Communication big data-driven sewage plant peak clipping and valley filling low-carbon scheduling method

Abstract

The invention discloses a communication big data driven sewage plant peak clipping and valley filling low-carbon scheduling method, and belongs to the technical field of environmental engineering intelligent control. The purpose of quantitative control over the adaptive relation between the sewage generation capacity and the bearing capacity of a treatment facility is achieved. The method comprises the steps of constructing a sewage plant inflow load prediction model training sample data set by adopting a sliding time window mechanism based on a sewage plant low-carbon dispatching underlying database, and constructing a sewage plant inflow load prediction model based on LSTM to obtain the hourly inflow flow of a sewage plant. The method comprises the following steps: constructing a matching degree index used for describing the operation condition of a sewage system in any hour, and designing a scheduling scheme to maximize the total matching degree of a sewage plant in a prediction period as a target, and obtaining different matching degrees before and after the scheduling scheme; and calculating the carbon emission before and after the scheduling scheme by using the carbon source per hour, the dephosphorization and dehydration agent feeding amount and the hour-level power consumption during the operation period of the sewage plant, and evaluating the carbon reduction effect of the scheduling scheme.

Description

Technical Field

[0001] This invention belongs to the field of intelligent control technology in environmental engineering, specifically relating to a low-carbon scheduling method for wastewater treatment plants driven by communication big data. Background Technology

[0002] With the acceleration of urbanization, the demand for wastewater treatment continues to grow. As a crucial component of urban infrastructure, wastewater treatment plants, while ensuring water environment safety, also generate significant energy consumption. The wastewater treatment industry has become a major source of greenhouse gas emissions. Research shows that carbon emissions from wastewater treatment account for approximately 1% to 3% of global anthropogenic carbon emissions, and this trend is continuously rising. Developing green and low-carbon operation and maintenance technologies has become an essential path for the wastewater industry to promote synergistic efficiency in pollution reduction and carbon reduction. In daily scenarios, due to the combined effects of intercity commuting, inter-regional commuting, and water usage habits, daily influent flows have peak and trough periods. Wastewater treatment facilities operate inefficiently during trough periods and are overloaded or even overflow during peak periods. On the one hand, traditional wastewater treatment plants generally use regulating tanks to passively respond to the daily load "valley-shaped" fluctuations, making it impossible to predict load changes, leading to frequent pump start-ups and shutdowns, resulting in increased carbon emissions. On the other hand, the unit treatment carbon emission intensity of wastewater treatment plants is significantly negatively correlated with the matching degree between actual operating load and design load. Long-term deviation from design conditions will lead to a significant decrease in energy efficiency and generate a large amount of unnecessary carbon emissions.

[0003] However, existing operation and maintenance scheduling methods primarily focus on process stability, compliance with emission standards, and operating costs, while rarely considering carbon reduction as an optimization objective. The scheduling process lacks quantitative indicators to characterize the relationship between wastewater generation capacity and the carrying capacity of treatment facilities, and scheduling scheme design relies heavily on experience-based judgment, failing to achieve precise peak-shaving and valley-filling control. Influent load forecasting often depends on historical averages or simple statistical models, making it difficult to accurately capture load fluctuations caused by dynamic factors such as intercity commuting and regional population tidal flows, resulting in significant discrepancies between forecasts and actual influent conditions. These limitations collectively mean that the potential for low-carbon operation in wastewater treatment plants has not been fully explored. Summary of the Invention

[0004] The problem this invention aims to solve is the quantitative control of the matching relationship between sewage generation capacity and the carrying capacity of treatment facilities. It proposes a low-carbon scheduling method for sewage treatment plants driven by communication big data.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] A low-carbon scheduling method for wastewater treatment plants driven by big data communication includes the following steps:

[0007] S1. Collect hourly influent flow data of the wastewater treatment plant, hourly mobile phone signal population data of the service area where the wastewater treatment plant is located, meteorological monitoring data for the corresponding time period, hourly rainfall and temperature data of the weather station corresponding to the wastewater treatment plant, hourly carbon source, phosphorus removal, dehydration agent dosage and hourly power consumption during the operation of the wastewater treatment plant, process the collected data to obtain the underlying database of low-carbon scheduling of the wastewater treatment plant.

[0008] S2. Based on the hourly influent flow data of the wastewater treatment plant obtained in step S1, the hourly mobile phone signal population data of the service area where the wastewater treatment plant is located, and the meteorological monitoring data of the corresponding time period, a training sample dataset for the wastewater treatment plant influent load prediction model is constructed using a sliding time window mechanism.

[0009] S3. Construct a wastewater treatment plant influent load prediction model based on LSTM. Use the wastewater treatment plant influent load prediction model training sample dataset obtained in step S2 to train the wastewater treatment plant influent load prediction model. The trained wastewater treatment plant influent load prediction model is used to predict the hourly influent flow rate of the wastewater treatment plant.

[0010] S4. Based on the hourly influent flow rate of the wastewater treatment plant obtained in step S3, construct a matching degree index to describe the operation of the wastewater system at any hour, and then design a scheduling scheme with the goal of maximizing the total matching degree of the wastewater treatment plant in the predicted period.

[0011] S5. For the different matching degrees obtained before and after the scheduling scheme, calculate the carbon emissions before and after the scheduling scheme based on the hourly carbon source, phosphorus removal, dewatering agent dosage and hourly power consumption during the operation of the wastewater treatment plant obtained in step S1, and evaluate the carbon reduction effect of the scheduling scheme.

[0012] Furthermore, the data processing method in step S1 is data cleaning and preprocessing, which handles erroneous values, missing values, and outliers of various types of data, uniformly deletes erroneous values, fills missing values ​​with the mean of the same period, and, based on the box plot, replaces high outliers with the third quartile and replaces low outliers with the first quartile.

[0013] Furthermore, the specific implementation method of step S2 is to construct training samples using the historical data of more than 2 consecutive years after the processing of step S1, including hourly influent flow data of sewage treatment plants, hourly mobile phone signal population data of the service area where sewage treatment plants are located, and meteorological monitoring data of the corresponding time period. The input of each sample is set as influent volume, population and meteorological data of 168 consecutive hours, and the sample label is the actual influent volume of the following 168 hours. The temporal mapping relationship between input features and prediction target is established.

[0014] Furthermore, the specific implementation method of step S3 includes the following steps:

[0015] S3.1. The wastewater treatment plant influent load prediction model based on LSTM adopts a three-channel parallel input architecture. The influent channel is configured with a 128-unit LSTM layer to extract the temporal features of the influent; the population channel is configured with a 64-unit LSTM layer to capture the tidal flow pattern of the population; and the meteorological channel is configured with a 64-unit LSTM layer to characterize the impact of meteorological factors on the influent. The outputs of the three channels are fused with a fully connected layer after feature splicing to generate the hourly influent prediction sequence for the next 168 hours. The expression of the wastewater treatment plant influent load prediction model output is obtained as follows:

[0016]

[0017] in, Output of the wastewater treatment plant influent load prediction model The time-series input data are for water inflow, population, and meteorological data, respectively. This is the LSTM feature extraction layer for the influent channel. For LSTM feature extraction of population channels, This is the LSTM feature extraction layer for the meteorological channel. For feature splicing operations, It is a fully connected layer;

[0018] S3.2. Divide the training sample dataset of the wastewater treatment plant influent load prediction model obtained in step S2 into a training set, a validation set, and a test set in chronological order, with a ratio of 7:1.5:1.5.

[0019] S3.3. Use samples from the training set to train the model. Select the mean squared error as the loss function, and the optimization objective is to minimize the sum of squares of the deviations between the predicted and actual water volume. The trained wastewater treatment plant influent load prediction model is then used to predict the hourly influent flow rate of the wastewater treatment plant.

[0020] Furthermore, the specific implementation method of step S4 includes the following steps:

[0021] S4.1. Construct a matching index to describe the operating status of the wastewater system at any given hour, with the following expression:

[0022]

[0023] in, Let be the matching degree of the region at hour t. Let i be the predicted influent flow rate of wastewater treatment plant i in hour t. The hourly influent flow rate of wastewater treatment plant i is designed as follows. The design scale conversion factor for wastewater treatment plant i;

[0024] Considering the design load of all wastewater treatment plants in the region, the scale conversion factor for wastewater treatment plant i is calculated, yielding the following expression:

[0025]

[0026] in, The design influent flow rate for wastewater treatment plant i;

[0027] S4.2. The design scheduling scheme aims to maximize the overall matching degree of the wastewater treatment plants during the forecast period, resulting in the following expression:

[0028]

[0029] in, This indicates that the wastewater treatment plant has the highest overall matching degree within the scheduling cycle. Let be the matching degree of wastewater treatment plant i in hour t. It is the scheduling cycle.

[0030] Furthermore, the specific implementation method of step S5 includes the following steps:

[0031] S5.1. Carbon emissions are defined as including 3 direct carbon emissions and 4 indirect carbon emissions. Direct carbon emissions include emissions generated during the removal of organic matter in wastewater treatment, emissions generated during the removal of total nitrogen (TN) in wastewater treatment, and emissions generated by the complete oxidation of external carbon sources. Indirect carbon emissions include emissions generated by electricity consumption, emissions from phosphorus removal agents, emissions from dehydration agents, and emissions from external carbon sources.

[0032] S5.2. The indirect carbon emissions are calculated using the emission factor method for each item, resulting in the following expression:

[0033]

[0034] Where j represents the j-th type of indirect carbon emissions. For the activity level of the j-th type of indirect carbon emission source, is the emission factor for the j-th type of indirect carbon emission source;

[0035] Based on the above formula, the formula for calculating indirect carbon emissions is as follows:

[0036]

[0037] in, Carbon emissions from electricity consumption To reduce carbon emissions from phosphorus removal agents, Carbon emissions from dehydration agents Carbon emissions are generated to produce carbon sources;

[0038] S5.3. Based on the hourly influent load prediction curve output by the trained wastewater treatment plant influent load prediction model obtained in step S3, design a scheduling scheme for the scheduling target based on the scheduling scheme in step S4. Based on identifying load peaks and valleys, lower the water level of the storage tank before the predicted peak period to temporarily store wastewater exceeding the treatment capacity to smooth out the peak; during the predicted valley period, release the stored wastewater to fill the load valley.

[0039] S5.4. Calculate the carbon emissions corresponding to the matching degree before and after scheduling to evaluate the carbon reduction effect of the scheduling scheme. The calculation formula is as follows:

[0040]

[0041]

[0042]

[0043] in, The total carbon emissions before the scheduling, The total carbon emissions after scheduling For direct carbon emissions before scheduling, For indirect carbon emissions before scheduling, For direct carbon emissions after scheduling. For indirect carbon emissions after scheduling, This refers to the amount of carbon emission reduction.

[0044] The beneficial effects of this invention are:

[0045] This invention presents a low-carbon scheduling method for wastewater treatment plants driven by big data communication, which constructs a quantifiable index for matching wastewater generation and treatment capacity (MDGT), fundamentally surpassing the limitations of traditional load rate evaluation. It can dynamically assess the fit between influent volume and designed treatment capacity at any given hourly scale. Traditional daily load rates only reflect the average situation throughout the day and cannot reveal instantaneous mismatches caused by intraday peak-valley fluctuations. MDGT, by comprehensively considering load deviations across all hours of the day, can more realistically and comprehensively depict the operating status of the wastewater treatment system, providing a scientific and accurate quantitative benchmark for scheduling optimization. An LSTM dynamic prediction model for influent load based on mobile phone signaling data has been established, achieving accurate prediction of hourly influent load for the next 7 days. The model integrates mobile phone signaling population data, meteorological data, and historical water volume data, effectively capturing load changes brought about by complex social activities such as cross-city commuting and regional population tides. Using the matching degree (MDGT) as a benchmark combined with high-precision load prediction, a refined "peak-shaving and valley-filling" scheduling strategy is formulated and implemented. This forward-looking and dynamic scheduling optimization ensures that the wastewater treatment system always operates within an optimal range that closely matches the design load, avoiding frequent equipment start-ups and shutdowns and inefficient operation. As a result, significant energy conservation and carbon reduction are achieved while ensuring treatment effectiveness.

[0046] The present invention discloses a communication big data-driven low-carbon scheduling method for wastewater treatment plants, which achieves peak shaving and valley filling of wastewater treatment volume through accurate prediction and intelligent control of influent load, thereby improving operational efficiency and realizing energy conservation and carbon reduction. Attached Figure Description

[0047] Figure 1 This is a flowchart of a low-carbon scheduling method for wastewater treatment plants driven by big data communication, as described in this invention.

[0048] Figure 2 This is a structural block diagram of a communication big data-driven low-carbon scheduling method for wastewater treatment plants.

[0049] Figure 3 This is an analysis curve showing the accuracy between actual and predicted values ​​of the LSTM-based wastewater treatment plant influent load prediction model of the present invention.

[0050] Figure 4 This is a water inflow prediction diagram for the present invention;

[0051] Figure 5 This is a schematic diagram illustrating the principle of allocation based on scheduling and allocation principles in this invention.

[0052] Figure 6 This is a diagram showing the scheduling results based on the scheduling allocation principle of this invention;

[0053] Figure 7This is a design diagram of the water storage tank scheme based on the scheduling and allocation principle of the present invention.

[0054] Figure 8 This is a comparison chart of the total carbon emission reductions achieved by the present invention based on the scheduling and allocation principle. Detailed Implementation

[0055] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only for explaining the invention and are not intended to limit the invention; that is, the described specific embodiments are merely a part of the embodiments of the invention, and not all of them. The components of the specific embodiments of the invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations, and the invention may also have other embodiments.

[0056] Therefore, the following detailed description of specific embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected specific embodiments of the invention. All other specific embodiments obtained by those skilled in the art based on these specific embodiments without inventive effort are within the scope of protection of this invention.

[0057] To further understand the invention's content, features, and effects, the following specific embodiments are provided, along with accompanying drawings. Figure 1 - Appendix Figure 8 Detailed explanation is as follows:

[0058] Example 1:

[0059] A low-carbon scheduling method for wastewater treatment plants driven by big data communication is characterized by the following steps:

[0060] S1. Collect hourly influent flow data of the wastewater treatment plant, hourly mobile phone signal population data of the service area where the wastewater treatment plant is located, meteorological monitoring data for the corresponding time period, hourly rainfall and temperature data of the weather station corresponding to the wastewater treatment plant, hourly carbon source, phosphorus removal, dehydration agent dosage and hourly power consumption during the operation of the wastewater treatment plant, process the collected data to obtain the underlying database of low-carbon scheduling of the wastewater treatment plant.

[0061] Furthermore, the data processing method in step S1 is data cleaning and preprocessing, which handles erroneous values, missing values, and outliers of various types of data, uniformly deletes erroneous values, fills missing values ​​with the mean of the same period, and, based on the box plot, replaces high outliers with the third quartile and replaces low outliers with the first quartile.

[0062] S2. Based on the hourly influent flow data of the wastewater treatment plant obtained in step S1, the hourly mobile phone signal population data of the service area where the wastewater treatment plant is located, and the meteorological monitoring data of the corresponding time period, a training sample dataset for the wastewater treatment plant influent load prediction model is constructed using a sliding time window mechanism.

[0063] Furthermore, the specific implementation method of step S2 is to construct training samples using the historical data of more than 2 consecutive years after the processing of step S1, including hourly influent flow data of sewage treatment plants, hourly mobile phone signal population data of the service area where sewage treatment plants are located, and meteorological monitoring data of the corresponding time period. The input of each sample is set as influent volume, population and meteorological data of 168 consecutive hours, and the sample label is the actual influent volume of the following 168 hours. The temporal mapping relationship between input features and prediction target is established.

[0064] S3. Construct a wastewater treatment plant influent load prediction model based on LSTM. Use the wastewater treatment plant influent load prediction model training sample dataset obtained in step S2 to train the wastewater treatment plant influent load prediction model. The trained wastewater treatment plant influent load prediction model is used to predict the hourly influent flow rate of the wastewater treatment plant.

[0065] Furthermore, the specific implementation method of step S3 includes the following steps:

[0066] S3.1. The wastewater treatment plant influent load prediction model based on LSTM adopts a three-channel parallel input architecture. The influent channel is configured with a 128-unit LSTM layer to extract the temporal features of the influent; the population channel is configured with a 64-unit LSTM layer to capture the tidal flow pattern of the population; and the meteorological channel is configured with a 64-unit LSTM layer to characterize the impact of meteorological factors on the influent. The outputs of the three channels are fused with a fully connected layer after feature splicing to generate the hourly influent prediction sequence for the next 168 hours. The expression of the wastewater treatment plant influent load prediction model output is obtained as follows:

[0067]

[0068] in, Output of the wastewater treatment plant influent load prediction model The time-series input data are for water inflow, population, and meteorological data, respectively. This is the LSTM feature extraction layer for the influent channel. For LSTM feature extraction of population channels, This is the LSTM feature extraction layer for the meteorological channel. For feature splicing operations, It is a fully connected layer;

[0069] Furthermore, the model employs a three-channel parallel input architecture: the inflow channel is configured with a 128-unit LSTM layer, which extracts long-term temporal dependency features of inflow through a gating mechanism; the population channel is configured with a 64-unit LSTM layer, which focuses on capturing the periodic patterns and temporal correlations of population tidal flows; and the meteorological channel is configured with a 64-unit LSTM layer, which is used to characterize the dynamic impact of meteorological factors such as temperature and rainfall on inflow. The output features of the three channels are concatenated to form a 256-dimensional fused feature vector (128+64+64), which is then mapped through a fully connected layer (containing a 128-dimensional hidden layer with ReLU activation) to generate hourly inflow prediction sequences for the next 168 hours.

[0070] S3.2. Divide the training sample dataset of the wastewater treatment plant influent load prediction model obtained in step S2 into a training set, a validation set, and a test set in chronological order, with a ratio of 7:1.5:1.5.

[0071] S3.3. Use samples from the training set to train the model. Select the mean squared error as the loss function, and the optimization objective is to minimize the sum of squares of the deviations between the predicted and actual water volume. The trained wastewater treatment plant influent load prediction model is then used to predict the hourly influent flow rate of the wastewater treatment plant.

[0072] Furthermore, during training, a validation set is used to monitor model performance. An early stopping method is implemented: training is stopped when the validation set loss no longer decreases significantly over several consecutive rounds, and the model parameters with the lowest validation set loss are saved to prevent overfitting. After training, the finally selected model is evaluated using an independent test set. Model inputs include: historical measured data (past seven days) including: actual hourly influent flow records of the wastewater treatment plant, actual hourly mobile phone signaling population records of the service area, actual hourly rainfall records of the corresponding weather station, and actual hourly temperature records of the corresponding weather station. Future projected data (next seven days) includes: hourly mobile phone signaling population projected values ​​of the service area (generated based on historical signaling data patterns, recent trends, or simple moving averages), hourly rainfall forecasts, and hourly temperature forecasts. Based on the above input data, the trained model outputs a sequence of hourly wastewater treatment plant influent flow predictions for the next 168 hours (7 days).

[0073] S4. Based on the hourly influent flow rate of the wastewater treatment plant obtained in step S3, construct a matching degree index to describe the operation of the wastewater system at any hour, and then design a scheduling scheme with the goal of maximizing the total matching degree of the wastewater treatment plant in the predicted period.

[0074] Furthermore, the specific implementation method of step S4 includes the following steps:

[0075] S4.1. Construct a matching index to describe the operating status of the wastewater system at any given hour, with the following expression:

[0076]

[0077] in, Let be the matching degree of the region at hour t. Let i be the predicted influent flow rate of wastewater treatment plant i in hour t. The hourly influent flow rate of wastewater treatment plant i is designed as follows. The design scale conversion factor for wastewater treatment plant i;

[0078] Considering the design load of all wastewater treatment plants in the region, the scale conversion factor for wastewater treatment plant i is calculated, yielding the following expression:

[0079]

[0080] in, The design influent flow rate for wastewater treatment plant i;

[0081] Furthermore, to clarify the operational status of a regional (urban) wastewater treatment plant at a specific moment (e.g., during peak water usage), an hourly evaluation is required. At this time, the influent load rate of the wastewater treatment plant becomes irrelevant and cannot depict hourly details. The Innovative Design Fit Index (MDGT, abbreviated as M) is used to describe the operational status of the wastewater system at any given hour.

[0082] S4.2. The design scheduling scheme aims to maximize the overall matching degree of the wastewater treatment plants during the forecast period, resulting in the following expression:

[0083]

[0084] in, This indicates that the wastewater treatment plant has the highest overall matching degree within the scheduling cycle. Let be the matching degree of wastewater treatment plant i in hour t. It is the scheduling cycle.

[0085] Furthermore, by combining the hourly influent load prediction curve output by the LSTM model, load peaks and valleys can be identified in advance. Based on this, a refined "peak shaving and valley filling" strategy can be formulated: before the predicted peak period, the water level of the storage tank is lowered to temporarily store the sewage that exceeds the treatment capacity to smooth out the peak; during the predicted valley period, the stored sewage is released to fill the load valley, so that the treatment system can operate stably in a more optimal load range.

[0086] S5. For the different matching degrees obtained before and after the scheduling scheme, calculate the carbon emissions before and after the scheduling scheme based on the hourly carbon source, phosphorus removal, dewatering agent dosage and hourly power consumption during the operation of the wastewater treatment plant obtained in step S1, and evaluate the carbon reduction effect of the scheduling scheme.

[0087] Carbon emissions are quantified based on the change in matching degree before and after "peak shaving and valley filling" scheduling.

[0088] Furthermore, the specific implementation method of step S5 includes the following steps:

[0089] S5.1. Carbon emissions are defined as including 3 direct carbon emissions and 4 indirect carbon emissions. Direct carbon emissions include emissions generated during the removal of organic matter in wastewater treatment, emissions generated during the removal of total nitrogen (TN) in wastewater treatment, and emissions generated by the complete oxidation of external carbon sources. Indirect carbon emissions include emissions generated by electricity consumption, emissions from phosphorus removal agents, emissions from dehydration agents, and emissions from external carbon sources.

[0090] S5.2. The indirect carbon emissions are calculated using the emission factor method for each item, resulting in the following expression:

[0091]

[0092] Where j represents the j-th type of indirect carbon emissions. For the activity level of the j-th type of indirect carbon emission source, is the emission factor for the j-th type of indirect carbon emission source;

[0093] Based on the above formula, the formula for calculating indirect carbon emissions is as follows:

[0094]

[0095] in, Carbon emissions from electricity consumption To reduce carbon emissions from phosphorus removal agents, Carbon emissions from dehydration agents Carbon emissions are generated to produce carbon sources;

[0096] S5.3. Based on the hourly influent load prediction curve output by the trained wastewater treatment plant influent load prediction model obtained in step S3, design a scheduling scheme for the scheduling target based on the scheduling scheme in step S4. Based on identifying load peaks and valleys, lower the water level of the storage tank before the predicted peak period to temporarily store wastewater exceeding the treatment capacity to smooth out the peak; during the predicted valley period, release the stored wastewater to fill the load valley.

[0097] S5.4. Calculate the carbon emissions corresponding to the matching degree before and after scheduling to evaluate the carbon reduction effect of the scheduling scheme. The calculation formula is as follows:

[0098]

[0099]

[0100]

[0101] in, The total carbon emissions before the scheduling, The total carbon emissions after scheduling For direct carbon emissions before scheduling, For indirect carbon emissions before scheduling, For direct carbon emissions after scheduling. For indirect carbon emissions after scheduling, This refers to the amount of carbon emission reduction.

[0102] The actual use case of this embodiment is as follows:

[0103] Taking a water purification plant in Shenzhen as an example, this paper describes the data collection and cleaning process at the plant. Table 1 below shows the five types of data collected, including historical data used for training and future prediction data used for model prediction.

[0104] Table 1

[0105]

[0106] The collected data underwent preprocessing using a three-stage cleaning strategy: erroneous values ​​were removed from clearly erroneous data. Missing values ​​were imputed; for example, missing population data for November 16, 2022, was imputed using the average of the two days before and after that date. Additionally, box plots were generated to identify and correct outliers.

[0107] An hourly wastewater load prediction model was constructed using Python. Pandas was used for data reading, and NumPy was employed for numerical computation. The final model was based on an LSTM. Historical time-series datasets were collected for model training, including hourly influent volumes from a water purification plant for two consecutive years, hourly mobile phone signal population data from the service area during the same period, and corresponding meteorological monitoring data (rainfall and temperature). A three-channel design (population, water volume, and meteorological data) was adopted to output hourly prediction curves. The predicted influent volumes were compared with actual values ​​to evaluate the model's stability. Figure 3 As shown;

[0108] Then, taking a water purification plant on November 10, 2022 as an example, the figure shows the hourly influent predicted by the prediction model, such as... Figure 4 As shown.

[0109] in accordance with Figure 5 The allocation is performed according to the schematic scheduling principle, and the matching degree is calculated based on the MDGT calculation formula to obtain the time points and flow rates of the inflow and release of the storage tank. Figure 6The actual scheduling results are shown in the diagram. The design load of Guanlan Water Purification Plant is 160,000 cubic meters / day. According to the formula, the matching degree before adjustment was 0.6964, and the matching degree after adjustment was 0.9577, which is an increase of 0.2613, and the relative improvement percentage is 37.52%.

[0110] The design of a regulation and storage tank operation plan is guided by calculated indicators to meet the designed allocation results. Specific plans include... Figure 7 The diagram illustrates how wastewater treatment plant operations and scheduling can achieve a state where the actual operating load is close to the design load, thereby improving operating efficiency and reducing carbon emissions.

[0111] An idealized scheduling test was conducted using all wastewater treatment plants in Shenzhen as the target, exploring the relationship between chemical consumption (phosphorus removal agents, carbon source agents, dehydration agents), power consumption, and matching degree. Indirect carbon emissions were calculated, and direct carbon emissions were calculated based on the ratio of indirect to direct carbon emissions. Finally, the total carbon emissions were obtained, and a comparison chart of carbon emissions before and after the adjustment in 2022 was plotted on a monthly basis, as shown below. Figure 8 As shown.

[0112] It should be noted that relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0113] Although this application has been described above with reference to specific embodiments, various modifications can be made and components can be replaced with equivalents without departing from the scope of this application. In particular, as long as there is no structural conflict, the features in the specific embodiments disclosed in this application can be combined with each other in any way. The lack of an exhaustive description of these combinations in this specification is merely for the sake of brevity and resource conservation. Therefore, this application is not limited to the specific embodiments disclosed herein, but includes all technical solutions falling within the scope of the claims.

Claims

1. A low-carbon scheduling method for wastewater treatment plants driven by big data communication, characterized in that, Includes the following steps: S1. Collect hourly influent flow data of the wastewater treatment plant, hourly mobile phone signal population data of the service area where the wastewater treatment plant is located, meteorological monitoring data for the corresponding time period, hourly rainfall and temperature data of the weather station corresponding to the wastewater treatment plant, hourly carbon source, phosphorus removal, dehydration agent dosage and hourly power consumption during the operation of the wastewater treatment plant, process the collected data to obtain the underlying database of low-carbon scheduling of the wastewater treatment plant. S2. Based on the hourly influent flow data of the wastewater treatment plant obtained in step S1, the hourly mobile phone signal population data of the service area where the wastewater treatment plant is located, and the meteorological monitoring data of the corresponding time period, a training sample dataset for the wastewater treatment plant influent load prediction model is constructed using a sliding time window mechanism. S3. Construct a wastewater treatment plant influent load prediction model based on LSTM. Use the wastewater treatment plant influent load prediction model training sample dataset obtained in step S2 to train the wastewater treatment plant influent load prediction model. The trained wastewater treatment plant influent load prediction model is used to predict the hourly influent flow rate of the wastewater treatment plant. S4. Based on the hourly influent flow rate of the wastewater treatment plant obtained in step S3, construct a matching degree index to describe the operation of the wastewater system at any hour, and then design a scheduling scheme with the goal of maximizing the total matching degree of the wastewater treatment plant in the predicted period. S5. For the different matching degrees obtained before and after the scheduling scheme, calculate the carbon emissions before and after the scheduling scheme based on the hourly carbon source, phosphorus removal, dewatering agent dosage and hourly power consumption during the operation of the wastewater treatment plant obtained in step S1, and evaluate the carbon reduction effect of the scheduling scheme.

2. The low-carbon scheduling method for wastewater treatment plants driven by big data communication as described in claim 1, characterized in that, The data processing method in step S1 is data cleaning and preprocessing, which handles erroneous values, missing values, and outliers of various data types, uniformly deletes erroneous values, fills missing values ​​with the mean of the same period, and replaces high outliers with the third quartile and low outliers with the first quartile based on the box plot.

3. A low-carbon scheduling method for wastewater treatment plants driven by big data communication, as described in claim 1 or 2, characterized in that... The specific implementation method of step S2 is to construct training samples using historical data of more than 2 consecutive years after the processing of step S1, including hourly influent flow data of sewage treatment plants, hourly mobile phone signal population data of the service area where the sewage treatment plant is located, and meteorological monitoring data of the corresponding time period. The input of each sample is set as influent volume, population and meteorological data of 168 consecutive hours, and the sample label is the actual influent volume of the following 168 hours. The temporal mapping relationship between input features and prediction target is established.

4. The low-carbon scheduling method for wastewater treatment plants driven by big data communication as described in claim 3, characterized in that, The specific implementation method of step S3 includes the following steps: S3.

1. The wastewater treatment plant influent load prediction model based on LSTM adopts a three-channel parallel input architecture. The influent channel is configured with a 128-unit LSTM layer to extract the temporal features of the influent; the population channel is configured with a 64-unit LSTM layer to capture the tidal flow pattern of the population; and the meteorological channel is configured with a 64-unit LSTM layer to characterize the impact of meteorological factors on the influent. The outputs of the three channels are fused with a fully connected layer after feature splicing to generate the hourly influent prediction sequence for the next 168 hours. The expression of the wastewater treatment plant influent load prediction model output is obtained as follows: ； in, Output of the wastewater treatment plant influent load prediction model The time-series input data are for water inflow, population, and meteorological data, respectively. This is the LSTM feature extraction layer for the influent channel. For LSTM feature extraction of population channels, This is the LSTM feature extraction layer for the meteorological channel. For feature splicing operations, It is a fully connected layer; S3.

2. Divide the training sample dataset of the wastewater treatment plant influent load prediction model obtained in step S2 into a training set, a validation set, and a test set in chronological order, with a ratio of 7:1.5:1.

5. S3.

3. Use samples from the training set to train the model. Select the mean squared error as the loss function, and the optimization objective is to minimize the sum of squares of the deviations between the predicted and actual water volume. The trained wastewater treatment plant influent load prediction model is then used to predict the hourly influent flow rate of the wastewater treatment plant.

5. The communication big data-driven low-carbon scheduling method for wastewater treatment plants, as described in claim 4, is characterized in that... The specific implementation method of step S4 includes the following steps: S4.

1. Construct a matching index to describe the operating status of the wastewater system at any given hour, with the following expression: ； in, Let be the matching degree of the region at hour t. Let i be the predicted influent flow rate of wastewater treatment plant i in hour t. The hourly influent flow rate for wastewater treatment plant i. The design scale conversion factor for wastewater treatment plant i; Considering the design load of all wastewater treatment plants in the region, the scale conversion factor for wastewater treatment plant i is calculated, yielding the following expression: ； in, The design influent flow rate for wastewater treatment plant i; S4.

2. The design scheduling scheme aims to maximize the overall matching degree of the wastewater treatment plants during the forecast period, resulting in the following expression: ； in, This indicates that the wastewater treatment plant has the highest overall matching degree within the scheduling cycle. Let be the matching degree of wastewater treatment plant i in hour t. It is the scheduling cycle.

6. The communication big data-driven low-carbon scheduling method for wastewater treatment plants, as described in claim 5, is characterized in that... The specific implementation method of step S5 includes the following steps: S5.

1. Carbon emissions are defined as including 3 direct carbon emissions and 4 indirect carbon emissions. Direct carbon emissions include emissions generated during the removal of organic matter in wastewater treatment, emissions generated during the removal of total nitrogen (TN) in wastewater treatment, and emissions generated by the complete oxidation of external carbon sources. Indirect carbon emissions include emissions generated by electricity consumption, emissions from phosphorus removal agents, emissions from dehydration agents, and emissions from external carbon sources. S5.

2. The indirect carbon emissions are calculated using the emission factor method for each item, resulting in the following expression: ； Where j represents the j-th type of indirect carbon emissions. For the activity level of the j-th type of indirect carbon emission source, is the emission factor for the j-th type of indirect carbon emission source; Based on the above formula, the formula for calculating indirect carbon emissions is as follows: ； in, Carbon emissions from electricity consumption Carbon emissions from phosphorus removal agents Carbon emissions from dehydration agents Carbon emissions are generated to produce carbon sources; S5.

3. Based on the hourly influent load prediction curve output by the trained wastewater treatment plant influent load prediction model obtained in step S3, design a scheduling scheme for the scheduling target based on the scheduling scheme in step S4. Based on identifying load peaks and valleys, lower the water level of the storage tank before the predicted peak period to temporarily store wastewater exceeding the treatment capacity to smooth out the peak; during the predicted valley period, release the stored wastewater to fill the load valley. S5.

4. Calculate the carbon emissions corresponding to the matching degree before and after scheduling to evaluate the carbon reduction effect of the scheduling scheme. The calculation formula is as follows: ； ； ； in, The total carbon emissions before the scheduling, The total carbon emissions after scheduling For direct carbon emissions before scheduling, For indirect carbon emissions before scheduling, For direct carbon emissions after scheduling. For indirect carbon emissions after scheduling, This refers to the amount of carbon emission reduction.

Images