Equipment Configuration Method and Device Based on Time Allocation Coefficient and Industry Feature Extraction

Abstract

This disclosure presents an embodiment of a device configuration method and apparatus based on time allocation coefficients and industry feature extraction. One specific implementation of the method includes: acquiring time feature data information; performing the following steps: determining point source time allocation coefficients; determining industry activity feature information; determining a set of point source time allocation coefficients and a set of industry activity feature information; determining a set of area source time allocation coefficients; generating a set of point source dynamic emission data and an area source dynamic emission data set; generating an overall dynamic inventory; and adaptively configuring cleaning equipment in the target emission reduction area to drive the cleaning equipment to perform cleaning operations in the target emission reduction area. This implementation can reduce system configuration redundancy, improve the accuracy of the dynamic inventory, reduce waste of cleaning resources and energy consumption of cleaning equipment; and reduce system complexity, improve the coherence, consistency, and stability of the overall dynamic inventory, thereby enabling accurate and efficient environmental decision-making.

Description

Technical Field

[0001] The embodiments disclosed herein relate to the field of computer technology, and more specifically to a device configuration method and apparatus based on time allocation coefficients and industry feature extraction. Background Technology

[0002] Dynamic inventory systems are crucial tools for tracing pollutant sources, assessing pollution levels, and developing control strategies in ecological and environmental monitoring and management. Currently, when generating dynamic inventory systems to configure cleanroom equipment, the common approach is to calculate point source and area source emissions separately using pre-defined static parameters and periodically updated data. For example, when obtaining emission data for a specific area, a fixed time allocation coefficient template, static industry activity characteristic parameters, and corresponding data access interfaces need to be configured on the data calculation terminal.

[0003] However, in practice, when using the above method to generate a dynamic inventory for configuring cleaning equipment, the following technical problems often arise: For each industry type and even each emission source (point source or area source), static parameters and calculation interfaces must be preset based on experience. This results in high system redundancy and difficulty in adapting to the dynamic and ever-changing production patterns of different industries. The preset static parameters and templates are updated late and cannot respond to the spatiotemporal fluctuations of production activities and emissions, leading to poor accuracy in the generated dynamic inventory. This results in insufficient cleaning efforts in high-emission areas and wasted cleaning resources in low-emission areas, consuming more cleaning equipment and energy. Furthermore, it can only perform independent calculations for a single type of point source or area source, and cannot integrate multi-source data or perform unified dynamic integration based on industry characteristics. This reduces the coherence, consistency, and stability of the overall dynamic inventory, making it difficult to support accurate and efficient environmental decision-making.

[0004] The information disclosed in this background section is only intended to enhance the understanding of the background of the inventive concept, and therefore may contain information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] The summary portion of this disclosure is intended to provide a brief overview of the concepts, which will be described in detail in the detailed description portion. This summary portion is not intended to identify key or essential features of the claimed technical solutions, nor is it intended to limit the scope of the claimed technical solutions.

[0006] Some embodiments of this disclosure propose a device configuration method, apparatus, device / terminal / server, and computer-readable medium based on time allocation coefficients and industry feature extraction to solve one or more of the technical problems mentioned in the background section above.

[0007] In a first aspect, some embodiments of this disclosure provide a device configuration method based on time allocation coefficients and industry feature extraction. The method includes: in response to receiving a cleaning instruction for a target emission reduction area, acquiring time feature data information corresponding to the target emission reduction area, wherein the time feature data information includes a set of area source data and a set of point source data; for each point source data in the point source data set, performing the following steps: determining a point source time allocation coefficient based on the point source data and a dynamic adjustment model corresponding to the point source data; determining industry activity feature information based on the point source data and an industry activity feature extraction model corresponding to the point source data; determining the obtained point source time allocation coefficients as a set of point source time allocation coefficients; and... The activity characteristic information of each industry is determined as the industry activity characteristic information set; for the above-mentioned area source data set and the time allocation coefficient dynamic adjustment model corresponding to each area source data in the above-mentioned area source data set, the area source time allocation coefficient set is determined; based on the above-mentioned point source data set, the above-mentioned area source data set, the above-mentioned point source time allocation coefficient set, the above-mentioned industry activity characteristic information set, and the above-mentioned area source time allocation coefficient set, a point source dynamic emission data set and an area source dynamic emission data set are generated; according to the above-mentioned point source dynamic emission data set and the above-mentioned area source dynamic emission data set, an overall dynamic inventory is generated; based on the above-mentioned overall dynamic inventory, the cleaning equipment in the above-mentioned target emission reduction area is adaptively configured to drive the above-mentioned cleaning equipment to perform cleaning operations in the above-mentioned target emission reduction area.

[0008] Secondly, some embodiments of this disclosure provide a device configuration apparatus based on time allocation coefficients and industry feature extraction. The apparatus includes: an acquisition unit configured to acquire time feature data information in response to receiving a cleaning instruction for a target emission reduction area, wherein the time feature data information includes a set of area source data and a set of point source data; an execution unit configured to perform the following steps for each point source data in the set of point source data: determining a point source time allocation coefficient based on the point source data and a dynamic adjustment model corresponding to the point source data; determining industry activity feature information based on the point source data and an industry activity feature extraction model corresponding to the point source data; and a first determination unit configured to determine the obtained point source time allocation coefficients as a set of point source time allocation coefficients and to determine the obtained industry activity feature information. The system comprises: a first determination unit, configured to determine a set of area source time allocation coefficients; a second determination unit, configured to dynamically adjust the model for the area source data set and the time allocation coefficients corresponding to each area source data in the area source data set; a first generation unit, configured to generate a set of point source dynamic emission data and a set of area source dynamic emission data based on the point source data set, the area source data set, the point source time allocation coefficient set, the industry activity characteristic information set, and the area source time allocation coefficient set; a second generation unit, configured to generate an overall dynamic list based on the point source dynamic emission data set and the area source dynamic emission data set; and a configuration unit, configured to adaptively configure the cleaning equipment in the target emission reduction area based on the overall dynamic list, so as to drive the cleaning equipment to perform cleaning operations in the target emission reduction area.

[0009] Thirdly, some embodiments of this disclosure provide an electronic device, including: one or more processors; and a storage device having one or more programs stored thereon, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the method described in any implementation of the first aspect above.

[0010] Fourthly, some embodiments of this disclosure provide a computer-readable medium having a computer program stored thereon, wherein the program, when executed by a processor, implements the method described in any of the implementations of the first aspect above.

[0011] The above embodiments of this disclosure have the following beneficial effects: the equipment configuration method based on time allocation coefficient and industry feature extraction of some embodiments of this disclosure can reduce the redundancy of system configuration, improve the accuracy of dynamic inventory, reduce the waste of cleaning resources and the energy consumption of cleaning equipment; and reduce the complexity of the system, improve the coherence, consistency and stability of the overall dynamic inventory, thereby enabling accurate and efficient environmental decision-making. Specifically, the high redundancy in system configuration leads to poor accuracy of dynamic inventories, resulting in longer processing times and higher resource consumption. It also reduces the overall coherence, consistency, and stability of the dynamic inventory, making it difficult to support accurate and efficient environmental decision-making. This is because for each industry type and even each emission source (point or area source), static parameters and calculation interfaces must be preset based on experience. This results in high system redundancy and difficulty adapting to the dynamic and ever-changing production patterns of different industries. The preset static parameters and templates are updated laggingly, failing to respond to the spatiotemporal fluctuations in production activities and emissions. This leads to poor accuracy in the generated dynamic inventory, resulting in insufficient cleaning efforts in high-emission areas and wasted cleaning resources in low-emission areas, consuming more cleaning equipment and energy. Furthermore, it can only perform independent calculations for a single type of point or area source, failing to integrate multi-source data and perform unified dynamic integration based on industry characteristics. This reduces the overall coherence, consistency, and stability of the dynamic inventory, making it difficult to support accurate and efficient environmental decision-making. Based on this, some embodiments of the device configuration method based on time allocation coefficients and industry feature extraction disclosed herein firstly acquire time feature data information in response to receiving a cleaning instruction for a target emission reduction area. This yields the time feature data information. Next, for each point source data in the aforementioned point source data set, the following steps are performed: First, based on the aforementioned point source data and a dynamic adjustment model of the corresponding time allocation coefficients, a point source time allocation coefficient is determined. This yields the point source time allocation coefficient. Then, based on the aforementioned point source data and a dynamic adjustment model of the corresponding industry activity feature extraction model, industry activity feature information is determined. This yields the industry activity feature information. Next, the obtained point source time allocation coefficients are defined as a point source time allocation coefficient set, and the obtained industry activity feature information is defined as an industry activity feature information set. This yields the point source time allocation coefficient set and the industry activity feature information set. Finally, for the aforementioned area source data set and a dynamic adjustment model of the time allocation coefficients corresponding to each area source data in the aforementioned area source data set, an area source time allocation coefficient set is determined. This yields the area source time allocation coefficient set. Subsequently, based on the aforementioned point source data set, area source data set, point source time allocation coefficient set, industry activity characteristic information set, and area source time allocation coefficient set, a point source dynamic emission data set and an area source dynamic emission data set are generated. Thus, the point source dynamic emission data set and the area source dynamic emission data set can be obtained.Next, based on the aforementioned dynamic emission data sets from point sources and area sources, a comprehensive dynamic inventory is generated. This yields the overall dynamic inventory. Finally, based on this comprehensive dynamic inventory, the cleaning equipment in the target emission reduction areas is adaptively configured to drive the equipment to perform cleaning operations in those areas. This reveals the type and quantity of cleaning equipment in the target emission reduction areas. Because this approach does not employ a fragmented, pre-defined static calculation process for point and area sources, but rather uses a unified machine learning framework to fuse real-time data with historical features and perform dynamic calculations, the consistency of point and area source calculation results across time and business logic is improved. This enhances the reliability of the overall dynamic inventory, shortens the time and intervention costs from data access to inventory generation, and reduces system configuration redundancy, improves the accuracy of the dynamic inventory, reduces waste of cleaning resources and energy consumption of cleaning equipment, and lowers system complexity. It also improves the coherence, consistency, and stability of the overall dynamic inventory, enabling precise and efficient environmental decision-making. Attached Figure Description

[0012] The above and other features, advantages, and aspects of the embodiments of this disclosure will become more apparent from the accompanying drawings and the following detailed description. Throughout the drawings, the same or similar reference numerals denote the same or similar elements. It should be understood that the drawings are schematic, and elements are not necessarily drawn to scale.

[0013] Figure 1 This is a flowchart of some embodiments of the equipment configuration method based on time allocation coefficient and industry feature extraction according to the present disclosure; Figure 2 This is a schematic diagram of the structure of some embodiments of the equipment configuration device based on time allocation coefficient and industry characteristic extraction according to the present disclosure; Figure 3 This is a schematic diagram of the structure of an electronic device suitable for implementing some embodiments of the present disclosure. Detailed Implementation

[0014] Embodiments of this disclosure will now be described in more detail with reference to the accompanying drawings. While some embodiments of this disclosure are shown in the drawings, it should be understood that this disclosure can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of this disclosure. It should be understood that the accompanying drawings and embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of protection of this disclosure.

[0015] It should also be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings. Unless otherwise specified, the embodiments and features described in this disclosure can be combined with each other.

[0016] It should be noted that the concepts of "first" and "second" mentioned in this disclosure are used only to distinguish different devices, modules or units, and are not used to limit the order of functions performed by these devices, modules or units or their interdependencies.

[0017] It should be noted that the terms "a" and "a plurality of" used in this disclosure are illustrative rather than restrictive, and those skilled in the art should understand that, unless otherwise expressly indicated in the context, they should be understood as "one or more".

[0018] The names of messages or information exchanged between multiple devices in the embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of such messages or information.

[0019] This disclosure will now be described in detail with reference to the accompanying drawings and embodiments.

[0020] Figure 1 A flow 100 of some embodiments of the device configuration method based on time allocation coefficients and industry feature extraction according to this disclosure is shown. This device configuration method based on time allocation coefficients and industry feature extraction includes the following steps: Step 101: In response to receiving a cleaning instruction for the target emission reduction area, obtain the time characteristic data information corresponding to the target emission reduction area.

[0021] In some embodiments, in response to receiving a cleaning instruction for a target emission reduction area, the executing entity (e.g., a computing device) of the equipment configuration method based on time allocation coefficients and industry characteristic extraction can obtain time characteristic data information corresponding to the target emission reduction area. The target emission reduction area can represent an area requiring cleaning. The cleaning instruction can represent an instruction triggered by a user terminal to clean the target emission reduction area. The time characteristic data information includes area source data sets and point source data sets. The time characteristic data information can represent data obtained after feature processing of activity level data collected at the current moment. The activity level data can represent data quantifying the scale and intensity of human economic activities, production processes, or natural phenomena. The activity level data can include data from various industries. For example, activity level data can be data from the transportation industry, total social electricity consumption, or monthly output data of enterprises. The area source data in the area source data set can represent data on pollution activities that are dispersed at a regional scale and difficult to monitor individually. The point source data in the point source data set can represent data on pollution activities whose specific locations and emission characteristics can be clearly identified. The area source data in the aforementioned area source dataset and the point source data in the aforementioned point source dataset correspond to different industries. In practice, firstly, the implementing entities can acquire the source data of the aforementioned point source dataset and area source dataset through automatic pollutant emission monitoring equipment, business databases, road network traffic monitoring systems, electricity consumption information collection systems, and satellite remote sensing. The aforementioned automatic pollutant emission monitoring equipment can be a CEMS (Consumer Energy Management System). The aforementioned business database can be an industrial enterprise operation monitoring platform. The area source data and point source data in the aforementioned area source dataset and point source dataset are obtained by classifying all data included in the aforementioned source dataset using a preset pollution source classification system. The aforementioned preset pollution source classification system can be the national economic industry classification. The area source data and point source data in the aforementioned area source dataset and point source dataset correspond to different categories. Next, according to the aforementioned preset pollution source classification system, all data included in the aforementioned source dataset are classified to obtain various pollution sources. Each of the aforementioned pollution sources can be a point source or an area source. Then, for each of the aforementioned pollution sources, the implementing entity can perform the following steps: First, based on the time scale corresponding to the industry characteristics of the pollution sources, the pollution sources are cleaned and aggregated to obtain the activity level characteristics corresponding to the pollution sources. For example, high-frequency road traffic data can be aggregated into hourly averages. Then, the activity level characteristics and the spatial allocation proxy data corresponding to the pollution sources are determined as point source data or area source data. The spatial allocation proxy data can represent point source coordinates, road network density, or population distribution maps. Finally, the obtained point source data and area source data are determined as time feature data information.

[0022] Optionally, before step 101, the aforementioned execution entity can first crawl and organize activity level data that meets preset selection criteria. These preset selection criteria can be data from the three months preceding the current time in the selection calendar. In practice, the execution entity can first crawl activity level data that meets the preset selection criteria via an API interface. Then, the raw activity data is converted into a tabular form using Python's Pandas library, resulting in an activity level data table. Next, the activity level data table is processed by calling a standard time series data processing library to obtain monthly, daily, and hourly aggregated value sets for each industry. The standard time series data processing function can be the `groupby()` function from Python's Pandas library. The monthly aggregated values ​​in the monthly aggregated value set represent the average value obtained by grouping the data of an industry in the activity level data by month and averaging the data within each group. The daily aggregated values ​​in the daily aggregated value set represent the average value obtained by grouping the data of an industry in the historical activity level data by day of the week and averaging the data within each group. The hourly aggregated values ​​in the aforementioned hourly aggregated value set represent the average values ​​obtained by grouping the data of an industry in the aforementioned historical activity level data by hour and averaging the data within each group. Then, each monthly aggregated value in the aforementioned monthly aggregated value set, each daily aggregated value in the aforementioned daily aggregated value set, and each hourly aggregated value in the aforementioned hourly aggregated value set are normalized to obtain monthly allocation coefficient sets, daily allocation coefficient sets, and hourly allocation coefficient sets. The monthly allocation coefficient in the aforementioned monthly allocation coefficient set represents the value obtained by dividing the monthly aggregated value corresponding to the aforementioned monthly allocation coefficient by the sum of all monthly aggregated values. The daily allocation coefficient in the aforementioned daily allocation coefficient set represents the value obtained by dividing the daily type aggregated value corresponding to the aforementioned daily allocation coefficient by the average of each daily type aggregated value. The hourly allocation coefficient in the aforementioned hourly allocation coefficient set represents the value obtained by dividing the hourly type aggregated value corresponding to the aforementioned hourly allocation coefficient by the average of each hourly type aggregated value. Finally, the obtained monthly allocation coefficients, daily allocation coefficients, and hourly allocation coefficients are determined as the base time allocation coefficients for that industry. Finally, using the national pollution source census technical guidelines or industry best practice manuals, the data on key activity levels for the corresponding industries are identified as initial industry activity characteristic indicators. For example, initial industry activity characteristic indicators could be industrial activity levels based on electricity consumption or traffic activity levels based on vehicle flow.

[0023] Step 102: For each point source data in the point source data set, perform the following steps: Step 1021: Based on the point source data and the time allocation coefficient of the corresponding point source data, dynamically adjust the model to determine the point source time allocation coefficient.

[0024] In some embodiments, the aforementioned execution entity can determine the point source time allocation coefficients based on the aforementioned point source data and the corresponding time allocation coefficients of the aforementioned point source data, dynamically adjusting the model. The aforementioned dynamic adjustment model of time allocation coefficients can represent a model pre-trained for the aforementioned point source data or the aforementioned area source data. The aforementioned dynamic adjustment model of time allocation coefficients can take point source data or area source data as input and output the monthly emission coefficients corresponding to the aforementioned point source data or area source data. The aforementioned dynamic adjustment model of time allocation coefficients can be an LSTM-based model. The aforementioned dynamic adjustment model of time allocation coefficients can adopt a batch training method, using mean squared error as the loss function, employing the Adam optimizer for backpropagation to update network parameters, with the initial learning rate set to 0.001 and the training epochs set to 100 epochs. The aforementioned dynamic adjustment model of time allocation coefficients can include an input layer, an LSTM layer, a fully connected layer, and an output layer (the output dimension can be the number of months, 12). The aforementioned input layer is used to receive point source data or area source data. The aforementioned LSTM layer, connected after the input layer, consists of two stacked LSTM units. Each LSTM unit includes a forget gate, an input gate, an output gate, and a unit state, used to extract time-dependent features from the input data. The hidden state dimension is set to 128. The aforementioned fully connected layer, connected after the LSTM layer, maps the 128-dimensional feature vector output by the LSTM layer to a 64-dimensional feature vector, using ReLU activation. The aforementioned output layer, connected after the fully connected layer, is a linearly activated fully connected layer with an output dimension of 12, corresponding to the emission coefficients from January to December. The aforementioned monthly emission coefficient can represent the normalized ratio of the actual activity intensity of the aforementioned point source data in the current month to the corresponding baseline monthly activity level, or the normalized ratio of the actual activity intensity of the aforementioned area source data in the current month to the corresponding baseline monthly activity level. The aforementioned baseline monthly activity level can represent the monthly average activity level of the aforementioned point source data or the aforementioned area source data over a complete year. The aforementioned point source time allocation coefficient can represent a dynamically changing weight factor corresponding to the aforementioned point source data.

[0025] In some optional implementations of certain embodiments, the execution entity can determine the point source time allocation coefficients by dynamically adjusting the model based on the point source data and the corresponding time allocation coefficients of the point source data through the following steps: The first step is to determine the monthly emission coefficients for the aforementioned point source data based on the point source data and the corresponding dynamic adjustment model for time allocation coefficients. In practice, the implementing entity can input the point source data into the corresponding dynamic adjustment model for time allocation coefficients to obtain the monthly emission coefficients for the aforementioned point source data.

[0026] The second step involves determining the point source time allocation coefficients based on the aforementioned monthly emission coefficients and the obtained daily and hourly emission coefficients corresponding to the point source data. The daily emission coefficients represent weighting factors indicating the regular fluctuations in pollution emissions from the point source or area source data on different days within a week. The hourly emission coefficients represent weighting factors indicating the regular fluctuations in pollution emissions from the point source or area source data within a 24-hour day. In practice, the implementing entity first obtains the daily and hourly emission coefficients corresponding to the point source data from a standard database. This standard database can be TCSES 144-2024. Then, the daily, hourly, and monthly emission coefficients corresponding to the point source data are input into the point source time allocation function to obtain the point source time allocation coefficients. Finally, these point source time allocation coefficients are replaced with the initial time allocation coefficients corresponding to the current time. The aforementioned initial time allocation coefficient can be the product of the monthly allocation coefficient, daily allocation coefficient, and hourly allocation coefficient corresponding to the current time, which are included in the aforementioned basic time allocation coefficient, and the ratio of 4.345238.

[0027] As an example, the point source time allocation function mentioned above can be: .

[0028] in, It can be used as a time allocation coefficient for point sources. It can be a monthly emission factor. It can be the daily emission factor. It can be the hourly emission factor. This can be a weekly average. The weekly average above could be 4.345238.

[0029] Step 1022: Determine industry activity characteristics based on point source data and the corresponding industry activity feature extraction model.

[0030] In some embodiments, the aforementioned execution entity can determine industry activity features based on the aforementioned point source data and the corresponding industry activity feature extraction model. The aforementioned industry activity feature extraction model can represent a model pre-trained on the aforementioned point source data capable of extracting industry activity features. The aforementioned industry activity feature extraction model can take the aforementioned point source data as input and output the corresponding industry activity features. The aforementioned industry activity features can represent the features obtained after feature extraction from the aforementioned point source data through the aforementioned industry activity feature extraction model. For example, in the transportation industry, industry activity features can be the average vehicle speed of the road network and the traffic congestion index. The aforementioned industry activity feature extraction model can be an LSTM-based model. The aforementioned industry activity feature extraction model can include an input layer, an LSTM layer, a Dropout layer, a fully connected layer, and an output layer. The input layer receives point source data. The LSTM layer, connected to the input layer, captures temporal dependencies in the point source data. The Dropout layer, connected to the LSTM layer, randomly discards some neurons during training with a preset probability (e.g., 0.5) to prevent overfitting. The fully connected layer, connected to the Dropout layer, integrates and reduces the dimensionality of the temporal dependencies extracted by the LSTM layer. The output layer, connected to the fully connected layer, outputs the industry activity features corresponding to the point source data. The industry activity feature extraction model can be trained in batches, using mean squared error or mean absolute error as the loss function. An adaptive moment estimation optimizer updates the model parameters with an initial learning rate of 0.001, setting the batch size to 64 and the maximum training epochs to 150. The dimension of the output layer can be determined by the number of categories of industry activity features. For example, the output dimension for the transportation industry is 2, corresponding to the average vehicle speed and traffic congestion index; the output dimension for the industrial industry is 1, corresponding to the production load rate.

[0031] Step 103: Determine the obtained point source time allocation coefficients as a set of point source time allocation coefficients, and determine the obtained industry activity feature information as a set of industry activity feature information.

[0032] In some embodiments, the aforementioned executing entity may determine the obtained point source time allocation coefficients as a set of point source time allocation coefficients, and determine the obtained industry activity feature information as a set of industry activity feature information.

[0033] Step 104: Dynamically adjust the model based on the source data set and the time allocation coefficient corresponding to each source data in the source data set to determine the source time allocation coefficient set.

[0034] In some embodiments, the aforementioned executing entity can determine a set of surface source time allocation coefficients based on the aforementioned surface source data set and the time allocation coefficients corresponding to each surface source data in the aforementioned surface source data set, dynamically adjusting the model. The surface source time allocation coefficients in the aforementioned set of surface source time allocation coefficients can represent a dynamically changing weight factor corresponding to the aforementioned surface source data. It should be noted that there can be a one-to-one correspondence between the surface source data in the surface source data set and the surface source time allocation coefficients in the aforementioned set of surface source time allocation coefficients. It should also be noted that the method for determining the surface source time allocation coefficients based on the aforementioned surface source data and the corresponding dynamic adjustment model of the time allocation coefficients for the aforementioned surface source data is the same as the method for determining the point source time allocation coefficients based on the aforementioned point source data and the corresponding dynamic adjustment model of the time allocation coefficients for the aforementioned point source data, and will not be repeated here; please refer to steps 1021-1022.

[0035] Step 105: Based on the point source data set, area source data set, point source time allocation coefficient set, industry activity characteristic information set, and area source time allocation coefficient set, generate the point source dynamic emission data set and the area source dynamic emission data set.

[0036] In some embodiments, the executing entity may generate a point source dynamic emission data set and a area source dynamic emission data set based on the point source data set, the area source data set, the point source time allocation coefficient set, the industry activity characteristic information set, and the area source time allocation coefficient set. The point source dynamic emission data in the point source dynamic emission data set can represent the pollutant emissions of the corresponding point source data within the current hour. The area source dynamic emission data in the area source dynamic emission data set can represent the pollutant emissions of the corresponding area source data within the current hour.

[0037] In some optional implementations of certain embodiments, the aforementioned executing entity may generate a point source dynamic emission data set and a surface source dynamic emission data set based on the aforementioned point source data set, the aforementioned area source data set, the aforementioned point source time allocation coefficient set, the aforementioned industry activity characteristic information set, and the aforementioned area source time allocation coefficient set through the following steps: The first step is to determine the set of annual emissions from point sources and the set of annual emissions from area sources based on the aforementioned set of point source data and the set of annual emissions from area sources. The annual emissions from point sources in the set of annual emissions from point sources can represent the annual emissions corresponding to the aforementioned point source data. The annual emissions from area sources in the set of annual emissions from area sources can represent the annual emissions corresponding to the aforementioned area source data. In practice, firstly, the implementing entity can obtain the set of annual emissions from point sources from a pre-set database. This pre-set database can be CEMS data. Then, the activity level corresponding to each annual emission from area sources in the set of annual emissions from area sources is obtained. Next, the emission factor corresponding to each annual emission from area sources in the set of annual emissions from area sources is obtained through a target guideline. This target guideline can be the IPCC National Greenhouse Gas Inventory Guidelines. Finally, each activity level is multiplied by its corresponding emission factor to obtain the set of annual emissions from area sources. It should be noted that the annual emissions from point sources in the set of annual emissions from point sources can have a one-to-one correspondence with the point source data in the aforementioned set of point source data. The annual emissions from non-point sources in the above set of annual emissions can correspond one-to-one with the non-point source data in the above set of non-point source data.

[0038] The second step involves generating dynamic emission datasets for both point sources and areas based on the aforementioned set of annual point source emissions, annual area source emissions, time allocation coefficients for point sources, time allocation coefficients for areas, and industry activity characteristic information. In practice, for each point source data point in the dataset, the executing entity can perform the following steps: First, determine the ratio of the annual point source emissions corresponding to the data point to the total number of hours throughout the year as the basic point source emissions. The total number of hours throughout the year can be 8760. Next, input the industry activity characteristic information into a pre-trained machine learning model to obtain the feature adjustment factor corresponding to the point source data. The pre-trained machine learning model can be a Gradient Boosting Decision Tree (GBDT) model. The pre-trained machine learning model can use the industry activity characteristic information as input and the feature adjustment factor as output. The aforementioned machine learning model can employ an additive regression tree architecture, consisting of an input layer (receiving feature vectors composed of industry activity feature information), an iterative ensemble layer (generating multiple CART regression trees sequentially, each CART regression tree fitting the residual between the current model's predicted value and the true value, where the current model's predicted value is obtained by summing the initial constant (set as the mean of the training set sample labels) and the outputs of all previous trees), and an output accumulation layer (summing the initial constant and the leaf node weights of each CART regression tree output by multiplying them by the learning rate). Each CART regression tree contains a root node, internal nodes, and leaf nodes. During training, the aforementioned machine learning model can use mean squared error as the loss function and employ early stopping to terminate training if the validation set loss does not decrease for 10 consecutive rounds. Then, the product of the aforementioned point source basic emissions, the corresponding point source time allocation coefficient, and the corresponding feature adjustment factor is determined as the dynamic emission data information of the aforementioned point source data. Next, the obtained dynamic emission data information is determined as a dynamic emission data information set. Then, for each area source data in the aforementioned area source data set, the executing entity may perform the following steps: First, determine the ratio of the annual area source emissions to the total number of hours throughout the year as the basic area source emissions. Then, determine the dynamic area source emissions data for the annual area source emissions by multiplying the basic area source emissions by the corresponding area source time allocation coefficient. Finally, determine the resulting set of dynamic area source emissions data as a comprehensive set of dynamic area source emissions data.

[0039] Step 106: Generate an overall dynamic inventory based on the point source dynamic emission data set and the area source dynamic emission data set.

[0040] In some embodiments, the executing entity can generate an overall dynamic inventory based on the aforementioned point source dynamic emission data set and the aforementioned area source dynamic emission data set. This overall dynamic inventory can represent a dynamic inventory that integrates the dynamic emission data from each point source and each area source. This dynamic inventory can represent an inventory that is continuously adjusted and updated over time.

[0041] In some optional implementations of certain embodiments, the aforementioned executing entity may generate an overall dynamic inventory based on the aforementioned point source dynamic emission data set and the aforementioned area source dynamic emission data set through the following steps: The first step is to construct a unified spatial grid covering the aforementioned target emission reduction areas. This unified spatial grid represents the same grid to which the dynamic emission data from each point source and each area source can be mapped. The unified spatial grid may include multiple grid cells. In practice, the implementing entity can write scripts using open-source libraries to create the unified spatial grid. Such open-source libraries can be Python's GeoPandas.

[0042] The second step involves spatially allocating the dynamic emission data from each point source based on the unified spatial grid and the set of geographic location information corresponding to each point source dynamic emission data, thereby obtaining the point source emission contribution data for each grid cell. The geographic location information in the aforementioned set of geographic location information can represent the geographic location (latitude and longitude) of the dynamic emission data from point sources or area sources. The point source emission contribution data can represent the sum of all point source dynamic emission data mapped to the corresponding grid cell. In practice, firstly, the executing entity can match the geographic location information corresponding to each point source dynamic emission data with the unified spatial grid to obtain the grid cell corresponding to each point source dynamic emission data. Then, for each grid cell in the unified spatial grid, the sum of all point source dynamic emission data corresponding to that grid cell is determined as the point source emission contribution data for that grid cell. Finally, the obtained point source emission contribution data is determined as the point source emission contribution data for each grid cell.

[0043] The third step involves spatially allocating the various non-point source dynamic emission data according to the unified spatial grid and the corresponding spatial representation data sets, thereby obtaining the non-point source emission contribution data for each grid unit. The spatial representation data in the aforementioned spatial representation data set can represent the geographical location and spatial relationships of the non-point source dynamic emission data. The non-point source emission contribution data can represent the sum of the non-point source dynamic emission data mapped to the corresponding grid unit. In practice, firstly, for each non-point source dynamic emission data, the executing entity can perform the following steps: First, select the spatial representation data corresponding to the non-point source dynamic emission data using a preset guideline. This preset guideline can be a technical guideline for emission inventory compilation. For example, if the non-point source type of the non-point source dynamic emission data is traffic non-point source, the corresponding spatial representation data can be road network density; if the non-point source type is residential non-point source, the corresponding spatial representation data can be population density data. Then, for each grid unit in the unified spatial grid, perform the following steps: First, normalize the spatial representation data of the grid unit to obtain the weight of the non-point source type corresponding to the grid unit. For example, spatial representation data could be road network density. If the total road length within grid cell C3 is 5 kilometers, and the total road length in the city is 1000 kilometers, then the weight of the area source type corresponding to grid cell C3 could be 0.005. Then, the product of the aforementioned area source dynamic emission data and the aforementioned weight is determined as the emission contribution data for the corresponding grid cell. Thus, the emission contribution data for each area source dynamic emission data corresponding to each grid cell in the aforementioned unified spatial grid can be obtained. Next, for each grid cell in the aforementioned unified spatial grid, the sum of the emission contribution data for each area source dynamic emission data corresponding to that grid cell is determined as the area source emission contribution data for that grid cell. Finally, the obtained area source emission contribution data is determined as the area source emission contribution data for each grid cell.

[0044] The fourth step involves generating a comprehensive dynamic emission inventory based on the point source emission contribution data and area source emission contribution data for each grid cell. In practice, firstly, the executing entity can determine the total emission contribution data for the corresponding grid cell by summing the point source emission contribution data and the area source emission contribution data for each grid cell. Then, a data processing library is used to encapsulate each grid cell in the unified spatial grid and its corresponding total emission contribution data. This data processing library can be the Python Pandas library or the NetCDF library.

[0045] Step 107: Based on the overall dynamic inventory, adaptively configure the cleaning equipment in the target emission reduction area to drive the cleaning equipment to perform cleaning operations in the target emission reduction area.

[0046] In some embodiments, the executing entity can adaptively configure the cleaning equipment in the target emission reduction area based on the overall dynamic inventory to drive the cleaning equipment to perform cleaning operations in the target emission reduction area. The cleaning equipment can be a sweeper truck, a mist cannon truck, or a water sprinkler truck. In practice, firstly, the executing entity can obtain the total emission contribution data of the target emission reduction area within a preset time window from the overall dynamic inventory. The time window can be 1-2 hours prior to the current time. Then, for each total emission contribution data, in response to determining that the total emission contribution data is greater than or equal to the target threshold, the grid cell corresponding to the total emission contribution data is identified as a hotspot area, and the main pollutant types and exceedance multiples of the hotspot area are obtained from the overall dynamic inventory. The exceedance multiple can represent the ratio between the emission intensity of the pollutant type and the target threshold. Here, the specific value of the target threshold is not limited and can be adjusted according to actual needs. For example, the target threshold can be 30 μg / (m²·h). The types of pollutants mentioned above can be PM2.5, PM2.5, etc. 10 And NOx. Next, for each of the above hotspot areas, the following steps are performed: First, in response to determining that the pollutant type of the above hotspot area is PM2.5 or PM2.5... 10 The system matches sweeper trucks or mist cannon trucks; in response to determining that the pollutant type in the aforementioned hotspot area is NOx, it matches electric sweeper trucks. Next, the product of the area of ​​the aforementioned hotspot area and the multiple by which the hotspot area exceeds the standard is determined as a first target value. Then, the product of the operational efficiency corresponding to the selected cleaning equipment in the aforementioned hotspot area and the available operational time is determined as a second target value. Next, the ratio between the first target value and the second target value is determined as a third target value. Then, the third target value is rounded up to obtain the number of cleaning equipment corresponding to the aforementioned hotspot area. Finally, for each of the aforementioned hotspot areas, the type and required quantity of cleaning equipment corresponding to the aforementioned hotspot area are selected from the cleaning equipment resource pool of the aforementioned target emission reduction area to drive the cleaning equipment to perform cleaning operations. The aforementioned cleaning equipment resource pool can represent the storage of each configurable cleaning equipment within the aforementioned target emission reduction area.

[0047] Optionally, after step 106, the aforementioned executing entity may also perform the following steps: The first step involves verifying the overall dynamic inventory and the acquired monitoring results information to obtain verification result information. The monitoring results information can represent directly measured environmental concentration data. The environmental concentration data can represent the measured concentration of a certain pollutant or substance in the environment. The verification result information represents the result obtained after evaluating the overall dynamic inventory using the monitoring results information. In practice, firstly, the implementing entity can obtain monitoring results information from ambient air quality monitoring stations. Then, it obtains the simulated values ​​of the monitoring points and time points corresponding to the monitoring results information from the overall dynamic inventory. Next, in response to determining that the root mean square error between the monitoring results information and the simulated values ​​is less than a preset error threshold, the verification is considered successful and recognized as verification result information. Here, the specific value of the preset error threshold is not limited; for example, the preset error threshold can be 0.5.

[0048] The second step involves updating the time allocation coefficient dynamic adjustment model and the industry activity feature extraction model in response to the determination that the above verification results meet the preset update conditions. The preset update conditions can be that the verification results of the overall dynamic list pass the verification. In practice, the executing entity can update the time allocation coefficient dynamic adjustment model and the industry activity feature extraction model by optimizing the training dataset, adjusting input features, tuning model hyperparameters, or introducing post-processing correction coefficients.

[0049] In addressing the technical problems mentioned above, and specifically for the second application scenario—where provinces and cities are required to report key industry emission data to the Ministry of Ecology and Environment monthly to dynamically determine compliance—generating a dynamic inventory often presents the following technical challenges: model updates rely on periodic full retraining, consuming significant computational resources and lengthening the update cycle. Furthermore, the mixing of old and new data during retraining causes the model to forget historical patterns, compromising the long-term comparability of the dynamic inventory data. This reduces the real-time performance of full training, which cannot adapt to frequent updates, resulting in poor accuracy of the generated dynamic inventory. Consequently, this leads to a waste of cleaning resources and increased energy consumption of cleaning equipment. Considering the following requirements for this application scenario—adapting to high-frequency dynamic updates, low overhead, and long-term model service stability—we have decided to adopt the following solution: Optionally, after step 107, the update steps for the above-mentioned dynamic adjustment model of time allocation coefficients and the above-mentioned industry activity feature extraction model are as follows: First, in response to determining that the updated emission data meets the preset update trigger condition, a set of representative samples that meet the preset selection condition is selected from the aforementioned activity level data, and the corresponding set of actual monthly emission coefficients and the set of actual industry activity characteristics are stored in the historical sample buffer. The updated emission data can represent newly added activity level data. The actual monthly emission coefficients in the set of actual monthly emission coefficients can represent the actual monthly emission coefficients corresponding to the industry. The actual industry activity characteristics in the set of actual industry activity characteristics can represent the actual industry activity characteristics corresponding to the industry. The preset update trigger condition can be that the update cycle of the updated emission data meets a preset time range. Here, the specific range of the preset time range is not limited. For example, the preset time range can be 0 to 30 days. The preset selection condition can be that the data with the largest time span are selected from the aforementioned historical activity level data. The representative samples in the set of representative samples can represent the data selected from the aforementioned activity level data that meet the preset selection condition. The historical sample buffer can represent a buffer capable of storing the aforementioned representative samples and the corresponding emission coefficient set and industry activity characteristic set. In practice, firstly, the aforementioned implementing entity can invoke a preset sample selection algorithm to filter out a set of representative samples that meet preset selection criteria from historical activity level data. This preset sample selection algorithm can be a clustering-based algorithm. Next, for each representative sample in the aforementioned representative sample set, the sample is input into the aforementioned dynamic adjustment model for time allocation coefficients and the aforementioned industry activity feature extraction model to obtain monthly emission coefficients and industry activity features. Then, the representative samples, monthly emission coefficients, and industry activity features are stored in a structured form in the aforementioned historical sample buffer.

[0050] Next, based on the updated emission data and the representation samples in the historical sample buffer, a time allocation coefficient residual model and an industry feature residual model are determined. The time allocation coefficient residual model represents a lightweight machine learning model used to learn and predict the difference between the ideal time allocation coefficient corresponding to the updated emission data and the coefficient predicted by the dynamic adjustment model of the time allocation coefficient. The industry feature residual model represents a lightweight machine learning model used to learn and extract new feature patterns that can supplement and enhance the characteristics of industry activities. In practice, firstly, the executing entity can initialize a lightweight neural network as the initial time allocation coefficient residual model. The input layer dimension, output layer dimension, and structure of the initial time allocation coefficient residual model are the same as those of the dynamic adjustment model of the time allocation coefficient. Next, based on the initial time allocation coefficient residual model, the following update steps are performed: First, based on each sample of the corresponding industry included in the updated emission data, the following steps are performed: First, the sample is input into the dynamic adjustment model of the time allocation coefficient to obtain the target predicted coefficient. Then, the sample is input into the time allocation coefficient residual model to obtain the target residual value. Next, the sum of the target predicted coefficient and the target residual value is determined as the first predicted value. Then, using the mean squared error function, the first loss value of the sample is determined based on the first predicted value and the actual time allocation coefficient of the sample. Finally, the product of the mean of all the obtained first loss values ​​and a first preset value is determined as the update coefficient loss. Here, the specific value of the first preset value is not limited. For example, the first preset value can be 0.47. Next, for each representative sample corresponding to the aforementioned industry included in the historical sample buffer, the following steps are performed: First, the representative sample is input into the dynamic adjustment model of the time allocation coefficient to obtain the target predicted coefficient. Then, the representative sample is input into the residual model of the time allocation coefficient to obtain the target residual value. Next, the sum of the target predicted coefficient and the target residual value is determined as the second predicted value. Then, using the mean squared error function, the second loss value of the representative sample is determined based on the second predicted value and the actual time allocation coefficient of the representative sample. Finally, the product of the mean of all the obtained second loss values ​​and a second preset value is determined as the historical coefficient loss. Here, the specific value of the second preset value is not limited. For example, the second preset value can be 0.53. Then, the updated coefficient loss and the historical coefficient loss are determined as the target coefficient loss. In response to the determination that the update count of the initial time allocation coefficient residual model does not meet the preset round condition, the initial time allocation coefficient residual model is updated using a gradient descent algorithm (such as Adam) to obtain the updated initial time allocation coefficient residual model.Then, the updated initial time allocation coefficient residual model is determined as the initial time allocation coefficient residual model, and the above update step is executed again. In response to the determination that the target coefficient loss for each of the 10 consecutive rounds does not meet the preset update condition, the initial time allocation coefficient residual model corresponding to the smallest target coefficient loss among the aforementioned target coefficient losses is determined as the time allocation coefficient residual model. The preset update condition can characterize that the average decrease in the target coefficient loss of adjacent rounds among the aforementioned target coefficient losses is less than or equal to a third preset value. Here, the specific value of the third preset value is not limited. For example, the third preset value can be 1e-5. In response to the determination that the target coefficient loss for each of the 10 consecutive rounds meets the above preset update condition, the initial time allocation coefficient residual model corresponding to the target coefficient loss for the last round among the aforementioned target coefficient losses is updated using a gradient descent algorithm (such as Adam) to obtain the updated initial time allocation coefficient residual model. Then, the updated initial time allocation coefficient residual model is determined as the initial time allocation coefficient residual model, and the above update step is executed again. The aforementioned execution entity can obtain the aforementioned industry feature residual model in the following way: First, initialize a lightweight neural network as the initial industry feature residual model. The input layer dimension and structure of the aforementioned initial industry feature residual model are the same as those of the aforementioned activity feature extraction model. Next, based on the aforementioned initial industry feature residual model, perform the following modification steps: First, based on each sample corresponding to the aforementioned industry included in the aforementioned updated emission data, perform the following steps: First, input the aforementioned sample into the aforementioned activity feature extraction model to obtain the target prediction feature. Then, input the aforementioned sample into the aforementioned initial industry feature residual model to obtain the target residual feature. Next, concatenate the aforementioned target prediction feature and the aforementioned target residual feature to obtain the target concatenated feature. Afterward, input the aforementioned target concatenated feature into a preset auxiliary industry classifier to obtain the predicted probability distribution of the aforementioned sample. The aforementioned preset auxiliary industry classifier can be a multilayer perceptron. The structure of the aforementioned preset auxiliary industry classifier can be: an input layer (the dimension of which is the sum of the dimensions of the target prediction feature and the target residual feature), a fully connected layer, a ReLU activation function, and an output layer (the dimension of which is the total number of industry categories). Secondly, the second reconstruction loss value of the above samples is determined using the cross-entropy loss function. Then, the cosine similarity between the above target predicted features and the above target residual features is negatively calculated to obtain the second distillation loss value. Next, the sum of the above second reconstruction loss value and the above second distillation loss value is determined as the second loss value. Finally, the product of the mean of each obtained second loss value and the above first preset value is determined as the update feature loss. Next, for each representation sample corresponding to the above industry included in the above historical sample buffer, the following steps are performed: First, the above representation sample is input into the above activity feature extraction model to obtain the target predicted features.Then, the aforementioned representation samples are input into the aforementioned initial industry feature residual model to obtain the target residual features. Next, the aforementioned target predicted features and the aforementioned target residual features are concatenated to obtain the target concatenated features. Then, the aforementioned target concatenated features are input into a preset auxiliary industry classifier to obtain the predicted probability distribution of the aforementioned representation samples. Next, the second reconstruction loss value of the aforementioned samples is determined using the cross-entropy loss function. Then, the cosine similarity between the aforementioned target predicted features and the aforementioned target residual features is negatively evaluated to obtain the second distillation loss value. Then, the sum of the aforementioned second reconstruction loss value and the aforementioned second distillation loss value is determined as the second loss value. Finally, the product of the mean of each obtained second loss value and a second preset value is determined as the historical feature loss. Then, the sum of the aforementioned update coefficient loss and the aforementioned historical coefficient loss is determined as the target feature loss. In response to the determination that the number of changes to the aforementioned initial industry feature residual model does not meet the preset round condition, the aforementioned initial industry feature residual model is updated using a gradient descent algorithm (such as Adam) to obtain the updated initial industry feature residual model. Then, the updated initial industry feature residual model is determined as the initial industry feature residual model, and the above change steps are executed again. In response to the determination that the target feature loss in each of the 10 consecutive rounds does not meet the preset update condition, the initial industry feature residual model corresponding to the target feature loss with the smallest target coefficient loss among the above targets is determined as the initial industry feature residual model. The above preset update condition can characterize that the average decrease in the target feature loss in adjacent rounds among the above target coefficient losses is less than or equal to a fourth preset value. Here, the specific value of the third preset value is not limited. For example, the third preset value can be 1e-5. In response to the determination that the target feature loss in each of the 10 consecutive rounds meets the above preset update condition, the initial industry feature residual model corresponding to the target feature loss in the last round among the above targets is updated using a gradient descent algorithm (such as Adam) to obtain the updated initial industry feature residual model. Then, the updated initial industry feature residual model is determined as the initial industry feature residual model, and the above change steps are executed again.

[0051] Then, the above-mentioned dynamic adjustment model of time allocation coefficients is determined as the coefficient teacher model, and the above-mentioned industry activity feature extraction model is determined as the feature teacher model.

[0052] Secondly, the aforementioned coefficient teacher model and the aforementioned time allocation coefficient residual model are fused, and the aforementioned feature teacher model and the aforementioned industry feature residual model are fused to obtain a coefficient joint prediction model and a feature joint prediction model. The coefficient joint prediction model can represent a logical model that combines the output of the aforementioned coefficient teacher model with the output of the aforementioned time allocation coefficient residual model. The feature joint prediction model can represent a logical model that combines the output of the aforementioned feature teacher model with the output of the aforementioned industry feature residual model. In practice, firstly, the implementing entity can input the aforementioned historical activity level data or the aforementioned updated emission data into the aforementioned coefficient teacher model to obtain a first predicted value. Then, the aforementioned historical activity level data or the aforementioned updated emission data is input into the aforementioned time allocation coefficient residual model to obtain a first residual value. Next, the sum of the aforementioned first predicted value and the aforementioned first residual value is determined as the output of the aforementioned coefficient joint prediction model. Secondly, the aforementioned historical activity level data or the aforementioned updated emission data is input into the aforementioned feature teacher model to obtain a first feature vector. Then, the aforementioned historical activity level data or the aforementioned updated emission data is input into the aforementioned industry feature residual model to obtain a second feature vector. Finally, the first feature vector and the second feature vector are concatenated to obtain the output of the joint feature prediction model.

[0053] Subsequently, based on the initial allocation coefficient model, the initial industry activity feature extraction model, the aforementioned joint prediction model of coefficients, and the aforementioned joint prediction model of features obtained from initialization, the target allocation coefficient model and the target feature extraction model are determined. The initial allocation coefficient model can represent the lightweight model obtained from initialization. The initial industry activity feature extraction model can represent a lightweight temporal feature extractor. The target allocation coefficient model can represent the model obtained after training the initial allocation coefficient model using the aforementioned joint prediction model of features. The target feature extraction model can represent the model obtained after training the initial industry activity feature extraction model using the aforementioned joint prediction model of features. In practice, firstly, the aforementioned execution entity can determine the network structure of the initial allocation coefficient model as the sequentially connected long short-term memory network layer, random deactivation layer, and fully connected output layer, obtaining the first network structure. Then, the sequentially connected long short-term memory network layer, attention aggregation layer, and fully connected output layer are determined as the network structure of the initial industry activity feature extraction model, obtaining the second network structure. Next, model instances corresponding to the aforementioned first network structure and the aforementioned second network structure are created, obtaining the initial allocation coefficient model and the initial industry activity feature extraction model. The parameters of the first and second network structures are obtained through random initialization. Next, updated emission data is input into the joint prediction model and the initial allocation coefficient model, and the difference between the predicted value output by the initial allocation coefficient model and the predicted value output by the joint prediction model is used as the first loss value. Then, the parameters of the initial allocation coefficient model are updated only through backpropagation to minimize the first loss value. Next, in response to determining that the first loss value is minimized in the target round, the initial allocation coefficient model is determined as the target allocation coefficient model. Here, the specific value of the target round is not limited; for example, the target round can be 200. Accordingly, the implementation of the target feature extraction model can be obtained through the following steps: First, updated emission data is input into the joint prediction model and the initial industry activity feature extraction model, and the difference between the predicted value output by the initial industry activity feature extraction model and the predicted value output by the joint prediction model is used as the second loss value. Next, in response to determining that the second loss value is minimized in the target round, the initial industry activity feature extraction model is determined as the target feature extraction model.

[0054] Finally, the aforementioned target allocation coefficient model and target feature extraction model are replaced with the aforementioned time allocation coefficient dynamic adjustment model and industry activity feature extraction model. Based on the time allocation coefficient dynamic adjustment model and industry activity feature extraction model, the cleaning equipment in the aforementioned target emission reduction area is adaptively configured to drive the cleaning equipment to perform cleaning operations in the aforementioned target emission reduction area. For specific implementation methods, please refer to the specific implementation methods of steps 102 to 107.

[0055] The above technical solution, as an inventive point of this disclosure, solves technical problem two: "It leads to the consumption of a large amount of computing resources and a prolonged update cycle, reduces the real-time performance of full training that cannot adapt to high-frequency updates, resulting in poor accuracy of the generated dynamic inventory, thus wasting cleaning resources and increasing the energy consumption of cleaning equipment." The reasons for this are as follows: Model updates rely on periodic full retraining, which consumes a large amount of computing resources and prolongs the update cycle. Furthermore, during retraining, the mixing of old and new data causes the model to forget historical patterns, destroying the long-term comparability of dynamic inventory data, reducing the real-time performance of full training that cannot adapt to high-frequency updates, resulting in poor accuracy of the generated dynamic inventory, thus wasting cleaning resources and increasing the energy consumption of cleaning equipment. To achieve this effect, the dynamic inventory implementation method disclosed herein employs a historical sample buffer to prevent knowledge forgetting, combines a lightweight time-allocation coefficient residual model and an industry-specific residual model to avoid full parameter updates, and utilizes knowledge distillation technology to fuse and compress new and old knowledge into an efficient model. This allows the model to adapt in real-time to monthly or even higher frequency data updates while ensuring continuous performance optimization. Consequently, it reduces computational resource consumption, shortens the update cycle, improves the real-time performance of adapting full training to high-frequency updates, enhances the accuracy of the generated dynamic inventory, and ultimately reduces the energy consumption of cleaning equipment.

[0056] In addressing the technical challenges of the aforementioned background technologies, and specifically for Scenario 3—emergency source tracing and collaborative decision-making for sudden cross-border air pollution transmission events—generating a comprehensive dynamic inventory often presents the following technical problems: The inability to continuously and with high resolution quantitatively track and predict the dynamic evolution of cross-border pollution sources (such as overseas wildfires or accidental emissions) necessitates reliance on historical data comparisons or single static models for rough assessments. This leads to a disconnect between simulated decision-making results and reality, loss of spatial details, and reduced accuracy of predictions across different regions and time periods. The lag and low resolution of this prediction method cause policy-making to rely on experience-based judgments, increasing trial-and-error costs and decision-making risks, thereby increasing the cycle and cost of scientific policy implementation and consuming significant computational resources and time. Considering the following requirements for this application scenario—adapting to high precision, high perception, and high visualization feedback—we have decided to adopt the following solution: Optionally, after step 106, the implementing entity can first acquire meteorological forecast data, production plan data, and traffic control plan data for the target emission reduction area that meet preset acquisition conditions. The preset acquisition conditions can be acquiring meteorological forecast data, production plan data, and traffic control plan data for a preset time period after the current time. Here, the specific range of the preset time period is not limited; for example, the preset time period can be 0~48h. The meteorological forecast data can represent the weather conditions predicted within the next 48 hours through meteorological observation and numerical simulation. The weather conditions can be temperature, wind speed, and precipitation. The production plan data can represent data used by the enterprise to plan, arrange, and control production activities during the production process. The production plan data can include production tasks, production progress, and material requirements. The traffic control plan data can represent various data used to plan, implement, and monitor traffic management measures. The traffic control plan data can include traffic flow, road condition information, and traffic light timing. In practice, the implementing entity can first acquire meteorological forecast data through the official website of the meteorological bureau. Then, it acquires production plan data through the MES system. Finally, obtain traffic control plan data from the official website of the traffic management department.

[0057] Next, based on the aforementioned meteorological forecast data, production plan data, traffic control plan data, and a pre-trained activity level prediction model, the activity level data of various pollution sources in the target emission reduction area are determined. The pre-trained activity level prediction model can be an LSTM model. The activity level data represents the activity level of the corresponding pollution sources. These pollution sources can be industrial, transportation, or residential. The pre-trained activity level prediction model takes the aforementioned meteorological forecast data, production plan data, and traffic control plan data as input and outputs the activity level data of various pollution sources. For example, if the pollution source is an industrial source and the corresponding production plan data is production line shutdown for maintenance, then the activity level data for the industrial source is 0. The pre-trained activity level prediction model can include an input layer, an LSTM layer, a fully connected layer, and an output layer (output dimension: the number of pollution source types). In practice, firstly, the implementing entity can use one-hot encoding to transform the traffic control plan data to obtain a basic encoding vector. Then, the target emission reduction area is divided into multiple geographic grids. Then, according to the specific rules of the aforementioned traffic control plan (such as restricted areas, vehicle types, and time periods), and combined with the road attributes and land use type information of each of the multiple geographic grids, the intensity coefficient of the impact of the control plan on the aforementioned geographic grids is determined. Next, the intensity coefficient is added as a new dimension and concatenated with the aforementioned basic coding vector to obtain a multi-dimensional numerical vector corresponding to each geographic grid. Then, for each of the multiple geographic grids, the corresponding meteorological forecast data, production plan data, and multi-dimensional numerical vector are input into the pre-trained activity level prediction model to obtain the activity level data of various pollution sources in the aforementioned geographic grids. Finally, the obtained activity level data of various pollution sources from multiple geographic grids are determined as the activity level data of various pollution sources in the target emission reduction area.

[0058] Then, based on the obtained activity level data of various pollution sources, the aforementioned meteorological forecast data, and the parameterized emission factor model, a set of parameterized emission factors is determined. The parameterized emission factor model can be a model that takes the aforementioned meteorological forecast data and key parameters corresponding to different pollution sources during the emission process as input, and outputs the parameterized emission factors of various pollution sources. For example, if the pollution source is a motor vehicle source, the corresponding key parameter could be average vehicle speed. If the pollution source is a dust source, the corresponding key parameters could be road surface humidity and dust load. The parameterized emission factor model can be a machine learning model based on gradient boosting trees. For example, the gradient boosting tree machine learning model could be XGBoost. The parameterized emission factor model can be composed of a preset number of decision trees ensembled. Here, the specific value of the preset number is not limited; for example, the preset number could be 100. The parameterized emission factor model can employ the gradient boosting algorithm, using mean squared error as the loss function, iteratively training multiple decision trees to fit the residuals, and using cross-validation to select the optimal hyperparameters (learning rate, maximum tree depth, minimum number of samples per leaf node). The parameterized emission factors in the aforementioned set of parameterized emission factors can characterize the parameterized emission factors of pollutants corresponding to each of the aforementioned pollution sources. In practice, firstly, for each type of pollution source, the implementing entity can perform the following steps: First, by querying a preset pollution source correspondence table, the various pollutants corresponding to the aforementioned pollution sources are determined. The preset pollution source correspondence table can characterize a table including various pollution sources and their corresponding pollutants. Then, the obtained pollutants, the key parameters corresponding to the aforementioned pollution sources, and the aforementioned meteorological forecast data are input into the aforementioned parameterized emission factor model to obtain the dynamic emission factors of the various pollutants included for the aforementioned pollution sources. Next, the obtained dynamic emission factors are determined as parameterized emission factor groups. Finally, the obtained parameterized emission factor groups are determined as the parameterized emission factor set.

[0059] Secondly, based on the obtained activity level data of various pollution sources and the aforementioned set of parameterized emission factors, the set of pollutant emissions for the target emission reduction area within a future preset time period is determined. The pollutant emissions in this set represent the emissions of each pollution source. The aforementioned preset time period represents a preset time period after the current moment. In practice, firstly, for each type of pollution source obtained, the implementing entity can perform the following steps: First, for each pollutant corresponding to the pollution source, the product of the pollution source's activity level data and the pollutant's dynamic emission factor is determined as the pollutant's emission amount. Then, each emission amount is determined as the emission set corresponding to the pollution source. Next, the obtained emission sets are determined as the total emission set. Then, the sum of each emission amount for the same pollutant in the total emission set is determined as the pollutant emission amount for that pollutant. Finally, the obtained pollutant emission amounts for each pollutant are determined as the pollutant emission set.

[0060] Subsequently, based on the aforementioned set of pollutant emissions, a sudden predicted emission inventory for the target area is generated. This sudden predicted emission inventory represents the emission inventory predicted for a future preset time period after spatiotemporal allocation of the aforementioned set of pollutant emissions. In practice, firstly, the implementing entity can perform spatiotemporal allocation of each pollutant emission in the aforementioned set of pollutant emissions according to the type of each pollution source: First, each pollutant emission in the aforementioned set of pollutant emissions is allocated to a designated geospatial grid in the target area. For example, if the pollution source type is industrial point source, point allocation is based on its geographic coordinates. If the pollution source type is traffic source, grid allocation is based on road network data; if the pollution source type is residential or dust, spatial interpolation allocation is based on population density distribution maps and land use type maps. Then, using time profiles matching the types of pollution sources, each pollutant emission in the aforementioned set of pollutant emissions is allocated to various time slices within the aforementioned future preset time period. For example, hourly traffic flow change profiles are applied to traffic sources, and continuous or intermittent production profiles are applied to industrial sources. Next, the spatiotemporally allocated pollutant emission sets are integrated and encoded using a scientific data format to obtain a structured data file. This scientific data format can be NetCDF. Finally, metadata is embedded into the data file to obtain a dynamic emissions inventory. This metadata may include the activity level data used, the meteorological data source, and the timestamp used to generate the dynamic emissions inventory.

[0061] Secondly, based on the aforementioned emergency emission inventory and atmospheric diffusion model, diffusion information for the various pollution sources within the aforementioned future preset time period is generated. The atmospheric diffusion model can be a computational fluid dynamics (CFD) model. This diffusion information characterizes the diffusion paths, concentration distributions, and impact ranges of the various pollution sources within the aforementioned target emission reduction area. In practice, the implementing entity can input the aforementioned emergency emission inventory into the aforementioned atmospheric diffusion model to obtain the diffusion information for the various pollution sources within the aforementioned future preset time period.

[0062] Finally, based on the aforementioned diffusion information, a dynamic pollution plume map is generated. This pollution plume map characterizes the spatial distribution of plumes formed by the diffusion of various pollution sources in the atmosphere. In practice, the implementing entity can generate the dynamic pollution plume map using atmospheric data visualization software based on the aforementioned diffusion information. The atmospheric data visualization software can be VERDI.

[0063] The above-mentioned technical solution, as an inventive point of this disclosure, solves technical problem three: "The simulation decision-making results become disconnected from reality, spatial details are lost, the accuracy of the prediction results in different regions and time periods is reduced, the cost of trial and error and the risk of decision-making are increased, and thus the cycle and cost of scientific policy implementation are increased, consuming a large amount of computing resources and time." The reasons for this disconnect and loss of spatial details in the simulation decision-making results, the increased cost of trial and error and the risk of decision-making, and the increased cycle and cost of scientific policy implementation, consuming a large amount of computing resources and time, are as follows: It is impossible to continuously and with high resolution quantitatively track and predict the dynamic evolution of cross-border pollution sources (such as overseas wildfires and accidental emissions). Only historical data comparison or a single static model can be used for rough evaluation, leading to a disconnect between the simulation decision-making results and reality, loss of spatial details, and reduced accuracy of the prediction results in different regions and time periods. The lag and low resolution of this prediction method cause the policy-making process to rely on experience-based judgment, increasing the cost of trial and error and the risk of decision-making, thus increasing the cycle and cost of scientific policy implementation, consuming a large amount of computing resources and time. To achieve this effect, the dynamic inventory implementation method disclosed herein integrates production planning data, traffic control plan data, and meteorological forecast data to drive an activity level prediction model and a parameterized emission factor model, generating a future emission inventory with high spatiotemporal resolution. This inventory is then coupled with an atmospheric diffusion model for rapid simulation and visualization, thereby enabling the prediction of pollution trends and the quantitative assessment of the effectiveness of measures. This improves the accuracy of prediction results across different regions and time periods, reduces trial-and-error costs and decision-making risks, and ultimately shortens the cycle and cost of scientific policy implementation, while reducing computational resources and time consumption.

[0064] The above embodiments of this disclosure have the following beneficial effects: the equipment configuration method based on time allocation coefficient and industry feature extraction of some embodiments of this disclosure can reduce the redundancy of system configuration, improve the accuracy of dynamic inventory, reduce the waste of cleaning resources and the energy consumption of cleaning equipment; and reduce the complexity of the system, improve the coherence, consistency and stability of the overall dynamic inventory, thereby enabling accurate and efficient environmental decision-making. Specifically, the high redundancy in system configuration leads to poor accuracy of dynamic inventories, resulting in longer processing times and higher resource consumption. It also reduces the overall coherence, consistency, and stability of the dynamic inventory, making it difficult to support accurate and efficient environmental decision-making. This is because for each industry type and even each emission source (point or area source), static parameters and calculation interfaces must be preset based on experience. This results in high system redundancy and difficulty adapting to the dynamic and ever-changing production patterns of different industries. The preset static parameters and templates are updated laggingly, failing to respond to the spatiotemporal fluctuations in production activities and emissions. This leads to poor accuracy in the generated dynamic inventory, resulting in insufficient cleaning efforts in high-emission areas and wasted cleaning resources in low-emission areas, consuming more cleaning equipment and energy. Furthermore, it can only perform independent calculations for a single type of point or area source, failing to integrate multi-source data and perform unified dynamic integration based on industry characteristics. This reduces the overall coherence, consistency, and stability of the dynamic inventory, making it difficult to support accurate and efficient environmental decision-making. Based on this, some embodiments of the device configuration method based on time allocation coefficients and industry feature extraction disclosed herein firstly acquire time feature data information in response to receiving a cleaning instruction for a target emission reduction area. This yields the time feature data information. Next, for each point source data in the aforementioned point source data set, the following steps are performed: First, based on the aforementioned point source data and a dynamic adjustment model of the corresponding time allocation coefficients, a point source time allocation coefficient is determined. This yields the point source time allocation coefficient. Then, based on the aforementioned point source data and a dynamic adjustment model of the corresponding industry activity feature extraction model, industry activity feature information is determined. This yields the industry activity feature information. Next, the obtained point source time allocation coefficients are defined as a point source time allocation coefficient set, and the obtained industry activity feature information is defined as an industry activity feature information set. This yields the point source time allocation coefficient set and the industry activity feature information set. Finally, for the aforementioned area source data set and a dynamic adjustment model of the time allocation coefficients corresponding to each area source data in the aforementioned area source data set, an area source time allocation coefficient set is determined. This yields the area source time allocation coefficient set. Subsequently, based on the aforementioned point source data set, area source data set, point source time allocation coefficient set, industry activity characteristic information set, and area source time allocation coefficient set, a point source dynamic emission data set and an area source dynamic emission data set are generated. Thus, the point source dynamic emission data set and the area source dynamic emission data set can be obtained.Next, based on the aforementioned dynamic emission data sets from point sources and area sources, a comprehensive dynamic inventory is generated. This yields the overall dynamic inventory. Finally, based on this comprehensive dynamic inventory, the cleaning equipment in the target emission reduction areas is adaptively configured to drive the equipment to perform cleaning operations in those areas. This reveals the type and quantity of cleaning equipment in the target emission reduction areas. Because this approach does not employ a fragmented, pre-defined static calculation process for point and area sources, but rather uses a unified machine learning framework to fuse real-time data with historical features and perform dynamic calculations, the consistency of point and area source calculation results across time and business logic is improved. This enhances the reliability of the overall dynamic inventory, shortens the time and intervention costs from data access to inventory generation, and reduces system configuration redundancy, improves the accuracy of the dynamic inventory, reduces waste of cleaning resources and energy consumption of cleaning equipment, and lowers system complexity. It also improves the coherence, consistency, and stability of the overall dynamic inventory, enabling precise and efficient environmental decision-making.

[0065] Further reference Figure 2 As an implementation of the methods shown in the above figures, this disclosure provides some embodiments of an equipment configuration device based on time allocation coefficients and industry feature extraction. These device embodiments are similar to... Figure 1 Corresponding to the method embodiments shown, the device can be specifically applied to various electronic devices.

[0066] like Figure 2As shown, a device configuration apparatus 200 based on time allocation coefficients and industry feature extraction in some embodiments includes: an acquisition unit 201, an execution unit 202, a first determination unit 203 and a second determination unit 204, a first generation unit 205, a second generation unit 206, and a configuration unit 207. The acquisition unit 201 is configured to acquire time feature data information in response to receiving a cleaning instruction for a target emission reduction area, wherein the time feature data information includes a set of area source data and a set of point source data; the execution unit 202 is configured to perform the following steps for each point source data in the point source data set: determining the point source time allocation coefficient based on the point source data and a dynamic adjustment model of the corresponding time allocation coefficient; determining industry activity feature information based on the point source data and an industry activity feature extraction model of the corresponding point source data; the first determination unit 203 is configured to determine the obtained point source time allocation coefficients as a set of point source time allocation coefficients and the obtained industry activity feature information as a set of industry activity feature information; the second determination unit 204… The system is configured to dynamically adjust the model for the aforementioned area source data set and the time allocation coefficient corresponding to each area source data in the aforementioned area source data set, and determine the area source time allocation coefficient set; the first generation unit 205 is configured to generate a point source dynamic emission data set and an area source dynamic emission data set based on the aforementioned point source data set, the aforementioned area source data set, the aforementioned point source time allocation coefficient set, the aforementioned industry activity characteristic information set, and the aforementioned area source time allocation coefficient set; the second generation unit 206 is configured to generate an overall dynamic list based on the aforementioned point source dynamic emission data set and the aforementioned area source dynamic emission data set; the configuration unit 207 is configured to adaptively configure the cleaning equipment in the aforementioned target emission reduction area based on the aforementioned overall dynamic list, so as to drive the aforementioned cleaning equipment to perform cleaning operations in the aforementioned target emission reduction area.

[0067] It is understandable that the units and references recorded in the equipment configuration device 200 based on time allocation coefficients and industry characteristics extraction are... Figure 1 The steps in the described method correspond to each other. Therefore, the operations, features, and beneficial effects described above for the method also apply to the dome camera position information generation device 200 and the units contained therein, and will not be repeated here.

[0068] The following is for reference. Figure 3 It shows a schematic diagram of the structure of an electronic device (such as a computing device) 300 suitable for implementing some embodiments of the present disclosure. Figure 3 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments of this disclosure.

[0069] like Figure 3As shown, the electronic device 300 may include a processing unit (e.g., a central processing unit, a graphics processing unit, etc.) 301, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 302 or a program loaded from a storage device 308 into a random access memory (RAM) 303. The RAM 303 also stores various programs and data required for the operation of the electronic device 300. The processing unit 301, ROM 302, and RAM 303 are interconnected via a bus 304. An input / output (I / O) interface 305 is also connected to the bus 304.

[0070] Typically, the following devices can be connected to I / O interface 305: input devices 306 including, for example, touchscreens, touchpads, keyboards, mice, cameras, microphones, accelerometers, gyroscopes, etc.; output devices 307 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 308 including, for example, magnetic tapes, hard disks, etc.; and communication devices 309. Communication device 309 allows electronic device 300 to communicate wirelessly or wiredly with other devices to exchange data. Although Figure 3 An electronic device 300 with various devices is shown; however, it should be understood that it is not required to implement or possess all of the devices shown. More or fewer devices may be implemented or possessed alternatively. Figure 3 Each box shown can represent a device or multiple devices as needed.

[0071] In particular, according to some embodiments of this disclosure, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, some embodiments of this disclosure include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication device 309, or installed from storage device 308, or installed from ROM 302. When the computer program is executed by processing device 301, it performs the functions defined in the methods of some embodiments of this disclosure.

[0072] It should be noted that, in some embodiments of this disclosure, the computer-readable medium may be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium may be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In some embodiments of this disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In some embodiments of this disclosure, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium can be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wires, optical fibers, RF (radio frequency), etc., or any suitable combination thereof.

[0073] In some implementations, clients and servers can communicate using any currently known or future-developed network protocol such as HTTP (Hypertext Transfer Protocol) and can interconnect with digital data communication (e.g., communication networks) of any form or medium. Examples of communication networks include local area networks (“LANs”), wide area networks (“WANs”), the Internet (e.g., the Internet of Things), and peer-to-peer networks (e.g., ad hoc peer-to-peer networks), as well as any currently known or future-developed networks.

[0074] The aforementioned computer-readable medium may be included in the aforementioned electronic device; or it may exist independently and not assembled into the electronic device. The aforementioned computer-readable medium carries one or more programs that, when executed by the electronic device, cause the electronic device to...

[0075] Computer program code for performing operations of some embodiments of this disclosure can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0076] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0077] The units described in some embodiments of this disclosure can be implemented in software or hardware. The described units can also be housed in a processor; for example, a processor may be described as including an acquisition unit, an execution unit, a first determination unit, a second determination unit, a first generation unit, a second generation unit, and a configuration unit. The names of these units do not necessarily limit the specific unit; for example, the acquisition unit may also be described as "a unit that acquires time characteristic data information in response to receiving a cleaning instruction for a target emission reduction area."

[0078] The functions described above in this document can be performed at least in part by one or more hardware logic components. For example, exemplary types of hardware logic components that can be used, without limitation, include: field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), system-on-a-chip (SoCs), complex programmable logic devices (CPLDs), and so on.

[0079] The above description is merely a selection of preferred embodiments of this disclosure and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of the invention involved in the embodiments of this disclosure is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-described inventive concept. For example, technical solutions formed by substituting the above-described features with (but not limited to) technical features with similar functions disclosed in the embodiments of this disclosure.

Claims

1. A device configuration method based on time allocation coefficients and industry characteristic extraction, comprising: In response to receiving a cleaning instruction for a target emission reduction area, the system acquires time characteristic data information corresponding to the target emission reduction area, wherein the time characteristic data information includes a set of area source data and a set of point source data. For each point source data in the aforementioned point source data set, perform the following steps: Based on the point source data and the corresponding time allocation coefficient of the point source data, the dynamic adjustment model is used to determine the point source time allocation coefficient. Based on the point source data and the corresponding industry activity feature extraction model for the point source data, industry activity feature information is determined. The obtained point source time allocation coefficients are defined as the point source time allocation coefficient set, and the obtained industry activity feature information is defined as the industry activity feature information set. For the set of surface data and the dynamic adjustment model of the time allocation coefficient corresponding to each surface data in the set of surface data, determine the set of surface time allocation coefficients; Based on the point source data set, the area source data set, the point source time allocation coefficient set, the industry activity characteristic information set, and the area source time allocation coefficient set, a point source dynamic emission data set and an area source dynamic emission data set are generated; Based on the point source dynamic emission data set and the area source dynamic emission data set, an overall dynamic inventory is generated; Based on the overall dynamic inventory, the cleaning equipment in the target emission reduction area is adaptively configured to drive the cleaning equipment to perform cleaning operations in the target emission reduction area.

2. The method of claim 1, wherein, The method further includes: Based on the overall dynamic list and the acquired monitoring results, the overall dynamic list is verified to obtain verification results. In response to determining that the verification result information meets the preset update conditions, the dynamic adjustment model of the time allocation coefficient and the industry activity feature extraction model are updated.

3. The method of claim 1, wherein, The dynamic adjustment model based on the point source data and the corresponding time allocation coefficients of the point source data determines the point source time allocation coefficients, including: Based on the point source data and the dynamic adjustment model of the time allocation coefficient corresponding to the point source data, the monthly emission coefficient corresponding to the point source data is determined. The point source time allocation coefficient is determined based on the monthly emission coefficient, the daily emission coefficient, and the hourly emission coefficient corresponding to the acquired point source data.

4. The method according to claim 1, wherein, The process of generating a dynamic emission data set for point sources and a dynamic emission data set for area sources based on the point source data set, the area source data set, the point source time allocation coefficient set, the industry activity characteristic information set, and the area source time allocation coefficient set includes: Based on the point source data set and the area source data set, determine the annual emission set of point sources and the annual emission set of area sources; Dynamic emission data information is generated based on the set of annual emissions from point sources, the set of annual emissions from area sources, the set of time allocation coefficients for point sources, and the set of time allocation coefficients for area sources.

5. A device configuration apparatus based on time allocation coefficients and industry characteristic extraction, comprising: The acquisition unit is configured to acquire time characteristic data information in response to receiving a cleaning instruction for the target emission reduction area, wherein the time characteristic data information includes a set of area source data and a set of point source data; The execution unit is configured to perform the following steps for each point source data in the point source data set: dynamically adjust the model based on the point source data and the time allocation coefficient corresponding to the point source data to determine the point source time allocation coefficient; and determine industry activity feature information based on the point source data and the industry activity feature extraction model corresponding to the point source data. The first determining unit is configured to determine the obtained point source time allocation coefficients as a set of point source time allocation coefficients and to determine the obtained industry activity feature information as a set of industry activity feature information. The second determining unit is configured to dynamically adjust the model for the area source data set and the time allocation coefficient corresponding to each area source data in the area source data set, and determine the area source time allocation coefficient set. The first generation unit is configured to generate a dynamic emission data set of point sources and a dynamic emission data set of area sources based on the point source data set, the area source data set, the point source time allocation coefficient set, the industry activity feature information set, and the area source time allocation coefficient set. The second generation unit is configured to generate an overall dynamic inventory based on the point source dynamic emission data set and the area source dynamic emission data set; The configuration unit is configured to adaptively configure the cleaning equipment in the target emission reduction area based on the overall dynamic inventory, so as to drive the cleaning equipment to perform cleaning operations in the target emission reduction area.

6. An electronic device, comprising: One or more processors; A storage device on which one or more programs are stored; When the one or more programs are executed by the one or more processors, the one or more processors implement the method as described in any one of claims 1 to 4.

7. A computer-readable medium having a computer program stored thereon, wherein, When the computer program is executed by a processor, it implements the method as described in any one of claims 1 to 4.

Images