Reverse osmosis micro wastewater control system based on AI algorithm prediction

Abstract

The invention discloses a reverse osmosis micro-wastewater control system based on AI algorithm prediction, and the system comprises a micro-wastewater discharge waterway which is used for generating purified water and treating wastewater; the single chip microcomputer control system is used for receiving the real-time data uploaded by the sensor group in the micro wastewater discharge waterway, uploading the real-time data to the data preprocessing module, receiving the calculation result of the random forest algorithm module and controlling the working process of the micro wastewater discharge waterway according to the received data; the data preprocessing module is used for preprocessing the data uploaded by the single chip microcomputer and uploading the preprocessed data to the random forest algorithm module; and the random forest algorithm module is used for receiving the data uploaded by the data preprocessing module, performing calculation and issuing a calculation result to the single chip microcomputer. On the premise of ensuring the stability of the treated water quality, wastewater discharge can be remarkably reduced, the utilization rate of water resources is increased, and contributions are made to water resource saving and environmental protection.

Description

Technical Field

[0001] The present invention relates to the field of water treatment, and in particular to a reverse osmosis micro-wastewater control system based on AI algorithm prediction. Background Art

[0002] In the field of reverse osmosis (RO) water treatment technology, improving water resource utilization and reducing wastewater discharge have always been the focus of research. Traditional reverse osmosis systems usually produce a certain amount of wastewater, which contains high levels of dissolved solids (TDS) and other impurities. Direct discharge not only wastes water resources, but may also pollute the environment.

[0003] In order to reduce wastewater discharge, some solutions have been proposed in the prior art. For example, some systems control the discharge of wastewater by adjusting the opening of the wastewater valve, but this method often relies on experience or simple control logic, and it is difficult to achieve the optimization of wastewater discharge. In addition, some technologies attempt to reprocess wastewater, such as removing impurities in wastewater through physical, chemical or biological methods, and then reuse it in the system, but this method is usually costly and the treatment effect may be unstable.

[0004] In recent years, with the development of artificial intelligence and machine learning technologies, some researchers have begun to try to apply these technologies to the field of water treatment to optimize the operating parameters of the system and improve the utilization rate of water resources. However, most applications are still in the exploratory stage and have not yet formed a mature and reliable technical solution.

[0005] In the prior art, the patent application closest to the present invention may be a reverse osmosis system that combines simple control logic and wastewater reprocessing technology. The system monitors the water quality parameters of the wastewater, such as TDS concentration, through sensors, and then adjusts the opening of the wastewater valve according to preset rules to control the discharge of wastewater. At the same time, the system is also equipped with a wastewater reprocessing unit to simply treat part of the wastewater and reuse it in the system.

[0006] However, there are the following problems and disadvantages in the prior art:

[0007] Simple control logic. The control logic of existing systems is often based on experience or simple rules, which makes it difficult to optimize wastewater discharge. When water quality parameters, system status, and waterway pressure change, the system may not be able to adjust operating parameters in time, resulting in reduced water resource utilization.

[0008] The wastewater reprocessing effect is unstable. The existing wastewater reprocessing technology may lead to unstable treatment effects due to water quality fluctuations, treatment equipment failures, etc., affecting the water quality of recycled water and the overall performance of the system, and significantly shortening the life of the RO reverse osmosis membrane.

[0009] Lack of intelligent prediction and optimization capabilities. Existing systems generally lack intelligent prediction and optimization capabilities and cannot actively adjust and optimize according to the real-time status and future needs of the system. Summary of the invention

[0010] In view of the above-mentioned deficiencies in the prior art, the present invention provides a reverse osmosis micro-wastewater control system based on AI algorithm prediction, which solves the problems of unstable wastewater treatment effect and lack of intelligent prediction and optimization capabilities in the prior art.

[0011] In order to achieve the above-mentioned invention object, the technical solution adopted by the present invention is: a reverse osmosis micro-wastewater control system based on AI algorithm prediction, comprising:

[0012] Micro wastewater discharge waterway, used to produce pure water and treat wastewater;

[0013] A single-chip control system is used to receive real-time data uploaded by a sensor group in a micro-wastewater discharge waterway, upload the real-time data to a data preprocessing module, receive calculation results of a random forest algorithm module, and control the working process of the micro-wastewater discharge waterway according to the received data;

[0014] Data preprocessing module: used to preprocess the data uploaded by the single-chip control system and upload the preprocessed data to the random forest algorithm module;

[0015] The random forest algorithm module is used to receive the data uploaded by the data preprocessing module and perform calculations, and send the calculation results to the single-chip control system.

[0016] Further: Preprocessing includes cleaning, denoising, normalization, smoothing and anomaly detection of data.

[0017] Further: the micro-wastewater discharge waterway includes a source water inlet, a first one-way valve, a pressure reducing valve, a primary filter element assembly, a solenoid valve C1, a flow meter F1, a first three-way valve, a resin, a second three-way valve, a booster pump C2, a reverse osmosis module, a second one-way valve, a high-pressure switch W7, a flow meter F2, a third three-way valve, and a pure water outlet connected in sequence; and a fourth three-way valve, a fifth three-way valve, a third one-way valve, a sixth three-way valve, a seventh three-way valve, and an adjustable flow limiting valve connected in sequence;

[0018] Among them, the input end of the fourth three-way valve is connected to the wastewater output end of the reverse osmosis module; the other input end of the first three-way valve is connected to the output end of the adjustable flow limiting valve; the other input end of the second three-way valve is connected to the first TDS / NTC sensor; the other input end of the third three-way valve is connected to the second TDS / NTC sensor; the other output end of the fourth three-way valve is connected to the solenoid valve KA; the other input end of the fifth three-way valve is connected to the pressure sensor; the other output end of the sixth three-way valve is connected to the pressure barrel; the other output end of the seventh three-way valve is connected to the domestic water outlet.

[0019] Furthermore: the sensor group includes a first TDS / NTC sensor, a second TDS / NTC sensor, a pressure sensor, a flow meter F1 and a flow meter F2 in the micro-wastewater discharge waterway.

[0020] Further: when the TDS value uploaded by the first TDS / NTC sensor to the single-chip control system is greater than or equal to the set threshold, the single-chip control system opens the solenoid valve KA, and the water in the pressure barrel flows through the sixth three-way valve, the seventh three-way valve, the adjustable flow limiting valve, the first three-way valve, the resin, the second three-way valve, the booster pump C2, the reverse osmosis module, the fourth three-way valve and the solenoid valve KA, thereby discharging all the water in the pressure barrel; when the TDS value uploaded by the first TDS / NTC sensor to the single-chip control system does not reach the set threshold, the solenoid valve KA remains closed, and the single-chip control system controls the water in the pressure barrel to pass through the sixth three-way valve, the seventh three-way valve, the adjustable flow limiting valve in turn, and flows to the reverse osmosis module after mixing with the source water through the first three-way valve. The first TDS / NTC sensor detects the TDS value of the mixed water and uploads the detection result to the single-chip control system to form a wastewater recycling treatment.

[0021] Further: The training process of the random forest algorithm module specifically includes the following steps:

[0022] A1. Collect the water pressure and flow data of the wastewater under the condition of stable operation of the micro-wastewater discharge waterway and maximum wastewater utilization, as well as the corresponding opening percentage of the adjustable flow limiting valve under the wastewater water pressure and flow, to form a training set; the water pressure of the wastewater is the data detected by the pressure sensor, and the flow rate of the wastewater is the data detected by the flow meter F1;

[0023] A1. Randomly select m samples with replacement from the training set to form a sample set;

[0024] A2. Calculate the mean of all samples in the current node on each feature in the random forest algorithm Its expression is:

[0025]

[0026] x i =[Pi ,F i ]

[0027] Where n is the number of samples in the current node; x i represents the characteristic value of the i-th record; f j Indicates the type of feature. For the current node, it includes two features: water pressure f1 and flow f2; x i [f j ] represents the value of the i-th sample on each feature; P i represents the water pressure of the i-th record, F i Indicates the flow of the i-th record;

[0028] A3. According to the mean The sample set S in the current node in the random forest algorithm is divided into two parts, S1 and S2, and its expression is:

[0029]

[0030] Among them, S1 is the sample set whose eigenvalue is less than or equal to the mean, and S2 is the sample set whose eigenvalue is greater than the mean;

[0031] A4. Calculate the square error reduction ΔE before and after each split point split, and its expression is:

[0032] ΔE=E before -E after

[0033]

[0034] Where E before is the square error before splitting, E after is the square error after splitting; y i Indicates the valve opening percentage corresponding to the i-th record; for all y in S i The average value of is all y in S1 i The average value of is all y in S2 i The split point is the mean of all samples in the current node on each feature.

[0035] Select the feature and split point with the largest square error reduction as the best split for the current node;

[0036] A5. Repeat steps A2 to A4 for each child node in the random forest algorithm until the stopping condition is met. When the tree in the random forest algorithm reaches the stopping condition, the current node is recorded as a leaf node and the predicted value of the leaf node is calculated. Its expression is:

[0037]

[0038] Among them, L is the sample set contained in the leaf node, |L| is the number of samples in the set L, and y i is the sample x i The target variable value of

[0039] A6. Calculate the average of all tree predictions As the final prediction result, its expression is:

[0040]

[0041] Where B is the number of regression trees, is the predicted value of the bth tree, that is, the predicted value of the leaf node

[0042]

[0043] A7. Evaluate and optimize the random forest algorithm and end the training.

[0044] Furthermore: the single-chip microcomputer control system uploads the received data to the data preprocessing module, the data preprocessing module processes the received data and then uploads it to the random forest algorithm module, the random forest algorithm module predicts the valve opening percentage of the adjustable flow limiting valve, and then sends the prediction result to the single-chip microcomputer control system, and the single-chip microcomputer control system controls the valve opening size of the adjustable flow limiting valve according to the prediction result.

[0045] The beneficial effects of the present invention are:

[0046] 1. The present invention introduces a random forest algorithm for real-time prediction and combines it with intelligent control of wastewater circulation and reprocessing technology. The present invention can minimize wastewater discharge and recycle wastewater after reprocessing, thereby significantly improving water resource utilization.

[0047] 2. The present invention can effectively avoid the damage of high-concentration impurities in the wastewater to equipment such as RO membranes, extend the service life of the equipment, and reduce maintenance costs by intelligently controlling the path and flow rate of the wastewater circulation and the operating parameters of the wastewater reprocessing unit.

[0048] 3. The present invention sets a threshold value for wastewater TDS, and discharges and resets the wastewater in a timely manner when the wastewater quality deteriorates, thereby ensuring the stability and reliability of the treated water quality and meeting the user's high requirements for water quality.

[0049] 4. By using a single-chip microcomputer as the core controller of the system, the present invention can collect, process and analyze the operating data of the system in real time, quickly respond to changes in the system status, automatically adjust the operating parameters of the system, and improve the system's adaptability and response speed.

[0050] 5. The intelligent control system of the present invention can automatically execute control logic, reduce manual intervention, reduce operation complexity, and improve system operation efficiency and reliability. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 It is a data flow diagram of the system proposed by the present invention;

[0052] Figure 2 This is a schematic diagram of the micro-wastewater discharge waterway. DETAILED DESCRIPTION

[0053] The specific implementation modes of the present invention are described below so that those skilled in the art can understand the present invention. However, it should be clear that the present invention is not limited to the scope of the specific implementation modes. For those of ordinary skill in the art, as long as various changes are within the spirit and scope of the present invention as defined and determined by the attached claims, these changes are obvious, and all inventions and creations utilizing the concept of the present invention are protected.

[0054] In one embodiment of the present invention, Figure 1 As shown, the reverse osmosis micro-wastewater control system provided by the present invention based on AI algorithm prediction includes:

[0055] Micro wastewater discharge waterway is used to produce pure water and treat wastewater.

[0056] The single-chip control system is used to receive the real-time data uploaded by the sensor group in the micro-wastewater discharge waterway, upload the real-time data to the data preprocessing module, receive the calculation results of the random forest algorithm module, and control the working process of the micro-wastewater discharge waterway according to the received data. The single-chip microcomputer will give an experience-based preset correspondence between water pressure, flow and the opening percentage of the adjustable flow limiting valve. Before receiving the correction result from the random forest algorithm module, the system adjusts the opening percentage of the adjustable flow limiting valve based on the preset threshold. The sensor group includes the first TDS / NTC sensor, the second TDS / NTC sensor, the pressure sensor, the flowmeter F1 and the flowmeter F2 in the micro-wastewater discharge waterway. The components in the sensor group are connected to the single-chip microcomputer control system through a data line, and the real-time data is transmitted to the single-chip microcomputer for processing. When the system fails or is abnormal, the single-chip microcomputer sends an alarm signal to remind the user to perform maintenance and repair.

[0057] Data preprocessing module: used to preprocess the data uploaded by the single-chip microcomputer, and upload the preprocessed data to the random forest algorithm module. Preprocessing includes data cleaning, denoising, normalization, smoothing and anomaly detection; data preprocessing can identify and process abnormal data points caused by sensor failure, communication errors or other reasons, and reduce data jitter caused by sensor noise or instantaneous fluctuations. An important responsibility of the data preprocessing module is to adjust the corresponding relationship data of water pressure, flow and wastewater valve opening percentage uploaded by the single-chip microcomputer control system under the TDS threshold preset based on experience in the early stage. For example: based on the initial threshold, case 1: when the water pressure is 0.3Mpa, the flow value may be 800ml / min; case 2: when the water pressure is 0.8Mpa, the flow value may be 300ml / min. Case 2 is unreasonable, and the random forest algorithm module will mark this data as abnormal for learning.

[0058] The random forest algorithm module is used to receive the data uploaded by the data preprocessing module and perform calculations, and send the calculation results to the microcontroller.

[0059] The single-chip microcomputer control system uploads the received data to the data preprocessing module, which processes the received data and uploads it to the random forest algorithm module. The random forest algorithm module predicts the valve opening percentage of the adjustable flow limiting valve, and then sends the prediction result to the single-chip microcomputer control system. The single-chip microcomputer control system adjusts the valve opening size of the adjustable flow limiting valve in advance according to the prediction result to cope with the sudden change in water pressure and stabilize the water pressure at the target value (0.5~0.6MPa).

[0060] like Figure 2 As shown, the micro-wastewater discharge waterway includes a source water inlet, a first one-way valve, a pressure reducing valve, a primary filter element combination, a solenoid valve C1, a flow meter F1, a first three-way valve, a resin, a second three-way valve, a booster pump C2, a reverse osmosis module, a second one-way valve, a high-pressure switch W7, a flow meter F2, a third three-way valve, and a pure water outlet connected in sequence; and a fourth three-way valve, a fifth three-way valve, a third one-way valve, a sixth three-way valve, a seventh three-way valve, and an adjustable flow limiting valve connected in sequence; wherein the input end of the fourth three-way valve is connected to the wastewater output end of the reverse osmosis module; the other input end of the first three-way valve is connected to the output end of the adjustable flow limiting valve; the other input end of the second three-way valve is connected to the first TDS / NTC sensor; the other input end of the third three-way valve is connected to the second TDS / NTC sensor; the other output end of the fourth three-way valve is connected to the solenoid valve KA; the other input end of the fifth three-way valve is connected to the pressure sensor; the other output end of the sixth three-way valve is connected to the pressure barrel; the other output end of the seventh three-way valve is connected to the domestic water outlet.

[0061] In this embodiment, when the system starts to run, the source water enters the micro-wastewater discharge waterway from the source water inlet, and the source water flows through the first one-way valve to ensure that the water flow can only move forward in one direction to prevent backflow, and then passes through the pressure reducing valve to reduce the water pressure to a level suitable for system processing, and then the source water enters the primary filter element combination (the primary filter element is combined according to the model of each water purifier) ​​to preliminarily filter out impurities in the water. The solenoid valve C1 controls whether the entire system is influent. After the solenoid valve C1 is opened, the source water flows through the first three-way valve to enter the ion exchange resin to reduce the calcium and magnesium ion content in the source water, thereby reducing the possibility of RO membrane crystallization; a flow meter F1 is provided at the first three-way valve to record the water flow; after the source water passes through the resin, it flows through the second three-way valve and the booster pump C2 to enter the reverse osmosis module, and the second three-way valve is provided with a first TDS / NTC sensor to detect the TDS value (soluble solid content value) of the water flowing through the second three-way valve, and the booster pump C2 is used to increase the water pressure to ensure that the water flowing through the booster pump C2 has enough pressure to be filtered through the reverse osmosis module. The reverse osmosis module contains a reverse osmosis membrane (RO membrane). After the water flows into the reverse osmosis module, it is deeply purified through the RO membrane to remove impurities such as soluble solids. The water that passes through the RO membrane is pure water, and the water that does not pass through the RO membrane has a high TDS value and is waste water. The pure water flows to the second one-way valve and passes through the high-pressure switch W7 to prevent the equipment from running under high pressure for a long time and avoid damage to the equipment due to high pressure. Finally, the pure water flows to the third three-way valve and flows out of the pure water outlet for direct drinking by users or for use in high-precision industrial laboratories. The flowmeter F2 and the second TDS / NTC sensor are provided at the third three-way valve. The flowmeter F2 is used to record the flow of pure water after filtration, and the second TDS / NTC sensor is used to detect the TDS value of pure water and its conductivity and temperature. Pure water with different TDS values ​​or conductivity and temperature can be obtained according to actual needs. Wastewater that does not pass through the RO membrane flows to the fourth three-way valve, where a solenoid valve KA is provided for discharging wastewater. When wastewater needs to be discharged, the solenoid valve KA automatically opens. When the conditions for discharging wastewater are not met, the solenoid valve KA closes, and wastewater continues to flow through the fifth three-way valve, the third one-way valve, and the sixth three-way valve into the pressure barrel. The pressure sensor at the fifth three-way valve is used to monitor and control the flow, pressure, and circulation path of wastewater in real time.

[0062] When the TDS value uploaded by the first TDS / NTC sensor to the single-chip control system is greater than or equal to the set threshold (the threshold can be adjusted according to the actual operation of the system and user needs to maximize the utilization of wastewater and optimize the system performance), the single-chip control system opens the solenoid valve KA, and the water in the pressure barrel flows through the sixth three-way valve, the seventh three-way valve, the adjustable flow limiting valve, the first three-way valve, the resin, the second three-way valve, the booster pump C2, the reverse osmosis module, the fourth three-way valve and the solenoid valve KA, thereby discharging all the water in the pressure barrel and restarting the cycle from an empty barrel to avoid the impact of deterioration of wastewater quality on system performance. ; When the TDS value uploaded by the first TDS / NTC sensor to the single-chip control system does not reach the set threshold, the solenoid valve KA remains closed, and the single-chip control system controls the water in the pressure barrel to pass through the sixth three-way valve, the seventh three-way valve, and the adjustable flow limiting valve in turn, and flows to the reverse osmosis module after mixing with the source water through the first three-way valve. The first TDS / NTC sensor detects the TDS value of the mixed water and uploads the detection result to the single-chip control system to form a wastewater circulation treatment. When the wastewater that does not reach the discharge threshold flows through the seventh three-way valve, it can also be used as domestic water for users to wash hands and vegetables. The adjustable flow limiting valve is used to adjust the water flow rate. The opening and closing size of the adjustable flow limiting valve determines the pressure of the wastewater and is used to control the flow rate of the wastewater. The single-chip control system adjusts the opening and closing size of the adjustable flow limiting valve in real time according to the data uploaded by the pressure sensor to avoid the situation where the source water cannot enter or the wastewater cannot go out, and at the same time avoids the rupture or shortening of the life of the RO membrane caused by the sudden change of the source water pressure.

[0063] The training process of the random forest algorithm module specifically includes the following steps:

[0064] A1. Collect the water pressure and flow data of the wastewater under the condition of stable operation of the micro-wastewater discharge waterway and maximum wastewater utilization, as well as the corresponding opening percentage of the adjustable flow limiting valve under the wastewater water pressure and flow, to form a training set; the water pressure of the wastewater is the data detected by the pressure sensor, and the flow rate of the wastewater is the data detected by the flow meter F1;

[0065] A1. Randomly select m samples with replacement from the training set to form a sample set;

[0066] A2. Calculate the mean of all samples in the current node on each feature in the random forest algorithm Its expression is:

[0067]

[0068] x i =[P i ,F i ]

[0069] Where n is the number of samples in the current node; x irepresents the characteristic value of the i-th record; f j Indicates the type of feature. For the current node, it includes two features: water pressure f1 and flow f2; x i [f j ] represents the value of the i-th sample on each feature; P i represents the water pressure of the i-th record, F i Indicates the flow of the i-th record;

[0070] A3. According to the mean The sample set S in the current node in the random forest algorithm is divided into two parts, S1 and S2, and its expression is:

[0071]

[0072] Among them, S1 is the sample set whose eigenvalue is less than or equal to the mean, and S2 is the sample set whose eigenvalue is greater than the mean;

[0073] A4. Calculate the square error reduction ΔE before and after each split point split, and its expression is:

[0074] ΔE=E before -E after

[0075]

[0076] Where E before is the square error before splitting, E after is the square error after splitting; y i Indicates the valve opening percentage corresponding to the i-th record; for all y in S i The average value of is all y in S1 i The average value of is all y in S2 i The split point is the mean of all samples in the current node on each feature.

[0077] Select the feature and split point with the largest square error reduction as the best split for the current node;

[0078] A5. Repeat steps A2 to A4 for each child node in the random forest algorithm until the stopping condition is met. When the tree in the random forest algorithm reaches the stopping condition, the current node is recorded as a leaf node and the predicted value of the leaf node is calculated. Its expression is:

[0079]

[0080] Among them, L is the sample set contained in the leaf node, |L| is the number of samples in the set L, and y i is the sample x i The target variable value of

[0081] A6. Calculate the average of all tree predictions As the final prediction result, its expression is:

[0082]

[0083] Where B is the number of regression trees, is the predicted value of the bth tree, that is, the predicted value of the leaf node

[0084]

[0085] A7. Evaluate and optimize the random forest algorithm and end the training.

[0086] In summary, the present invention realizes the micro-wastewater discharge and high-efficiency operation of the reverse osmosis water treatment system by introducing the random forest algorithm and intelligent control technology, significantly improves the utilization rate of water resources, optimizes the system performance, ensures the stability of the treated water quality, improves the adaptability and response speed of the system, realizes intelligent management, and has significant advantages and positive effects.

Claims

1. A reverse osmosis micro-wastewater control system based on AI algorithm prediction, characterized in that: include: Micro wastewater discharge waterway, used to produce pure water and treat wastewater; A single-chip control system is used to receive real-time data uploaded by a sensor group in a micro-wastewater discharge waterway, upload the real-time data to a data preprocessing module, receive calculation results of a random forest algorithm module, and control the working process of the micro-wastewater discharge waterway according to the received data; Data preprocessing module: used to preprocess the data uploaded by the single-chip control system and upload the preprocessed data to the random forest algorithm module; The random forest algorithm module is used to receive the data uploaded by the data preprocessing module and perform calculations, and send the calculation results to the single-chip control system.

2. According to claim 1, a reverse osmosis micro-wastewater control system based on AI algorithm prediction is characterized in that: The preprocessing includes cleaning, denoising, normalization, smoothing and anomaly detection of data.

3. According to claim 1, a reverse osmosis micro-wastewater control system based on AI algorithm prediction is characterized in that: The micro-wastewater discharge waterway includes a source water inlet, a first one-way valve, a pressure reducing valve, a primary filter element assembly, a solenoid valve C1, a flow meter F1, a first three-way valve, a resin, a second three-way valve, a booster pump C2, a reverse osmosis module, a second one-way valve, a high-pressure switch W7, a flow meter F2, a third three-way valve, and a pure water outlet connected in sequence; and a fourth three-way valve, a fifth three-way valve, a third one-way valve, a sixth three-way valve, a seventh three-way valve, and an adjustable flow limiting valve connected in sequence; Wherein, the input end of the fourth three-way valve is connected to the wastewater output end of the reverse osmosis module; The other input end of the first three-way valve is connected to the output end of the adjustable flow limiting valve; The other input end of the second three-way valve is connected to the first TDS / NTC sensor; The other input end of the third three-way valve is connected to the second TDS / NTC sensor; The other output end of the fourth three-way valve is connected to the solenoid valve KA; The other input end of the fifth three-way valve is connected to a pressure sensor; The other output end of the sixth three-way valve is connected to the pressure barrel; The other output end of the seventh three-way valve is connected to the domestic water outlet.

4. According to claim 3, a reverse osmosis micro-wastewater control system based on AI algorithm prediction is characterized in that: The sensor group includes a first TDS / NTC sensor, a second TDS / NTC sensor, a pressure sensor, a flow meter F1 and a flow meter F2 in the micro wastewater discharge waterway.

5. According to claim 4, a reverse osmosis micro-wastewater control system based on AI algorithm prediction is characterized in that: When the TDS value uploaded by the first TDS / NTC sensor to the single-chip control system is greater than or equal to the set threshold, the single-chip control system opens the solenoid valve KA, and the water in the pressure barrel flows through the sixth three-way valve, the seventh three-way valve, the adjustable flow limiting valve, the first three-way valve, the resin, the second three-way valve, the booster pump C2, the reverse osmosis module, the fourth three-way valve and the solenoid valve KA, thereby discharging all the water in the pressure barrel; when the TDS value uploaded by the first TDS / NTC sensor to the single-chip control system does not reach the set threshold, the solenoid valve KA remains closed, and the single-chip control system controls the water in the pressure barrel to pass through the sixth three-way valve, the seventh three-way valve, the adjustable flow limiting valve in turn, and flows to the reverse osmosis module after mixing with the source water through the first three-way valve. The first TDS / NTC sensor detects the TDS value of the mixed water and uploads the detection result to the single-chip control system, forming a wastewater recycling treatment.

6. According to claim 3, a reverse osmosis micro-wastewater control system based on AI algorithm prediction is characterized in that: The training process of the random forest algorithm module specifically includes the following steps: A1. Collect the water pressure and flow data of the wastewater under the condition of stable operation of the micro-wastewater discharge waterway and maximum wastewater utilization, as well as the corresponding opening percentage of the adjustable flow limiting valve under the wastewater water pressure and flow, to form a training set; the water pressure of the wastewater is the data detected by the pressure sensor, and the flow rate of the wastewater is the data detected by the flow meter F1; A1. Randomly select m samples with replacement from the training set to form a sample set; A2. Calculate the mean of all samples in the current node on each feature in the random forest algorithm Its expression is: x i =[P i ,F i ] Where n is the number of samples in the current node; x i represents the characteristic value of the i-th record; f j Indicates the type of feature. For the current node, it includes two features: water pressure f1 and flow f2; x i [f j ] represents the value of the i-th sample on each feature; P i represents the water pressure of the i-th record, F i Indicates the flow of the i-th record; A3. According to the mean The sample set S in the current node in the random forest algorithm is divided into two parts, S1 and S2, and its expression is: Among them, S1 is the sample set whose eigenvalue is less than or equal to the mean, and S2 is the sample set whose eigenvalue is greater than the mean; A4. Calculate the square error reduction ΔE before and after each split point split, and its expression is: ΔE=E before -E after Where E before is the square error before splitting, E after is the square error after splitting; y i Indicates the valve opening percentage corresponding to the i-th record; for all y in S i The average value of is all y in S1 i The average value of is all y in S2 i The split point is the mean of all samples in the current node on each feature. Select the feature and split point with the largest square error reduction as the best split for the current node; A5. Repeat steps A2 to A4 for each child node in the random forest algorithm until the stopping condition is met. When the tree in the random forest algorithm reaches the stopping condition, the current node is recorded as a leaf node and the predicted value of the leaf node is calculated. Its expression is: Among them, L is the sample set contained in the leaf node, |L| is the number of samples in the set L, and y i is the sample x i The target variable value of A6. Calculate the average of all tree predictions As the final prediction result, its expression is: Where B is the number of regression trees, is the predicted value of the bth tree, that is, the predicted value of the leaf node A7. Evaluate and optimize the random forest algorithm and end the training.

7. According to claim 3, a reverse osmosis micro-wastewater control system based on AI algorithm prediction is characterized in that: The single-chip microcomputer control system uploads the received data to the data preprocessing module. The data preprocessing module processes the received data and then uploads it to the random forest algorithm module. The random forest algorithm module predicts the valve opening percentage of the adjustable flow limiting valve, and then sends the prediction result to the single-chip microcomputer control system. The single-chip microcomputer control system controls the valve opening size of the adjustable flow limiting valve according to the prediction result.

Images