Sewage pre-evacuation treatment method and device for drainage pipe network

Abstract

The invention relates to the technical field of sewage pre-evacuation, and discloses a sewage pre-evacuation treatment method and device for a drainage pipe network, and the method comprises the steps: obtaining a water level value, a tryptophan fluorescence intensity value, a humic acid-like acid content value, an ammonia nitrogen concentration value and a pipe network slope value of a forebay of a target pump station; based on a water level value, a tryptophan fluorescence intensity value, a humic acid-like acid content value, an ammonia nitrogen concentration value and a pipe network slope value of a forebay of a target pump station, a pipe network volume predicted value, a pollution ratio predicted value and an ammonia nitrogen concentration predicted value are determined by utilizing an LSTM model, and finally, an ammonia nitrogen concentration prediction value is calculated based on the pipe network volume predicted value, the pollution ratio predicted value and the ammonia nitrogen concentration predicted value. Dynamically selecting and controlling water pumping equipment to pre-evacuate sewage of the target drainage pipe network; wherein the water pumping equipment comprises a rainwater pump, an intercepting pump and a storage pond gate. The pre-evacuation pump type is reasonably adjusted according to the pollution degree of sewage, and the regulation and control precision of pre-evacuation treatment is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of sewage pre-evacuation, and particularly relates to a sewage pre-evacuation treatment method and device for a drainage pipe network. BACKGROUND

[0002] The drainage pipe network is a network of infrastructure for collecting and transporting rainwater and sewage. The operation state of the drainage pipe network is easily affected by weather changes and fluctuations in water quantity in the pipe network on dry days. The pre-evacuation of the stagnant sewage in the drainage pipe network can avoid water overflow caused by sudden rainfall or pipe leakage, thereby reducing the risk of sewage overflow, waterlogging and water environmental pollution.

[0003] However, the pre-evacuation treatment method for the related drainage pipe network does not reasonably adjust the pre-evacuation pump type according to the pollution degree of the sewage, so that the control precision of the pre-evacuation treatment is insufficient. SUMMARY

[0004] The present application provides a sewage pre-evacuation treatment method and device for a drainage pipe network to solve the problem of insufficient control precision of the pre-evacuation treatment.

[0005] In a first aspect, the present application provides a sewage pre-evacuation treatment method for a drainage pipe network, which comprises: obtaining a water level value, a tryptophan fluorescence intensity value, a humic acid-like content value, an ammonia nitrogen concentration value and a pipe network slope value of a target pump station forebay; determining a pipe network volume prediction value, a pollution proportion prediction value and an ammonia nitrogen concentration prediction value by using an LSTM model based on the water level value, the tryptophan fluorescence intensity value, the humic acid-like content value, the ammonia nitrogen concentration value and the pipe network slope value of the target pump station forebay; dynamically selecting and controlling a water pumping device for the sewage pre-evacuation treatment of the target drainage pipe network based on the pipe network volume prediction value, the pollution proportion prediction value and the ammonia nitrogen concentration prediction value; wherein the water pumping device comprises a rainwater pump, a intercepting pump and a storage tank gate.

[0006] The sewage pre-evacuation treatment method for the drainage pipe network provided in this embodiment determines the pipe network volume prediction value, the pollution proportion prediction value and the ammonia nitrogen concentration prediction value by using the LSTM model based on the water level value, the tryptophan fluorescence intensity value, the humic acid-like content value, the ammonia nitrogen concentration value and the pipe network slope value of the target pump station forebay, realizes the accurate prediction of the pollution concentration value and the volume value, and dynamically selects and controls the water pumping device for the sewage pre-evacuation treatment of the target drainage pipe network based on the pipe network volume prediction value, the pollution proportion prediction value and the ammonia nitrogen concentration prediction value, reasonably adjusts the pre-evacuation pump type, and improves the control precision of the pre-evacuation treatment.

[0007] In one optional implementation, the LSTM model includes: a first LSTM layer, a second LSTM layer, a random mask layer, and a fully connected layer; based on the water level, tryptophan fluorescence intensity, humic acid content, ammonia nitrogen concentration, and pipe network slope of the target pumping station forebay, the LSTM model is used to determine the predicted pipe network volume, predicted pollution percentage, and predicted ammonia nitrogen concentration, including: Data preprocessing was performed on the water level, tryptophan fluorescence intensity, humic acid content, ammonia nitrogen concentration, and pipe network slope of the target pumping station forebay to obtain the pipe network characteristic sequence; The first LSTM layer is used to perform temporal integration and preliminary feature extraction on the pipeline network feature sequence to obtain basic temporal features; The second LSTM layer is used to filter and fuse the basic temporal features to obtain a deep fused feature vector; The deep fusion feature vector is processed by random masking layer to obtain the feature vector after random masking. The feature vectors after random masking are mapped to the predicted values ​​of pipeline volume, pollution ratio, and ammonia nitrogen concentration using a fully connected layer.

[0008] The wastewater pre-emptying treatment method for drainage pipe networks provided in this embodiment preprocesses the water level, tryptophan fluorescence intensity, humic acid content, ammonia nitrogen concentration, and pipe network slope of the target pump station forebay to obtain a standardized pipe network feature sequence, laying a high-quality data foundation for subsequent temporal feature extraction. The first LSTM layer is used to perform temporal integration and preliminary feature extraction on the pipe network feature sequence, effectively capturing the basic temporal dependencies between data. The second LSTM layer is used to filter and fuse the basic temporal features, strengthening the correlation of key features and generating a more representative deep fusion feature vector. The deep fusion feature vector is randomly masked using a random masking layer to suppress overfitting and improve the generalization ability and robustness of the feature vector. Finally, the randomly masked feature vector is mapped using a fully connected layer, realizing the simultaneous and efficient prediction of pipe network volume, pollution ratio, and ammonia nitrogen concentration.

[0009] In one optional implementation, a fully connected layer is used to map the feature vector after random masking to the predicted values ​​of pipeline volume, pollution ratio, and ammonia nitrogen concentration, respectively, including: Obtain the pipeline reference volume, volume prediction bias term, slope correction coefficient, and volume prediction weight matrix. Based on the pipeline reference volume, volume prediction bias term, slope correction coefficient, volume prediction weight matrix, and feature vector processed by random masking, calculate the pipeline volume prediction value. Obtain the maximum eigenvalue of humic acid fluorescence intensity, the pollution proportion prediction bias term, and the pollution proportion prediction weight matrix. Based on the maximum eigenvalue of humic acid fluorescence intensity, the pollution proportion prediction bias term, the pollution proportion prediction weight matrix, and the feature vector after random masking, calculate the pollution proportion prediction value. Obtain the ammonia nitrogen concentration prediction weight matrix, ammonia nitrogen concentration prediction bias term, and ammonia nitrogen concentration baseline offset. Based on the ammonia nitrogen concentration prediction weight matrix, ammonia nitrogen concentration prediction bias term, ammonia nitrogen concentration baseline offset, and the feature vector after random masking, calculate the predicted ammonia nitrogen concentration value.

[0010] The wastewater pre-emptying treatment method for drainage pipe networks provided in this embodiment calculates the predicted pipe network volume based on the pipe network baseline volume, volume prediction bias term, slope correction coefficient, volume prediction weight matrix, and feature vector processed by random masking. This improves the accuracy and scenario adaptability of volume prediction. Based on the maximum eigenvalue of humic acid fluorescence intensity, pollution proportion prediction bias term, pollution proportion prediction weight matrix, and feature vector processed by random masking, the predicted pollution proportion value is calculated. This strengthens the correlation of pollution characteristics and improves the pertinence and reliability of pollution proportion prediction. Finally, based on the ammonia nitrogen concentration prediction weight matrix, ammonia nitrogen concentration prediction bias term, ammonia nitrogen concentration baseline offset, and feature vector processed by random masking, the predicted ammonia nitrogen concentration value is calculated. This compensates for the baseline deviation and improves the accuracy and stability of ammonia nitrogen concentration prediction.

[0011] In one optional implementation, based on predicted network volume, predicted pollution percentage, and predicted ammonia nitrogen concentration, the system dynamically selects and controls pumping equipment to pre-emptively pump wastewater from the target drainage network, including: Obtain the warning volume value and the rainfall probability value, compare the predicted pipeline volume value with the warning volume value, and compare the rainfall probability value with the rainfall warning value; If the predicted pipeline volume is greater than or equal to the warning volume, and the rainfall probability is less than the rainfall warning value, then the predicted pollution percentage will be compared with the first preset threshold and the second preset threshold respectively; wherein the first preset threshold is less than the second preset threshold. If the predicted pollution percentage is greater than the second preset threshold, the predicted ammonia nitrogen concentration will be compared with the ammonia nitrogen concentration warning value. Based on the comparison between the predicted ammonia nitrogen concentration and the warning value of ammonia nitrogen concentration, the pumping equipment is controlled to pre-pump the sewage in the target drainage network.

[0012] The wastewater pre-emptive treatment method for drainage pipe networks provided in this embodiment compares the predicted pipe network volume with the warning volume value and the rainfall probability value with the rainfall warning value, providing a precise basis for subsequent drainage pipe network regulation. In scenarios where the pipe network volume exceeds the warning level and the rainfall risk is low, the method compares the predicted pollution ratio with the first and second preset thresholds to achieve precise quantitative determination of the pollution level. The method compares the predicted ammonia nitrogen concentration with the ammonia nitrogen concentration warning value to further verify the predicted ammonia nitrogen concentration for high pollution ratio scenarios, focusing on core pollution indicators and improving the pertinence and scientific nature of regulation decisions. Finally, based on the comparison results of ammonia nitrogen concentration and warning value, the method triggers the pre-emptive operation of the pumping equipment, achieving precise prevention and control of high-pollution-risk wastewater and ensuring pipe network safety.

[0013] In one optional implementation, based on a comparison between the predicted ammonia nitrogen concentration and the warning ammonia nitrogen concentration, the pumping equipment is controlled to pre-pump the target drainage network, including: If the predicted ammonia nitrogen concentration is greater than the warning value, the sewage pipe network will be pre-emptively emptied by selecting the gate of the storage tank. If the predicted ammonia nitrogen concentration is less than or equal to the warning value, then an intercepting pump should be selected to pre-emptively pump the sewage from the drainage network.

[0014] The wastewater pre-emption treatment method for the drainage network provided in this embodiment pre-emptively removes wastewater from the drainage network when the predicted ammonia nitrogen concentration is greater than the ammonia nitrogen concentration warning value. This quickly intercepts high-concentration wastewater, avoiding the pollution risks caused by direct discharge of wastewater. When the predicted ammonia nitrogen concentration is less than or equal to the ammonia nitrogen concentration warning value, a flow interceptor pump is selected to pre-emptively remove wastewater from the drainage network. This method treats conventional wastewater with low energy consumption, reducing operation and maintenance costs.

[0015] In one alternative implementation, the method further includes: If the predicted pollution percentage is less than the first preset threshold, then rainwater pumps will be used to pre-pump the sewage from the drainage network. Alternatively, if the predicted pollution percentage is greater than or equal to the first preset threshold and less than or equal to the second preset threshold, then an intercepting pump is selected to pre-emptively pump sewage from the drainage network.

[0016] The wastewater pre-emption treatment method for the drainage network provided in this embodiment selects a rainwater pump to pre-emption the drainage network when the predicted pollution ratio is less than a first preset threshold. This method treats low-pollution rainwater with low energy consumption and reduces equipment wear. When the predicted pollution ratio is greater than or equal to the first preset threshold and less than or equal to the second preset threshold, a diversion pump is selected to pre-emption the drainage network. This method balances the relationship between treatment efficiency and energy consumption and is more suitable for pre-emption treatment under medium pollution loads.

[0017] In a second aspect, the present invention provides a sewage pre-emptying treatment device for a drainage pipe network, the device comprising: The acquisition module is used to acquire the water level, tryptophan fluorescence intensity, humic acid content, ammonia nitrogen concentration, and pipeline slope of the target pumping station forebay. The determination module is used to determine the predicted values ​​of pipeline volume, pollution ratio, and ammonia nitrogen concentration based on the water level of the target pump station forebay, tryptophan fluorescence intensity, humic acid content, ammonia nitrogen concentration, and pipeline slope using an LSTM model. The pre-emptying module is used to dynamically select and control pumping equipment to pre-emptively empty the sewage in the target drainage network based on the predicted values ​​of the pipeline volume, pollution ratio, and ammonia nitrogen concentration. The pumping equipment includes rainwater pumps, intercepting pumps, and regulating tank gates.

[0018] Thirdly, the present invention provides an electronic device, comprising: a memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the computer instructions to perform the sewage pre-emptying treatment method of the drainage pipe network described in the first aspect or any corresponding embodiment.

[0019] Fourthly, the present invention provides a computer-readable storage medium storing computer instructions for causing a computer to execute the sewage pre-emptying treatment method for a drainage network according to the first aspect or any corresponding embodiment described above.

[0020] Fifthly, the present invention provides a computer program product, including computer instructions for causing a computer to execute the sewage pre-emptying treatment method for a drainage network according to the first aspect or any corresponding embodiment described above. Attached Figure Description

[0021] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0022] Figure 1 This is a schematic diagram of an application scenario according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the first process of a sewage pre-emptive treatment method for a drainage network according to an embodiment of the present invention; Figure 3This is a schematic diagram of a second process for a sewage pre-emptive treatment method for a drainage network according to an embodiment of the present invention; Figure 4 This is a schematic diagram of the workflow of an LSTM model according to an embodiment of the present invention; Figure 5 This is a schematic diagram of the third process of the sewage pre-emptive treatment method for drainage pipe network according to an embodiment of the present invention; Figure 6 This is a schematic diagram of the threshold control strategy logic according to an embodiment of the present invention; Figure 7 This is a schematic flowchart of the pre-evacuation control method according to an embodiment of the present invention; Figure 8 This is a structural block diagram of a sewage pre-emption treatment device for a drainage network according to an embodiment of the present invention. Figure 9 This is a schematic diagram of the hardware structure of an electronic device according to an embodiment of the present invention. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0024] It is understood that before using the technical solutions disclosed in the various embodiments of the present invention, users should be informed of the types, scope of use, and usage scenarios of the personal information involved in the present invention and their authorization should be obtained in accordance with relevant laws and regulations through appropriate means.

[0025] As an optional application scenario of this invention, such as Figure 1 As shown, the sewage pre-vacuuming treatment device for this drainage network may include at least one terminal device and at least one server. Figure 1 The system is illustrated in the example, which includes a computer 101, a mobile terminal 102, and a server 103, and the terminal devices such as the computer 101 and the mobile terminal 102 are connected to the server 103 through a network 110.

[0026] Specifically, the terminal device can be a smartphone, tablet, laptop, PDA, desktop computer, game console, smart TV, smart wearable device, in-vehicle terminal, VR (Virtual Reality) device, AR (Augmented Reality) device, etc. Server 103 can be a standalone physical server, a server cluster, a distributed system, or a cloud server providing cloud services. Network 110 can be a wired or wireless network, examples of which include, but are not limited to, the Internet, corporate intranet, local area network, wide area network, mobile communication network, and combinations thereof.

[0027] Drainage systems face numerous challenges. For example, during dry weather, drainage networks often suffer from problems such as sewage retention, low network capacity utilization, and pollution discharge. However, most drainage systems lack accurate prediction and effective control of network operation during dry weather, making it difficult to achieve pollution control while ensuring network capacity optimization.

[0028] While some related technologies involve the optimization and scheduling of drainage systems, they are mostly based on historical data or simple rules, which cannot adapt to complex and ever-changing urban drainage scenarios. In particular, they lack accurate predictions of drainage conditions in future periods, making it difficult to precisely control the timing and intensity of pre-emptive operations. This can easily lead to over-emptying or under-emptying, which not only affects the efficiency of the pipe network but may also cause unnecessary energy waste or pollution overflows.

[0029] Furthermore, most of the related technologies for regulating drainage networks during dry weather focus only on optimizing a single objective, such as simply pursuing the maximum utilization of network volume while ignoring the importance of pollution control, or only considering the reduction of pollution emissions without making full use of network volume, resulting in low system operating efficiency. Moreover, the related technologies do not make full use of real-time data and cannot adjust control strategies in a timely manner based on real-time monitored water quality, water quantity and other data, thus making it difficult to cope with emergencies.

[0030] According to an embodiment of the present invention, a method for pre-emptying sewage in a drainage network is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0031] This embodiment provides a method for pre-emptying sewage in a drainage network, which can be used in the aforementioned electronic equipment. Figure 2 This is a flowchart of a sewage pre-emptive treatment method for a drainage network according to an embodiment of the present invention, as follows: Figure 2 As shown, the process includes the following steps: Step S201: Obtain the water level, tryptophan fluorescence intensity, humic acid content, ammonia nitrogen concentration, and pipeline slope of the target pump station forebay.

[0032] Specifically, as the terminal node of the drainage pipe network system, the pumping station's forebay is connected to the drainage pipe network. The water level data obtained by measurement can directly characterize the system's volume status, making it easy to calculate the remaining space of the pipe network through the water level. In terms of water pollution monitoring, since the pumping station is a key node for water flow convergence, its monitoring data can accurately reflect the overall pollution status of the pipe network.

[0033] Furthermore, considering the differences between pumping stations in different regions and the fact that most pumping station forebays are cylindrical or cuboid in structure, monitoring equipment is deployed on the cylindrical or cuboid structure of the pumping station forebay, including fluorescence sensors, water level gauges, and online ammonia nitrogen detectors. Simultaneously, to optimize the treatment effect of high-concentration wastewater in the regulating tank, water level gauges and online ammonia nitrogen detectors are also installed in the regulating tank. The fluorescence sensors are used to monitor the real-time fluorescence intensity of tryptophan and the content of humic acid. Changes in the tryptophan fluorescence intensity can determine the concentration and source of domestic wastewater. Humic acid mainly originates from industrial wastewater, agricultural non-point source pollution, and the decomposition of organic matter in the natural environment. The water level gauges are used to monitor... The pump station water level is calculated and converted into the current water volume of the pipeline network, providing a crucial data foundation for subsequent control operations. The online ammonia nitrogen detector is used to monitor the ammonia nitrogen concentration in the pipeline network water in real time, providing key data support for determining the degree of pollution. Among them, the tryptophan fluorescence intensity is used to identify the source of pollution. If the tryptophan fluorescence intensity is high (domestic sewage is the main pollutant), the sewage can be treated according to the preset parameters of the "domestic sewage treatment process" (such as aeration time) when it is transported to the sewage treatment plant. If the tryptophan fluorescence intensity is low, the sewage treatment plant should be prompted to adjust the treatment plan in advance (such as adding industrial pollutant degradation agents) to avoid a decrease in the treatment efficiency of the sewage treatment plant.

[0034] Furthermore, fluorescent sensors, water level gauges, and online ammonia nitrogen concentration detectors are deployed on the cylindrical / cubic structure of the pump station forebay. The fluorescent sensor is a TSF-3000 model (a fluorescent sensor type), capable of real-time monitoring of tryptophan fluorescence intensity (sensitivity up to 0.1 qu, used to indicate the concentration and source of domestic sewage) and humic acid content (detection accuracy 0.01 mg / L, reflecting groundwater infiltration). The water level gauge is an immersion hydrostatic level gauge with a measurement range of 0-10 meters (accuracy ±0.5%FS). By monitoring the water level in the pump station forebay and combining it with the pipe slope and pipe network geometric parameters (pipe diameter D, pipe length L), the real-time water level data is converted into the current volume of the pipe network, providing data support for subsequent control. If the pipe network is an inclined cylinder with a slope of 8°, a pipe diameter D=1000mm, and a pipe length L=500m, then the horizontal equivalent length of the inclined pipe is... When the water volume is calculated based on the arc-shaped cross section, when h前池 When the ammonia nitrogen level is 0.5D, the water filling ratio is approximately 30%. The ammonia nitrogen online monitoring instrument selected is the Medex digital ammonia nitrogen sensor MDS-C1000NH, which adopts an online ion-selective electrode, has strong stability, can work stably for a long time, reduces frequent maintenance and calibration, and is suitable for long-term continuous monitoring of drainage pipe networks. The measurement method is immersion measurement, which is convenient for installation at locations such as the forebay of pumping stations and the inlet of regulating tanks.

[0035] For example, on the main section of the pump station forebay (a cylindrical structure, 2.5 meters in diameter, 4 meters in height, with a volume of approximately 19.63 cubic meters), a monitoring module is installed perpendicular to the water flow direction. This forebay is connected to the upstream pipe network via a DN1000 pipe (a type of pipe), 500 meters long, with a cross-sectional area of ​​0.785㎡. Its water level data directly reflects the volume status of the entire pipe network system. The fluorescence sensor is a TSF-3000 type (a type of fluorescence sensor) installed in an immersion manner, with the probe extending 1.5 meters below the water surface to the center of the forebay's water flow. It monitors the tryptophan fluorescence intensity (0-100 qu, accuracy 0.1 qu) and humic acid content (0-50 mg / L, accuracy 0.01 mg / L) in real time. The tryptophan fluorescence intensity value... The detection range is 0-100 qu, with an accuracy of 0.1 qu. When the concentration of domestic sewage increases, the fluorescence intensity of tryptophan (a protein decomposition product) increases, which can be used to determine the source and concentration changes of sewage (e.g., during peak periods of domestic sewage drainage, the fluorescence intensity is usually between 20-40 qu). The detection range of humic acid content is 0-50 mg / L, with an accuracy of 0.01 mg / L. Humic acid mainly comes from industrial wastewater or groundwater infiltration. When its content increases (e.g., >5 mg / L), it indicates that there may be external water mixing. The submersible hydrostatic level gauge is installed at the lowest point of the bottom of the forebay, with a measurement range of 0-10 meters (accuracy ±0.5%FS). The remaining space of the pipe network is fed back in real time through the water level-volume conversion formula; the calculation formula for the water level-volume conversion is: (1) in, This is the water filling ratio function.

[0036] Furthermore, data transmission: Monitoring data is transmitted to the cloud server in real time via a 4G (Fourth-Generation Mobile Communication Technology) wireless module, with a sampling interval of 5 minutes to ensure the real-time and continuous nature of the data. Inside the cylindrical structure of the storage tank, a TSF-3000 fluorescent sensor (immersed 1-1.5 meters below the water surface), an online ammonia nitrogen detector (measuring range 1-1000 mg / L), and a submersible hydrostatic level gauge (installed at the lowest point of the tank bottom) are installed perpendicular to the water flow direction. The data is transmitted to the cloud server via a 4G wireless module, with a sampling interval of 5 minutes.

[0037] Step S202: Based on the water level, tryptophan fluorescence intensity, humic acid content, ammonia nitrogen concentration, and pipe slope of the target pumping station forebay, the predicted pipe network volume, pollution ratio, and ammonia nitrogen concentration are determined using an LSTM model.

[0038] Specifically, an LSTM (Long Short-Term Memory) model was built based on the TensorFlow framework (a machine learning framework). The training data for the LSTM model consisted of 3 years of historical data on the target pump station's forebay water level, historical tryptophan fluorescence intensity, historical humic acid content, historical ammonia nitrogen concentration, and historical pipeline slope (approximately 17,520 samples). These data were divided into a training set (12,264 samples) and a validation set (5,256 samples) in a 7:3 ratio, covering different slopes (5°, 10°, and 15°), seasons, and weekday or holiday scenarios. The Adam optimizer (Adaptive Moment Estimation) was used with an initial learning rate of 0.001, using Mean Squared Error (MSE) as the loss function. It iterated for 50 rounds with an early stopping mechanism (termination occurred if the validation set loss did not decrease for 5 consecutive rounds). The batch size was 32. The expression for the loss function MSE is as follows: (2) in, For the true value, These are predicted values.

[0039] Furthermore, the LSTM model is deployed on a cloud server in SavedModel format (a storage format for deep learning models). It receives real-time data every 5 minutes through an API (Application Programming Interface), outputs prediction results for the next 5 hours, and establishes a monthly iteration mechanism to fine-tune the model (retraining 10 times) each month using newly added actual running data to continuously optimize prediction accuracy.

[0040] Step S203: Based on the predicted values ​​of pipeline volume, pollution ratio, and ammonia nitrogen concentration, dynamically select and control pumping equipment to pre-pump the sewage from the target drainage pipeline; wherein, the pumping equipment includes rainwater pumps, intercepting pumps, and regulating tank gates.

[0041] Specifically, the pumping equipment is installed at the end of the pumping station's back pool. The back pool is designed with a cylindrical or cuboid structure to ensure that the equipment installation is compatible with the water flow path, enabling precise control of the water in the pipe network. The pumping equipment includes three core types of equipment: intercepting pumps, rainwater pumps, and regulating tank gates. These three types use interlocking control logic, allowing only one path to be opened each time the system runs, thus preventing pollution control failure caused by mixing of different water bodies. The intercepting pump is a WQ type (a pump model) with a design flow rate of 20 m³ / h, installed at the beginning of the pipeline connecting the back pool and the sewage treatment plant, used to transport highly polluted sewage. The rainwater pump is an ISG type (a pump model) with a flow rate set at 50 m³ / h, deployed on the discharge pipeline leading to the river, activated for low-pollution water bodies to quickly release the pipe network volume. The regulating tank gate is installed at the connection channel between the back pool and the regulating tank, opened in cases of high pollution and high ammonia nitrogen levels to introduce sewage into the regulating tank for temporary storage.

[0042] Furthermore, after the equipment is installed, it is necessary to conduct joint debugging, that is, to verify the interlocking function of the intercepting pump, rainwater pump, and regulating tank gate to ensure that there is no overlapping operation during the switching process; to test the equipment response speed under different pollution scenarios, such as the rainwater pump start-up time should be ≤30 seconds under low pollution conditions, to simulate pump failure and rainfall warning scenarios, and to check the reliability of the backup pump start-up and program stop, so that the pumping equipment can meet the operation requirements of "accurate execution according to the prediction results and automatic response in case of emergencies".

[0043] The wastewater pre-emptive treatment method for drainage pipe networks provided in this embodiment uses an LSTM model to determine the predicted values ​​of pipe network volume, pollution ratio, and ammonia nitrogen concentration based on the water level of the target pump station forebay, tryptophan fluorescence intensity, humic acid content, ammonia nitrogen concentration, and pipe network slope. This achieves accurate prediction of pollution concentration and volume values. Based on these values, the method dynamically selects and controls the pumping equipment to pre-emptive treat the wastewater in the target drainage pipe network, reasonably adjusts the pre-emptive pump type, and improves the control accuracy of the pre-emptive treatment.

[0044] This embodiment provides a method for pre-emptying sewage in a drainage network, which can be used in the aforementioned electronic equipment. Figure 3 This is a flowchart of a sewage pre-emptive treatment method for a drainage network according to an embodiment of the present invention, as follows: Figure 3 As shown, the process includes the following steps: Step S301: Obtain the water level, tryptophan fluorescence intensity, humic acid content, ammonia nitrogen concentration, and pipe network slope of the target pumping station forebay. For details, please refer to [link to relevant documentation]. Figure 2 Step S201 of the illustrated embodiment will not be described again here.

[0045] Step S302: Based on the water level, tryptophan fluorescence intensity, humic acid content, ammonia nitrogen concentration, and pipe slope of the target pumping station forebay, the predicted pipe network volume, pollution ratio, and ammonia nitrogen concentration are determined using an LSTM model.

[0046] Among them, such as Figure 4 As shown, the LSTM model adopts a progressive structure of "data preprocessing → two-layer LSTM layer → random mask layer → fully connected layer" (the random mask layer is a regularization layer based on the Dropout mechanism, used to suppress overfitting). The workflow of the LSTM model includes: acquiring historical pipeline operation data, including at least the water level, tryptophan fluorescence intensity, humic acid content, ammonia nitrogen concentration, and pipeline slope of the target pump station forebay; data preprocessing and feature construction: cleaning and normalizing the historical pipeline operation data, extracting time-series features (such as sliding window statistics and trend terms), and constructing the LSTM input sequence (i.e., the pipeline feature sequence); LSTM model training and optimization: based on the preprocessed data, building a multi-layer LSTM network, combined with a loss function, such as Mean Absolute Error (MAE). The model uses the Adam algorithm for iterative training to optimize model parameters, and real-time pipeline network data is input for prediction and decision output. After training, the model predicts future states (such as pollution peaks and water level changes) and generates control decisions by combining dual threshold rules. Finally, the actual operation data after the control is implemented is collected and fed back for model verification and retraining to continuously optimize prediction accuracy. The core objective of the LSTM model is to learn the pipeline network patterns through historical time series data and output the key state prediction values ​​for the next 5 hours (i.e., pipeline volume prediction value, pollution ratio prediction value, and ammonia nitrogen concentration prediction value).

[0047] Specifically, step S302 includes: Step S3021: Perform data preprocessing on the water level, tryptophan fluorescence intensity, humic acid content, ammonia nitrogen concentration, and pipe network slope of the target pump station forebay to obtain the pipe network characteristic sequence.

[0048] Specifically, the original input data consisted of five time-series parameters from the past seven hours, sampled at 15-minute intervals (28 time steps in total). These parameters included the pump station forebay water level (m), tryptophan fluorescence intensity (qu), humic acid content (mg / L), pipe network slope θ (°), and ammonia nitrogen concentration (mg / L). During preprocessing, outliers were removed using the 3σ principle, missing values ​​were filled using linear interpolation, and all parameters were normalized to the [0,1] interval. A slope correction coefficient was also introduced to unify the volume calculation dimension for different slope scenarios, ultimately forming an input feature sequence in the format (N,28,5) (where N is the number of samples). The normalization calculation formula and the slope correction coefficient are detailed below. The calculation formulas are as follows: (3) (4) in, The original input data after normalization. The minimum value of the original input data. This represents the maximum value of the original input data.

[0049] Furthermore, the pipeline network feature sequence consists of historical 7-hour time-series data, with a time interval of 15 minutes, comprising 28 time steps (denoted as T=28). Each time step integrates 5 types of core monitoring parameters, forming an input feature matrix with a dimension of 28×5, which can then be converted into the pipeline network feature sequence X. in； The expressions for the input feature matrix and the pipeline feature sequence are as follows: (5) (6) in, For each sampling time step within the historical 7 hours; for The water level in the pump station forebay at any given time, in meters. Normalized water level data for pipe networks with different slopes ensures data comparability across scenarios with structural differences. for The fluorescence intensity value of tryptophan at any given time, expressed in qu, is used to characterize the concentration and source of domestic sewage. for The fluorescence intensity value of humic acid at any given time, expressed in mg / L, is used to reflect the input of external pollution sources such as industrial wastewater and agricultural non-point source pollution. for The ammonia nitrogen concentration at any given time, expressed in mg / L, is used to quantify the pollution load on water bodies.

[0050] Step S3022: Use the first LSTM layer to perform temporal integration and preliminary feature extraction on the pipeline feature sequence to obtain basic temporal features.

[0051] Specifically, the number of neurons in the first LSTM layer is set to 128 to meet the feature capacity requirements of multi-dimensional temporal data, and the activation function is the hyperbolic tangent function. To enhance the model's ability to capture nonlinear time-series trends such as water level fluctuations and changes in pollution concentration, the output mode is set to return_sequences=True, and the final output dimension is [missing information]. Basic time series characteristics The basic time series characteristics include preliminary time series characteristics of the pipeline network operation status within the past 7 hours; among them, the expression of the hyperbolic tangent function is: t (7) Step S3023: Use the second LSTM layer to filter and fuse the basic temporal features to obtain a deep fused feature vector.

[0052] Specifically, based on the basic time-series features of the first layer output, we further explore the deep correlation between "water level change - pollution accumulation - time cycle", focusing on capturing the long-term time-series patterns under complex scenarios such as the periodicity of day and night water consumption peaks and sudden discharge of industrial wastewater, forming deep information that integrates global trends and local mutation features (i.e., deep fusion feature vector).

[0053] Furthermore, the basic temporal features input to the output of the first LSTM layer are... (dimension) The second LSTM layer employs a gating mechanism (forget gate, input gate, and output gate) to filter, update, and fuse features. The filtering and fusion steps include: 1) Forget gate calculation: used to filter historical features that are effective for the current prediction (e.g., the diurnal water usage cycle); 2) Input gate and candidate cell state calculation: used to update the cell state at the current time step (e.g., sudden changes in pollution concentration caused by industrial wastewater); 3) Cell state update: used to store long-term temporal correlation information; 4) Output gate and layer output calculation: used to output the deep features at the current time step. The calculation formulas for the forget gate, input gate, and candidate cell state output gate are as follows: (8) (9) (10) (11) (12) (13) in, , These are the weight matrices for the input gate and the candidate cell state, respectively. , These are the corresponding bias terms. For input gate output, Candidate cell state, This is element-level multiplication. for Cellular state at any given moment This is the output gate weight matrix. For the output gate bias term, For output gate output, for The deep fusion feature vector output by the second LSTM layer at time step [time]. Here is the forget gate weight matrix. For the forget gate bias term, For activation function, for The output features of the first LSTM layer at time step 1 for Input feature vectors at any time.

[0054] Furthermore, the number of neurons in the second LSTM layer is set to 128, consistent with the first LSTM layer, to ensure the continuity of the feature dimensions; the activation function is the tanh function to maintain the non-linear expressive power of the features; the output mode is set to return_sequences=False, and the final output is a deep fusion feature vector H2 with a dimension of 1×128, which integrates the long-term correlation features of the historical 7-hour pipeline operation data.

[0055] Step S3024: Use a random mask layer to perform random masking on the deep fusion feature vector to obtain the feature vector after random masking.

[0056] Specifically, a random masking layer (a regularization layer based on the Dropout mechanism) is used to suppress the model's overfitting to noise information in the training data, improving the model's generalization ability and anti-interference capability under abnormal pipeline conditions (such as temporary industrial wastewater discharge and fluctuations in monitoring data). The dropout rate of the random masking layer is set to 0.2 (verified through 50 rounds of model iterations to achieve an optimal balance between feature retention rate and overfit suppression effect). The deep fusion feature vector output by the second LSTM layer is subjected to random masking processing to obtain the feature vector after random masking; wherein, the feature vector after random masking processing... The calculation formula is: (14) In the above formula, Dropout is the core implementation mechanism of the random mask layer, which achieves regularization by randomly masking some neuron connections.

[0057] in, This is a binary mask matrix whose elements follow a Bernoulli distribution, meaning each element is... The probability is 1. The probability is 0, 20% of the neuron connections are randomly masked, and during the model training phase, the outputs of the retained neurons are processed according to... Scaling is performed to ensure that the feature distribution is consistent between the training and inference phases, and to avoid feature mean shift during inference.

[0058] Step S3025: The feature vector after random masking is mapped to the predicted value of pipeline volume, the predicted value of pollution ratio and the predicted value of ammonia nitrogen concentration using a fully connected layer.

[0059] Specifically, the fully connected layer uses a linear activation function, combined with the physical characteristics of the pipeline network and the water quality standard parameters, to derive various prediction indicators (i.e., predicted pipeline volume, predicted pollution ratio, and predicted ammonia nitrogen concentration).

[0060] In some optional implementations, step S3025 above includes: Step a1: Obtain the pipeline reference volume, volume prediction bias term, slope correction coefficient, and volume prediction weight matrix. Based on the pipeline reference volume, volume prediction bias term, slope correction coefficient, volume prediction weight matrix, and the feature vector after random masking, calculate the pipeline volume prediction value.

[0061] Specifically, the pipeline network volume forecast is the pipeline network volume forecast for the next 5 hours, in meters (m²). 3 The formula for calculating the predicted pipeline volume is: (15) in, This is the volume prediction weight matrix. This is the volume prediction bias term. The pipeline reference volume is used to calibrate the impact of pipeline slope on volume calculation. By combining physical parameters and data characteristics, the error of the volume prediction value is controlled within a certain range. Within; of which, the reference volume of the pipeline network is determined by the pipe diameter (Unit: m), Pipe Length (Unit: m) The calculation formula for the reference volume of the pipeline network is as follows: (16) Step a2: Obtain the maximum eigenvalue of humic acid fluorescence intensity, the pollution proportion prediction bias term, and the pollution proportion prediction weight matrix. Based on the maximum eigenvalue of humic acid fluorescence intensity, the pollution proportion prediction bias term, the pollution proportion prediction weight matrix, and the eigenvector after random masking, calculate the pollution proportion prediction value.

[0062] Specifically, the predicted percentage of pollution in the next 5 hours. The formula for calculating (i.e., the predicted pollution percentage) is: (17) in, For the pollution proportion prediction weight matrix, This is the bias term for the predicted proportion of pollution. It represents the maximum characteristic value of humic acid fluorescence intensity, and RU, The pollution percentage calibration coefficient is used to align the predicted results with the pollution threshold. and Matching ensures the accuracy of pollution classification, making the prediction error of pollution proportion value ≤ ±8%.

[0063] Step a3: Obtain the ammonia nitrogen concentration prediction weight matrix, ammonia nitrogen concentration prediction bias term, and ammonia nitrogen concentration baseline offset. Based on the ammonia nitrogen concentration prediction weight matrix, ammonia nitrogen concentration prediction bias term, ammonia nitrogen concentration baseline offset, and the feature vector after random masking, calculate the predicted ammonia nitrogen concentration value.

[0064] Specifically, the predicted ammonia nitrogen concentration for the next 5 hours. (i.e., predicted ammonia nitrogen concentration), in mg / L. The formula for calculating the predicted ammonia nitrogen concentration is: (18) in, The weight matrix for predicting ammonia nitrogen concentration. This is the bias term for predicting ammonia nitrogen concentration. The ammonia nitrogen concentration baseline offset is calculated from historical monitoring data of the pipeline network during dry weather. It is taken as the average ammonia nitrogen concentration during dry weather and is used to calibrate the deviation between the predicted results and actual operating conditions, thus reducing the error of the predicted ammonia nitrogen concentration. .

[0065] Step S303: Based on the predicted values ​​of pipe network volume, pollution ratio, and ammonia nitrogen concentration, dynamically select and control pumping equipment to pre-pump the sewage from the target drainage pipe network; wherein, the pumping equipment includes rainwater pumps, intercepting pumps, and regulating tank gates. For details, please refer to... Figure 2 Step S203 of the illustrated embodiment will not be described again here.

[0066] The wastewater pre-emptying treatment method for drainage pipe networks provided in this embodiment preprocesses the water level, tryptophan fluorescence intensity, humic acid content, ammonia nitrogen concentration, and pipe network slope of the target pump station forebay to obtain a standardized pipe network feature sequence, laying a high-quality data foundation for subsequent temporal feature extraction. The first LSTM layer is used to perform temporal integration and preliminary feature extraction on the pipe network feature sequence, effectively capturing the basic temporal dependencies between data. The second LSTM layer is used to filter and fuse the basic temporal features, strengthening the correlation of key features and generating a more representative deep fusion feature vector. The deep fusion feature vector is randomly masked using a random masking layer to suppress overfitting and improve the generalization ability and robustness of the feature vector. Finally, the randomly masked feature vector is mapped using a fully connected layer, realizing the simultaneous and efficient prediction of pipe network volume, pollution ratio, and ammonia nitrogen concentration.

[0067] This embodiment provides a method for pre-emptying sewage in a drainage network, which can be used in the aforementioned electronic equipment. Figure 5This is a flowchart of a sewage pre-emptive treatment method for a drainage network according to an embodiment of the present invention, as follows: Figure 5 As shown, the process includes the following steps: Step S501: Obtain the water level, tryptophan fluorescence intensity, humic acid content, ammonia nitrogen concentration, and pipe network slope of the target pumping station forebay. For details, please refer to [link to relevant documentation]. Figure 3 Step S302 of the illustrated embodiment will not be described again here.

[0068] Step S502: Based on the water level in the target pumping station forebay, tryptophan fluorescence intensity, humic acid content, ammonia nitrogen concentration, and pipe network slope, the LSTM model is used to determine the predicted pipe network volume, pollution percentage, and ammonia nitrogen concentration. For details, please refer to [link to details]. Figure 3 Step S302 of the illustrated embodiment will not be described again here.

[0069] Step S503: Based on the predicted values ​​of pipe network volume, pollution ratio, and ammonia nitrogen concentration, dynamically select and control pumping equipment to pre-pump the sewage from the target drainage pipe network; wherein, the pumping equipment includes rainwater pumps, intercepting pumps, and regulating tank gates.

[0070] Specifically, step S503 includes: Step S5031: Obtain the warning volume value and the rainfall probability value, compare the predicted pipeline volume value with the warning volume value, and compare the rainfall probability value with the rainfall warning value.

[0071] Specifically, the warning volume value V 警戒 It is the core criterion for initiating pre-evacuation operations, calculated using pipeline structure parameters; among them, the warning volume value V 警戒 The calculation formula is: V 警戒 =0.56×(V 斜管总 +V 前池总 (19) Among them, V 斜管总 V is the total volume of the inclined tube section, calculated based on an inclined cylindrical structure. 前池总 The total volume of the pump station forebay is calculated directly based on its cylindrical or cuboid geometric dimensions (diameter, side length, and height). For example, the volume of a cylindrical forebay is calculated by multiplying its base area and height. Based on statistical analysis of the safe volume of the pipeline network before historical rainfall, a regression model determines 0.56 as the optimal correction coefficient. The formula for calculating the total volume of the inclined pipe section is as follows: V 斜管总 = (20) Step S5032: If the predicted value of the pipeline volume is greater than or equal to the warning volume value, and the rainfall probability value is less than the rainfall warning value, then the predicted value of the pollution proportion is compared with the first preset threshold and the second preset threshold respectively; wherein, the first preset threshold is less than the second preset threshold.

[0072] Specifically, the pre-evacuation procedure needs to meet two conditions simultaneously: first, the pipeline network volume V predicted by the LSTM model for the next 5 hours. 预测,t+5 Reaching or exceeding V 警戒 Secondly, the 24-hour rainfall forecast obtained through the meteorological API interface must meet the requirement that the rainfall probability value is less than 10% (i.e., the rainfall warning value) to ensure that the pipeline space is released in advance before rainfall to avoid sewage overflow due to insufficient volume during rainfall; among which, the first preset threshold is 0.62 and the second preset threshold is 1.

[0073] Furthermore, before pre-evacuation, the pre-evacuation time needs to be calculated, and the target volume V needs to be... 目标 Set to 0.3V 总 When calculating the evacuation time, a pollution correction factor of 0.5 is introduced. The formula for calculating the pre-evacuation time is: (twenty one) Where Q is the flow rate of the selected pump type. This represents the current forebay volume of the target pumping station. The calculation of the current forebay volume needs to consider the forebay water level h. 前池 Scenario-specific processing, when h 前池 When ≤D, the volume of water in the inclined tube is calculated using the formula for the volume of the arc-shaped section, Vinclined tube, and the volume of water in the forebay is superimposed. When h 前池 When the volume is greater than D, the water in the inclined tube is calculated based on the full tube volume V. 斜管总 The volume is calculated and then added to the volume of water in the forebay to ensure that the volume assessment is consistent with the actual pipe network status.

[0074] Furthermore, when the pipe network is an inclined cylinder and the water level... At that time, the water filling ratio function is derived based on the formula for the area of ​​the arc, and the expression of the water filling ratio function is: (twenty two) in, For example, when x = 0.5, that is... When x = 0.5D, the calculated f(x) is approximately 0.3 (i.e., 30%). When x = 1, i.e. When =D, f(x) = 1 (full tube state).

[0075] Step S5033: If the predicted pollution percentage is greater than the second preset threshold, then the predicted ammonia nitrogen concentration is compared with the ammonia nitrogen concentration warning value.

[0076] Specifically, if the predicted pollution percentage is less than a first preset threshold, then a rainwater pump is selected to pre-pump the sewage from the drainage network, i.e., when... When the concentration is less than 0.62, it is determined to be a low-pollution state (such as groundwater infiltration or slightly polluted rainwater). The rainwater pump is started to quickly discharge the rainwater into the river, and the pipeline volume is released first.

[0077] Furthermore, if the predicted pollution percentage is greater than or equal to the first preset threshold and less than or equal to the second preset threshold, then an intercepting pump is selected to pre-emptively pump sewage from the drainage network, i.e., when 0.62 ≤ When the concentration is ≤1, it is determined to be a moderate pollution state, and the interception pump is started to transport the wastewater to the sewage treatment plant.

[0078] Step S5034: Based on the comparison between the predicted ammonia nitrogen concentration and the warning ammonia nitrogen concentration, control the pumping equipment to pre-pump the sewage from the target drainage network.

[0079] In some optional implementations, step S5034 above includes: Step b1: If the predicted ammonia nitrogen concentration is greater than the warning value, then the gate of the storage tank is selected to pre-pump the sewage into the drainage network.

[0080] Specifically, when When ≥1, it is judged as a high pollution state, and further decision needs to be made based on ammonia nitrogen concentration. If β 预测 >40mg / L (high pollution and high ammonia nitrogen), open the gate of the regulating tank to introduce the sewage into the regulating tank for physical sedimentation (the warning volume V of the regulating tank is set to 70% of its total volume, and the gate will be automatically closed when this value is reached to prevent overflow).

[0081] Step b2: If the predicted ammonia nitrogen concentration is less than or equal to the ammonia nitrogen concentration warning value, then select an intercepting pump to pre-emptively pump the sewage from the drainage network.

[0082] Specifically, if β 预测 For concentrations ≤40 mg / L (high pollution but controllable ammonia nitrogen), start the interceptor pump to deliver the pollutant to the wastewater treatment plant at a flow rate of 20 m³ / h to avoid direct discharge of pollutants and control the load on the wastewater treatment plant.

[0083] The wastewater pre-emptive treatment method for drainage pipe networks provided in this embodiment compares the predicted pipe network volume with the warning volume value and the rainfall probability value with the rainfall warning value, providing a precise basis for subsequent drainage pipe network regulation. In scenarios where the pipe network volume exceeds the warning level and the rainfall risk is low, the method compares the predicted pollution ratio with the first and second preset thresholds to achieve precise quantitative determination of the pollution level. The method compares the predicted ammonia nitrogen concentration with the ammonia nitrogen concentration warning value to further verify the predicted ammonia nitrogen concentration for high pollution ratio scenarios, focusing on core pollution indicators and improving the pertinence and scientific nature of regulation decisions. Finally, based on the comparison results of ammonia nitrogen concentration and warning value, the method triggers the pre-emptive operation of the pumping equipment, achieving precise prevention and control of high-pollution-risk wastewater and ensuring pipe network safety.

[0084] The following specific embodiment illustrates the specific steps of a pre-emptive treatment of sewage in a drainage pipe network.

[0085] Example 1: The current network volume V of a certain drainage network 当前 =3000 m³, target volume V 目标 =0.3V 总 =1500 m³ (V 总 =5000m³), pipeline slope θ=10°, i.e. slope correction factor k=1 / cos10°≈1.015, pollution proportion prediction value α 预测 =0.8, and a flow interceptor pump (flow rate Q = 20 m³ / h) is selected. The specific steps of the LSTM-based pre-emptive pumping method for dry weather drainage networks with volumetric and pollution dual control include: 1) such as Figure 6 As shown, during the pre-evacuation process, process monitoring and adjustment need to track changes in the pipeline network status in real time to ensure the stability and controllability of the pre-evacuation operation. Specific measures include: Real-time data acquisition: The monitoring module acquires data such as actual water level, fluorescence intensity, and ammonia nitrogen concentration every 5 minutes, and compares them with LSTM predicted values ​​(i.e., pipeline network volume prediction, pollution ratio prediction, and ammonia nitrogen concentration prediction). An early warning is triggered when the deviation exceeds 10%. Emergency response mechanism: If the weather forecast is updated to indicate rainfall in the next 24 hours (rainfall probability ≥ 30%), the system immediately stops the evacuation procedure, opens the overflow valve to prevent the pipeline network from collapsing due to negative pressure, and if the pollution ratio prediction value α in the next 5 hours is not met, the system will trigger an early warning. 预测，t+5 When the concentration drops to <0.62, the pump automatically switches from the interceptor pump to the rainwater pump to accelerate the release of volume. If the equipment (such as the interceptor pump) fails, the backup pump will start automatically within 10 seconds and send an SMS alarm (including the faulty equipment number and location information) to the maintenance personnel via the 4G module.

[0086] 2) Storage tank management: When the pipeline network volume V is expected to be within the next 5 hours... 预测，t+5When the warning volume (70% of the total volume) is reached, the gate will automatically close; after 24 hours of sedimentation, the ammonia nitrogen concentration in the supernatant will be measured. If the predicted ammonia nitrogen concentration β for the next five hours is... 预测，t+5 ≤40 mg / L, the wastewater is discharged into the river via rainwater pumps, and the remaining wastewater is sent to the wastewater treatment plant via interceptor pumps.

[0087] 3) Rules for Termination and Extension of Vacuuming: Early Termination of Vacuuming: Real-time monitoring of the actual volume V of the pipeline network. 实际 When V 实际 ≤V 目标 (Target volume) and remain stable for 30 minutes (to avoid misjudgment due to water level fluctuations), the system automatically shuts down the currently running pump unit and terminates the evacuation procedure; Extend the evacuation time: If the pre-evacuation procedure runs for the preset time (such as 52 hours in the example), and still meets V 实际 >V 目标 Furthermore, if the weather forecast indicates no rainfall in the next 24 hours, the system automatically triggers the extension mechanism, recalculating the extension duration T based on the current pollution percentage α. 延长 (The formula is the same as the original calculation of the vacuum time, substitute the real-time V) 实际 α 实际 Continue operating the corresponding pump set; after the extension, check every hour until V. 实际 ≤V 目标 Or receive a rainfall warning; Post-termination status maintenance: Regardless of whether the termination is early or on time, the monitoring module continues to collect data every 15 minutes after termination. If the LSTM model predicts the pipeline volume V for the next 5 hours... 预测，t+5 Again ≥V 警戒 If there is no rainfall, the pumping procedure can be restarted. If a rainfall warning is issued, immediately switch to the normal flood prevention status of the pipeline network.

[0088] Example 2: After the combined sewer system in an old urban area was installed and commissioned, a warning volume V was set based on historical operating data of the pumping stations in the area. 警戒The total volume of the inclined tube (V) is 560 m³, the total volume of the forebay (V) is 800 m³, the total volume of the forebay (V) is 100 m³, the calculated warning volume (V) is 504 m³, which is corrected to 560 m³ under actual operating conditions. The pollution threshold α (i.e., the first preset threshold) is 0.62, the ammonia nitrogen concentration threshold β is 40 mg / L, the first correction factor (humic acid / tyrosine) is 2.0, and the second correction factor (tryptophan) is 0.4. Therefore, during a dry period (72 consecutive hours without rainfall), the operational data monitored by the LSTM model is as follows: Below: Historical 7-hour water level data shows a fluctuating upward trend, rising from 1.2m to 1.8m (sampled at 15-minute intervals, totaling 28 data sets). The average tryptophan fluorescence intensity is 25qu, fluctuating between 22-28qu. The average humic acid content is 3.5mg / L, fluctuating between 3.2-3.8mg / L. The average ammonia nitrogen concentration is 32mg / L, fluctuating between 30-34mg / L. The pipe network slope θ=10°, with a slope correction factor of approximately 1.015. Figure 7 As shown, the LSTM model outputs the prediction results for the next 5 hours: the pipeline volume V for the next 5 hours. 预测，t+5 =580m³ (≥560m³), predicted pollution percentage α for the next 5 hours 预测，t+5 =0.75 (≥0.62) and the predicted ammonia nitrogen concentration β for the next five hours 预测，t+5 =38mg / L (≤40mg / L).

[0089] Meanwhile, the probability of rainfall in the next 24 hours, obtained through the meteorological API, is 5% (no rainfall), meeting the pre-emptive evacuation trigger condition. The current pipeline volume V 当前 =650m³, target volume V 目标 For a volume of 270 m³, using a flow-stop pump (flow rate Q = 20 m³ / h), and a pollution correction factor of 0.375, the formula for calculating the pre-evacuation time is: T=(650-270) / (20×0.375×1.015)≈(380) / (7.6125)≈50(23) Meanwhile, the system starts the vacuuming procedure 52 hours in advance, because of α 预测，t+5 =0.75 (0.62≤α≤1) and β 预测，t+5 =38mg / L (≤40mg / L), start the intercepting pump (WQ type, 20m³ / h), keep the regulating tank gate and rainwater pump closed, and run until the 15th hour. The real-time water level drops to the corresponding volume of 500m³, tryptophan fluorescence intensity is 23qu, humic acid is 3.3mg / L, and ammonia nitrogen is 35mg / L. The parameters are stable, so continue to run the intercepting pump until the 30th hour. The LSTM model updates the prediction: α 预测When the concentration drops to 0.61 (<0.62), the system automatically switches to a rainwater pump (ISG type, 50m³ / h) to accelerate volume release. After running for the preset 52 hours, the real-time monitoring shows the actual network volume is 280m³ (still greater than the target volume of 270m³). At this point, a review via meteorological API confirms the 24-hour rainfall probability is still 8% (no rainfall), meeting the conditions for extending the pumping trigger. The current actual pollution percentage α<0.60 (<0.62, low pollution state) remains at 0.60 (<0.62), and the rainwater pump (flow rate Q = 50m³ / h) continues to operate. The network slope θ = 10° (slope correction coefficient k ≈ 1.015). The current actual volume Vactual = 280m³, and the target volume V…… 目标 =270m³, substituting into the evacuation time formula, with a pollution correction factor of 0.3, the formula for calculating the extension of the pre-evacuation time T is: T 延长 =(280-270) / (50×0.3×1.015)≈10 / (15.225)≈0.656 hours (24) The system automatically extended the operation of the rainwater pump for 40 minutes (approximately 0.656 hours), during which the actual volume was checked every 10 minutes. At the end of the extended period, the actual volume dropped to 268 m³ (less than or equal to 270 m³) and remained stable for 15 minutes without fluctuation. The system then shut down the rainwater pump, officially terminating the pre-emptive pumping operation. After this pre-emptive pumping operation, the pipeline network maintained a low water level before rainfall, and there was no direct discharge of sewage or load impact on the sewage treatment plant.

[0090] The beneficial effects corresponding to the above embodiments are as follows: 1) It solves the problems of lack of accurate prediction and inability to balance volume optimization and pollution control in the regulation of drainage pipe networks during dry weather in related technologies. It can pre-emptively pump out and perform routine maintenance on the pipe network during dry weather to ensure that the pipe network is in a low water volume and low pollution state before rainfall, reserve storage space, and avoid direct discharge of sewage and load shock to sewage treatment plants.

[0091] 2) Adjust the evacuation time dynamically according to the degree of pollution to ensure that the evacuation process releases sufficient pipeline volume while reasonably controlling the pollution load.

[0092] 3) The pollution correction factor reduces the pumping rate as the pollution level increases, thus preventing highly polluted wastewater from flooding into the wastewater treatment plant.

[0093] 4) By using the LSTM model to analyze the dynamic patterns of the pipeline network and combining the volume-pollution dual thresholds, a "prediction-decision-execution" closed loop is formed. The solution can dynamically adjust the pump type and flow rate according to the degree of pollution, and automatically switch modes or activate the backup pump when encountering rainfall, abnormal pollution ratio, or pump failure, thereby accurately improving the control accuracy and pipeline network operation stability.

[0094] This embodiment also provides a sewage pre-emptive treatment device for a drainage network, which is used to implement the above embodiments and preferred embodiments; details already described will not be repeated. As used below, the term "module" can refer to a combination of software and / or hardware that performs a predetermined function. Although the device described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0095] This embodiment provides a sewage pre-emptying treatment device for drainage pipe networks, such as... Figure 8 As shown, it includes: The acquisition module 801 is used to acquire the water level, tryptophan fluorescence intensity, humic acid content, ammonia nitrogen concentration, and pipeline slope of the target pump station forebay.

[0096] The determination module 802 is used to determine the predicted values ​​of pipeline volume, pollution ratio, and ammonia nitrogen concentration based on the water level, tryptophan fluorescence intensity, humic acid content, ammonia nitrogen concentration, and pipeline slope of the target pumping station forebay using an LSTM model.

[0097] The pre-emptying module 803 is used to dynamically select and control pumping equipment to pre-emptively empty the sewage in the target drainage network based on the predicted values ​​of the pipeline volume, pollution ratio, and ammonia nitrogen concentration; the pumping equipment includes rainwater pumps, intercepting pumps, and regulating tank gates.

[0098] In some alternative implementations, the determining module 802 includes: The preprocessing unit is used to preprocess the water level, tryptophan fluorescence intensity, humic acid content, ammonia nitrogen concentration, and pipeline slope of the target pumping station forebay to obtain the pipeline characteristic sequence.

[0099] The extraction unit is used to perform temporal integration and preliminary feature extraction on the pipeline feature sequence using the first LSTM layer to obtain basic temporal features.

[0100] The fusion unit is used to filter and fuse the basic temporal features using the second LSTM layer to obtain a deep fused feature vector.

[0101] The masking unit is used to perform random masking on the deep fusion feature vector using a random masking layer to obtain the feature vector after random masking.

[0102] The mapping unit is used to map the feature vector after random masking to the predicted values ​​of pipeline volume, pollution ratio, and ammonia nitrogen concentration using a fully connected layer.

[0103] In some optional implementations, the mapping unit includes: The first calculation subunit is used to obtain the pipeline reference volume, volume prediction bias term, slope correction coefficient and volume prediction weight matrix, and calculate the pipeline volume prediction value based on the pipeline reference volume, volume prediction bias term, slope correction coefficient, volume prediction weight matrix and the feature vector after random masking.

[0104] The second calculation subunit is used to obtain the maximum eigenvalue of humic acid fluorescence intensity, the pollution proportion prediction bias term, and the pollution proportion prediction weight matrix. Based on the maximum eigenvalue of humic acid fluorescence intensity, the pollution proportion prediction bias term, the pollution proportion prediction weight matrix, and the eigenvector after random masking, the pollution proportion prediction value is calculated.

[0105] The third calculation subunit is used to obtain the ammonia nitrogen concentration prediction weight matrix, the ammonia nitrogen concentration prediction bias term, and the ammonia nitrogen concentration baseline offset. Based on the ammonia nitrogen concentration prediction weight matrix, the ammonia nitrogen concentration prediction bias term, the ammonia nitrogen concentration baseline offset, and the feature vector after random masking, the predicted ammonia nitrogen concentration value is calculated.

[0106] In some alternative implementations, the determining module 802 includes: The first comparison unit is used to compare the predicted pollution percentage with a first preset threshold and a second preset threshold respectively if the predicted pipeline volume is greater than or equal to the warning volume and the rainfall probability value is less than the rainfall warning value; wherein the first preset threshold is less than the second preset threshold.

[0107] The second comparison unit is used to compare the predicted value of ammonia nitrogen concentration with the warning value of ammonia nitrogen concentration if the predicted value of pollution proportion is greater than the second preset threshold.

[0108] The control unit is used to control the pumping equipment to pre-pump the target drainage network based on the comparison between the predicted ammonia nitrogen concentration and the warning ammonia nitrogen concentration.

[0109] In some alternative implementations, the control unit includes: The first pre-emptive drainage subunit is used to select the regulating tank gate to perform pre-emptive drainage treatment on the drainage network if the predicted ammonia nitrogen concentration is greater than the ammonia nitrogen concentration warning value.

[0110] The second pre-evacuation subunit is used to select an intercepting pump to pre-evacuate the sewage from the drainage network if the predicted ammonia nitrogen concentration is less than or equal to the ammonia nitrogen concentration warning value.

[0111] In some alternative implementations, the determining module 802 further includes: The first pre-emptive unit is used to select a rainwater pump to pre-emptive the sewage in the drainage network if the predicted pollution percentage is less than a first preset threshold.

[0112] The second pre-emptive pumping system is used to select an intercepting pump to pre-emptively pump sewage from the drainage network if the predicted pollution percentage is greater than or equal to the first preset threshold and less than or equal to the second preset threshold.

[0113] The sewage pre-emptying treatment device for drainage pipe networks provided in this embodiment of the invention can execute the sewage pre-emptying treatment method for drainage pipe networks provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects for executing the method. Further functional descriptions of the above modules and units are the same as in the corresponding embodiments described above, and will not be repeated here.

[0114] Figure 9 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention.

[0115] The following is a detailed reference. Figure 9 This diagram illustrates a structural schematic suitable for implementing an electronic device according to embodiments of the present invention. The electronic device may include a processor (e.g., a central processing unit, graphics processor, etc.) 901, which can perform various appropriate actions and processes according to a program stored in read-only memory (ROM) 902 or a program loaded from memory 908 into random access memory (RAM) 903. The RAM 903 also stores various programs and data required for the operation of the electronic device. The processor 901, ROM 902, and RAM 903 are interconnected via a bus 904. An input / output (I / O) interface 905 is also connected to the bus 904.

[0116] Typically, the following devices can be connected to I / O interface 905: input devices 906 including, for example, touchscreens, touchpads, keyboards, mice, cameras, microphones, accelerometers, gyroscopes, etc.; output devices 907 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; memory devices 908 including, for example, magnetic tapes, hard disks, etc.; and communication devices 908. Communication device 908 allows electronic devices to communicate wirelessly or wiredly with other devices to exchange data. Although Figure 9 Electronic devices with various devices are shown, but it should be understood that it is not required to implement or have all of the devices shown, and more or fewer devices may be implemented or have instead.

[0117] In particular, according to embodiments of the present invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of the present invention include a computer program product comprising a computer program carried on a non-transitory 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 a communication device 908, or installed from a memory 908, or installed from a ROM 902. When the computer program is executed by the processor 901, it performs the functions defined in the wastewater pre-emptying treatment method for a drainage network according to embodiments of the present invention.

[0118] Figure 9 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of use of the embodiments of the present invention.

[0119] This invention also provides a computer-readable storage medium. The methods described above according to embodiments of the invention can be implemented in hardware or firmware, or implemented as computer code that can be recorded on a storage medium, or implemented as computer code downloaded via a network and originally stored on a remote storage medium or a non-transitory machine-readable storage medium and then stored on a local storage medium. Thus, the methods described herein can be processed by software stored on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. The storage medium can be a magnetic disk, optical disk, read-only memory, random access memory, flash memory, hard disk, or solid-state drive, etc.; further, the storage medium can also include combinations of the above types of memory. It is understood that the computer, processor, microprocessor controller, or programmable hardware includes storage components capable of storing or receiving software or computer code. When the software or computer code is accessed and executed by the computer, processor, or hardware, the sewage pre-emption treatment method for the drainage network shown in the above embodiments is implemented.

[0120] A portion of this invention can be applied as a computer program product, such as computer program instructions, which, when executed by a computer, can invoke or provide the methods and / or technical solutions according to the invention through the operation of the computer. Those skilled in the art will understand that the forms in which computer program instructions exist in a computer-readable medium include, but are not limited to, source files, executable files, installation package files, etc. Correspondingly, the ways in which computer program instructions are executed by a computer include, but are not limited to: the computer directly executing the instructions, or the computer compiling the instructions and then executing the corresponding compiled program, or the computer reading and executing the instructions, or the computer reading and installing the instructions and then executing the corresponding installed program. Here, the computer-readable medium can be any available computer-readable storage medium or communication medium accessible to a computer.

[0121] Although embodiments of the invention have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of the invention, and such modifications and variations all fall within the scope defined by the appended claims.

Claims

1. A method for pre-emptive treatment of sewage in a drainage pipe network, characterized in that, The method includes: Obtain the water level, tryptophan fluorescence intensity, humic acid content, ammonia nitrogen concentration, and pipe network slope of the target pumping station forebay; Based on the water level, tryptophan fluorescence intensity, humic acid content, ammonia nitrogen concentration, and pipe slope of the target pumping station forebay, the predicted pipe volume, pollution ratio, and ammonia nitrogen concentration are determined using an LSTM model. Based on the predicted value of the pipeline network volume, the predicted value of the pollution ratio, and the predicted value of the ammonia nitrogen concentration, the pumping equipment is dynamically selected and controlled to pre-pump the sewage from the target drainage pipeline network; wherein, the pumping equipment includes rainwater pumps, intercepting pumps, and regulating tank gates.

2. The method according to claim 1, characterized in that, The LSTM model includes: a first LSTM layer, a second LSTM layer, a random mask layer, and a fully connected layer; the determination of predicted network volume, predicted pollution percentage, and predicted ammonia nitrogen concentration based on the water level, tryptophan fluorescence intensity, humic acid content, ammonia nitrogen concentration, and network slope of the target pumping station forebay using the LSTM model includes: The water level of the target pump station forebay, the tryptophan fluorescence intensity, the humic acid content, the ammonia nitrogen concentration, and the pipe network slope were preprocessed to obtain a pipe network characteristic sequence. The first LSTM layer is used to perform temporal integration and preliminary feature extraction on the pipeline feature sequence to obtain basic temporal features; The second LSTM layer is used to filter and fuse the basic temporal features to obtain a deep fused feature vector. The deep fusion feature vector is randomly masked using the random mask layer to obtain the feature vector after random masking. The fully connected layer is used to map the feature vector after random masking to the predicted value of the pipeline volume, the predicted value of the pollution ratio, and the predicted value of the ammonia nitrogen concentration, respectively.

3. The method according to claim 2, characterized in that, The step of mapping the feature vector after random masking using the fully connected layer to the predicted pipeline volume, the predicted pollution percentage, and the predicted ammonia nitrogen concentration, respectively, includes: Obtain the pipeline reference volume, volume prediction bias term, slope correction coefficient, and volume prediction weight matrix. Based on the pipeline reference volume, the volume prediction bias term, the slope correction coefficient, the volume prediction weight matrix, and the feature vector after random masking, calculate the pipeline volume prediction value. Obtain the maximum eigenvalue of humic acid fluorescence intensity, the pollution proportion prediction bias term, and the pollution proportion prediction weight matrix. Based on the maximum eigenvalue of humic acid fluorescence intensity, the pollution proportion prediction bias term, the pollution proportion prediction weight matrix, and the feature vector after random masking, calculate the pollution proportion prediction value. Obtain the ammonia nitrogen concentration prediction weight matrix, ammonia nitrogen concentration prediction bias term, and ammonia nitrogen concentration baseline offset. Based on the ammonia nitrogen concentration prediction weight matrix, the ammonia nitrogen concentration prediction bias term, the ammonia nitrogen concentration baseline offset, and the feature vector after random masking, calculate the predicted ammonia nitrogen concentration value.

4. The method according to claim 1, characterized in that, The step of dynamically selecting and controlling pumping equipment to pre-emptively pump wastewater from the target drainage network based on the predicted network volume, the predicted pollution percentage, and the predicted ammonia nitrogen concentration includes: Obtain the warning volume value and the rainfall probability value, compare the predicted pipeline volume value with the warning volume value, and compare the rainfall probability value with the rainfall warning value; If the predicted value of the pipeline volume is greater than or equal to the warning volume value, and the rainfall probability value is less than the rainfall warning value, then the predicted value of the pollution proportion is compared with the first preset threshold and the second preset threshold respectively; wherein, the first preset threshold is less than the second preset threshold. If the predicted pollution percentage is greater than the second preset threshold, the predicted ammonia nitrogen concentration is compared with the ammonia nitrogen concentration warning value. Based on the comparison between the predicted ammonia nitrogen concentration and the warning ammonia nitrogen concentration, the pumping equipment is controlled to perform pre-emptive pumping of sewage into the target drainage network.

5. The method according to claim 4, characterized in that, The step of controlling the pumping equipment to pre-pump the target drainage network based on the comparison between the predicted ammonia nitrogen concentration and the warning value of ammonia nitrogen concentration includes: If the predicted ammonia nitrogen concentration is greater than the warning value for ammonia nitrogen concentration, then the gate of the regulating reservoir is selected to perform pre-emptive sewage treatment on the drainage network. If the predicted ammonia nitrogen concentration is less than or equal to the warning value for ammonia nitrogen concentration, then the intercepting pump is selected to perform pre-emptive sewage treatment on the drainage network.

6. The method according to claim 4, characterized in that, The method further includes: If the predicted pollution percentage is less than the first preset threshold, then the rainwater pump is selected to pre-pump the sewage from the drainage network. Alternatively, if the predicted pollution percentage is greater than or equal to the first preset threshold and less than or equal to the second preset threshold, then the intercepting pump is selected to perform pre-emptive sewage treatment on the drainage network.

7. A sewage pre-vacuuming treatment device for a drainage pipe network, characterized in that, The device includes: The acquisition module is used to acquire the water level, tryptophan fluorescence intensity, humic acid content, ammonia nitrogen concentration, and pipeline slope of the target pumping station forebay. The determination module is used to determine the predicted values ​​of pipeline volume, pollution ratio, and ammonia nitrogen concentration based on the water level, tryptophan fluorescence intensity, humic acid content, ammonia nitrogen concentration, and pipeline slope of the target pumping station forebay using an LSTM model. The pre-emptying module is used to dynamically select and control pumping equipment to pre-emptively empty the sewage in the target drainage network based on the predicted value of the pipeline volume, the predicted value of the pollution ratio, and the predicted value of the ammonia nitrogen concentration; wherein the pumping equipment includes a rainwater pump, a intercepting pump, and a regulating tank gate.

8. An electronic device, characterized in that, include: The system includes a memory and a processor, which are interconnected. The memory stores computer instructions, and the processor executes the computer instructions to perform the sewage pre-emptive treatment method for the drainage network as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions for causing the computer to execute the sewage pre-emptying treatment method for the drainage network as described in any one of claims 1 to 6.

10. A computer program product, characterized in that, Includes computer instructions for causing a computer to perform the sewage pre-emptying treatment method for the drainage network as described in any one of claims 1 to 6.

Images