AI hierarchical monitoring method and system for reverse osmosis operation state based on pure water system

Abstract

This invention is applicable to the field of digital boiler water treatment, providing an AI-based hierarchical monitoring method and system for the operation status of reverse osmosis systems in pure water systems. The specific method includes: real-time acquisition of molecular vibrational spectra of contaminants on the reverse osmosis membrane surface using a Raman spectroscopy probe, matching them with a pre-set contaminant database for contaminant type identification; monitoring the differential pressure distribution of each filter element in the security filter using a differential pressure sensor array, calculating the differential pressure non-uniformity to predict the remaining lifespan of the filter element; acquiring the surface temperature distribution of the membrane module using an infrared thermal imager, analyzing the temperature difference between the permeate and concentrate sides to assess permeate uniformity; monitoring the input power and output parameters of the high-pressure pump based on a Hall effect sensor, establishing a real-time energy efficiency mapping model; and using a pre-set algorithm to fuse and analyze multi-source monitoring data to generate hierarchical operation and maintenance decisions. This method and system, through the fusion of multi-source sensors and AI, can achieve intelligent monitoring of the entire reverse osmosis system chain, and realize intelligent early warning of contamination and filter element lifespan prediction, further reducing reagent costs and equipment energy consumption.

Description

Technical Field

[0001] This invention belongs to the field of digital boiler water treatment, and particularly relates to an AI-based hierarchical monitoring system for the reverse osmosis operation status of a pure water system. Background Technology

[0002] In the operation of reverse osmosis systems for pure water, traditional monitoring methods typically focus on explicit indicators such as conductivity, pressure, and flow rate to control the basic operating status of the system. However, the efficient and stable operation of reverse osmosis systems is also significantly affected by some implicit characteristics that cannot be directly monitored by conventional instruments. Furthermore, their cumulative effects can lead to membrane element performance degradation, decreased product water quality, and increased system energy consumption. Current technologies lack systematic monitoring methods for these implicit characteristics, making it difficult to identify potential faults in their early stages, thus affecting the long-term stable operation of the system.

[0003] To address the aforementioned issues, current technologies lack a comprehensive solution integrating the monitoring and control of latent characteristics, and cannot detect them promptly using conventional parameters. Therefore, there is an urgent need for a monitoring and control method capable of covering multi-dimensional latent characteristics to improve the operational reliability and maintenance efficiency of reverse osmosis systems. Summary of the Invention

[0004] The purpose of this invention is to provide an AI-based hierarchical monitoring method and system for the operation status of reverse osmosis systems based on pure water systems, aiming to solve the problems mentioned in the background art.

[0005] This invention is implemented as follows: On the one hand, based on the AI-based hierarchical monitoring method for the reverse osmosis operation status of a pure water system, the following steps are adopted:

[0006] The molecular vibrational spectra of pollutants on the surface of the reverse osmosis membrane are collected in real time using a Raman spectroscopy probe, and the pollutant type is identified by matching it with a preset pollutant database.

[0007] The differential pressure distribution of each filter element in the security filter is monitored by a differential pressure sensor array, and the differential pressure non-uniformity is calculated to predict the remaining life of the filter element.

[0008] The surface temperature distribution of the membrane module is obtained by using an infrared thermal imager, and the temperature difference between the permeate side and the concentrate side is analyzed to assess the uniformity of permeate production.

[0009] A real-time energy efficiency mapping model is established based on Hall effect sensor to monitor the input power and output parameters of high-pressure pump.

[0010] A preset algorithm is used to fuse and analyze multi-source monitoring data to generate hierarchical operation and maintenance decisions;

[0011] In the pollution type identification process, the degree of pollution is quantitatively characterized by a pollution index. The calculation process of the pollution index is as follows:

[0012] ;

[0013] In the formula, For the first The spectral intensity of each characteristic peak For the first The contamination weight coefficients corresponding to each characteristic peak This represents the number of characteristic peaks.

[0014] Furthermore, the step of using a differential pressure sensor array to monitor the differential pressure distribution of each filter element in the security filter and calculating the differential pressure non-uniformity to predict the remaining lifespan of the filter element specifically includes:

[0015] Collect differential pressure data for each filter element unit. Calculate differential pressure data average pressure difference and standard deviation ;

[0016] The average pressure difference and standard deviation The calculation process is as follows:

[0017]

[0018] ;

[0019] In the formula, This represents the total number of filter elements.

[0020] Define the remaining lifespan of the filter element. :

[0021] ;

[0022] In the formula, This is the preset threshold for differential pressure unevenness.

[0023] Furthermore, the step of acquiring the surface temperature distribution of the membrane module using an infrared thermal imager and analyzing the temperature difference between the permeate and concentrate sides to assess the uniformity of the permeate specifically includes:

[0024] Collect the temperature distribution on the product water side. Temperature distribution on the concentrate side Calculate the maximum temperature difference ;

[0025] The maximum temperature difference The calculation process is as follows:

[0026] ;

[0027] In the formula, These are the temperature measurement point numbers for the product water side and the concentrate side, respectively.

[0028] When the maximum temperature difference When the temperature difference exceeds the preset warning value, a warning message for water production uniformity is triggered.

[0029] Furthermore, the establishment of a real-time energy efficiency mapping model based on monitoring the input power and output parameters of the high-pressure pump using a Hall effect sensor specifically includes:

[0030] Real-time calculation of high-pressure pump efficiency ;

[0031] The efficiency of the high-pressure pump The calculation process is as follows:

[0032] ;

[0033] In the formula, For water density, It is the acceleration due to gravity. For the high-pressure pump head, For real-time traffic, Input power;

[0034] When the high-pressure pump efficiency The efficiency drop exceeds the preset threshold. At that time, an energy consumption anomaly warning message is triggered.

[0035] Furthermore, the step of using a preset algorithm to fuse and analyze multi-source monitoring data specifically includes:

[0036] Construct a multi-parameter correlation model;

[0037] The input variables of the multi-parameter correlation model include: raw water hardness. Operational recovery rate Scale inhibitor dosage concentration ;

[0038] Among them, the operational recovery rate The calculation method is as follows:

[0039] ;

[0040] In the formula, For water production flow rate, This refers to the influent flow rate;

[0041] The output variables of the multi-parameter association model include:

[0042] Pollution trend forecast ;

[0043] Optimal dosage ;

[0044] An LSTM neural network is used to train the historical data, and the training objective function is:

[0045] ;

[0046] In the formula, These are the model's predicted values. These are actual monitored values. The length of the time series. For index variables of time series;

[0047] Based on the trained LSTM network, the input time series is... The normalized parameters, output the future Pollution prediction values ​​at any time ,and The range of values ​​is ;

[0048] when the future Pollution prediction values ​​at any time Calculate the optimal pesticide dosage if the pollution index exceeds the threshold. and with the optimal dosage Add the medicine;

[0049] The optimal dosage of the drug The calculation process is as follows:

[0050] ;

[0051] In the formula, 2 represents the basic dosage, 4 represents the minimum dosage under low pollution conditions, and 2 represents the maximum increment.

[0052] ;

[0053] In the formula, 75 is the recovery rate adjustment factor, 0.01 is the baseline recovery rate, and 75 is the correction factor.

[0054] ;

[0055] In the formula, 292 is the hardness correction factor, and 292 is the raw water hardness reference value;

[0056] .

[0057] Furthermore, the multi-parameter association model specifically includes:

[0058] Calculate the Langerile index ;

[0059] The Langerile index The calculation process is as follows:

[0060]

[0061] ;

[0062] In the formula, It is the solubility product of calcium carbonate. dissociation constant of carbonic acid This refers to the calcium ion concentration. carbonate concentration, Bicarbonate concentration, This represents the carbon dioxide concentration.

[0063] Furthermore, the method of using a preset algorithm to fuse and analyze multi-source monitoring data specifically includes:

[0064] Calculate and generate equipment health index ;

[0065] The equipment health index The calculation process is as follows:

[0066] ;

[0067] in,

[0068] Corresponding membrane fouling index, For pollution weighting coefficients, This is the normalized value of the pollution index;

[0069] Corresponding to filter life, This is the filter element weighting coefficient. This is the normalized value of the remaining lifetime;

[0070] Corresponding to the efficiency of the high-pressure pump, This is the weighting coefficient for the efficiency of the high-pressure pump. The normalized value for efficiency reduction;

[0071] Corresponding temperature uniformity index, Temperature weighting coefficient, This is the normalized value for temperature difference;

[0072] And health index The weight allocation is based on the following fault handling priorities:

[0073] Membrane fouling 40%, filter life 20%, high-pressure pump efficiency 30%, temperature uniformity index 10%.

[0074] Furthermore, the generation of hierarchical operation and maintenance decisions specifically includes:

[0075] When the pollution index When the preset pollution index threshold is reached at 30%-50%, a first-level instruction for citric acid rinsing of the membrane surface is generated and sent.

[0076] When the remaining lifespan of the filter element is... At that time, a secondary filter replacement command is generated and sent;

[0077] When the high-pressure pump reaches the efficiency vibration protection condition or health index When <0.3, generate and send the Phase II backup startup Level 3 command;

[0078] The specific conditions for efficiency vibration protection are as follows:

[0079] Based on the rated speed of the high-pressure pump, the effective value of vibration velocity is obtained by using an IEPE accelerometer installed in the horizontal or vertical direction of the high-pressure pump. Vibration velocity amplitude at 1X frequency Acceleration in the characteristic frequency band of rolling bearing outer ring fault ;

[0080] The following conditions must be met simultaneously for efficient vibration protection:

[0081] .

[0082] As a further aspect of the present invention, the generation of hierarchical operation and maintenance decisions specifically includes:

[0083] when At that time, a command is generated and sent to adjust the scale inhibitor dosage to 4-6 ppm;

[0084] If the membrane surface experiences the maximum temperature difference within the target time period after being cleaned with citric acid. If the temperature does not drop below 1.0℃, it is determined to be colloidal contamination, and a flushing instruction for sodium hydroxide and sodium hypochlorite is generated and sent.

[0085] An AI-based hierarchical monitoring system for the reverse osmosis operation status of a pure water system, comprising:

[0086] The real-time acquisition module is used to acquire the molecular vibrational spectra of contaminants on the surface of the reverse osmosis membrane in real time using a Raman spectroscopy probe;

[0087] The matching and identification module is used to match a preset pollution database to identify the type of pollution.

[0088] The array monitoring module is used to monitor the differential pressure distribution of each filter element in the security filter using an array of differential pressure sensors.

[0089] The calculation module is used to calculate the differential pressure unevenness to predict the remaining life of the filter element;

[0090] The acquisition and analysis module is used to acquire the surface temperature distribution of the membrane module through an infrared thermal imager, and analyze the temperature difference between the permeate side and the concentrate side to evaluate the uniformity of permeate production.

[0091] The mapping model module is used to establish a real-time energy efficiency mapping model based on the monitoring of the input power and output parameters of the high-pressure pump using Hall effect sensors.

[0092] The fusion analysis module is used to fuse and analyze multi-source monitoring data using a preset algorithm;

[0093] The generation module is used to generate hierarchical operation and maintenance decisions.

[0094] The present invention provides an AI-based hierarchical monitoring method and system for the operation status of a reverse osmosis system. This method and system integrate multi-source sensors with AI to achieve intelligent monitoring of the entire reverse osmosis system chain, and realize intelligent early warning of pollution and prediction of filter life. At the same time, it further reduces reagent costs and equipment energy consumption, and solves the problems of insufficient collaborative control in traditional systems, thereby effectively improving water production efficiency and safety. Attached Figure Description

[0095] Figure 1 This is the main flowchart of an AI-based hierarchical monitoring method for the reverse osmosis operation status of a pure water system.

[0096] Figure 2 This is a flowchart of the AI-based hierarchical monitoring method for the reverse osmosis operation status of a pure water system, which uses a preset algorithm to fuse and analyze multi-source monitoring data.

[0097] Figure 3 This is a flowchart of the first embodiment of the AI-based hierarchical monitoring method for the reverse osmosis operation status of a pure water system, which generates hierarchical operation and maintenance decisions.

[0098] Figure 4 This is a flowchart of the second embodiment of the AI-based hierarchical monitoring method for the reverse osmosis operation status of a pure water system, which generates hierarchical operation and maintenance decisions.

[0099] Figure 5 This is the main structure diagram of an AI-based hierarchical monitoring system for the reverse osmosis operation status of a pure water system. Detailed Implementation

[0100] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0101] The specific implementation of the present invention will be described in detail below with reference to specific embodiments.

[0102] The present invention provides an AI-based hierarchical monitoring method and system for the reverse osmosis operation status of a pure water system, which solves the technical problems in the background art.

[0103] like Figure 1 The diagram shown is the main flowchart of an AI-based hierarchical monitoring method for the reverse osmosis operating status of a pure water system, provided in an embodiment of the present invention. The AI-based hierarchical monitoring method for the reverse osmosis operating status of a pure water system includes:

[0104] Step S100: The molecular vibrational spectra of pollutants on the surface of the reverse osmosis membrane are acquired in real time using a Raman spectroscopy probe, and the pollutant type is identified by matching it with a preset pollutant database;

[0105] Step S200: Use a differential pressure sensor array to monitor the differential pressure distribution of each filter element in the security filter, and calculate the differential pressure non-uniformity to predict the remaining life of the filter element;

[0106] Step S300: Obtain the surface temperature distribution of the membrane module using an infrared thermal imager, and analyze the temperature difference between the permeate side and the concentrate side to assess the permeate uniformity.

[0107] Step S400: Based on the Hall effect sensor, monitor the input power and output parameters of the high-pressure pump and establish a real-time energy efficiency mapping model;

[0108] Step S500: Use a preset algorithm to fuse and analyze multi-source monitoring data to generate hierarchical operation and maintenance decisions;

[0109] In the process of pollution type identification, the degree of pollution is quantitatively characterized by the pollution index. The calculation process of the pollution index is as follows:

[0110] ;

[0111] In the formula, For the first The spectral intensity of each characteristic peak For the first The contamination weight coefficients corresponding to each characteristic peak The number of characteristic peaks;

[0112] In this embodiment, a full-chain status monitoring system is constructed across multiple technological dimensions. Raman spectroscopy is used to identify the types of contaminants on the membrane surface in real time. A differential pressure sensor array is combined to quantify the degree of clogging of the security filter cartridge and predict its remaining lifespan. Infrared thermal imaging is used to analyze the temperature difference between the permeate and concentrate sides of the membrane module to assess uniformity. A real-time energy consumption-efficiency model of the high-pressure pump is established based on the Hall effect and vibration sensors. For membrane module monitoring, a HORIBA XploRA miniature Raman spectroscopy probe can be used, with a sampling frequency of 1 time / minute and a spectral resolution of 1-5 cm⁻¹. It is installed on the side of the first-stage reverse osmosis pressure vessel via a flange interface, with an optical fiber transmission distance ≤2m, ensuring real-time data acquisition at 1000-1800cm. The membrane surface contaminant spectrum is measured in 10-band. A contamination database pre-stores characteristic peak parameters for calcium carbonate, iron oxides, etc. Threshold settings can be based on critical contamination states where permeate flow decreases by 10% or conductivity increases by 10%. An early warning is triggered when the contamination index exceeds 1.5 times the baseline value. A Rosemount 3051S differential pressure sensor array is used, axially arranged inside the eight filter cartridges of the Φ350×1650mm security filter, with a sampling frequency of 10Hz. The preset differential pressure non-uniformity threshold can be set to 0.03MPa (i.e., 30% of the average differential pressure). When the differential pressure of a filter cartridge deviates from the average value by more than this threshold, a lifespan prediction model is established based on historical filter cartridge contamination data. An FLIRA615 infrared thermal imager is used, fixed above the reverse osmosis unit with a bracket, its monitoring angle perpendicular to the pressure vessel surface, covering the entire surface of the R8040C membrane module. The determination of the preset temperature difference warning value between the permeate and concentrate sides is based on experimental data on the reverse osmosis operating temperature range and the local temperature rise during membrane scaling. Furthermore, the AI ​​algorithm integrates operating parameters such as raw water hardness and recovery rate to calculate the equipment health index, thereby generating hierarchical operation and maintenance decisions and linking the Langerier index to regulate the scale inhibitor dosage, realizing intelligent management of the entire process from pollution warning and equipment protection to chemical optimization.

[0113] In a preferred embodiment of the present invention, the method of using a differential pressure sensor array to monitor the differential pressure distribution of each filter element in a security filter and calculating the differential pressure non-uniformity to predict the remaining lifespan of the filter element specifically includes:

[0114] Collect differential pressure data for each filter element unit. Calculate differential pressure data average pressure difference and standard deviation ;

[0115] Mean pressure difference and standard deviation The calculation process is as follows:

[0116]

[0117] ;

[0118] In the formula, This represents the total number of filter elements.

[0119] Define the remaining lifespan of the filter element. :

[0120] ;

[0121] In the formula, The preset pressure differential non-uniformity threshold is used;

[0122] In this embodiment, differentiated maintenance of filter cartridges is achieved through array-type differential pressure monitoring. The sensor used is the GEDruck PTX7517, which is threaded and installed inside the filter cartridge inlet of the security filter head. Each filter cartridge corresponds to one measuring point, with a total of eight sensors. In the differential pressure uniformity calculation, the average differential pressure... The baseline value is based on the condition of a new filter cartridge, and the standard deviation is... Reflecting the dispersion of filter element clogging, when When the pressure is ≥0.03MPa (equivalent to 60% of the initial pressure difference), local blockage can be identified. In practical applications, the filter replacement cycle is optimized from a fixed 3 months to dynamic adjustment, and replacement work orders are generated in advance.

[0123] In a preferred embodiment of the present invention, the surface temperature distribution of the membrane module is acquired using an infrared thermal imager, and the temperature difference between the permeate side and the concentrate side is analyzed to assess the uniformity of the permeate. Specifically, this includes:

[0124] Collect the temperature distribution on the product water side. Temperature distribution on the concentrate side Calculate the maximum temperature difference ;

[0125] Maximum temperature difference The calculation process is as follows:

[0126] ;

[0127] In the formula, These are the temperature measurement point numbers for the product water side and the concentrate side, respectively.

[0128] When the maximum temperature difference When the temperature difference exceeds the preset warning value, a warning message for water production uniformity is triggered.

[0129] In this embodiment, the thermal imaging monitoring system adopts a fixed installation scheme: the FLIR A615 thermal imager is mounted on top of the reverse osmosis frame via an L-shaped bracket, 1.5m away from the membrane module, and the field of view covers six R8040C pressure vessels arranged side by side. The temperature measurement range is set to 5-50℃, the emissivity is calibrated to 0.95, and a temperature field cloud map is generated every 10 minutes.

[0130] Maximum temperature difference The calculations are based on the thermodynamic characteristics of the product water and concentrate sides: Under normal operation, the temperature on the product water side is 0.5-1.0℃ lower than that on the concentrate side due to heat absorption during permeation. When membrane fouling obstructs water flow, the flow velocity on the concentrate side decreases locally, and frictional heat generation causes the temperature to rise. Experimental data show that when the fouled area of ​​a certain membrane element reaches 20%, At ≥1.5℃, the water production begins to decrease by 5%, so 1.5℃ can be set as the preset temperature difference warning value.

[0131] As a preferred embodiment of the present invention, the establishment of a real-time energy efficiency mapping model based on monitoring the input power and output parameters of a high-pressure pump using a Hall effect sensor specifically includes:

[0132] Real-time calculation of high-pressure pump efficiency ;

[0133] High pressure pump efficiency The calculation process is as follows:

[0134] ;

[0135] In the formula, For water density, It is the acceleration due to gravity. For the high-pressure pump head, For real-time traffic, Input power;

[0136] When the high-pressure pump efficiency The efficiency drop exceeds the preset threshold. When this occurs, an energy consumption anomaly warning message is triggered;

[0137] It should be understood that the monitoring solution is designed for the German Wilo 80MVLA45-80 high-pressure pump: the Hall effect sensor is selected as LEM LA 125-P, which is installed at the motor inlet terminal, with a measurement accuracy of 0.5%, and collects three-phase current and voltage data in real time; the IEPE vibration sensor adopts PCB 352C65, which is attached to the pump bearing housing in water or vertical direction through a magnetic base, with a sampling frequency of 10kHz.

[0138] Water density in efficiency calculation Take 1000 kg / m³, head Take the rated value of 150m and the real-time flow rate. Data is collected in real time via an electromagnetic flowmeter. A preset efficiency degradation threshold is set. It can be set to 5%, a preset efficiency degradation threshold. The determination is based on the pump performance curve: when the efficiency drops from 85% to 80% of the rated value, the net positive suction head (NPSH) in the pump approaches the critical value, at which point the effective value of the vibration velocity is... Typically ≥4.5 mm / s.

[0139] The vibration protection threshold is set in three levels: ≥7.1mm / s, 1X frequency amplitude ≥3.0mm / s, bearing outer ring failure frequency BPFO (106Hz) acceleration ≥25m / s².

[0140] Table 1 Equipment and Sensor Configuration Table

[0141]

[0142] like Figure 2 As shown, in a preferred embodiment of the present invention, the fusion analysis of multi-source monitoring data using a preset algorithm specifically includes:

[0143] Step S501: Construct a multi-parameter correlation model;

[0144] The input variables for the multi-parameter correlation model include: raw water hardness. Operational recovery rate Scale inhibitor dosage concentration ;

[0145] Among them, the operational recovery rate The calculation method is as follows:

[0146] ;

[0147] In the formula, For water production flow rate, This refers to the influent flow rate;

[0148] The output variables of the multi-parameter association model include:

[0149] Pollution trend forecast Optimal dosage of reagent ;

[0150] Step S502: Train the historical data using an LSTM neural network;

[0151] The training objective function is:

[0152] ;

[0153] Step S503: Based on the trained LSTM network, the input time series is... The normalized parameters, output the future Pollution prediction values ​​at any time ,and The range of values ​​is ;

[0154] Step S504: When the future Pollution prediction values ​​at any time Calculate the optimal pesticide dosage if the pollution index exceeds the threshold. and with the optimal dosage Add the medicine;

[0155] Optimal dosage The calculation process is as follows:

[0156] ;

[0157] In the formula, 2 represents the basic dosage, 4 represents the minimum dosage under low pollution conditions, and 2 represents the maximum increment.

[0158] ;

[0159] In the formula, 75 is the recovery rate adjustment factor, 0.01 is the baseline recovery rate, and 75 is the correction factor.

[0160] ;

[0161] In the formula, 292 is the hardness correction factor, and 292 is the raw water hardness reference value;

[0162]

[0163] In this embodiment, the hardness of the raw water is used as the reference. Operational recovery rate Current scale inhibitor dosage concentration Using historical data as input, the system learns the correlation patterns between "water quality-chemicals-pollution trends" to predict the appropriate scale inhibitor dosage for current operating conditions, thereby improving the predicted pollution trend values. Optimal, close to the system's stable operating range.

[0164] As another preferred embodiment of the present invention, the multi-parameter correlation model further includes:

[0165] Calculate the Langerile index ;

[0166] Langley index The calculation process is as follows:

[0167]

[0168] ;

[0169] In the formula, It is the solubility product of calcium carbonate. dissociation constant of carbonic acid This refers to the calcium ion concentration. carbonate concentration, Bicarbonate concentration, This represents the carbon dioxide concentration.

[0170] As another preferred embodiment of the present invention, the fusion analysis of multi-source monitoring data using a preset algorithm further includes:

[0171] Step S505: Calculate and generate the equipment health index ;

[0172] Equipment Health Index The calculation process is as follows:

[0173] ;

[0174] in,

[0175] Corresponding membrane fouling index, For pollution weighting coefficients, This is the normalized value of the pollution index;

[0176] Corresponding to filter life, This is the filter element weighting coefficient. This is the normalized value of the remaining lifetime;

[0177] Corresponding to the efficiency of the high-pressure pump, This is the weighting coefficient for the efficiency of the high-pressure pump. The normalized value for efficiency reduction;

[0178] Corresponding temperature uniformity index, Temperature weighting coefficient, This is the normalized value for temperature difference;

[0179] And health index The weight allocation is based on the following fault handling priorities:

[0180] Membrane fouling 40%, filter life 20%, high-pressure pump efficiency 30%, temperature uniformity index 10%.

[0181] like Figure 3As shown, in another preferred embodiment of the present invention, generating hierarchical operation and maintenance decisions specifically includes:

[0182] Step S511: When the pollution index When the preset pollution index threshold is reached at 30%-50%, a first-level instruction for citric acid rinsing of the membrane surface is generated and sent.

[0183] Step S512: When the remaining lifespan of the filter element is... At that time, a secondary filter replacement command is generated and sent;

[0184] Step S513: When the high-pressure pump reaches the efficiency vibration protection condition or health index When <0.3, generate and send the Phase II backup startup Level 3 command;

[0185] The specific conditions for efficiency vibration protection are as follows:

[0186] Based on the rated speed of the high-pressure pump, the effective value of vibration velocity is obtained by using an IEPE accelerometer installed in the horizontal or vertical direction of the high-pressure pump. Vibration velocity amplitude at 1X frequency Acceleration in the characteristic frequency band of rolling bearing outer ring fault ;

[0187] The following conditions must be met simultaneously for efficient vibration protection:

[0188] ;

[0189] In practical application, the first-level instruction (membrane flushing) is as follows: When the fouling index reaches the threshold of 30%-50%, corresponding to a 5%-10% decrease in permeate flow, low-pressure citric acid flushing is recommended. This utilizes the weak acidity of citric acid (2% concentration, 35℃, matching the membrane material's tolerance and contaminant dissolution characteristics) to dissolve carbonate scale and metal oxides adhering to the membrane surface, restoring the uniformity of water flow across the membrane. The second-level instruction (filter cartridge replacement) is as follows: When the remaining lifespan of the filter cartridge reaches a certain percentage... ≤0.3, meaning remaining lifespan <72 hours, generates and sends a level 2 filter replacement command; Level 3 warning (pump shutdown or multiple parameters reaching threshold): when efficiency drops by more than 5% and vibration exceeds the standard or multiple parameters reach the threshold, generates and sends a level 3 command to start the second-phase backup system. Within 10 seconds, closes the inlet and outlet valves of the faulty high-pressure pump and starts the second-phase backup reverse osmosis unit.

[0190] like Figure 4 As shown, in another preferred embodiment of the present invention, the generation of hierarchical operation and maintenance decisions further includes:

[0191] Step S521: When At that time, a command is generated and sent to adjust the scale inhibitor dosage to 4-6 ppm;

[0192] Step S522: If the membrane surface experiences the maximum temperature difference within the target time period after being cleaned with citric acid... If the temperature does not drop below 1.0℃, it is determined to be colloidal contamination, and a flushing instruction for sodium hydroxide and sodium hypochlorite is generated and sent.

[0193] In this embodiment, when applied, when At that time, the model automatically adjusts the dosage of the pesticide to 4-6 ppm, with an adjustment step of 0.5 ppm. This threshold is based on field experiments: when At this point, the calcium carbonate saturation on the concentrate side reaches 80%, and continued operation will easily lead to scaling. Set a target time period (such as a monitoring period of 1-2 hours after flushing) and monitor the maximum temperature difference again. If the maximum temperature difference does not drop below 1.0℃, it indicates that the membrane fouling is not simply carbonate or metal oxide fouling, and is therefore classified as colloidal fouling. Colloidal fouling (such as organic colloids, silica colloids, etc.) has strong adsorption and stability, and is difficult to remove effectively by conventional acid washing. In this case, a sodium hydroxide and sodium hypochlorite flushing command is generated and sent. Sodium hydroxide (adjusting the pH to 12) disrupts the colloidal stability, while sodium hypochlorite (effective chlorine concentration 500ppm) oxidizes and decomposes the colloidal organic matter. The synergistic effect achieves efficient cleaning of colloidal pollutants on the membrane surface, thereby restoring the uniformity and permeability of the membrane module's permeate.

[0194] like Figure 5 As shown, in another preferred embodiment of the present invention, an AI-based hierarchical monitoring system for the reverse osmosis operation status of a pure water system is provided, the system comprising:

[0195] The real-time acquisition module 100 is used to acquire the molecular vibrational spectrum of pollutants on the surface of the reverse osmosis membrane in real time through a Raman spectroscopy probe.

[0196] The matching and identification module 200 is used to match a preset pollution database to identify the type of pollution.

[0197] The array monitoring module 300 is used to monitor the differential pressure distribution of each filter element unit of the security filter using a differential pressure sensor array;

[0198] Calculation module 400 is used to calculate differential pressure unevenness to predict the remaining life of the filter element;

[0199] The acquisition and analysis module 500 is used to acquire the surface temperature distribution of the membrane module through an infrared thermal imager, and analyze the temperature difference between the permeate side and the concentrate side to evaluate the uniformity of permeate.

[0200] The mapping model module 600 is used to monitor the input power and output parameters of a high-pressure pump based on a Hall effect sensor and establish a real-time energy efficiency mapping model.

[0201] The fusion analysis module 700 is used to perform fusion analysis on multi-source monitoring data using a preset algorithm;

[0202] The generation module 800 is used to generate hierarchical operation and maintenance decisions.

[0203] In this embodiment, during application, the real-time acquisition module 100 acquires the molecular vibrational spectra of pollutants on the reverse osmosis membrane surface in real time using a Raman spectroscopy probe; the matching and identification module 200 matches the pollutant type against a preset pollutant database; the array monitoring module 300 uses a differential pressure sensor array to monitor the differential pressure distribution of each filter element in the security filter; the calculation module 400 calculates the differential pressure non-uniformity to predict the remaining lifespan of the filter element; the acquisition and analysis module 500 acquires the surface temperature distribution of the membrane module using an infrared thermal imager and analyzes the temperature difference between the permeate and concentrate sides to assess the uniformity of permeate; the input power and output parameters of the high-pressure pump are monitored based on a Hall effect sensor; the mapping model module 600 establishes a real-time energy efficiency mapping model; the fusion analysis module 700 uses a preset algorithm to perform fusion analysis on multi-source monitoring data; and the generation module 800 generates hierarchical operation and maintenance decisions.

[0204] The above embodiments of the present invention provide an AI-based hierarchical monitoring method for the operation status of a reverse osmosis system based on a pure water system, and also provide an AI-based hierarchical monitoring system for the operation status of a reverse osmosis system based on a pure water system. This system constructs a full-link status monitoring system across technical dimensions, utilizes Raman spectroscopy to identify membrane surface contaminant types in real time, combines a differential pressure sensor array to quantify the degree of clogging of the security filter cartridge and predict its remaining lifespan, analyzes the temperature difference between the permeate and concentrate sides of the membrane module using infrared thermal imaging to assess uniformity, and establishes a real-time energy consumption-efficiency model for the high-pressure pump based on the Hall effect and vibration sensors. For membrane module monitoring, a HORIBA XploRA miniature Raman spectroscopy probe can be used, with a sampling frequency of 1 time / minute and a spectral resolution of 1-5 cm⁻¹. It is installed on the side of the first-stage reverse osmosis pressure vessel via a flange interface, with an optical fiber transmission distance ≤2m, ensuring real-time data acquisition at 1000-1800cm. The membrane surface contaminant spectrum is measured in 10-band. The contamination database pre-stores characteristic peak parameters for calcium carbonate, iron oxides, etc. Threshold settings can be based on critical contamination states where permeate flow decreases by 10% or conductivity increases by 10%. An early warning is triggered when the contamination index exceeds 1.5 times the baseline value. A Rosemount 3051S differential pressure sensor array is used, axially arranged inside the eight filter cartridges of the Φ350×1650mm security filter, with a sampling frequency of 10Hz. The preset differential pressure non-uniformity threshold can be set to 0.03MPa (i.e., 30% of the average differential pressure). When the differential pressure of a filter cartridge deviates from the average value by more than this threshold, a lifespan prediction model is established based on historical filter cartridge contamination data. An infrared thermal imager, FLIR A615, is fixed above the reverse osmosis unit with a bracket, its monitoring angle perpendicular to the pressure vessel surface, covering the entire surface of the R8040C membrane module. The determination of the preset temperature difference warning value between the permeate and concentrate sides is based on experimental data on the reverse osmosis operating temperature range and the local temperature rise during membrane scaling. Furthermore, the AI ​​algorithm integrates operating parameters such as raw water hardness and recovery rate to calculate the equipment health index, thereby generating hierarchical operation and maintenance decisions and linking the Langerier index to regulate the dosage of scale inhibitor, realizing intelligent management of the entire process from pollution warning and equipment protection to reagent optimization. This method and system, through the fusion of multi-source sensors and AI, can realize intelligent monitoring of the entire reverse osmosis system, and achieve intelligent early warning of pollution and prediction of filter life. At the same time, it further reduces reagent costs and equipment energy consumption, and solves the problems of insufficient collaborative control in traditional systems, thereby effectively improving permeate efficiency and safety.

[0205] In order for the above methods and systems to operate smoothly, the system may include more or fewer components than those described above, or combine certain components, or different components, in addition to the various modules mentioned above. For example, it may include input / output devices, network access devices, buses, processors, and memory.

[0206] The processor can be a central processing unit, or other general-purpose processors, digital signal processors, application-specific integrated circuits (ASICs), off-the-shelf programmable gate arrays (OPGs) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor. This processor is the control center of the system, connecting various parts via various interfaces and lines.

[0207] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0208] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention. Therefore, the scope of protection of this patent should be determined by the appended claims.

[0209] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for AI-based hierarchical monitoring of the reverse osmosis operation status of a pure water system, characterized in that, The following steps are adopted: The molecular vibrational spectra of pollutants on the surface of the reverse osmosis membrane are collected in real time using a Raman spectroscopy probe, and the pollutant type is identified by matching it with a preset pollutant database. The differential pressure distribution of each filter element in the security filter is monitored by a differential pressure sensor array, and the differential pressure non-uniformity is calculated to predict the remaining life of the filter element. The surface temperature distribution of the membrane module is obtained by using an infrared thermal imager, and the temperature difference between the permeate side and the concentrate side is analyzed to assess the uniformity of permeate production. A real-time energy efficiency mapping model is established based on Hall effect sensor to monitor the input power and output parameters of high-pressure pump. A preset algorithm is used to fuse and analyze multi-source monitoring data to generate hierarchical operation and maintenance decisions; In the pollution type identification process, the degree of pollution is quantitatively characterized by a pollution index. The calculation process of the pollution index is as follows: ； In the formula, For the first The spectral intensity of each characteristic peak For the first The contamination weight coefficients corresponding to each characteristic peak This represents the number of characteristic peaks.

2. The AI-based hierarchical monitoring method for the reverse osmosis operation status of a pure water system according to claim 1, characterized in that, The method of using a differential pressure sensor array to monitor the differential pressure distribution of each filter element in the security filter and calculating the differential pressure non-uniformity to predict the remaining life of the filter element specifically includes: Collect differential pressure data for each filter element unit. Calculate differential pressure data average pressure difference and standard deviation ; The average pressure difference and standard deviation The calculation process is as follows: ； ； In the formula, This represents the total number of filter elements. Define the remaining lifespan of the filter element. : ； In the formula, This is the preset threshold for differential pressure unevenness.

3. The AI-based hierarchical monitoring method for the reverse osmosis operation status of a pure water system according to claim 2, characterized in that, The step of acquiring the surface temperature distribution of the membrane module using an infrared thermal imager and analyzing the temperature difference between the permeate and concentrate sides to assess the uniformity of the permeate specifically includes: Collect the temperature distribution on the product water side. Temperature distribution on the concentrate side Calculate the maximum temperature difference ; The maximum temperature difference The calculation process is as follows: ； In the formula, These are the temperature measurement point numbers for the product water side and the concentrate side, respectively. When the maximum temperature difference When the temperature difference exceeds the preset warning value, a warning message for water production uniformity is triggered.

4. The AI-based hierarchical monitoring method for the reverse osmosis operation status of a pure water system according to claim 3, characterized in that, The specific steps for establishing a real-time energy efficiency mapping model based on monitoring the input power and output parameters of a high-pressure pump using a Hall effect sensor include: Real-time calculation of high-pressure pump efficiency ; The efficiency of the high-pressure pump The calculation process is as follows: ； In the formula, For water density, It is the acceleration due to gravity. For the high-pressure pump head, For real-time traffic, Input power; When the high-pressure pump efficiency The efficiency drop exceeds the preset threshold. At that time, an energy consumption anomaly warning message is triggered.

5. The AI-based hierarchical monitoring method for the reverse osmosis operation status of a pure water system according to claim 1, characterized in that, The specific steps of using a preset algorithm to fuse and analyze multi-source monitoring data include: Construct a multi-parameter correlation model; The input variables of the multi-parameter correlation model include: raw water hardness. Operational recovery rate Scale inhibitor dosage concentration ; Among them, the operational recovery rate The calculation method is as follows: ； In the formula, For water production flow rate, This refers to the influent flow rate; The output variables of the multi-parameter association model include: Pollution trend forecast Optimal dosage of reagent ; An LSTM neural network is used to train the historical data, and the training objective function is: ； In the formula, These are the model's predicted values. These are actual monitored values. The length of the time series. For index variables of time series; Based on the trained LSTM network, the input time series is... The normalized parameters, output the future Pollution prediction values ​​at any time ,and The range of values ​​is ; when the future Pollution prediction values ​​at any time Calculate the optimal pesticide dosage if the pollution index exceeds the threshold. and with the optimal dosage Add the medicine; The optimal dosage of the drug The calculation process is as follows: ； In the formula, 2 represents the basic dosage, 4 represents the minimum dosage under low pollution conditions, and 2 represents the maximum increment. ； In the formula, , where 75 is the baseline recovery rate and 0.01 is the correction factor; ； In the formula, 292 is the hardness correction factor, and 292 is the raw water hardness reference value; 。 6. The AI-based hierarchical monitoring method for the reverse osmosis operation status of a pure water system according to claim 5, characterized in that, The multi-parameter correlation model specifically also includes: Calculate the Langerile index ; The Langerile index The calculation process is as follows: ； ； In the formula, It is the solubility product of calcium carbonate. dissociation constant of carbonic acid This refers to the calcium ion concentration. carbonate concentration, Bicarbonate concentration, This represents the carbon dioxide concentration.

7. The AI-based hierarchical monitoring method for the reverse osmosis operation status of a pure water system according to claim 4, characterized in that, The method of using a preset algorithm to fuse and analyze multi-source monitoring data further includes: Calculate and generate equipment health index ; The equipment health index The calculation process is as follows: ； in, Corresponding membrane fouling index, This is the pollution weighting coefficient. This is the normalized value of the pollution index; Corresponding to filter life, This is the filter element weighting coefficient. This is the normalized value of the remaining lifetime; Corresponding to the efficiency of the high-pressure pump, This is the weighting coefficient for the efficiency of the high-pressure pump. The normalized value for efficiency reduction; Corresponding temperature uniformity index, Temperature weighting coefficient, This is the normalized value for temperature difference; And health index The weight allocation is based on the following fault handling priorities: Membrane fouling 40%, filter life 20%, high-pressure pump efficiency 30%, temperature uniformity index 10%.

8. The AI-based hierarchical monitoring method for the reverse osmosis operation status of a pure water system according to claim 4, characterized in that, The generation of hierarchical operation and maintenance decisions specifically includes: When the pollution index When the preset pollution index threshold is reached at 30%-50%, a first-level instruction for citric acid rinsing of the membrane surface is generated and sent. When the remaining lifespan of the filter element is... At that time, a secondary filter replacement command is generated and sent; When the high-pressure pump reaches the efficiency vibration protection condition or health index When <0.3, generate and send the Phase II backup startup Level 3 command; The specific conditions for efficiency vibration protection are as follows: Based on the rated speed of the high-pressure pump, the effective value of vibration velocity is obtained by using an IEPE accelerometer installed in the horizontal or vertical direction of the high-pressure pump. Vibration velocity amplitude at 1X frequency Acceleration in the characteristic frequency band of rolling bearing outer ring fault ; The following conditions must be met simultaneously for efficient vibration protection: 。 9. The AI-based hierarchical monitoring method for the reverse osmosis operation status of a pure water system according to claim 6 or 8, characterized in that, The generation of hierarchical operation and maintenance decisions specifically includes: when At that time, a command is generated and sent to adjust the scale inhibitor dosage to 4-6 ppm; If the membrane surface experiences the maximum temperature difference within the target time period after being cleaned with citric acid. If the temperature does not drop below 1.0℃, it is determined to be colloidal contamination, and a flushing instruction for sodium hydroxide and sodium hypochlorite is generated and sent.

10. An AI-based hierarchical monitoring system for the reverse osmosis operation status of a pure water system, characterized in that: The system employs the AI-based hierarchical monitoring method for the reverse osmosis operating status of a pure water system as described in any one of claims 1-9, wherein the system comprises: The real-time acquisition module is used to acquire the molecular vibrational spectra of contaminants on the surface of the reverse osmosis membrane in real time using a Raman spectroscopy probe; The matching and identification module is used to match a preset pollution database to identify the type of pollution. The array monitoring module is used to monitor the differential pressure distribution of each filter element in the security filter using an array of differential pressure sensors. The calculation module is used to calculate the differential pressure unevenness to predict the remaining life of the filter element; The acquisition and analysis module is used to acquire the surface temperature distribution of the membrane module through an infrared thermal imager, and analyze the temperature difference between the permeate side and the concentrate side to evaluate the uniformity of permeate production. The mapping model module is used to establish a real-time energy efficiency mapping model based on the monitoring of the input power and output parameters of the high-pressure pump using Hall effect sensors. The fusion analysis module is used to fuse and analyze multi-source monitoring data using a preset algorithm; The generation module is used to generate hierarchical operation and maintenance decisions.

Images