if the data of a certain enterprise is complete and the outliers are removed, it will be included in the enterprise data set to be supplemented.

[0158] Finally, the data to be supplemented are supplemented. By analyzing the actual data, it can be seen that the missing rate of data of each enterprise is inconsistent with the maximum consecutive missing days, and the universality of supplementing data using a single method is not high. In the traditional power loss processing, a small amount of missing data is supplemented by the adjacent value (previous moment) electricity. This method is simple and convenient but has low accuracy and low applicability for long-term continuous missing data. Therefore, certain improvements need to be made to meet the data requirements of this study. Enter the data of the enterprises to be supplemented in turn, calculate their maximum consecutive missing days (MMD) indicators, and use different methods to fill them.

[0159] When MMD = 1, considering that the power consumption will not change suddenly in a short time, the adjacent value supplement method is used to fill x n =x n -1;

[0160] When 2≤MMD≤7, the traditional time series prediction method moving average method is used to supplement the missing data. Specifically, at the place where the missing data occurs, the maximum number of consecutive missing days i∈(2,7) is calculated and filled

[0161] When 8≤MMD≤30, the historical similarity value method is used to supplement the data. Extract the data of the past year, use the similar day data in the historical data to supplement the data, and construct the data into a data set S = {A, B, C, D, E}, T = {T 1 ,T 2 ,T 3 ,...,T n}', T∈A,B,C,D,E, column A in the data set represents electricity consumption data, and column B is obtained based on whether the electricity consumption is missing. Column B is the missing label (0 represents missing, 1 represents missing). Based on the assumption that the electricity consumption in the corresponding time period of the missing electricity data has a relatively similar periodic fluctuation, the day label is calibrated according to the holiday, date and week attributes. Column C is the holiday label (0 represents a holiday, 1 represents a working day), column D is the date label (1:31), and column E is the week data (0:6). The data is divided into missing data set S' and complete data set S' according to the different values ​​of column B. The single missing data f in S' is m ,b m ,c m ,d m ,e m}, and calculate similar days with the complete data set S' Right i Arrange from large to small to get d j ', the number of similar days selected has a significant impact on data completion. A single similar day is accidental, and too many similar days will increase data uncertainty. Based on the 95% confidence interval test method, this method sets the non-similar days to account for 95% of the total number of days, and the similar day distance threshold d limit =d m ', where m is j*95% and rounded, dynamically select d i <d limit N similar days. Fill in a j is d j The corresponding daily electricity consumption, data f is moved to S", and repeated until S' is empty. The complete data set S" is repaired with enterprise data as required, and the enterprise data repair is completed. For specific supplementary examples, see Figure 7 If the electricity consumption data is missing for more than one month, the enterprise data is considered unreliable and the corresponding enterprise data is discarded.

[0162] The reliability of the supplementary results is verified by using data evaluation indicators. Among the evaluation indicators, similarity (R) represents the similarity of the two curves, and RMSE represents the average deviation of the two curves. A certain enterprise with complete data is selected, and a data set with 10% random missing data is artificially established, and the missing data is supplemented by the above method. The RMSE (0.49) and R (0.98) of the supplemented data and the original data both show good performance, indicating that the power data preprocessing method proposed in this study can better achieve the optimization of power data.

[0163] Through data optimization and cleaning, hourly power data of 27 cement enterprises and 4 coking enterprises in the pilot cities were obtained. Data on the process conditions of enterprises and the output of products by process were collected simultaneously. Enterprise CEMS data was obtained from the provincial ecological environment department, and pollutant emission inventory data was obtained from the MEIC platform of Tsinghua University.

[0164] In this embodiment, the above data may be allocated in a manner of 70% of the data being used as a training set and 30% being used as a validation set.

[0165] The method of this application is verified as a whole below, and the specific verification method and effect are as follows:

[0166] In the cement industry, clinker-cement joint dual-process enterprises allocate purchased electricity according to the ratio of clinker process / grinding process = 0.684211. The comparison of the 2019 atmospheric pollutant emissions of three typical clinker-cement joint enterprises calculated based on the "electricity-production-pollution" model and the MEIC inventory is shown in Figures 8 to 11 .contrast Figures 8 to 10 It can be seen that the absolute errors of enterprises A and C are both less than 1%, and the absolute error of enterprise B is the largest, but still within 20%. Fig.11 It can be seen that the model calculation error of cement dual-process enterprises mainly comes from the clinker process. The possible reason is that electricity consumption is concentrated in the grinding process, and the correlation between electricity and grinding emissions is high, while the clinker process requires the use of a certain amount of fossil energy, resulting in a lower correlation between electricity and emissions. Overall, the atmospheric pollutant emission accounting model of cement enterprises based on power big data can better calculate the atmospheric pollutant emission level of enterprises.

[0167] The errors between the pollutant emission model calculation results of the four enterprises in the coking industry and the MEIC inventory results are shown in Table 8. It can be seen that the absolute percentage error between the overall prediction results of the coking industry and the MEIC inventory is within 20%, and the pollutant emission prediction model based on power big data can better characterize the atmospheric pollutant emission characteristics of the coking industry.

[0168] The following is an explanation with a specific embodiment, which is as follows:

[0169] Take a dual-process cement enterprise in a pilot city (with an annual clinker production capacity of about 1.5×106 t, the annual grinding capacity is about 1.8×10 6 t) of the annual atmospheric pollutant emission characteristics in 2019. By inputting the 2019 outsourced electricity data of the enterprise provided by the State Grid Corporation of China Big Data Center into the model, the atmospheric pollutant emission accounting of the enterprise is obtained as follows: Fig.11 As shown, CO, SO 2 , NOx, TSP, PM 10 、PM 2.5 The annual emission accounting values ​​of BC and OC are 32312.13t, 2615.87t, 26.02t, 845.68t, 469.26t, 194.65t, 1.76t and 2.85t respectively, and the absolute error with the MEIC inventory results of the pilot cities is within 1%. Fig.11 From the data A in the figure, we can see that the monthly emissions of air pollutants of the company in 2019 were relatively low in the first and third quarters, and then decreased and increased after the second quarter, with the increase mainly concentrated in the fourth quarter, which is consistent with the trend of cement prices first falling and then rising in 2019 shown by the data of the China Cement Network Cement Big Data Platform (https: / / data.ccement.com / ). In addition, it is obvious from the figure that the monthly emissions of various air pollutants of the company have a consistent fluctuation trend, so it can be seen that the monthly emissions of various air pollutants of the company have a certain typical pollutant (such as PM 2.5 ) reflects the overall change pattern of atmospheric pollutant emissions from cement enterprises. The month with the highest pollutant emissions is November, and the month with the lowest is February. However, since the emission in April, the second lowest month, is very close to that in February and has two more production days than February, April can be selected as the month with the lowest emissions for the convenience of comparing the daily emission changes. Fig.11 From B in Figure 1, we can see that the company’s PM in November and April 2.5 The daily emissions vary greatly and fluctuate significantly, reflecting that the company's production status in April was mainly suspended or limited, while the low production status in November was mainly normal production or overload production, which was significantly affected by the fluctuation of cement prices. Fig.11 From the C in the figure, we can see that the pollutant emissions of this enterprise mainly come from the clinker process, because the energy consumption of clinker production is large and more fossil fuels are used. The results of this study show that electricity data can provide an important basis for the high-time resolution accounting of atmospheric pollutant emissions from cement enterprises. Therefore, by monitoring the electricity consumption of this enterprise, it can effectively assist the supervision of atmospheric pollutant emissions and the identification of pollution sources and pollution processes, and also provide an effective way and more value space for the application of power big data in the field of environmental supervision.

[0170] This application has the following advantages:

[0171] (1) The data preprocessing method based on data extraction, data classification, and data cleaning steps, as well as the data supplementation method based on the maximum consecutive missing days (MMD) indicator proposed in this application, can better optimize the continuity of power data. It has been verified that the optimization results of artificially established random missing data have a similarity R of 0.98 with the original data, and the root mean square error RMSE is only 0.49.

[0172] (2) Based on the electricity consumption data, production capacity and output value data of industry enterprises and process-level pollutant emission factors, the “electricity-production-pollution” emission accounting method was constructed to calculate the CO and SO emissions of enterprises in the cement and coking industries. 2 , NOx, TSP, PM 10 、PM 2.5 , BC, and OC specific emissions. Among them, the relative error between the emission accounting results of the three cement dual-process enterprises and the MEIC inventory results is between -19.30% and 0.55%, and the relative error between the model accounting values ​​of atmospheric pollutant emissions in different processes and the MEIC inventory results is between -19.31% and 0.55%; the relative error between the model accounting values ​​of atmospheric pollutant emissions of the four coking enterprises and the MEIC inventory results is between -18.56% and -16.63%. The model accounting error of the cement industry comes from the high proportion of fossil fuels used in the main pollution-discharging processes, and the model accounting error of the coking industry mainly comes from insufficient training set samples. It is recommended to expand the training sample size of the "electricity-production-pollution" model of each industry to gradually improve the accuracy of model prediction.

[0173] (3) The results of the atmospheric pollutant emission accounting of a cement dual-process enterprise in the pilot city in 2019 show that its monthly emissions were relatively low in the first and third quarters, and began to show a "U"-shaped distribution trend of first decreasing and then increasing in the second quarter. The increase was mainly concentrated in the fourth quarter, with April and November being the lowest and highest emission months respectively. The PM in April and November was 2.5 The daily emissions vary greatly and fluctuate significantly. The maximum daily emissions are 0.86t, corresponding to an overloaded production state, and the minimum daily emissions are 0.04t, corresponding to a stop / limited production state. The emission contribution of the clinker process accounts for 85% to 100%.

[0174] (4) Electricity big data runs through all aspects of production and life. It takes advantage of its strong real-time, high resolution and wide collection range to build an underlying methodology for electricity-assisted supervision, explore the coupling relationship between regional and enterprise pollution emissions and electricity consumption data, and achieve dynamic monitoring of pollutant emissions in regions, parks, industries and enterprises, and precise supervision of the implementation of pollution prevention and control measures.

[0175] The present application also provides a cement and coking enterprise atmospheric pollutant emission monitoring device based on power data, wherein the enterprise atmospheric pollutant emission calculation device comprises a production load acquisition module, an enterprise different process annual capacity data acquisition module, an enterprise each process daily total output acquisition module and an enterprise atmospheric pollutant emission acquisition module, wherein:

[0176] The production load acquisition module is used to obtain the production load of the enterprise;

[0177] The annual capacity data acquisition module for different processes of an enterprise is used to obtain the annual capacity data of different processes of an enterprise;

[0178] The daily total output acquisition module for each process of the enterprise is used to acquire the daily total output of each process of the enterprise according to the production load of the enterprise and the annual production capacity data of different processes of the enterprise;

[0179] An enterprise's atmospheric pollutant emission acquisition module is used to obtain one or more of the daily atmospheric pollutant emissions of each process of the enterprise, the annual atmospheric pollutant emissions of each process of the enterprise, and the annual atmospheric pollutant emissions of each process of the enterprise based on the total daily output of each process of the enterprise.

[0180] It should be noted that the above explanations of the method embodiment are also applicable to the system of this embodiment and will not be repeated here.

[0181] The present application also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and capable of running on the processor. When the processor executes the computer program, it implements the above-mentioned cement and coking enterprise atmospheric pollutant emission monitoring method based on power data.

[0182] The present application also provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it can implement the above-mentioned cement and coking enterprise atmospheric pollutant emission monitoring method based on power data.

[0183] Figure 2 It is an exemplary structural diagram of an electronic device capable of implementing a method for monitoring atmospheric pollutant emissions from cement and coking enterprises based on power data provided in accordance with an embodiment of the present application.

[0184] like Figure 2As shown, the electronic device includes an input device 501, an input interface 502, a central processor 503, a memory 504, an output interface 505, and an output device 506. The input interface 502, the central processor 503, the memory 504, and the output interface 505 are interconnected through a bus 507, and the input device 501 and the output device 506 are connected to the bus 507 through the input interface 502 and the output interface 505, respectively, and then connected to other components of the electronic device. Specifically, the input device 501 receives input information from the outside, and transmits the input information to the central processor 503 through the input interface 502; the central processor 503 processes the input information based on the computer executable instructions stored in the memory 504 to generate output information, temporarily or permanently stores the output information in the memory 504, and then transmits the output information to the output device 506 through the output interface 505; the output device 506 outputs the output information to the outside of the electronic device for use by the user.

[0185] That is to say, Figure 2 The electronic device shown can also be implemented as comprising: a memory storing computer executable instructions; and one or more processors, which can implement the combination when executing the computer executable instructions. Figure 1 The paper describes a method for monitoring atmospheric pollutant emissions from cement and coking enterprises based on electricity data.

[0186] In one embodiment, Figure 2 The electronic device shown can be implemented to include: a memory 504 configured to store executable program code; and one or more processors configured to run the executable program code stored in the memory 504 to execute the cement and coking enterprise atmospheric pollutant emission monitoring method based on power data in the above-mentioned embodiment.

[0187] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0188] The memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0189] Computer readable media include permanent and non-permanent, removable and non-removable, and the media can be implemented 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 technology, read-only compact disk read-only memory (CD-ROM), data versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device.

[0190] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present application. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and a module, a program segment or a part of a code includes one or more executable instructions for realizing the specified logical function. It should also be noted that in some implementations as replacements, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes identified in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or the total flow chart can be implemented with a dedicated hardware-based system that performs the specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0191] The processor referred to in this embodiment may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or the processor may be any conventional processor, etc.

[0192] The memory can be used to store computer programs and / or modules. The processor realizes various functions of the system / terminal device by running or executing the computer programs and / or modules stored in the memory and calling the data stored in the memory. The memory can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, an application required for at least one function (such as a sound playback function, an image playback function, etc.), etc.; the data storage area can store data created according to the use of the mobile phone (such as audio data, a phone book, etc.), etc. In addition, the memory can include a high-speed random access memory, and can also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0193] In this embodiment, if the module / unit integrated in the system / terminal device is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present invention implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may include: any entity or system that can carry computer program code, recording medium, U disk, mobile hard disk, disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electric carrier signal, telecommunication signal and software distribution medium. It should be noted that the content contained in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. Although the present application is disclosed as above in terms of a preferred embodiment, it is not intended to limit the present application. Any technical personnel in this field may make possible changes and modifications without departing from the spirit and scope of the present application. Therefore, the scope of protection of the present application shall be based on the scope defined by the claims of the present application.

[0194] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment or an embodiment in combination with software and hardware. Moreover, the present application may adopt 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.) that contain computer-usable program code.

[0195] In addition, it is obvious that the word "comprising" does not exclude other units or steps. The multiple units, modules or systems stated in the system claims can also be implemented by one unit or the overall system through software or hardware.

[0196] Although the present invention has been described in detail above with general descriptions and specific embodiments, it is obvious to those skilled in the art that some modifications or improvements may be made thereto based on the present invention. Therefore, these modifications or improvements made without departing from the spirit of the present invention all fall within the scope of protection claimed by the present invention.

Claims

1. A method for monitoring atmospheric pollutant emissions from cement and coking enterprises based on power data, characterized in that: The cement and coking enterprise atmospheric pollutant emission monitoring method based on power data includes: Obtain the production load of the enterprise; Obtain annual production capacity data for different processes of the enterprise; Obtain the total daily output of each process of the enterprise based on the production load of the enterprise and the annual production capacity data of different processes of the enterprise; According to the daily total output of each process of the enterprise, obtain one or more of the daily atmospheric pollutant emissions of each process of the enterprise, the annual atmospheric pollutant emissions of each process of the enterprise, and the annual atmospheric pollutant emissions of each process of the enterprise.

2. The method for monitoring atmospheric pollutant emissions from cement and coking enterprises based on power data according to claim 1, characterized in that: The obtaining of the production load of the enterprise includes: Obtain the daily electricity consumption information of the enterprise; Obtaining the trained load forecasting model; The enterprise's daily electricity consumption information and electricity consumption status are input into the trained load forecasting model to obtain the enterprise's production load.

3. The method for monitoring atmospheric pollutant emissions from cement and coking enterprises based on power data according to claim 2, characterized in that: Before obtaining the trained load forecasting model, obtaining the production load of the enterprise further includes: The load forecasting model is trained, wherein the training of the load forecasting model comprises: Acquire an electricity consumption database, wherein the electricity consumption database includes daily electricity consumption data, production capacity data, and output data of multiple enterprises; Performing cluster analysis on the daily electricity consumption data of each enterprise, thereby obtaining each electricity consumption clustering interval of each enterprise and the enterprise electricity consumption data within each electricity consumption clustering interval, wherein one electricity consumption clustering interval corresponds to one electricity consumption state; The daily electricity consumption data, production capacity data, and output data of each enterprise after clustering are divided to form a training set and a test set; Obtain load forecasting model; The load forecasting model is trained by the training set and the trained load forecasting model is tested by the test set.

4. The method for monitoring atmospheric pollutant emissions from cement and coking enterprises based on power data according to claim 3, characterized in that: The total daily output of each process of the enterprise is obtained by the following formula: in, i, j, and k represent the enterprise, process, and date, respectively; C represents the output accounting result, in tons; G represents the annual production capacity data of different processes of the enterprise, in tons; F represents production load (%).

5. The method for monitoring atmospheric pollutant emissions from cement and coking enterprises based on power data according to claim 4, characterized in that: The daily atmospheric pollutant emissions of each process of the enterprise are obtained according to the daily total output of each process of the enterprise through the following formula: E i,j,k =∑(C i,j,k ×EF×(1-η)); where E is the daily atmospheric pollutant emission of each process of the enterprise, EF is the emission factor, C is the output accounting result, η is the pollutant control and emission reduction efficiency, i, j, k represent the enterprise, process and date respectively.

6. The method for monitoring atmospheric pollutant emissions from cement and coking enterprises based on power data according to claim 3, characterized in that: The obtaining of the power consumption database comprises: Obtain industry power data; Calculate the missing rate data set of the included enterprises and calculate the industry missing boundary L, find the missing rate data of all enterprises in the industry, and sort them from small to large to obtain the data set C miss ={C1, C2, ..., C n }, where n is the number of enterprises, find the quartile Q1=C (n+1) / 4 、Q2=C (n+1) / 2 、Q3=C 3(n+1) / 4 、Q4=C n , and calculate the upper limit of the missing rate L = Q3 + 1.5 (Q3-Q1), and divide the enterprise data into complete enterprise data, enterprise data to be cleaned, and discarded enterprise data according to the data missing rate Q; Eliminate outliers in complete enterprise data and enterprise data to be supplemented, and limit the data range using the Laida criterion Will Assign a null value and remove it; at the same time, combine the moving window W of length 5 = {x n-2 , x n-1 , x n , x n+1 , x n+2 } to identify the magnitude of data jumps, such as x n >>x n-2 , x n-1 , x n , x n+1 , x n+2 , assign null values ​​to outliers and remove x n Outliers < 0, in the above data classification process, if the data of a certain enterprise is complete and the outliers are eliminated, they will be included in the enterprise data set to be supplemented; To supplement the enterprise data to be supplemented, input the enterprise data to be supplemented in sequence, calculate the maximum consecutive missing days (MMD) index, and use different methods to fill in according to the maximum consecutive missing days (MMD) as follows: When MMD = 1, considering that the power consumption will not change suddenly in a short time, the adjacent value supplement method is used to fill x n =x n-1 ; When 2≤MMD≤7, the moving average method is used to supplement the missing data. Specifically, at the place where the missing data occurs, the maximum number of consecutive missing days i∈(2,7) is calculated and filled When 8≤MMD≤30, the historical similarity value method is used to supplement the data, extract the data of the past year, and use the similar day data in the historical data to supplement the data. The data is constructed into a data set S={A,B,C,D,E}, T={T1,T2,T3,...,T n }', T∈A,B,C,D,E, column A in the data set represents the power consumption data, and column B is obtained based on whether the power consumption is missing. Column B is the missing label (0 represents missing, 1 represents missing). Based on the assumption that the power consumption in the corresponding time period of the missing power data has a relatively similar periodic fluctuation, the day label is calibrated according to the holiday, date and week attributes. Column C is the holiday label (0 represents a holiday, 1 represents a working day), column D is the date label (1:31), and column E is the week data (0:6). The data set is divided into missing data set S' and complete data set S' according to the different values ​​of column B. The single missing data f in S' is m ,b m ,c m ,d m ,e m }, and calculate similar days with the complete data set S' Right i Arrange from large to small to get d j ', the number of similar days selected has a significant impact on data completion. A single similar day is accidental, and too many similar days will increase data uncertainty. According to the 95% confidence interval test method, the number of non-similar days is set to 95% of the total number of days, and the similar day distance threshold d limit =d m ', where m is j*95% and then rounded, and d is selected dynamically i <d limit N similar days, fill in a j is d j For the corresponding daily electricity consumption, data f is moved to S' and repeated until S' is empty. The complete data set S' is used to repair the enterprise data as needed. If the electricity consumption data for more than one month is missing, the enterprise data is considered unreliable and the corresponding enterprise data is discarded.

7. A cement and coking enterprise atmospheric pollutant emission monitoring device based on power data, characterized in that: The cement and coking enterprise atmospheric pollutant emission monitoring device based on power data includes: A production load acquisition module, wherein the production load acquisition module is used to acquire the production load of the enterprise; An enterprise's different process annual capacity data acquisition module, the enterprise's different process annual capacity data acquisition module is used to acquire the enterprise's different process annual capacity data; A daily total output acquisition module for each process of the enterprise, wherein the daily total output acquisition module for each process of the enterprise is used to acquire the daily total output of each process of the enterprise according to the production load of the enterprise and the annual production capacity data of different processes of the enterprise; An enterprise's atmospheric pollutant emission acquisition module is used to obtain one or more of the daily atmospheric pollutant emissions of each process of the enterprise, the annual atmospheric pollutant emissions of each process of the enterprise, and the annual atmospheric pollutant emissions of each process of the enterprise based on the total daily output of each process of the enterprise.

Images