A digital-based intelligent wastewater discharge control method

Abstract

This invention discloses a digitally-based intelligent wastewater discharge method, comprising the following steps: acquiring a historical dataset, which includes historical wastewater data and historical optimal opening / closing size data of wastewater valves; training a feedforward neural network using the historical wastewater data as input and the historical optimal opening / closing size data of wastewater valves as output to obtain a trained wastewater monitoring model; inputting real-time monitoring data into the wastewater monitoring model and outputting optimal opening / closing size data of wastewater valves; and dynamically adjusting the opening / closing size of the wastewater valves according to the optimal opening / closing size data and adjustment commands. This invention dynamically adjusts the opening / closing size of wastewater valves using a digital method, ensuring wastewater is discharged in optimal condition, avoiding the resource waste and poor treatment effect caused by traditional fixed wastewater valve installations, thereby improving the overall efficiency of wastewater treatment.

Description

Technical Field

[0001] This invention belongs to the field of wastewater treatment and relates to a control method, specifically a digital-based intelligent wastewater discharge control method. Background Technology

[0002] In the field of wastewater treatment, RO (reverse osmosis) technology has been widely used. It effectively removes various pollutants from wastewater through semi-permeable membranes, thereby obtaining high-quality purified water. However, RO systems generate a certain amount of concentrated wastewater during operation, and the treatment of this wastewater becomes a significant issue. Traditional wastewater treatment methods typically involve discharging wastewater through fixed wastewater valves. However, this method lacks flexibility and precision, failing to dynamically adjust the valve opening based on actual conditions, resulting in poor wastewater treatment efficiency and resource waste.

[0003] In recent years, with the rapid development of digital technology, some researchers have begun to apply it to wastewater treatment, hoping to optimize the process through intelligent methods. However, existing digital wastewater treatment technologies still have some problems. First, these technologies typically only monitor and control the on / off state of wastewater valves, failing to achieve precise adjustment of the valve opening size. Second, existing digital technologies lack real-time feedback and dynamic adjustment capabilities, unable to adjust the wastewater valve opening size promptly based on the actual operating conditions of the RO system and changes in wastewater quality. Summary of the Invention

[0004] In view of the above-mentioned shortcomings in the prior art, the present invention provides a digital-based intelligent wastewater discharge control method, which solves the problem that the existing digital technology lacks the ability to provide real-time feedback and dynamic adjustment, and cannot adjust the opening and closing of the wastewater valve in a timely, rapid and accurate manner according to the actual operation of the RO system and changes in wastewater quality.

[0005] To achieve the above-mentioned objectives, the technical solution adopted by this invention is as follows:

[0006] A digital-based intelligent wastewater discharge control method includes the following steps:

[0007] S1: Obtain historical datasets, which include historical wastewater data and historical data on the optimal opening and closing sizes of wastewater valves;

[0008] S2: Use historical wastewater data as input and historical wastewater valve optimal opening and closing size data as output to train the feedforward neural network and obtain a trained wastewater monitoring model.

[0009] S3: Input real-time monitoring data into the wastewater monitoring model and output the optimal opening and closing size data of the wastewater valve;

[0010] S4: Dynamically adjust the opening and closing size of the wastewater valve according to the optimal opening and closing size data and adjustment instructions.

[0011] The beneficial effects of the above scheme are:

[0012] (1) This invention dynamically adjusts the opening and closing of the wastewater valve through a digital method, which can respond in real time to changes in the operating conditions of the RO system and ensure that the wastewater is discharged in the best condition. This precise control avoids the resource waste and poor treatment effect caused by the traditional fixed setting of wastewater valve, thereby improving the overall efficiency of wastewater treatment.

[0013] (2) This invention optimizes the opening and closing of the wastewater valve through an intelligent control algorithm, which can minimize wastewater discharge while ensuring the normal operation of the RO system. This not only helps to save water resources, but also reduces energy and material consumption in the wastewater treatment process, and realizes the effective utilization of resources.

[0014] (3) This invention enables operators to promptly identify and resolve potential faults in the RO system and wastewater valves. This preventative maintenance strategy extends the system's lifespan, reduces unexpected downtime, and thus improves the stability and reliability of the entire wastewater treatment system.

[0015] (4) This invention allows operators to intuitively and easily view the real-time operating status of the RO system, the opening and closing status of the wastewater valve, and early warning information. Simultaneously, the opening and closing size of the wastewater valve can be dynamically adjusted at any time according to adjustment commands, providing operators with greater flexibility and convenience.

[0016] Furthermore, prior to step S3, the method also includes:

[0017] Initial data is acquired using high-precision sensors;

[0018] The initial data is preprocessed and features are extracted to obtain real-time monitoring data.

[0019] The beneficial effects of the above-mentioned further solutions are as follows: preprocessing can help clean the data, remove duplicate values, missing values, and outliers, and improve the accuracy and completeness of the data; preprocessing can extract and select the most relevant features, reduce data dimensionality and noise, and improve the training efficiency and prediction accuracy of the model; standardizing or normalizing the data can eliminate the differences in dimensionality between different features, ensuring the stability and convergence of model training; preprocessing can reduce data noise through methods such as smoothing and clustering, improving the generalization ability and prediction accuracy of the model; preprocessed data is more suitable for training and testing machine learning models, which can improve the performance and effectiveness of the model.

[0020] Furthermore, step S4 specifically includes:

[0021] Receive optimal opening and closing data and adjustment commands from the wastewater valve;

[0022] When only the optimal opening and closing size data of the wastewater valve is received, the opening and closing size of the wastewater valve is dynamically adjusted according to the optimal opening and closing size data. Alternatively, when only the adjustment command is received, the opening and closing size of the wastewater valve is dynamically adjusted according to the adjustment command. Or, when both the optimal opening and closing size data of the wastewater valve and the adjustment command are received, the opening and closing size of the wastewater valve is dynamically adjusted according to the adjustment command.

[0023] The beneficial effect of the above-mentioned further solution is that the opening and closing size of the wastewater valve can be dynamically adjusted according to the optimal opening and closing size data and adjustment commands, which increases the flexibility of the system.

[0024] Furthermore, the historical wastewater data in S1 and the real-time monitoring data in S3 include: source water TDS, wastewater valve orifice size, water production rate, flow rate, and time.

[0025] Furthermore, the feedforward neural network in step S2 includes an input layer, a hidden layer, and an output layer.

[0026] Furthermore: the input layer includes 5 nodes; the hidden layer contains multiple neurons, and a non-linear activation function is used to transform and transmit the data; the output layer uses a linear activation function to output the optimal opening and closing size of the wastewater valve.

[0027] Furthermore, in step S2, when training the feedforward neural network, the backpropagation algorithm and gradient descent optimizer are used to minimize the error between the predicted value and the true value.

[0028] The beneficial effects of the above-mentioned further solutions are: the backpropagation algorithm and gradient descent optimizer can help the neural network gradually reduce errors during training, enabling the neural network to converge to better parameter values ​​and improve the accuracy of prediction.

[0029] Furthermore: the feedforward neural network has L layers, and the weight matrix of the l-th layer is W. l The deviation vector is b l The activation function is σ.

[0030] Furthermore: the forward propagation formula for the l-th layer is:

[0031] z l =W l a l-1 +b l

[0032] a l =σ(z) l )

[0033] Among them, z l It is the weighted input of the l-th layer, a l It is the activation output of the l-th layer, a l-1 It is the activation output of the layer preceding the l-th layer.

[0034] Furthermore: the error δ of the backpropagation algorithm l The calculation formula is:

[0035]

[0036] For l <L:

[0037] δ l =((W) l+1 ) T δ l+1 )⊙σ'(z l )

[0038] Where L represents the number of layers in the feedforward neural network, l represents the l-th layer of the feedforward neural network, and W l+1 Let z represent the weight matrix of the (l+1)th layer. l It is the weighted input of the l-th layer.

[0039] The beneficial effects of the above-mentioned further solutions are as follows: by calculating the impact of errors on various parameters, the backpropagation algorithm can automatically adjust the weights and biases in the neural network, enabling the neural network to better fit the training data; the backpropagation algorithm and gradient descent optimizer can take advantage of parallel computing to accelerate the training speed of the neural network and improve efficiency; by minimizing the error between the predicted value and the true value, the backpropagation algorithm and gradient descent optimizer can help the neural network improve its generalization ability, thereby better adapting to new data. Attached Figure Description

[0040] Figure 1 This is a flowchart of a digitally based intelligent wastewater discharge control method.

[0041] Figure 2 This is a diagram of a feedforward neural network structure.

[0042] Figure 3 This is a flow chart of a water treatment system.

[0043] Figure 4 This is a schematic diagram of the user interface for a digital-based intelligent wastewater discharge control method. Detailed Implementation

[0044] The present invention will be further described below with reference to the accompanying drawings and specific embodiments.

[0045] like Figure 1As shown, a digital-based intelligent wastewater discharge control method includes the following steps:

[0046] S1: Obtain historical datasets, which include historical wastewater data and historical data on the optimal opening and closing sizes of wastewater valves.

[0047] In step S1, the historical wastewater data may include: source water TDS, wastewater valve orifice size, water production rate, flow rate, and time. The data format for each historical wastewater data is: source water TDS (unit: mg / L) | wastewater valve orifice size (unit: mm) | water production rate (referring to how much pure water was produced) | flow rate (referring to how much water passed before the RO membrane malfunctioned (unit: ml)) | time (referring to how much time passed before the RO membrane malfunctioned (unit: minutes)).

[0048] S2: Using historical wastewater data as input and historical wastewater valve optimal opening and closing size data as output, train the feedforward neural network to obtain a trained wastewater monitoring model.

[0049] In step S2, the structure of the feedforward neural network is as follows: Figure 2 As shown, the input layer receives the historical dataset, with the data format as follows: source water TDS (mg / L) | wastewater valve orifice size (mm) | permeate rate (how much pure water was produced) | flow rate (how much water passed before the RO membrane malfunctioned (ml)) | time (how much time elapsed before the RO membrane malfunctioned (minutes)). The hidden layer contains multiple neurons that transform and transmit the data using a non-linear activation function (ReLU) to capture the complex relationships within the data. The output layer predicts the optimal opening and closing size of the wastewater valve; this value is a continuous variable, therefore a linear activation function is used.

[0050] During training, backpropagation and the gradient descent optimizer (Adam) are used to minimize the error between the predicted and actual values. In backpropagation, the feedforward neural network has L layers, and the weight matrix of the l-th layer is W. l The deviation vector is b l The activation function is σ.

[0051] For the l-th layer, its forward propagation formula can be expressed as:

[0052] z l =W l a l-1 +b l

[0053] a l =σ(z) l )

[0054] Among them, z l It is the weighted input of the l-th layer, a l It is the activation output of the l-th layer, a l-1 It is the activation output of the layer preceding the l-th layer.

[0055] The error δ of the backpropagation algorithm l The calculation formula is:

[0056]

[0057] For l <L:

[0058] δ l =((W) l+1 ) T δ l+1 )⊙σ'(z l )

[0059] Where L represents the number of layers in the feedforward neural network, l represents the l-th layer of the feedforward neural network, and W l+1 Let z represent the weight matrix of the (l+1)th layer. l It is the weighted input of the l-th layer.

[0060] The gradient calculation for weights and biases can be expressed as:

[0061]

[0062]

[0063] In the Adam optimizer, the update process for each parameter θ (which can be a weight W or a bias b) is as follows:

[0064] initialization:

[0065] m0 = 0

[0066] v0 = 0

[0067] t=0

[0068] In each step t:

[0069]

[0070] m t =β1m t-1 +(1-β1)g t

[0071]

[0072]

[0073]

[0074]

[0075] Where β1, β2, η, and ε are the hyperparameters of the Adam optimizer, typically set to β1 = 0.9, β2 = 0.999, and ε = 10. -8 η is the learning rate, which can be adjusted during optimization.

[0076] During training, the network weights and biases are continuously adjusted through multiple iterations to optimize the model's predictive performance.

[0077] After training, the model is validated using a validation set, and the model with the best performance is selected as the final prediction model.

[0078] S3: Input real-time monitoring data into the wastewater monitoring model and output the optimal opening and closing size of the wastewater valve.

[0079] Before step S3, initial data can be acquired in real time using a high-precision sensor, and the initial data can be preprocessed and feature extracted to obtain real-time monitoring data.

[0080] For example, this invention can be deployed in the intelligent control system of an RO reverse osmosis membrane wastewater treatment system that dynamically adjusts wastewater valves based on digital methods, such as... Figure 3 As shown, Figure 3 This is a complete water treatment system flowchart, depicting the entire process from influent to pure water output.

[0081] In the filtration process, when water flows through the reverse osmosis filter, the invention deployed on the cloud platform can calculate the most suitable dynamic wastewater valve setting to control the amount of wastewater, thereby affecting the working pressure, water production rate and recovery rate of the reverse osmosis filter, so as to maximize the utilization of the reverse osmosis filter and produce more pure water and less wastewater.

[0082] The RO reverse osmosis membrane wastewater treatment system (hereinafter referred to as the RO system) based on a digitally-driven dynamic adjustment wastewater valve mainly includes an RO reverse osmosis membrane unit, a wastewater valve, a real-time monitoring system, an intelligent control system, and a user interface. The RO reverse osmosis membrane unit uses standard RO reverse osmosis membrane modules to remove dissolved salts, organic matter, and other pollutants from the wastewater. The inlet of the RO unit is connected to the raw water tank, and the outlet is connected to the purified water collection tank and the wastewater discharge pipe, respectively. The wastewater valve is installed on the wastewater discharge pipe of the RO unit to control the wastewater discharge flow rate. The wastewater valve uses an electric actuator that can dynamically adjust its opening and closing size according to received control commands. Sensors are installed at the inlet, outlet, and wastewater discharge pipe of the RO unit to monitor key parameters such as water quality, flow rate, and pressure in real time. The monitoring data is transmitted to the intelligent control system in real time via a data cable for processing and analysis. The intelligent control system incorporates the intelligent control algorithm proposed in this invention, calculates the optimal opening and closing size of the wastewater valve based on the real-time monitoring data, and drives the electric actuator of the wastewater valve to make corresponding adjustments by outputting control commands. The user interface is as follows: Figure 4 As shown, an LCD screen can be used. Figure 4 The interface can display the real-time operating status of the RO system, the opening and closing status of the wastewater valve, and early warning information. The interface is equipped with operation buttons, allowing operators to manually adjust the opening and closing size of the wastewater valve or switch to automatic control mode.

[0083] The RO system achieves dynamic adjustment of the wastewater valve opening and closing through digital methods. This adjustment method can automatically adjust the wastewater discharge flow rate based on the real-time operating status of the RO system and changes in water quality, thereby improving the efficiency and effectiveness of wastewater treatment. Simultaneously, the introduction of a real-time monitoring system and an intelligent control system enhances the system's stability and reliability, and reduces maintenance costs for operators.

[0084] S4: Dynamically adjust the opening and closing size of the wastewater valve according to the optimal opening and closing size data and adjustment instructions.

[0085] Optionally, step S4 may specifically be: receiving optimal opening and closing size data and adjustment instructions for the wastewater valve; when only the optimal opening and closing size data for the wastewater valve is received, dynamically adjusting the opening and closing size of the wastewater valve according to the optimal opening and closing size data; or, when only the adjustment instructions are received, dynamically adjusting the opening and closing size of the wastewater valve according to the adjustment instructions; or, when both the optimal opening and closing size data and the adjustment instructions are received, dynamically adjusting the opening and closing size of the wastewater valve according to the adjustment instructions.

[0086] Optionally, based on the trained wastewater monitoring model, a function can be defined to find the optimal pore size within a given pore size range to maximize a custom performance evaluation criterion. The specific steps are as follows:

[0087] S5: Initialization function: the optimal pore size best_pore_size is initialized to None, and the optimal score best_score is initialized to negative infinity (indicating that no valid solution has been found yet).

[0088] S6: Search process: For each aperture size p within the range [pmin, pmax), traverse the range with a step size Δp:

[0089] S61: Use the previously trained model to predict the current water production rate Wp and RO membrane lifetime Lp.

[0090] S62: Performance Evaluation

[0091] S621: Calculate the water production rate score SW using the following formula:

[0092] SW=-∣Wp-Wt∣

[0093] Where Wt is the target water production rate (the closer SW is to the target water production rate, the higher the score).

[0094] S622: Calculate the RO membrane lifetime score SL using the following formula:

[0095] SL=-|L p -L t |÷L t

[0096] Among them, L t The target RO membrane lifetime is the percentage of deviation from the target RO membrane lifetime; the closer it is to the target, the higher the score, but it has been normalized.

[0097] S623: Calculate the total water production score (Stotal). The formula is as follows:

[0098] Stotal = SW + 0.5 × SL

[0099] Where SW is the water production rate score and SL is the RO membrane life score.

[0100] This invention comprehensively considers the scores of water production rate and RO membrane life, and the goal of this invention is to maximize the score (i.e., to make the predicted water production rate and RO membrane life as close as possible to the target water production rate and RO membrane life).

[0101] S7: Update the optimal solution. If the current total score (Stotal) is higher than the previous best score (best_score), update best_score to Stotal and set best_pore_size to the current pore size (p).

[0102] S8: Returns the results. The function ultimately returns the best pore size (best_pore_size) and the corresponding best score (best_score).

[0103] Those skilled in the art will recognize that the embodiments described herein are intended to help the reader understand the principles of the invention, and should be understood that the scope of protection of the invention is not limited to such specific statements and embodiments. Those skilled in the art can make various other specific modifications and combinations based on the technical teachings disclosed in this invention without departing from the spirit of the invention, and these modifications and combinations are still within the scope of protection of the invention.

Claims

1. A digitally-based intelligent wastewater discharge control method, characterized in that, The control method includes: S1: Obtain historical datasets, which include historical wastewater data and historical wastewater valve optimal opening and closing size data; S2: Using the historical wastewater data as input and the optimal opening and closing size data of the historical wastewater valve as output, train the feedforward neural network to obtain a trained wastewater monitoring model. S3: Input the real-time monitoring data into the wastewater monitoring model and output the optimal opening and closing size data of the wastewater valve; S4: Based on the optimal opening and closing size data and adjustment command of the wastewater valve, dynamically adjust the opening and closing size of the wastewater valve, specifically including: Receive the optimal opening / closing size data of the wastewater valve and the adjustment command; When only the optimal opening and closing size data of the wastewater valve is received, the opening and closing size of the wastewater valve is dynamically adjusted according to the optimal opening and closing size data of the wastewater valve; or, when only the adjustment command is received, the opening and closing size of the wastewater valve is dynamically adjusted according to the adjustment command; or, when both the optimal opening and closing size data of the wastewater valve and the adjustment command are received, the opening and closing size of the wastewater valve is dynamically adjusted according to the adjustment command. The historical wastewater data in S1 and the real-time monitoring data in S3 include: source water TDS, wastewater valve orifice size, water production rate, water flow rate, and time. The specific method for outputting the optimal opening and closing size of the wastewater valve in the wastewater monitoring model is as follows: Define a function to find the optimal aperture size within a given aperture size range; Initialization functions: the optimal pore size (best_pore_size) is initialized to None, and the optimal score (best_score) is initialized to negative infinity; For each aperture size p within the range Within, traversal is performed with a step size Δp, specifically including: Use the trained model to predict the current water production rate. and RO membrane life L p Calculate the water production rate score based on the current water production rate. : in, Target water production rate; RO membrane life score is calculated based on RO membrane life. : in, Target RO membrane life; Based on water production rate score and RO membrane life score Calculate the total water production score (Stotal): in, To calculate the water production rate score, The RO membrane lifetime score is given. Update the optimal solution. If the current total score (Stotal) is higher than the previous best score (best_score), then update best_score to Stotal and set best_pore_size to the current pore size (p). Returns the optimal pore size (best_pore_size) and the corresponding optimal score (best_score).

2. The method according to claim 1, characterized in that, Before step S3, the method further includes: Initial data is acquired using high-precision sensors; The initial data is preprocessed and features are extracted to obtain the real-time monitoring data.

3. The method according to claim 1, characterized in that, The feedforward neural network in step S2 includes an input layer, a hidden layer, and an output layer.

4. The method according to claim 3, characterized in that, The input layer includes 5 nodes; the hidden layer contains multiple neurons and uses a non-linear activation function to transform and transmit data; the output layer uses a linear activation function to output the optimal opening and closing size data of the wastewater valve.

5. The method according to claim 1, characterized in that, In step S2, when training the feedforward neural network, the backpropagation algorithm and gradient descent optimizer are used to minimize the error between the predicted value and the true value.

6. The method according to claim 5, characterized in that, The number of layers in the feedforward neural network is , No. The weight matrix of the layer is The deviation vector is The activation function is .

7. The method according to claim 6, characterized in that, The first The forward propagation formula for a layer is: in, It is the first Weighted input of the layer, It is the first Layer activation output, It is the first The activation output of the layer preceding the layer.

8. The method according to claim 5, characterized in that, The error of the backpropagation algorithm The calculation formula is: for : in, This indicates the number of layers in the feedforward neural network. Indicates the first Layered feedforward neural network, Indicates the first The weight matrix of the layer, It is the first Weighted input for the layer.

Images