Method and device for classifying carbon emission data of thermal power plant and electronic equipment

Abstract

The invention discloses a method and device for classifying carbon emission data of a thermal power plant and electronic equipment, and the method comprises the steps: obtaining a plurality of weighted Euclidean distances, and screening out candidate Euclidean distances meeting a preset distance condition from the weighted Euclidean distances; performing trend weighting processing on the candidate Euclidean distance to generate a corresponding trend weighted Euclidean distance; performing matching processing on the candidate Euclidean distance and the trend weighted Euclidean distance, and determining a target Euclidean distance corresponding to the first time sequence and the second time sequence; and based on the target Euclidean distance, using a nearest neighbor classifier to divide the subsequence categories of the thermal power plant carbon emission data, realizing classification of the thermal power plant carbon emission data, and based on a classification result, analyzing and controlling the production data related to the thermal power plant carbon emission. The device comprises a processor and a memory. The electronic equipment comprises a processor, a memory, a network interface, a display screen and an input device which are connected through a system bus. According to the method, different types of thermal power plant carbon emission data can be classified, and the safety and reliability of industrial production are improved.

Description

Technical Field

[0001] This invention relates to the field of thermal power plants, and more particularly to a method, apparatus, and electronic equipment for classifying carbon emission data from thermal power plants. Background Technology

[0002] As a critical infrastructure for ensuring energy security, thermal power plants are a major source of greenhouse gas emissions due to the combustion of fossil fuels. Optimizing the operating conditions of thermal power plants to maximize energy efficiency and minimize carbon emissions has become a core issue in the industry in order to address global climate change. Effectively classifying carbon emission data from thermal power plants under different operating conditions is a technological prerequisite for achieving this goal and a foundation for intelligent operation and maintenance and energy efficiency diagnosis.

[0003] Carbon emission data from thermal power plants is stored in time series format. Current technologies for measuring time series similarity often focus on the overall shape or statistical characteristics of the series, neglecting the dynamic trends within subsequences—the "trend information." These differences in internal dynamic trends are crucial for diagnosing unit operating status, but traditional measurement methods cannot capture these subtle dynamic changes and lack a mechanism that effectively integrates the structural complexity of the series with complementary features such as dynamic trends in a low-complexity manner. For thermal power plant carbon emission data, this means an inability to effectively identify abnormal operating conditions characterized by high energy consumption and high emissions, and difficulty in providing early warnings of emission fluctuations caused by minor equipment changes—key differences hidden within the dynamic trends of the data series.

[0004] Therefore, when existing technologies are applied to the classification of carbon emission data in thermal power plants, their accuracy and reliability in terms of operating condition diagnosis and anomaly early warning are insufficient to meet the needs of modern industrial intelligent management. Summary of the Invention

[0005] This invention provides a method, apparatus, and electronic device for classifying carbon emission data from thermal power plants. This invention can classify different categories of carbon emission data from thermal power plants, improving the safety and reliability of industrial production. Details are described below:

[0006] A first aspect: a method for classifying carbon emission data from thermal power plants, the method comprising:

[0007] Obtain multiple weighted Euclidean distances and filter out candidate Euclidean distances that meet the preset distance conditions;

[0008] The candidate Euclidean distances are subjected to trend-weighted processing to generate the corresponding trend-weighted Euclidean distances;

[0009] The candidate Euclidean distance and trend-weighted Euclidean distance are matched to determine the target Euclidean distance between the first time series and the second time series.

[0010] Based on the target Euclidean distance, the nearest neighbor classifier is used to divide the subsequence categories of carbon emission data from thermal power plants, thereby classifying the carbon emission data of thermal power plants. Based on the classification results, the production data of thermal power plants are analyzed and controlled.

[0011] The acquisition of multiple weighted Euclidean distances is as follows:

[0012] Acquire a first time series and a second time series of carbon emission data from thermal power plants, and extract information from multiple sensing points from the first time series; based on the information from multiple sensing points, determine multiple sensing sub-sequences corresponding to the first time series;

[0013] Based on multiple sensing subsequences, multiple location subsequences are determined from the second time series; the Euclidean distance between the multiple sensing subsequences and the corresponding subsequence of each location subsequence is determined.

[0014] The complexity of the Euclidean distances of multiple subsequences is weighted to obtain multiple weighted Euclidean distances.

[0015] The process of selecting candidate Euclidean distances that meet the preset distance conditions is as follows:

[0016] For all Euclidean distances D′ eval Sort the sequences and select the smallest n perceptual subsequences and positional subsequences for the next step of calculation.

[0017] The trend-weighted Euclidean distance is generated by performing trend-weighted processing on the candidate Euclidean distances.

[0018] Determine the standard deviations of multiple perceptual subsequences and their corresponding multiple positional subsequences; obtain preset trend configuration information of the multiple standard deviations and preset trend encoding conditions, and determine the trend encoding value corresponding to the standard deviations based on the preset trend configuration information; determine the trend weight based on the trend encoding value.

[0019] The candidate Euclidean distance and the trend weight are weighted to generate the trend-weighted Euclidean distance.

[0020] The process of performing trend-weighted processing on candidate Euclidean distances to generate corresponding trend-weighted Euclidean distances is as follows:

[0021] Given any subsequence S = {s0, s1, ..., s2} l-1 Given a preset step size parameter step, the generation process of the trend encoding sequence E(S) = {e0, e1, ...} is as follows:

[0022] Δ k =s k+step -sk

[0023] In the formula, s k Here, is the value at position k of the subsequence, step is the preset step size parameter, and s k+step finger subsequences s k The value of the step position after the position, Δ k This represents the difference between adjacent steps of points in the subsequence;

[0024]

[0025] θ1=0.5·σ valid (S)

[0026] θ2=1.0·σ valid (S)

[0027] In the formula, σ valid (S) is the standard deviation of sequence S, e k The trend difference ratio is calculated using the encoded values ​​of the subsequences.

[0028]

[0029] In the formula, E1 and E2 represent the trend coding sequences of the two subsequences, respectively, and K valid It is the set of indices of all valid corresponding positions in the two encoded sequences; (E1)k and (E2)k are the encoded values ​​of the two encoded sequences at the k-th position; N valid This represents the total number of valid corresponding positions; the trend influence term d is obtained. trend :

[0030] d trend (S A ,S B ) = ED(S A ,S B )·Δ ratio (E(S A ),E(S B ))

[0031] In the formula, d trend (S A ,S B ) is the Euclidean distance weighted by the trend difference ratio, ED(S) A ,S B ) represents the original Euclidean distance, Δ ratio (E(S A ),E(S B )) is the aforementioned trend difference ratio.

[0032] The subsequence classification of carbon emission data from thermal power plants based on target Euclidean distance using a nearest neighbor classifier is as follows:

[0033] After calculating the candidate window, two Euclidean distances are obtained: one weighted by complexity and the other by trend. The distance between the perceptual subsequence and the positional subsequence is then calculated using preset weights.

[0034]

[0035] In the formula, d comb (S A ,S B ) represents the perceptual subsequence S A and positional subsequence S B The final combined distance between them; D′ eval (S A ,S B ) represents the complexity-weighted Euclidean distance, d trend (S A ,S B α is the trend-weighted Euclidean distance. c and α t These are the weights of the complexity-weighted Euclidean distance and the trend-weighted Euclidean distance, respectively; the Euclidean distance matching results of multi-feature fusion of the first and second time series are calculated:

[0036]

[0037] In the formula, D(T1,T2) is the distance result between the first time series T1 and the second time series T2; L(S) is the length of the sequence S, (1) refers to the first time series, PIS represents that the subsequence is a perceptual subsequence, and i refers to the i-th perceptual subsequence. The i-th perceptual subsequence of the first time series, d comb For subsequence and The final combined distance between them, (1→2) refers to the first time series guiding the second time series to divide into subsequences, PCS represents that the subsequence is a positional subsequence, and j refers to the j-th positional subsequence. This refers to the positional subsequence of the second time series, which is divided according to the first time series. This refers to the positional subsequence of the first time series, which is divided according to the second time series. Refers to the j-th perceptual subsequence of the second time series. It refers to the k-th perceptual subsequence of the first time series. It refers to the kth perceptual subsequence of the second time series.

[0038] A second aspect is an apparatus for classifying carbon emission data from thermal power plants, the apparatus comprising: a processor and a memory, the memory storing program instructions, the processor invoking the program instructions stored in the memory to cause the apparatus to perform the method described in any of the first aspects.

[0039] Third aspect, a computer-readable storage medium storing a computer program, the computer program including program instructions that, when executed by a processor, cause the processor to perform the method described in any one of the first aspects.

[0040] The beneficial effects of the technical solution provided by this invention are:

[0041] 1. This invention uses complexity-weighted Euclidean distance for preliminary screening, prioritizing key features in the data. This solves the problem of mismatches caused by local noise or non-critical fluctuations when using simple distance metrics in existing technologies. It can classify carbon emission data from different types of thermal power plants, improving the safety and reliability of industrial production.

[0042] 2. This invention introduces trend information analysis, which is beneficial for grasping the overall dynamic behavior of the sequence. It solves the shortcomings of existing technologies that only focus on point-to-point morphological similarity and ignore macro-level changing trends, thereby enhancing the robustness and interpretability of the matching results; improving the classification accuracy of carbon emission data from thermal power plants under complex operating conditions and reducing abnormal interference.

[0043] 3. This invention integrates the morphological and trend characteristics of carbon emission data from thermal power plants, achieving a synergistic consideration of microscopic morphology and macroscopic trends. It overcomes the limitations of existing technologies that rely on a single feature dimension for matching, and improves the comprehensive discrimination capability of matching. In the application scenario of carbon emission analysis and processing in thermal power plants, it can analyze carbon emission data from thermal power plants from multiple perspectives, comprehensively judge the data category, and provide technical support for achieving intelligent and efficient plant operation. Attached Figure Description

[0044] Figure 1 A flowchart of a method for classifying carbon emission data from thermal power plants;

[0045] Figure 2 A schematic diagram illustrating the correspondence between the perceived subsequence and the positional subsequence;

[0046] Figure 3 This is a schematic diagram illustrating the average rank of the algorithms.

[0047] Figure 4 This is a schematic diagram of a device for classifying carbon emission data from thermal power plants.

[0048] Figure 5A block diagram of an electronic device for classifying carbon emission data from thermal power plants;

[0049] Figure 6 Another block diagram of an electronic device for classifying carbon emission data in thermal power plants. Detailed Implementation

[0050] To make the objectives, technical solutions, and advantages of the present invention clearer, the embodiments of the present invention will be described in further detail below.

[0051] Therefore, in order to solve the problems existing in the background technology, there is an urgent need in this field for a new time series similarity measurement method that can deeply explore and quantify the internal dynamic trend of the sequence, and effectively integrate this trend information with other key features such as the complexity of the sequence, so as to achieve more accurate similarity measurement while ensuring low time complexity.

[0052] Example 1

[0053] This invention proposes a method for classifying carbon emission data from thermal power plants. (See also...) Figure 1 The specific steps of this method to solve the problem are as follows:

[0054] S101: Acquire the first and second time series of carbon emission data from thermal power plants, and extract information from multiple sensing points from the first time series;

[0055] S102: Based on information from multiple sensing points, determine multiple sensing sub-sequences corresponding to the first time series;

[0056] S103: Based on multiple sensing subsequences, determine multiple location subsequences from the second time series;

[0057] S104: Determine the Euclidean distance between multiple perceptual subsequences and their respective corresponding subsequences at each location;

[0058] S105: Perform complexity weighting on the Euclidean distances of multiple subsequences to obtain multiple weighted Euclidean distances;

[0059] S106: Select candidate Euclidean distances that meet the preset distance conditions from multiple weighted Euclidean distances;

[0060] S107: Perform trend-weighted processing on the candidate Euclidean distances to generate the corresponding trend-weighted Euclidean distances;

[0061] S108: Perform matching processing on candidate Euclidean distances and trend-weighted Euclidean distances to determine the target Euclidean distance between the first time series and the second time series;

[0062] S109: Based on the target Euclidean distance, the nearest neighbor classifier is used to divide the subsequence categories of carbon emission data from thermal power plants, thereby classifying the carbon emission data of thermal power plants and analyzing and controlling the production data of thermal power plants based on the classification results.

[0063] In summary, by accurately classifying time-series data in the carbon emission data of thermal power plants, this invention can effectively achieve intelligent analysis of thermal power plant production data and improve production efficiency.

[0064] Example 2

[0065] The specific formula will be used below. Figure 2 The following example will be used to provide a detailed description of the solution in Embodiment 1, as detailed below:

[0066] In step S101 and in this embodiment of the invention, the first time series can refer to sensor data in the industrial field, including: gas concentration sensor data, equipment vibration sensor data, temperature data, electrocardiogram, etc. This embodiment of the invention uses carbon emission data from a thermal power plant as an example for illustration, but in specific implementation, this embodiment of the invention does not impose any limitations on this. The second time series refers to another set of data arranged in chronological order that needs to be used to calculate the distance from the first time series.

[0067] First, the first and second time series need to be normalized. The calculation formula is as follows:

[0068]

[0069]

[0070] In the formula, μ(I) represents the mean of the indices of the first or second time series, and n represents the total length of the first or second time series; σ(I) represents the standard deviation of the indices of the first or second time series, and i represents the original time indices traversed from 0 to n-1; x i It is the new horizontal coordinate value obtained after standardizing the original time index i.

[0071] Sensing points are extracted from the normalized first or second time series. These sensing points are a set of sparse key data points, and their selection is not based on uniform sampling, but rather through an iterative geometric optimization process to ensure that the most morphological information is captured with the fewest points. The calculation formula is as follows:

[0072]

[0073] In the formula, P j This is the data point currently being examined, x j These are the standardized coordinate values, tj This is the original ordinate value. 'a' represents the segment(P) of the line segment. s ,P e The slope of ) is c′, which is a parameter related to the intercept of the line segment. s and P e These are two currently identified adjacent sensing points, initially representing the first or last data points of either the first or second time series. The sensing points are the two known sensing points P. s and P e Between d(P) j ,segment(P s ,P e The largest point. Through multiple iterations, a predetermined number of all sensing points are obtained.

[0074] In step S102, the sensing subsequence of the first time series is obtained using sensing points. The construction of the sensing subsequence employs a sliding window strategy, where the sequence segment covered by three consecutive sensing points is used as the sensing subsequence. Its mathematical expression is as follows:

[0075]

[0076] In the formula, Let T1 represent the j-th perceptual subsequence, T1 be the original first time series, and pip be the pip subsequence. j and pip j+2 These are the indices of the starting and ending sensing points that define the boundaries of this subsequence. The sliding window starts at j=0 and moves one position forward each time until the end of the window reaches the last point in the sensing point list. Through this process, all sensing subsequences of the first time series are obtained.

[0077] In step S103, Figure 2 This diagram illustrates the correspondence between the perceived subsequence and the position subsequence. Based on the perceived subsequence of the first time series, the position subsequence of the second time series for each perceived subsequence is calculated. The position of the position subsequence in the second time series is calculated using the following formula:

[0078]

[0079] In the formula, j is the index of the sensing point, (1) refers to the first time series, For the j-th sensing point in the first time series, Let w be the (j+2)th sensing point of the first time series, w be a pre-set parameter, and N2 be the length of the second time series.

[0080] startpos j and endpos jThe w represents the start and end points of the positional subsequence. A perceptual subsequence can correspond to multiple positional subsequences depending on the value of w.

[0081] In step S104, it is necessary to calculate the Euclidean distance between the aforementioned perceptual subsequence and the positional subsequence. This embodiment of the invention employs a pre-computation method to improve computational efficiency, using an acceleration matrix to calculate the Euclidean distance between the perceptual subsequence and the positional subsequence. The calculation formula is as follows:

[0082] PDM[i,k]=(t′ 1,i -t′ 2,k ) 2

[0083] In the formula, PDM[i,k] is the element in the i-th row and k-th column of the matrix, and t′ 1,i and t′ 2,k These are the values ​​at indices i and k of the first and second time series after z-score standardization.

[0084]

[0085] In the formula, any sensing subsequence is calculated using a pre-computed PDM matrix. with positional subsequence The square of the Euclidean distance between them ED 2 It can be calculated quickly using an acceleration matrix.

[0086] In step S105, the Euclidean distance calculated in the previous step needs to be weighted to obtain a more refined similarity judgment. This embodiment of the invention uses a complexity-weighted mechanism for calculation, and the calculation formula is as follows:

[0087]

[0088] In the formula, s k+1 and s k These are two adjacent data points in a subsequence S. CI(S) represents the complexity index of the first or second time series. The larger the value, the more drastic the fluctuations within the series, and the higher the complexity.

[0089]

[0090] In the formula, R c This is the complexity ratio, and its value is always greater than or equal to 1. S A and S B These are the first and second subsequences, respectively. When R... c When the value is close to 1, it indicates that the complexity of the two subsequences is very close, R. cThe larger the value, the greater the difference in their complexity. `max()` and `min()` are functions that retrieve the maximum and minimum values, respectively.

[0091] D′ eval (S A ,S B ) = ED(S A ,S B )·R c

[0092] In the formula, D′ eval It is the Euclidean distance weighted by complexity ratio, ED(S) A ,S B R is the original Euclidean distance between subsequences A and B. c It is the complexity ratio.

[0093] In step S106, for all Euclidean distances D′ eval Sort the sequences and select the smallest n perceptual subsequences and positional subsequences for the next step of calculation. n is a pre-set parameter.

[0094] In step S107, a trend-weighted mechanism is used to perform a secondary correction on the candidate Euclidean distance. Given any subsequence S = {s0, s1, ..., s...} 1-1 Given a preset step size parameter step, the generation process of the trend encoding sequence E(S) = {e0, e1, ...} is as follows:

[0095] Δ k =s k+step -s k .

[0096] In the formula, s k Here, is the value at position k of the subsequence, step is the preset step size parameter, and s k+step finger subsequences s k The value of the step position after the position, Δ k It represents the difference between adjacent steps of points in the subsequence.

[0097]

[0098] θ1=0.5·σ valid (S)

[0099] θ2=1.0·σ valid (S)

[0100] In the formula, σ valid (S) is the standard deviation of sequence S, e kThe subsequence is encoded using the above formula. The subsequence is then converted into encoded values. Next, the trend difference ratio will be calculated using these encoded values:

[0101]

[0102] In the formula, E1 and E2 represent the trend coding sequences of the two subsequences, respectively, and K valid It is the set of indices of all valid corresponding positions in the two encoded sequences. (E1) k and (E2) k N is the encoded value of two encoded sequences at the k-th position. valid This represents the total number of valid corresponding positions. The constant 4 in the denominator is the maximum possible difference in trend coding values, used to normalize the ratio to the range [0,1]. Subsequently, this trend difference ratio is used as a penalty factor and multiplied by the original Euclidean distance to obtain the trend influence term d. trend :

[0103] d trend (S A ,S B ) = ED(S A ,S B )·Δ ratio (E(S A ),E(S B ))

[0104] In the formula, d trend (S A ,S B ) is the Euclidean distance weighted by the trend difference ratio, ED(S) A ,S B ) represents the original Euclidean distance, Δ ratio (E(S A ),E(S B The aforementioned trend difference ratio is used. After this calculation, the Euclidean distance is corrected for the trend perspective.

[0105] In step S108, after the candidate window calculation, two Euclidean distances are obtained: complexity-weighted and trend-weighted. The distance between the perceptual subsequence and the positional subsequence is then calculated using preset weights.

[0106]

[0107] In the formula, d comb (S A ,S B ) represents the perceptual subsequence S A and positional subsequence S B The final combined distance between them, D′. eval (S A ,SB ) represents the complexity-weighted Euclidean distance, d trend (S A ,S B α is the trend-weighted Euclidean distance. c and α t These are the weights for complexity-weighted Euclidean distance and trend-weighted Euclidean distance, respectively. Having obtained the final combined distance of the subsequences, the next step is to calculate the Euclidean distance matching result of the multi-feature fusion of the first and second time series:

[0108]

[0109] In the formula, D(T1,T2) represents the distance between the first time series T1 and the second time series T2. L(S) is the length of the sequence S, (1) refers to the first time series, PIS represents that the subsequence is a perceptual subsequence, and i refers to the i-th perceptual subsequence. The i-th perceptual subsequence of the first time series, d comb For subsequence and The final combined distance between them, (1→2) refers to the first time series guiding the second time series to divide into subsequences, PCS represents that the subsequence is a positional subsequence, and j refers to the j-th positional subsequence. This refers to the positional subsequence of the second time series, which is divided according to the first time series. This refers to the positional subsequence of the first time series, which is divided according to the second time series. Refers to the j-th perceptual subsequence of the second time series. It refers to the k-th perceptual subsequence of the first time series. This refers to the k-th perceptual subsequence of the second time series. The second term of the formula is calculated in the same way as the first term, except that the first and second time series are interchanged. This formula uses a symmetric calculation method to calculate the distance between T1 and T2.

[0110] In summary, the embodiments of the present invention use complexity-weighted Euclidean distance for preliminary screening, giving priority to key features in the data. This solves the problem that existing technologies using simple distance metrics are prone to mismatches due to local noise or non-critical fluctuations. It can classify carbon emission data from different types of thermal power plants, thereby improving the safety and reliability of industrial production.

[0111] Example 3

[0112] This invention uses 70 datasets from the publicly available UCR time series classification dataset and one gas turbine sensor dataset to verify the performance of the proposed method for classifying carbon emission data from thermal power plants. For all data, a pre-defined training and test set was used. The average rank of 15 algorithms was compared, and the results are as follows: Figure 3 As shown, the method of this invention achieved the lowest average rank value, ranking first among all algorithms. This fully demonstrates that the multi-feature fusion strategy proposed in this embodiment of the invention has the strongest overall discriminative ability and classification performance.

[0113] The Gas Turbine dataset originates from gas turbine data in a thermal power plant. The data is collected by multiple sensors placed within the system to measure the concentrations of carbon monoxide (CO) and nitrogen oxides (NOx) in emissions in real time. This is a binary classification dataset; based on the changes in sensor data over a period of time, it determines whether the current emission from the system is carbon monoxide or nitrogen oxides.

[0114] Table 1 shows the classification accuracy information of the Gas Turbine dataset in the algorithm of this invention.

[0115]

[0116]

[0117] As can be seen from the results in Table 1, the embodiments of the present invention have a high classification accuracy and can achieve the classification of carbon emission data.

[0118] Example 4

[0119] An apparatus for classifying carbon emission data from thermal power plants, see [link to apparatus]. Figure 4 The device includes a processor and a memory, the memory storing program instructions, and the processor calling the program instructions stored in the memory to cause the device to perform the following method steps in Embodiment 1:

[0120] Obtain multiple weighted Euclidean distances and filter out candidate Euclidean distances that meet the preset distance conditions;

[0121] The candidate Euclidean distances are subjected to trend-weighted processing to generate the corresponding trend-weighted Euclidean distances;

[0122] The candidate Euclidean distance and trend-weighted Euclidean distance are matched to determine the target Euclidean distance between the first time series and the second time series.

[0123] Based on the target Euclidean distance, the nearest neighbor classifier is used to divide the subsequence categories of carbon emission data from thermal power plants, thereby classifying the carbon emission data of thermal power plants. Based on the classification results, the production data of thermal power plants are analyzed and controlled.

[0124] The acquisition of multiple weighted Euclidean distances is as follows:

[0125] Acquire a first time series and a second time series of carbon emission data from thermal power plants, and extract information from multiple sensing points from the first time series; based on the information from multiple sensing points, determine multiple sensing sub-sequences corresponding to the first time series;

[0126] Based on multiple sensing subsequences, multiple location subsequences are determined from the second time series; the Euclidean distance between the multiple sensing subsequences and the corresponding subsequence of each location subsequence is determined.

[0127] The complexity of the Euclidean distances of multiple subsequences is weighted to obtain multiple weighted Euclidean distances.

[0128] The process of selecting candidate Euclidean distances that meet the preset distance conditions is as follows:

[0129] For all Euclidean distances D′ eval Sort the sequences and select the smallest n perceptual subsequences and positional subsequences for the next step of calculation.

[0130] Specifically, the candidate Euclidean distances are subjected to trend-weighted processing to generate the corresponding trend-weighted Euclidean distances:

[0131] Determine the standard deviations of multiple perceptual subsequences and their corresponding multiple positional subsequences; obtain preset trend configuration information of the multiple standard deviations and preset trend encoding conditions, and determine the trend encoding value corresponding to the standard deviations based on the preset trend configuration information; determine the trend weight based on the trend encoding value.

[0132] The candidate Euclidean distance and the trend weight are weighted to generate the trend-weighted Euclidean distance.

[0133] The process of performing trend-weighted processing on candidate Euclidean distances to generate corresponding trend-weighted Euclidean distances is as follows:

[0134] Given any subsequence S = {s0, s1, ..., s2} 1-1 Given a preset step size parameter step, the generation process of the trend encoding sequence E(S) = {e0, e1, ...} is as follows:

[0135] Δ k =s k+step -s k

[0136] In the formula, s k Here, is the value at position k of the subsequence, step is the preset step size parameter, and s k+step finger subsequences sk The value of the step position after the position, Δ k This represents the difference between adjacent steps of points in the subsequence;

[0137]

[0138] θ1=0.5·σ valid (S)

[0139] θ2=1.0·σ valid (S)

[0140] In the formula, σ valid (S) is the standard deviation of sequence S, e k The trend difference ratio is calculated using the encoded values ​​of the subsequences.

[0141]

[0142] In the formula, E1 and E2 represent the trend coding sequences of the two subsequences, respectively, and K valid It is the set of indices of all valid corresponding positions in the two encoded sequences; (E1)k and (E2)k are the encoded values ​​of the two encoded sequences at the k-th position; N valid This represents the total number of valid corresponding positions; the trend influence term d is obtained. trend :

[0143] d trend (S A ,S B ) = ED(S A ,S B )·Δ ratio (E(S A ),E(S B ))

[0144] In the formula, d trend (S A ,S B ) is the Euclidean distance weighted by the trend difference ratio, ED(S) A ,S B ) represents the original Euclidean distance, Δ ratio (E(S A ),E(S B )) is the aforementioned trend difference ratio.

[0145] Among them, based on the target Euclidean distance, the nearest neighbor classifier is used to divide the subsequence categories of carbon emission data from thermal power plants into the following categories:

[0146] After calculating the candidate window, two Euclidean distances are obtained: one weighted by complexity and the other by trend. The distance between the perceptual subsequence and the positional subsequence is then calculated using preset weights.

[0147]

[0148] In the formula, d comb (S A ,S B ) represents the perceptual subsequence S A and positional subsequence S B The final combined distance between them; D′ eval (S A ,S B ) represents the complexity-weighted Euclidean distance, d trend (S A ,S B α is the trend-weighted Euclidean distance. c and α t These are the weights of the complexity-weighted Euclidean distance and the trend-weighted Euclidean distance, respectively; the Euclidean distance matching results of multi-feature fusion of the first and second time series are calculated:

[0149]

[0150] In the formula, D(T1,T2) is the distance result between the first time series T1 and the second time series T2; L(S) is the length of the sequence S, (1) refers to the first time series, PIS represents that the subsequence is a perceptual subsequence, and i refers to the i-th perceptual subsequence. The i-th perceptual subsequence of the first time series, d comb For subsequence and The final combined distance between them, (1→2) refers to the first time series guiding the second time series to divide into subsequences, PCS represents that the subsequence is a positional subsequence, and j refers to the j-th positional subsequence. This refers to the positional subsequence of the second time series, which is divided according to the first time series. This refers to the positional subsequence of the first time series, which is divided according to the second time series. Refers to the j-th perceptual subsequence of the second time series. It refers to the k-th perceptual subsequence of the first time series. It refers to the kth perceptual subsequence of the second time series.

[0151] In summary, the embodiments of the present invention use complexity-weighted Euclidean distance for preliminary screening, giving priority to key features in the data. This solves the problem that existing technologies using simple distance metrics are prone to mismatches due to local noise or non-critical fluctuations. It can classify carbon emission data from different types of thermal power plants, thereby improving the safety and reliability of industrial production.

[0152] Example 5

[0153] Figure 5 This is a block diagram illustrating an electronic device for classifying carbon emission data from a thermal power plant, according to an exemplary embodiment. The electronic device can be a terminal, and its internal structure diagram can be as follows: Figure 5 As shown, the electronic device includes a processor, memory, network interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The network interface is used to communicate with external terminals via a network connection. When executed by the processor, the computer program implements a method for classifying carbon emission data from thermal power plants. The display screen can be a liquid crystal display (LCD) or an e-ink display. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad mounted on the device's casing, or an external keyboard, touchpad, or mouse.

[0154] Those skilled in the art will understand that Figure 5 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the electronic device to which the present application is applied. The specific electronic device may include more or fewer components than shown in the figure, or combine certain components, or have different component arrangements.

[0155] Figure 6 This is a block diagram of an electronic device for a system for classifying carbon emission data from a thermal power plant, according to an exemplary embodiment. The electronic device may be a server. The electronic device includes a processor, memory, and a network interface connected via a system bus. The processor provides computing and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The network interface of the electronic device is used to communicate with external terminals via a network connection. When the computer program is executed by the processor, it implements a system for classifying carbon emission data from a thermal power plant.

[0156] Those skilled in the art will understand that Figure 6 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the electronic device to which the present application is applied. The specific electronic device may include more or fewer components than shown in the figure, or combine certain components, or have different component arrangements.

[0157] In an exemplary embodiment, an electronic device is also provided, comprising: a processor; and a memory for storing processor-executable instructions; wherein the processor is configured to execute the instructions to implement a method for classifying carbon emission data from a thermal power plant as described in the embodiments of this application.

[0158] In an exemplary embodiment, a computer-readable storage medium is also provided, which, when executed by a processor of an electronic device, enables the electronic device to perform a method for classifying carbon emission data from a thermal power plant, as described in this application. The computer-readable storage medium may be a ROM, random access memory (RAM), CD-ROM, magnetic tape, floppy disk, or optical data storage device, etc.

[0159] In an exemplary embodiment, a computer program product containing instructions is also provided, which, when run on a computer, causes the computer to execute a method for classifying carbon emission data from thermal power plants according to an embodiment of this application.

[0160] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. This computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and RAMbus dynamic RAM (RDRAM), etc.

[0161] Unless otherwise specified, the model numbers of the various devices in this embodiment of the invention are not limited, and any device that can perform the above functions is acceptable.

[0162] Those skilled in the art will understand that the accompanying drawings are merely schematic diagrams of a preferred embodiment, and the sequence numbers of the above embodiments of the present invention are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0163] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for classifying carbon emission data from thermal power plants, characterized in that, The method includes: Obtain multiple weighted Euclidean distances and filter out candidate Euclidean distances that meet the preset distance conditions; The candidate Euclidean distances are subjected to trend-weighted processing to generate the corresponding trend-weighted Euclidean distances; The candidate Euclidean distance and trend-weighted Euclidean distance are matched to determine the target Euclidean distance between the first time series and the second time series. Based on the target Euclidean distance, the nearest neighbor classifier is used to divide the subsequence categories of carbon emission data from thermal power plants, thereby classifying the carbon emission data of thermal power plants. Based on the classification results, the production data of thermal power plants are analyzed and controlled.

2. The method for classifying carbon emission data from thermal power plants according to claim 1, characterized in that, The method for obtaining multiple weighted Euclidean distances is as follows: Acquire first and second time series of carbon emission data from thermal power plants, and extract information from multiple sensing points from the first time series; Based on information from multiple sensing points, multiple sensing sub-sequences corresponding to the first time series are determined. Based on multiple sensing subsequences, multiple location subsequences are determined from the second time series; Determine the Euclidean distance between multiple perceptual subsequences and the corresponding subsequence at each location; The complexity of the Euclidean distances of multiple subsequences is weighted to obtain multiple weighted Euclidean distances.

3. The method for classifying carbon emission data from thermal power plants according to claim 1, characterized in that, The candidate Euclidean distances that meet the preset distance conditions are selected from them as follows: For all Euclidean distances D′ eval Sort the sequences and select the smallest n perceptual subsequences and positional subsequences for the next step of calculation.

4. The method for classifying carbon emission data from thermal power plants according to claim 1, characterized in that, The trend-weighted Euclidean distance is generated by performing trend-weighted processing on the candidate Euclidean distances. Determine the standard deviation of multiple sensing subsequences and their corresponding multiple location subsequences; Obtain preset trend configuration information of multiple standard deviations and preset trend coding conditions, and determine the trend coding value corresponding to the standard deviation based on the preset trend configuration information; The trend weight is determined based on the trend code value; The candidate Euclidean distance and the trend weight are weighted to generate the trend-weighted Euclidean distance.

5. A method for classifying carbon emission data from thermal power plants according to claim 4, characterized in that, The process of performing trend-weighted processing on the candidate Euclidean distances to generate the corresponding trend-weighted Euclidean distances is as follows: Given any subsequence S = {s0, s1, ..., s2} 1-1 Given a preset step size parameter step, the generation process of the trend encoding sequence E(S) = {e0, e1, ...} is as follows: D k =s k+step -s k In the formula, s k Here, is the value at position k of the subsequence, step is the preset step size parameter, and s k+step finger subsequences s k The value of the step position after the position, Δ k This represents the difference between adjacent steps of points in the subsequence; θ1=0.5·σ valid (S) θ2=1.0·σ valid (S) In the formula, σ valid (S) is the standard deviation of sequence S, e k The trend difference ratio is calculated using the encoded values ​​of the subsequences. In the formula, E1 and E2 represent the trend coding sequences of the two subsequences, respectively, and K valid It is the set of indices of all valid corresponding positions in the two encoded sequences; (E1) k and (E2) k It is the encoded value of two encoded sequences at the k-th position; N valid This represents the total number of valid corresponding positions; the trend influence term d is obtained. trend : d trend (S A ,S B )=ED(S A ,S B )·Δ ratio (E(S A ),E(S B )) In the formula, d trend (S A ,S B ) is the Euclidean distance weighted by the trend difference ratio, ED(S) A ,S B ) represents the original Euclidean distance, Δ ratio (E(S A ),E(S B )) is the aforementioned trend difference ratio.

6. The method for classifying carbon emission data from thermal power plants according to claim 1, characterized in that, The subsequence categories of carbon emission data from thermal power plants, based on the target Euclidean distance and using a nearest neighbor classifier, are as follows: After calculating the candidate window, two Euclidean distances are obtained: one weighted by complexity and the other by trend. The distance between the perceptual subsequence and the positional subsequence is then calculated using preset weights. In the formula, d comb (S A ,S B ) represents the perceptual subsequence S A and positional subsequence S B The final combined distance between them; D′ eval (S A ,S B ) represents the complexity-weighted Euclidean distance, d trend (S A ,S B α is the trend-weighted Euclidean distance. c and α t These are the weights of complexity-weighted Euclidean distance and trend-weighted Euclidean distance, respectively. Calculate the Euclidean distance matching results of multi-feature fusion of the first and second time series: In the formula, D(T1,T2) is the distance result between the first time series T1 and the second time series T2; L(S) is the length of the sequence S, (1) refers to the first time series, PIS represents that the subsequence is a perceptual subsequence, and i refers to the i-th perceptual subsequence. The i-th perceptual subsequence of the first time series, d comb For subsequence and The final combined distance between them, (1→2) refers to the first time series guiding the second time series to divide into subsequences, PCS represents that the subsequence is a positional subsequence, and j refers to the j-th positional subsequence. This refers to the positional subsequence of the second time series, which is divided according to the first time series. This refers to the positional subsequence of the first time series, which is divided according to the second time series. Refers to the j-th perceptual subsequence of the second time series. It refers to the k-th perceptual subsequence of the first time series. It refers to the kth perceptual subsequence of the second time series.

7. A device for classifying carbon emission data from thermal power plants, characterized in that, The device includes a processor and a memory, the memory storing program instructions, the processor invoking the program instructions stored in the memory to cause the device to perform the method according to any one of claims 1-6.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, the computer program including program instructions that, when executed by a processor, cause the processor to perform the method described in any one of claims 1-6.

9. An electronic device, characterized in that, The electronic device includes a processor, a memory, a network interface, a display screen, and an input device connected via a system bus; the processor invokes program instructions stored in the memory to cause the device to perform the method described in any one of claims 1-6.

Images