A method and system for accurately quantifying sewage outlet flow based on a neural network model
The neural network model accurately predicts the flow of sewage outlets, which solves the problems of uncertain drainage time and difficulty in quantitative flow, and realizes high-precision and real-time flow monitoring, which is suitable for a variety of environmental conditions.
Patent Information
- Application Number
- CN202411144360.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-20
- Publication Date
- 2024-11-12
- Estimated Expiration
- 2044-08-20
AI Technical Summary
In the prior art, the drainage time is uncertain, the quantification of sewage flow is difficult, and the lack of low-cost monitoring equipment and technology to achieve accurate quantification.
The precise quantitative method of sewage outlet flow based on the neural network model is adopted. Through data acquisition, factor analysis, neuron transmission function processing and BP neural network model training, a traffic prediction model is established, and the weight is updated through the gradient descent method to achieve accurate real-time traffic. predict.
The accuracy of the discharge outlet flow is improved, and the average absolute percentage error of the flow is less than 5%, which enhances the stability and real-time nature of the system, and solves the shortcomings of traditional methods in comprehensive consideration of multiple environmental factors and data processing.
Smart Images

Figure CN118656656B_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to but is not limited to the field of water quality monitoring technology, and in particular relates to a method and system for accurately quantifying the flow rate of a sewage outlet based on a neural network model. Background Art
[0002] Water environment safety has a significant impact on human production and life. As the last barrier for wastewater to be discharged into rivers and lakes, monitoring of sewage outlets into rivers and seas is imminent. However, the drainage time is uncertain and the flow rate is difficult to quantify. Sewage flow is the basis for determining the amount of sewage and pollutants entering the river. Currently, there is a lack of low-cost equipment and technology to achieve accurate quantification.
[0003] At present, accurate quantification of monitoring data on sewage outlet flow is a pain point in the industry, and there is no reliable technology to break through. Patent CN202211700544.X is aimed at the waste of resources caused by continuous monitoring, and the wastewater intermittent sewage outlet flow triggering method and device based on flow threshold is optimized by machine learning algorithm on historical flow data to obtain the minimum flow threshold for monitoring, thereby realizing precise supervision of wastewater outlets; at the same time, patent CN202311137538.2 is used in the confidence interval algorithm, through real-time monitoring of the instantaneous flow of wastewater discharged from the target river outlet, and then the historical flow data of the target river outlet is analyzed through machine learning algorithm to determine the 95% confidence interval of its flow. However, these are just re-optimization of historical flow data through monitoring data, and cannot guarantee the reliability of front-end monitoring data.
[0004] At present, the commonly used flow monitoring technologies for sewage outlets into rivers in China include ultrasonic flowmeters and electromagnetic flowmeters. Among them, ultrasonic technology has high accuracy in sewage outlet monitoring, can accurately measure flow velocity and flow, and is not affected by pipe material and diameter, and is suitable for various water quality conditions. However, the cost of ultrasonic technology is high, including equipment purchase and maintenance costs, which may increase the monitoring cost. Electromagnetic flowmeters also have certain advantages in sewage outlet monitoring. They can stably measure flow velocity and flow, and have good adaptability to pipe materials. Compared with ultrasonic technology, the cost of electromagnetic flowmeters is relatively low, and to a certain extent, they can resist external interference, which improves the stability and reliability of monitoring. However, considering that sewage outlets into rivers and seas are not always in a full pipe state, if the pipe is not completely filled, the liquid may form an irregular flow pattern, which may affect the current induction in the magnetic field. In addition, in a half-filled state, the liquid may contain bubbles or solid particles. These impurities will change the current distribution in the electromagnetic field, resulting in measurement errors, thereby affecting the measurement of flow, making the accurate quantification of electromagnetic flowmeters a huge challenge. Thanks to its powerful data processing capabilities and pattern recognition technology, machine learning is increasingly used in environmental monitoring, which can significantly enhance the predictive ability, accuracy and efficiency of environmental monitoring. Neural network models are an important type of algorithm in machine learning. They can learn and simulate the nonlinear relationship between input data and output data. Properly trained neural networks can make accurate predictions for unseen input data and have good generalization capabilities. In addition, they are easy to implement, can be processed in parallel, have self-learning capabilities, are highly adaptable and scalable, making them a powerful tool for solving complex problems.
[0005] In view of the above analysis, the technical problems that need to be urgently solved in the existing technology are: the drainage time is uncertain and the flow rate is difficult to quantify. The sewage flow rate is the basis for determining the amount of sewage and pollutants entering the river. At present, there is a lack of low-cost monitoring equipment and technology to achieve its accurate quantification. Summary of the invention
[0006] In response to the problems existing in the prior art, the present invention provides a method and system for accurately quantifying the flow rate at a sewage outlet based on a neural network model, which can significantly improve the accuracy of the flow rate at the sewage outlet.
[0007] The present invention is implemented as follows: a method for accurately quantifying the flow rate of a sewage outlet based on a neural network model comprises the following steps.
[0008] Step 1: Collect data on water flows with different pipe diameters and different filling degrees under certain temperature, humidity, electromagnetic environment and other conditions.
[0009] Step 2: Use factorial experimental design to analyze the significance of the impact of various factors such as temperature, humidity, pipe diameter, fullness and electromagnetic environment on water flow, screen the significant influencing factors, set parameter weights and thresholds, and establish a training set.
[0010] Step three: preprocess the training set data, use the neuron's transfer function - Sigmoid function, normalize it to [0, 1], and process the sample data through linear transformation.
[0011] Step 4: Use the BP (Back Propagation) neural network model to screen the optimal structure and network parameters of the model to establish a sewage outlet water flow prediction model.
[0012] Step 5: Input sample data, calculate layer by layer, and sum the results based on parameter weights to get the output prediction value.
[0013] Step 6: Compare the predicted value with the actual value, use the mean absolute percentage error and determination coefficient to evaluate the error of the output predicted value, and use the gradient descent method to update the weights and biases to see if they meet the end conditions.
[0014] Step 7, repeat steps 4 to 6, combining the error control end mode and the number of control interruption modes, until a convergence state is reached.
[0015] Step eight, finally use the established BP neural network to compensate and correct the data to ensure accurate real-time prediction of the sewage outlet flow.
[0016] Furthermore, in step one, data is collected for water flows with different pipe diameters and different filling degrees under certain temperature, humidity, electromagnetic environment and other conditions: all data measurements are collected using standard equipment, and the equipment is calibrated before use to ensure the accuracy of the collected data.
[0017] Furthermore, the parameters of weights and thresholds to be set in step 2 include but are not limited to temperature, humidity, pipe diameter, fullness, conductivity, and electromagnetic environment.
[0018] Furthermore, the amount of training set data in step 3 is greater than 1000, and the neuron output is calculated layer by layer;
[0019] The neural network model in step 4 adopts a supervised learning process.
[0020] Furthermore, the layer-by-layer calculation in step five includes input values such as temperature, humidity, pipe diameter, fullness, conductivity, and electromagnetic environment parameters, and uses a layer-by-layer calculation method to multiply the output value of the previous layer by its corresponding weight, and then activate the output prediction value through the Sigmoid function after the sum is calculated.
[0021] Furthermore, in step seven, by comparing the actual values, the convergence condition is that the number of training steps is 1000-100000 times, and the training target error condition is 0.01-0.0001.
[0022] In step eight, actual water sample testing is performed and the established compensation correction is used to verify the test accuracy, and the final average absolute percentage error of the flow rate is less than 5%.
[0023] Another object of the present invention is to provide a sewage outlet flow accurate quantification system based on a neural network model for realizing the sewage outlet flow accurate quantification method based on a neural network model, comprising:
[0024] The data acquisition module is used to collect data on water flows with different pipe diameters and different filling degrees under certain temperature, humidity and electromagnetic environment conditions.
[0025] The training set establishment module is used to use factorial experimental design to analyze the significance of the impact of various factors such as temperature, humidity, pipe diameter, fullness and electromagnetic environment on water flow, and to screen significant influencing factors, set parameter weights and thresholds, and establish a training set.
[0026] The training set preprocessing module is used to preprocess the training set data. It uses the neuron's transfer function - the Sigmoid function, which needs to be normalized to [0, 1] and processes the sample data through linear transformation.
[0027] The flow prediction model building module is used to use the BP (Back Propagation) neural network model to screen the optimal structure and network parameters of the model to establish a sewage outlet water flow prediction model.
[0028] The output prediction value acquisition module is used to input sample data, calculate layer by layer and sum the results in combination with parameter weights to obtain the output prediction value.
[0029] The prediction value comparison module is used to compare the prediction value with the actual value, use the mean absolute percentage error and determination coefficient to evaluate the error of the output prediction value, and use the gradient descent method to update the weights and biases to see whether the end conditions are met.
[0030] The compensation and correction module is used to use the established BP neural network to compensate and correct the data to ensure accurate real-time prediction of the sewage outlet flow.
[0031] Another object of the present invention is to provide a computer device, which includes a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the method for accurately quantifying sewage outlet flow based on a neural network model.
[0032] Another object of the present invention is to provide a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, the processor executes the steps of the method for accurately quantifying sewage outlet flow based on a neural network model.
[0033] Another object of the present invention is to provide an information data processing terminal, which includes the sewage outlet flow precise quantification system based on the neural network model.
[0034] In combination with the above technical solutions and the technical problems solved, the advantages and positive effects of the technical solutions to be protected by the present invention are as follows:
[0035] First, the technology to be protected by the present invention utilizes the BP neural network model to realize accurate real-time prediction of the flow rate at the sewage outlet, and the average absolute percentage error of the flow rate is less than 5%. The established system method has good stability, wide applicability, and good application prospects.
[0036] The technical solution of the present invention solves multiple technical problems in the prior art and achieves significant technical progress.
[0037] Existing water flow prediction methods often lack comprehensive consideration of multiple environmental factors, resulting in low prediction accuracy. The present invention can more accurately capture the impact of various factors on water flow by performing comprehensive data collection on water flows with different pipe diameters and different filling degrees under conditions such as temperature, humidity, and electromagnetic environment, thereby improving the input data quality of the prediction model.
[0038] Existing methods tend to ignore certain significant influencing factors when processing multidimensional data, resulting in insufficient robustness of the prediction model. The present invention uses factorial experimental design to analyze the significance of each factor, screens out significant influencing factors, sets parameter weights and thresholds, and establishes a training set containing important influencing factors. This process significantly improves the reliability and effectiveness of the model.
[0039] Traditional flow prediction models have deficiencies in data preprocessing and model training, which can easily lead to overfitting or underfitting. The present invention establishes an efficient water flow prediction model by normalizing and linearly transforming the training set data and using the BP neural network model to screen the optimal structure and network parameters. The model can calculate layer by layer and sum the results in combination with parameter weights, ensuring the accuracy of the prediction results.
[0040] The existing technology has certain limitations in terms of model prediction accuracy and real-time performance. The present invention uses the gradient descent method to continuously update weights and biases, combined with error control and frequency control interruption methods to ensure that the model reaches a convergence state, and finally uses the BP neural network to compensate and correct the data, thereby achieving accurate real-time prediction of the sewage outlet flow. This method not only improves the accuracy of the prediction, but also enhances the real-time performance and reliability of the system.
[0041] While solving the problems of the prior art, the present invention has achieved significant technical progress and provided an efficient and reliable method for accurately predicting water flow.
[0042] The technical solution of the present invention can significantly improve the accuracy of the monitoring process of the front-end flow meter and greatly improve the data reliability, which is of great significance in the field of water quality monitoring. At the same time, it can be expanded to other fields where flow meters are used, and has extremely high commercial value.
[0043] At present, accurate quantification of monitoring data on sewage outlet flow is a pain point in the industry, and there is no reliable technology to break through. Some current patented technologies, aimed at ineffective monitoring of continuous monitoring, propose to use machine learning methods to re-optimize based on historical flow data, and establish a monitoring relationship between the wastewater monitoring device and the target sewage outlet into the river. However, this cannot guarantee the reliability of the front-end monitoring data. Based on this, the present invention targets the monitoring process of the front-end flow meter, and uses a method and system for real-time compensation of data using a BP neural network model to ensure accurate real-time monitoring of the sewage outlet flow. BRIEF DESCRIPTION OF THE DRAWINGS
[0044] Figure 1 It is a flow chart of a method for accurately quantifying the flow rate at a sewage outlet based on a neural network model provided in an embodiment of the present invention.
[0045] Figure 2 It is a structural diagram of a sewage outlet flow accurate quantification system based on a neural network model provided in an embodiment of the present invention.
[0046] Figure 3 It is a neural network training prediction flow chart provided by an embodiment of the present invention. DETAILED DESCRIPTION
[0047] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.
[0048] like Figure 1 As shown, the method for accurately quantifying the flow rate of a sewage outlet based on a neural network model provided in an embodiment of the present invention includes the following steps.
[0049] S101, collecting data on water flows with different pipe diameters and different filling degrees under certain temperature, humidity, electromagnetic environment and other conditions.
[0050] S102, using factorial experimental design to analyze the significance of the impact of various factors such as temperature, humidity, pipe diameter, fullness and electromagnetic environment on water flow, and screen the significant influencing factors, set parameter weights and thresholds, and establish a training set.
[0051] S103, preprocessing the training set data, using the transfer function of the neuron - Sigmoid function, needs to be normalized to [0, 1], and the sample data is processed by linear transformation.
[0052] S104, using the BP (Back Propagation) neural network model, the optimal structure and network parameters of the model are screened to establish a sewage outlet water flow prediction model.
[0053] S105, input sample data, calculate layer by layer and sum the results in combination with parameter weights to obtain an output prediction value.
[0054] S106, compare the predicted value with the actual value, use the mean absolute percentage error and the coefficient of determination to evaluate the error of the output predicted value, and use the gradient descent method to update the weight and bias to see whether the end condition is met.
[0055] S107, repeat S104-S106, combining the error control end mode and the number control interruption mode, until a convergence state is reached.
[0056] S108, finally using the established BP neural network to compensate and correct the data to ensure accurate real-time prediction of the sewage outlet flow.
[0057] The present invention provides a method for accurately quantifying the flow rate of a sewage outlet based on a neural network model, which includes the following key steps.
[0058] First, in step S101, the system collects detailed data on water flows with different pipe diameters and different filling degrees under certain temperature, humidity, and electromagnetic environment conditions. Through this multi-dimensional data collection, it is possible to fully understand the impact of various environmental factors on water flow, thereby providing basic data for subsequent analysis.
[0059] Next, in step S102, the collected data is analyzed using a factorial experimental design method to determine the significant effects of factors such as temperature, humidity, pipe diameter, fullness, and electromagnetic environment on water flow. By screening out significant influencing factors and setting corresponding parameter weights and thresholds, a training set containing important influencing factors is established to provide an effective data basis for the training of the neural network model.
[0060] In steps S103 and S104, the training set data is preprocessed, the data is normalized to the [0,1] interval using the Sigmoid function, and the sample data is processed by linear transformation. Then, the BP (Back Propagation) neural network model is used to screen the optimal structure and network parameters of the model, and a neural network model for sewage outlet flow prediction is established. The output prediction value is finally obtained by combining layer-by-layer calculation and parameter weights.
[0061] Finally, in steps S105 to S108, the predicted value is compared with the actual value, the mean absolute percentage error and the coefficient of determination are used to evaluate the accuracy of the prediction model, and the weights and biases are continuously updated by the gradient descent method until the preset convergence state is reached. After the model training is completed, the established BP neural network is used to compensate and correct the actual data, thereby achieving accurate real-time prediction of the sewage outlet flow. This method not only improves the accuracy of the prediction, but also ensures the real-time and reliability of the system.
[0062] Furthermore, in step one, data is collected for water flows with different pipe diameters and different filling degrees under certain temperature, humidity, electromagnetic environment and other conditions: all data measurements are collected using standard equipment, and the equipment is calibrated before use to ensure the accuracy of the collected data.
[0063] Furthermore, the parameters for setting weights and thresholds in S102 include but are not limited to temperature, humidity, pipe diameter, fullness, conductivity, and electromagnetic environment.
[0064] Furthermore, the amount of training set data in S103 is greater than 1000, and neuron outputs are calculated layer by layer.
[0065] Furthermore, the neural network model in S104 adopts a supervised learning process.
[0066] Furthermore, the layer-by-layer calculation in S105 includes multiplying the output value of the previous layer by its corresponding weight value for input values such as temperature, humidity, pipe diameter, fullness, conductivity, and electromagnetic environment parameters by layer-by-layer calculation, and activating the output prediction value through the Sigmoid function after summing the values.
[0067] Furthermore, in S107, by comparing the actual values, the convergence condition is that the number of training steps is 1000-100000 times, and the training target error condition is 0.01-0.0001.
[0068] Furthermore, in S108, the actual water sample test is performed, and the established compensation correction is used to verify the test accuracy, and the final flow average absolute percentage error is less than 5%.
[0069] like Figure 2 As shown, the sewage outlet flow accurate quantification system based on the neural network model provided by the embodiment of the present invention includes:
[0070] The data acquisition module is used to collect data on water flows with different pipe diameters and different filling degrees under certain temperature, humidity and electromagnetic environment conditions.
[0071] The training set establishment module is used to use factorial experimental design to analyze the significance of the impact of various factors such as temperature, humidity, pipe diameter, fullness and electromagnetic environment on water flow, and to screen significant influencing factors, set parameter weights and thresholds, and establish a training set.
[0072] The training set preprocessing module is used to preprocess the training set data. It uses the neuron's transfer function - the Sigmoid function, which needs to be normalized to [0, 1] and processes the sample data through linear transformation.
[0073] The flow prediction model building module is used to use the BP neural network model to screen the optimal structure and network parameters of the model to establish a sewage outlet water flow prediction model.
[0074] The output prediction value acquisition module is used to input sample data, calculate layer by layer and sum the results in combination with parameter weights to obtain the output prediction value.
[0075] The prediction value comparison module is used to compare the prediction value with the actual value, use the mean absolute percentage error and determination coefficient to evaluate the error of the output prediction value, and use the gradient descent method to update the weights and biases to see whether the end conditions are met.
[0076] The compensation and correction module is used to use the established BP neural network to compensate and correct the data to ensure accurate real-time prediction of the sewage outlet flow.
[0077] In order to make the disclosure of the present invention more complete, it is described below through more specific embodiments.
[0078] Example 1.
[0079] (1) Data collection is performed on water flows with different pipe diameters and different filling degrees under certain temperature, humidity, electromagnetic environment and other conditions. All data measurements are collected using standard equipment, and the equipment is calibrated before use to ensure the accuracy of the collected data.
[0080] (2) Factorial experimental design was used to analyze the significance of the impact of various factors such as humidity, pipe diameter, fullness, conductivity, and electromagnetic environment on water flow, and the significant influencing factors were screened, parameter weights and thresholds were set, and a training set was established.
[0081] (3) Ensure that the amount of training set data is greater than 1000. Preprocess the training set data using the neuron's transfer function - the Sigmoid function. Normalize it to [0, 1] and process the sample data through linear transformation.
[0082] (4) The BP (Back Propagation) neural network model adopts a supervised learning process to screen the optimal structure and network parameters of the model to establish a sewage outlet water flow prediction model.
[0083] (5) Input sample data, including input values of temperature, humidity, pipe diameter, fullness, conductivity, electromagnetic environment parameters, etc., calculate layer by layer, multiply the output value of the previous layer by its corresponding weight, sum it up, and activate the output prediction value through the Sigmoid function.
[0084] (6) Compare the predicted value with the actual value, use the mean absolute percentage error and determination coefficient to evaluate the error of the output prediction value, and use the gradient descent method to update the weights and biases.
[0085] (7) Repeat (4)-(6) until the convergence condition is that the number of training steps is 1000 and the training target error condition is 0.001.
[0086] (8) Finally, the established BP neural network is used to compensate and correct the data to verify the test accuracy, and the final average absolute percentage error of the flow is less than 5%. At the same time, the steps of the sewage outlet flow accurate quantitative method based on the neural network model are executed by using a computer device and a readable storage medium, through an information data processing terminal, which includes an accurate quantitative compensation system.
[0087] Example 2.
[0088] (1) Data collection is performed on water flows with different pipe diameters and different filling degrees under certain temperature, humidity, electromagnetic environment and other conditions. All data measurements are collected using standard equipment, and the equipment is calibrated before use to ensure the accuracy of the collected data.
[0089] (2) Factorial experimental design was used to analyze the significance of the impact of various factors such as humidity, pipe diameter, fullness, and conductivity on water flow, and the significant influencing factors were screened, parameter weights and thresholds were set, and a training set was established.
[0090] (3) Ensure that the amount of training set data is greater than 1000. Preprocess the training set data using the neuron's transfer function - the Sigmoid function. Normalize it to [0, 1] and process the sample data through linear transformation.
[0091] (4) The BP (Back Propagation) neural network model adopts a supervised learning process to screen the optimal structure and network parameters of the model to establish a sewage outlet water flow prediction model.
[0092] (5) Input sample data, including input values of temperature, humidity, pipe diameter, fullness, conductivity, electromagnetic environment parameters, etc., calculate layer by layer, multiply the output value of the previous layer by its corresponding weight, sum it up, and activate the output prediction value through the Sigmoid function.
[0093] (6) Compare the predicted value with the actual value, use the mean absolute percentage error and determination coefficient to evaluate the error of the output prediction value, and use the gradient descent method to update the weights and biases.
[0094] (7) Repeat (4)-(6) until the convergence condition is that the number of training steps is 10,000 and the training target error condition is 0.0001.
[0095] (8) Finally, the established BP neural network is used to compensate and correct the data to verify the test accuracy, and the final average absolute percentage error of the flow is less than 5%. At the same time, the steps of the sewage outlet flow accurate quantitative method based on the neural network model are executed by using a computer device and a readable storage medium, through an information data processing terminal, which includes an accurate quantitative compensation system.
[0096] The present invention relates to a method and system for accurately quantifying the flow rate of a sewage outlet based on a neural network model, including an information data processing device, a terminal and an accurate quantitative compensation method system. A stable compensation system is formed, which can significantly improve the accuracy of sewage outlet flow monitoring.
[0097] In order to illustrate the accuracy of prediction based on the neural network model, a pipe with a diameter of 100 mm was tested under different filling conditions to collect information data. The compensation method system established by the present invention is used to compare the displayed value with the compensated value, and the relative error is preliminarily calculated. The comparative analysis results are shown in Table 1. The results show that when the filling degree is less than 1, the test error of the flow rate is large, and after the system compensation optimization, the relative error is between 0.7% and 3.3%. It can be seen that the established compensation system has good prediction accuracy, especially for the flow value under the condition of not full pipe.
[0098] Table 1 Comparison between actual environmental values and compensation values
[0099]
[0100] It should be noted that the embodiments of the present invention can be implemented by hardware, software, or a combination of software and hardware. The hardware part can be implemented using dedicated logic; the software part can be stored in a memory and executed by an appropriate instruction execution system, such as a microprocessor or dedicated design hardware. It can be understood by a person of ordinary skill in the art that the above-mentioned devices and methods can be implemented using computer executable instructions and / or contained in a processor control code, such as a carrier medium such as a disk, CD or DVD-ROM, a programmable memory such as a read-only memory (firmware), or a data carrier such as an optical or electronic signal carrier. Such code is provided on the carrier medium. The device and its modules of the present invention can be implemented by hardware circuits such as very large-scale integrated circuits or gate arrays, semiconductors such as logic chips, transistors, etc., or programmable hardware devices such as field programmable gate arrays, programmable logic devices, etc., can also be implemented by software executed by various types of processors, and can also be implemented by a combination of the above-mentioned hardware circuits and software, such as firmware.
[0101] The above description is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any modifications, equivalent substitutions and improvements made by any technician familiar with the technical field within the technical scope disclosed by the present invention and within the spirit and principle of the present invention should be covered by the protection scope of the present invention.
Claims
1. A method for accurately quantifying the flow rate of a sewage outlet based on a neural network model, characterized in that: The steps include: Step 1: collect data on water flows with different pipe diameters and different filling degrees under certain temperature, humidity and electromagnetic environment conditions; Step 2: Use factorial experimental design to analyze the significance of the impact of temperature, humidity, pipe diameter, fullness and electromagnetic environment on water flow, screen the significant influencing factors, set parameter weights and thresholds, and establish a training set; Step 3: Preprocess the training set data, use the sigmoid function, the transfer function of the neuron, normalize it to [0, 1], and process the sample data through linear transformation; Step 4: Using the BP neural network model, the optimal structure and network parameters of the model are screened to establish a sewage outlet flow prediction model; Step 5: Input sample data, calculate layer by layer, and sum the results based on parameter weights to get the output prediction value; Step 6: Compare the predicted value with the actual value, use the mean absolute percentage error and determination coefficient to evaluate the error of the output predicted value, and use the gradient descent method to update the weights and biases to determine whether the end condition is met; Step 7, repeat steps 4-6, combining the error control end mode and the number of control interruption modes, until a convergence state is reached; Step eight, use the established BP neural network to compensate and correct the data to ensure accurate real-time prediction of the sewage outlet flow.
2. The method for accurately quantifying the flow rate of a sewage outlet based on a neural network model as claimed in claim 1, characterized in that: In the step 1, data is collected for water flows with different pipe diameters and different filling degrees under certain temperature, humidity and electromagnetic environment conditions: all data measurements are collected using standard equipment, and the equipment is calibrated before use to ensure the accuracy of the collected data.
3. The method for accurately quantifying the flow rate of a sewage outlet based on a neural network model as claimed in claim 1, characterized in that: The parameters for setting weights and thresholds in step 2 include, but are not limited to, temperature, humidity, pipe diameter, fullness, conductivity, and electromagnetic environment.
4. The method for accurately quantifying the flow rate of a sewage outlet based on a neural network model according to claim 1 is characterized in that: The amount of training set data in step 3 is greater than 1000, and the neuron output is calculated layer by layer; The neural network model in step 4 has a supervised learning process.
5. The method for accurately quantifying the flow rate of a sewage outlet based on a neural network model as claimed in claim 1, characterized in that: The layer-by-layer calculation in step 5 includes inputting parameter values of temperature, humidity, pipe diameter, fullness, conductivity, and electromagnetic environment, and multiplying the output value of the previous layer by its corresponding weight by layer-by-layer calculation, and activating the output prediction value through the Sigmoid function after summing the values.
6. The method for accurately quantifying the flow rate of a sewage outlet based on a neural network model according to claim 1, characterized in that: In step 7, by comparing the actual values, the convergence condition is that the number of training steps is 1000-100000 times, and the training target error condition is 0.01-0.0001; In step eight, actual water sample testing is performed and the established compensation correction is used to verify the test accuracy, and the final average absolute percentage error of the flow rate is less than 5%.
7. A sewage outlet flow accurate quantification system based on a neural network model that implements the method as described in any one of claims 1 to 6, characterized in that: include: The data acquisition module is used to collect data on water flows with different pipe diameters and different filling degrees under certain temperature, humidity and electromagnetic environment conditions; The training set establishment module is used to analyze the significance of the influence of various factors such as temperature, humidity, pipe diameter, fullness and electromagnetic environment on water flow rate by factorial experimental design, and to screen the significant influencing factors, set parameter weights and thresholds, and establish a training set; The training set preprocessing module is used to preprocess the training set data. It uses the transfer function of the neuron - the Sigmoid function, which needs to be normalized to [0, 1] and processes the sample data through linear transformation; The flow prediction model establishment module is used to use the BP neural network model to screen the optimal structure and network parameters of the model to establish a sewage outlet water flow prediction model; The output prediction value acquisition module is used to input sample data, calculate layer by layer, and sum the results based on the parameter weights to obtain the output prediction value; The prediction value comparison module is used to compare the prediction value with the actual value, use the mean absolute percentage error and determination coefficient to evaluate the error of the output prediction value, and use the gradient descent method to update the weight and bias to see whether the end condition is met; The compensation and correction module is used to use the established BP neural network to compensate and correct the data to ensure accurate real-time prediction of the sewage outlet flow.
8. A computer device, comprising a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the method for accurately quantifying the flow rate of a sewage outlet based on a neural network model as described in any one of claims 1 to 6.
9. A computer-readable storage medium storing a computer program, which, when executed by a processor, enables the processor to execute the steps of the method for accurately quantifying the flow rate of a sewage outlet based on a neural network model as described in any one of claims 1 to 6.
10. An information data processing terminal, comprising the sewage outlet flow accurate quantification system based on the neural network model as claimed in claim 7.
