A smart multimodal fusion water treatment closed-loop control system and method

Abstract

This invention discloses an intelligent multimodal fusion water treatment closed-loop control system and method, belonging to the technical field of water treatment control systems. The control system includes a chemical working platform, a dual-core intelligent control module, a multimodal data acquisition module, and a cloud server. Through multimodal data fusion, intelligent prediction algorithms, a dual-core architecture, and cloud-based LLM diagnostics, this invention achieves automated, high-precision control and intelligent decision-making in the water treatment titration process, reducing reagent consumption and improving the safety and operational efficiency of water treatment reactions.

Description

Technical Field

[0001] This invention belongs to the field of water treatment control system technology, specifically relating to an intelligent multimodal fusion water treatment closed-loop control system and method. Background Technology

[0002] With the acceleration of modern industrial production and urbanization, the demand for water treatment is increasing, especially in the fields of industrial wastewater treatment and municipal sewage treatment. Precise control of wastewater pH is crucial for ensuring that treated wastewater meets discharge standards, reducing treatment costs, and ensuring reaction safety. Solution pH control is a core operation widely used in water treatment processes such as neutralization and heavy metal precipitation. It involves precisely adding acidic or alkaline agents to the wastewater system to induce a neutralization reaction, stabilizing the wastewater pH within a preset range. Traditional water treatment pH closed-loop control systems often rely on a single pH sensor and simple PID algorithms to achieve constant or linear variable-rate addition. While simple to implement and providing direct feedback, these systems have many inherent drawbacks and are difficult to adapt to the high precision, low consumption, and high safety requirements of modern water treatment.

[0003] First, traditional systems lack a mechanism for predicting titration endpoints in advance, making them prone to reagent overshoot, resulting in high consumption of water treatment reagents, high operating costs, and non-compliance with environmental protection requirements. Second, the system only focuses on macroscopic pH changes and cannot monitor the internal temperature gradient of the wastewater, which may lead to local thermal runaway during strongly exothermic reactions in the water treatment process (such as acid-base neutralization reactions), resulting in poor safety. Third, traditional systems are easily affected by on-site factors such as ambient light, temperature, and humidity, and the data from sensors such as pH meters has low reliability, leading to unstable water treatment effects. In addition, existing equipment lacks intelligent diagnostic capabilities for the water treatment process, relies heavily on the experience of operation and maintenance personnel, and is difficult to achieve remote optimization and intelligent operation and maintenance, failing to meet the development trend of large-scale and intelligent industrial water treatment.

[0004] To address the aforementioned issues, there is an urgent need to develop a closed-loop water treatment control system that integrates multimodal sensing, intelligent prediction, safety adjustment, and remote diagnostic functions to achieve minimized consumption of water treatment chemicals, high-precision control, and high-safety operation. Summary of the Invention

[0005] This invention addresses the shortcomings of existing water treatment pH closed-loop control systems, such as high reagent consumption, low control accuracy, poor safety, simplistic endpoint determination, and lack of intelligent diagnostic capabilities. It aims to provide an intelligent multimodal fusion water treatment closed-loop control system and method. Through multimodal data fusion, intelligent predictive algorithms, a dual-core architecture, and cloud-based LLM diagnostics, it achieves automated, high-precision control and intelligent decision-making in the water treatment titration process, reducing reagent consumption and improving the safety and operational efficiency of water treatment reactions.

[0006] To achieve the above objectives, the present invention adopts the following technical solution:

[0007] In a first aspect, the present invention provides an intelligent multimodal fusion water treatment closed-loop control system, including a chemical working platform adapted to water treatment scenarios, a dual-core intelligent control module, a multimodal data acquisition module, and a cloud server;

[0008] The chemical working platform includes a layered experimental chamber, which is divided into an electronic control layer, a fluid control layer, and a chemical experimental layer from top to bottom by a corrosion-resistant partition. A dual-core intelligent control module is installed on the electronic control layer. The fluid control layer is equipped with a peristaltic pump, a reagent storage tank, and an ultrasonic atomizing plate drive board. The chemical experimental layer is equipped with a water treatment reactor. The reagent storage tank is connected to the water treatment reactor through the peristaltic pump. A magnetic stirrer is installed on the water treatment reactor. An ultrasonic atomizing plate is installed inside the water treatment reactor, and the ultrasonic atomizing plate is electrically connected to the ultrasonic atomizing plate drive board.

[0009] The dual-core intelligent control module includes a first microcontroller and a second microcontroller. The first microcontroller is connected to the multimodal data acquisition module via an I2C bus for real-time data processing, environmental adaptive calibration, and low-level real-time control. The second microcontroller is connected to the first microcontroller via UART / SPI and communicates with the cloud server for encrypted data storage, remote transmission, and optimized parameter reception, thereby realizing intelligent decision-making and control of the water treatment process.

[0010] The peristaltic pump, magnetic stirrer, and ultrasonic atomizing plate drive board are respectively controlled by the first microcontroller to achieve precise dripping of water treatment agents, uniform stirring of wastewater and agents, and emergency cooling.

[0011] The multimodal data acquisition module includes a pH sensor, a TDS sensor, a thermal imaging lens, a visible spectrum sensor, and a ToF ranging module. The pH sensor, TDS sensor, thermal imaging lens, and visible spectrum sensor are all installed in the chemical experiment layer and are used to collect data on pH value, TDS change, spatial temperature gradient of the reaction system, and solution color in the water treatment reactor during the water treatment process, respectively. The ToF ranging module is installed in the fluid control layer and is used to collect data on liquid level changes in the reagent storage tank. The thermal imaging lens, visible spectrum sensor, and ToF ranging module are all electrically connected to the first microcontroller, and the pH sensor and TDS sensor are both electrically connected to the first microcontroller through an analog-to-digital converter.

[0012] The cloud server is used to receive encrypted data, call a pre-trained large language model to perform comprehensive evaluation and expert-level diagnosis of the data, generate an expert-level intelligent diagnostic report for the water treatment process, and output optimized control parameters.

[0013] Furthermore, the electronic control layer of the layered experimental box is also equipped with a human-computer interaction central control screen, which is electrically connected to the first microcontroller.

[0014] Furthermore, the multimodal data acquisition module also includes an environmental sensing unit, which includes an ambient light sensor and a temperature, humidity and pressure sensor, used to collect data on the light, temperature, humidity and pressure of the environment around the water treatment reactor. The ambient light sensor and the temperature, humidity and pressure sensor are all electrically connected to the first microcontroller.

[0015] Furthermore, it also includes an external storage device, which is used to record local logs and historical operating data to facilitate iterative updates of control data. The external storage device is electrically connected to the second microcontroller.

[0016] Furthermore, the pH sensor is installed above the water treatment reactor by suspension, with the probe submerged below the wastewater surface and maintaining a safe distance from the magnetic stir bar placed inside the water treatment reactor; the TDS sensor is made of corrosion-resistant material and immersed in the wastewater solution, and is spaced apart from the pH sensor; the thermal imaging lens and the visible spectrum sensor are installed in a non-contact manner through a transparent window, covering the key areas of the water treatment reactor, and are installed 3-8 cm away from the liquid surface.

[0017] Secondly, the present invention provides an intelligent multimodal fusion water treatment closed-loop control method, implemented based on the aforementioned intelligent multimodal fusion water treatment closed-loop control system, comprising the following steps:

[0018] S1: Input the target pH range, target pollutant concentration standard, TDS prediction threshold and safe temperature gradient threshold parameters for water treatment through the human-computer interaction control screen;

[0019] S2: Start the system. The peristaltic pump adds the treatment agent to the wastewater at the initial speed. The magnetic stirrer starts. The multimodal data acquisition module simultaneously collects wastewater pH value, TDS change data, reaction system spatial temperature gradient, wastewater color, agent storage tank liquid level change, and ambient temperature, humidity / light / pressure data, and transmits them to the first microcontroller.

[0020] S3: The first microcontroller performs environmental adaptive calibration, calculates the TDS change slope d(TDS) / dt, spatial temperature gradient GT, and color change rate dC / dt in real time, and performs reagent minimization titration control and safety-uniformity adaptive adjustment of the water treatment process based on the above data;

[0021] S4: The first microcontroller determines the endpoint of the water treatment reaction based on the fusion strategy of pH and target pollutant concentration. If the endpoint is not reached, it returns to step S3 to continue closed-loop control. If the endpoint is reached, it stops the peristaltic pump and magnetic stirrer.

[0022] S5: After the water treatment data is processed by the first microcontroller, it is transmitted to the second microcontroller via UART / SPI, stored in an external storage device and encrypted, and then uploaded to the cloud server via Wi-Fi. The cloud server calls a pre-trained large language model to perform a comprehensive evaluation and expert-level diagnosis of the water treatment data, generates an expert-level intelligent diagnostic report for the water treatment process, and outputs optimized control parameters. The optimized control parameters are transmitted to the second microcontroller in encrypted form, and then transmitted to the first microcontroller to complete the parameter update.

[0023] Further, in step S3, the environmental adaptive calibration specifically involves: the first microcontroller calling the environmental temperature, humidity, and light data collected by the multimodal data acquisition module to compensate and calibrate the raw readings related to water treatment collected by the pH sensor and the visible spectrum sensor, thereby offsetting the interference of environmental factors on the sensor data and improving the detection accuracy of water treatment parameters;

[0024] Further, in step S3, the reagent minimization titration control specifically involves: when d(TDS) / dt approaches the preset TDS prediction threshold, the first microcontroller controls the peristaltic pump to enter the nonlinear deceleration microdroplet mode; simultaneously, the ToF ranging module monitors the liquid level change ΔH in the reagent storage tank, calculates the actual reagent consumption volume ΔV, and performs real-time calibration of the flow coefficient K of the peristaltic pump based on ΔV.

[0025] Further, in step S3, the safety-uniformity adaptive adjustment specifically involves: if the spatial temperature gradient GT exceeds the safety temperature gradient threshold, the first microcontroller immediately controls the peristaltic pump to stop dripping the agent and increases the driving power of the ultrasonic atomizing plate to cool down and dissipate heat, avoiding local thermal runaway caused by strong exothermic reactions during water treatment; the mixing uniformity of wastewater and agent is diagnosed based on the color change rate dC / dt, and the rotation speed of the magnetic stirrer is adjusted in real time to ensure that the agent and wastewater are mixed quickly and uniformly, thereby improving the water treatment effect.

[0026] Further, in step S4, the fusion strategy is as follows: when the pH sensor detection value falls within the target pH range for water treatment input in step S1, and the target pollutant concentration C calculated by the visible spectrum sensor using the Lambert-Beer model reaches the preset wastewater discharge standard, the first microcontroller confirms that the water treatment reaction has ended.

[0027] Compared with the prior art, the beneficial technical effects of the present invention are as follows:

[0028] 1. This invention achieves highly reliable determination of pH titration endpoint through multimodal fusion, breaking through the limitations of traditional single-factor determination. It adopts the dual conditions of pH reaching the standard and the target pollutant concentration reaching the standard through spectral analysis, which verifies the thoroughness of the water treatment reaction from a chemical perspective, avoids misjudgment caused by pH fluctuations, and ensures that the treated wastewater is discharged in a stable manner that meets the standards.

[0029] 2. The present invention significantly reduces the consumption of water treatment chemicals. By introducing a TDS slope prediction and ToF flow calibration mechanism, the peristaltic pump is automatically controlled to enter the micro-droplet mode before the reaction endpoint, and the dripping flow rate is calibrated in real time to prevent chemical over-flushing. This can reduce the consumption of water treatment chemicals by 10%-30% and greatly reduce operating costs.

[0030] 3. The safety of the water treatment process of the present invention is greatly improved. The temperature gradient of the reaction system is monitored in real time by a thermal imaging lens. In case of abnormality, the ultrasonic atomizing plate is linked to quickly cool down, effectively suppressing local thermal runaway in strong exothermic reactions. Combined with the color change rate to diagnose the mixing uniformity, the safety hazards caused by insufficient local reaction are avoided, which is suitable for the high safety requirements of industrial water treatment.

[0031] 4. The present invention improves the level of intelligence and operation and maintenance efficiency. The dual-core heterogeneous architecture realizes the efficient separation of real-time control and remote communication, ensuring the stability of system operation. The cloud server LLM intelligent diagnosis provides expert-level operation and maintenance reports, supports remote fault analysis, maintenance prediction and reagent optimization suggestions, reduces the dependence on on-site professionals, and conforms to the development trend of intelligent and unmanned industrial water treatment.

[0032] 5. The present invention has strong environmental adaptability and stable water treatment effect. The environmental sensing unit realizes the adaptive calibration of the sensor, which effectively counteracts the interference of on-site environmental factors such as temperature, humidity and light, and improves data reliability and control accuracy. The layered corrosion-resistant structure is suitable for the complex environment of corrosive wastewater and agents in the water treatment process, extends the service life of the equipment and ensures long-term operational stability. Attached Figure Description

[0033] Figure 1 This is a schematic diagram of the structure of an intelligent multimodal fusion water treatment closed-loop control system;

[0034] Figure 2 This is a block diagram illustrating the control principle of an intelligent multimodal fusion water treatment closed-loop control system.

[0035] Figure 3 A flowchart of an intelligent multimodal fusion closed-loop control method for water treatment;

[0036] Figure 4 This is a flowchart of the intelligent titration method for minimizing pharmaceutical preparations in this invention;

[0037] Figure 5 This is a flowchart of the adaptive adjustment method for mixing uniformity and safety in this invention. Detailed Implementation

[0038] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0039] like Figure 1 and 2 As shown, this embodiment provides an intelligent multimodal fusion water treatment closed-loop control system, including a chemical working platform adapted to water treatment scenarios, a dual-core intelligent control module, a multimodal data acquisition module, and a cloud server;

[0040] The chemical working platform includes a layered experimental chamber, which is divided into an electronic control layer, a fluid control layer, and a chemical experimental layer from top to bottom by a corrosion-resistant partition. A dual-core intelligent control module is installed on the electronic control layer. The fluid control layer is equipped with a peristaltic pump, a reagent storage tank, and an ultrasonic atomizing plate drive board. The chemical experimental layer is equipped with a water treatment reactor. The reagent storage tank is connected to the water treatment reactor through the peristaltic pump. A magnetic stirrer is installed on the water treatment reactor. An ultrasonic atomizing plate is installed inside the water treatment reactor, and the ultrasonic atomizing plate is electrically connected to the ultrasonic atomizing plate drive board.

[0041] The dual-core intelligent control module includes a first microcontroller and a second microcontroller. The first microcontroller is connected to the multimodal data acquisition module via an I2C bus for real-time data processing, environmental adaptive calibration, and low-level real-time control. The second microcontroller is connected to the first microcontroller via UART / SPI and communicates with the cloud server for encrypted data storage, remote transmission, and optimized parameter reception, thereby realizing intelligent decision-making and control of the water treatment process.

[0042] The peristaltic pump, magnetic stirrer, and ultrasonic atomizing plate drive board are respectively controlled by the first microcontroller to achieve precise dripping of water treatment agents, uniform stirring of wastewater and agents, and emergency cooling.

[0043] The multimodal data acquisition module includes a pH sensor, a TDS sensor, a thermal imaging lens, a visible spectrum sensor, and a ToF ranging module. The pH sensor, TDS sensor, thermal imaging lens, and visible spectrum sensor are all installed in the chemical experiment layer and are used to collect data on pH value, TDS change, spatial temperature gradient of the reaction system, and solution color in the water treatment reactor during the water treatment process, respectively. The ToF ranging module is installed in the fluid control layer and is used to collect data on liquid level changes in the reagent storage tank. The thermal imaging lens, visible spectrum sensor, and ToF ranging module are all electrically connected to the first microcontroller, and the pH sensor and TDS sensor are both electrically connected to the first microcontroller through an analog-to-digital converter.

[0044] The cloud server is used to receive encrypted data, call a pre-trained large language model to perform comprehensive evaluation and expert-level diagnosis of the data, generate an expert-level intelligent diagnostic report for the water treatment process, and output optimized control parameters.

[0045] This invention utilizes multi-sensor fusion algorithms and Large Language Model (LLM) intelligent diagnostics as a foundation to explore a closed-loop water treatment control system characterized by low reagent consumption, high safety, high precision, and high intelligence. It fully considers various key factors in the control of chemical reactions (such as neutralization reactions) and inputs the following information into the intelligent closed-loop control system:

[0046] Chemical consumption information: Liquid level change information of chemical storage tanks obtained through the ToF ranging module is used for real-time flow calibration.

[0047] Titration endpoint information: pH information of the solution is obtained through a pH sensor, and the change rate of total dissolved solids (TDS) is obtained through a TDS sensor, which is used for endpoint prediction.

[0048] Mixing homogeneity and safety information: The spatial temperature gradient information of the reaction system is obtained through a thermal imaging lens, and the color change information is obtained through a visible spectrum sensor, which is used to assist in diagnosis and safety regulation.

[0049] Environmental interference information: Ambient light, temperature, humidity, and pressure information are acquired through ambient light / temperature, humidity, and pressure sensors to calibrate the raw readings of the pH and TDS sensors in real time. Based on this multimodal information, the system will autonomously and intelligently perform closed-loop titration control and upload the data to the cloud for LLM intelligent diagnostics and parameter optimization through a dual-core architecture.

[0050] In this invention, the liquid-contacting components of the chemical working platform are made of corrosion-resistant materials to meet the contact requirements between corrosive wastewater and reagents in water treatment processes. The water treatment reactor and reagent storage tank are made of borosilicate glass or polypropylene, while the liquid-contacting parts of the peristaltic pump pipes, fluid pipelines, and pH probe are made of polytetrafluoroethylene or polyethylene. The outer side of the layered experimental chamber is made of a lightweight aluminum metal frame combined with a transparent polycarbonate protective cover. The corrosion-resistant partitions serving as the fluid control layer and the chemical experiment layer are equipped with flow guide grooves, and the outer side of the layered experimental chamber is provided with an external discharge interface that communicates with the flow guide grooves.

[0051] In this invention, a peristaltic pump is used for precise dispensing of reagents (such as acid and alkali solutions). A stepper motor is controlled using PWM pulse modulation technology to adjust the titration speed. In the control logic, the system combines TDS slope prediction and ToF flow calibration data to automatically perform nonlinear speed reduction control, accurately reducing the titration speed to micro-droplet mode near the endpoint. The liquid level H in the reagent storage tank is monitored in real time by a ToF ranging module to calculate the actual reagent consumption volume ΔV. The system then calibrates and corrects the peristaltic pump's flow coefficient K in real time based on ΔV to ensure absolute accuracy of the titration amount.

[0052] In this invention, the ultrasonic atomizing sheet serves as a safety actuator, utilizing real-time monitoring of the spatial temperature gradient G obtained through thermal imaging.T If G T If the safety threshold is exceeded, the system will increase the PWM drive power of the ultrasonic atomizing plate to disperse local heat by using the airflow generated by atomization and rapid evaporation, thereby suppressing thermal runaway and improving safety.

[0053] In this invention, the key analog signals generated by the pH sensor and TDS sensor are digitally acquired through an analog-to-digital converter (ADC) to ensure the highest possible signal resolution. The ADC used is the ADS1115, a 16-bit precision ADC characterized by low power consumption and high accuracy, suitable for space- and power-constrained sensor measurement applications. It supports precise measurement of microvolt-level signals and has a maximum sampling rate of 860 SPS. The pH sensor used is the PH4502C, the TDS sensor is the WAVGAT TDS module, the first microcontroller is the STM32F407VET6, the second microcontroller is the ESP8266, the peristaltic pump is the Lifu 101ADB, the ToF ranging module is the MINIT DVP0501C1, the magnetic stirrer is the MS3, and the thermal imaging lens is the MLX90640BAA.

[0054] Specifically, the electronic control layer of the layered experimental box is also equipped with a human-computer interaction central control screen, which is electrically connected to the first microcontroller.

[0055] Specifically, the multimodal data acquisition module further includes an environmental sensing unit, which includes an ambient light sensor and a temperature, humidity and pressure sensor, used to collect data on the light, temperature, humidity and pressure of the environment around the water treatment reactor. The ambient light sensor and the temperature, humidity and pressure sensor are all electrically connected to the first microcontroller.

[0056] In this invention, the ambient light sensor is model VEML7700, the temperature, humidity and barometric pressure sensor is model BME280, and the visible spectrum sensor is model AS7341.

[0057] Specifically, it also includes an external storage device used to record local logs and historical operating data, facilitating iterative updates of control data. The external storage device is electrically connected to the second microcontroller. The external storage device can be an SD card or a Flash chip.

[0058] Specifically, the pH sensor is installed above the water treatment reactor by suspension, with the probe submerged below the wastewater surface and maintaining a safe distance from the magnetic stir bar placed inside the water treatment reactor; the TDS sensor is made of corrosion-resistant material and immersed in the wastewater solution, and is spaced apart from the pH sensor; the thermal imaging lens and the visible spectrum sensor are installed in a non-contact manner through a transparent window, covering the key areas of the water treatment reactor, and are installed 3-8 cm away from the liquid surface.

[0059] like Figure 3 As shown, based on the above-mentioned intelligent multimodal fusion water treatment closed-loop control system, this embodiment also provides an intelligent multimodal fusion water treatment closed-loop control method, including the following steps:

[0060] S1: Input the target pH range, target pollutant concentration standard, TDS prediction threshold and safe temperature gradient threshold parameters for water treatment through the human-computer interaction control screen;

[0061] S2: Start the system. The peristaltic pump adds the treatment agent to the wastewater at the initial speed. The magnetic stirrer starts. The multimodal data acquisition module simultaneously collects wastewater pH value, TDS change data, reaction system spatial temperature gradient, wastewater color, agent storage tank liquid level change, and ambient temperature, humidity / light / pressure data, and transmits them to the first microcontroller.

[0062] S3: The first microcontroller performs environmental adaptive calibration, calculates the TDS change slope d(TDS) / dt, spatial temperature gradient GT, and color change rate dC / dt in real time, and performs reagent minimization titration control and safety-uniformity adaptive adjustment of the water treatment process based on the above data;

[0063] S4: The first microcontroller determines the endpoint of the water treatment reaction based on the fusion strategy of pH and target pollutant concentration. If the endpoint is not reached, it returns to step S3 to continue closed-loop control. If the endpoint is reached, it stops the peristaltic pump and magnetic stirrer.

[0064] S5: After the water treatment data is processed by the first microcontroller, it is transmitted to the second microcontroller via UART / SPI, stored in an external storage device and encrypted, and then uploaded to the cloud server via Wi-Fi. The cloud server calls a pre-trained large language model to perform a comprehensive evaluation and expert-level diagnosis of the water treatment data, generates an expert-level intelligent diagnostic report for the water treatment process, and outputs optimized control parameters. The optimized control parameters are transmitted to the second microcontroller in encrypted form, and then transmitted to the first microcontroller to complete the parameter update.

[0065] Specifically, in step S3, the environmental adaptive calibration is as follows: the first microcontroller calls the environmental temperature, humidity and light data collected by the multimodal data acquisition module to compensate and calibrate the raw readings related to water treatment collected by the pH sensor and the visible spectrum sensor, so as to counteract the interference of environmental factors on the sensor data and improve the detection accuracy of water treatment parameters.

[0066] Combination Figure 4 As shown, in step S3, the reagent minimization titration control specifically means that when d(TDS) / dt approaches the preset TDS prediction threshold, the first microcontroller controls the peristaltic pump to enter the nonlinear deceleration microdroplet mode.

[0067] In a preferred embodiment, the first microcontroller collects the instantaneous total dissolved solids concentration (TDS) of the wastewater at a fixed sampling period Δt using a TDS sensor, and calculates the TDS change rate d(TDS) / dt using two adjacent sampling values, specifically:

[0068] d(TDS) / dt=(TDS(t)-TDS(t-Δt)) / dt

[0069] Wherein, Δt ranges from 1 to 5 s. The predicted threshold τ_TDS of the TDS change rate is pre-calibrated based on the target working condition. When the following formula is satisfied, it is determined that d(TDS) / dt is close to the preset TDS predicted threshold:

[0070] ∣d(TDS) / dt-τ_TDS∣ / τ_TDS≤δ_TDS

[0071] Wherein, δ_TDS is the allowable relative deviation coefficient, preferably 0.1 to 0.3, more preferably 0.2. To avoid misjudgment caused by instantaneous noise, preferably, the first microcontroller determines that d(TDS) / dt is close to τ_TDS only when the TDS sensor meets the above conditions in N consecutive samplings (e.g., N=3 to 5 times).

[0072] Simultaneously, the ToF ranging module monitors the liquid level change ΔH in the drug storage tank, calculates the actual drug consumption volume ΔV, and calibrates the flow coefficient K of the peristaltic pump in real time based on ΔV. K_flow is the value of the flow coefficient K of the peristaltic pump at the current time.

[0073] In a preferred embodiment, the flow coefficient K of the peristaltic pump is defined and calculated as follows: K = ΔV / (Δt × n), where Δt is the time interval and n is the operating speed of the peristaltic pump.

[0074] ΔH is the difference in liquid level height before and after measurement by the ToF ranging module. Its expression is: ΔH=h(t)-h(t-Δt), where h(t) represents the liquid level height measured by the ToF ranging module at time t.

[0075] The expression for the actual volume of reagent consumption ΔV is: ΔV = S×ΔH, where S is the liquid surface area of the reagent storage tank.

[0076] Calibration trigger condition: For example, start calibration "after adding a certain amount of reagent" or "when it is detected that K deviates from the calibrated value by more than ±x%".

[0077] In the present invention, the change slope d(TDS) / dt of the TDS signal is used as an early warning signal for the titration end point. When the slope approaches zero, the control system immediately instructs the peristaltic pump to enter the micro-drop mode of non-linear deceleration, so as to accurately stop dropping before the end point.

[0078] Combined with Figure 5 As shown, in step S3, the safety-uniformity adaptive adjustment is specifically as follows: If the spatial temperature gradient G_T exceeds the safety temperature gradient threshold, the first microcontroller immediately controls the peristaltic pump to stop dropping the reagent, and increases the driving power of the ultrasonic atomizing sheet for cooling and heat dissipation, so as to avoid local thermal runaway caused by strong exothermic reactions during the water treatment process;

[0079] In a preferred embodiment, the thermal imaging lens used in the safety detection cycle is a two-dimensional thermal imaging sensor, which collects the temperature distribution of the monitoring area. The temperature data of the thermal imaging array at time t is recorded as (t). The first microcontroller first calculates the maximum temperature T_max and the average temperature T_avg in this area, where T_max is the maximum value among all pixels in the current frame, and T_avg is the arithmetic average of the temperatures of all pixels in the current frame.

[0080] Based on T_max and T_avg, a spatial temperature gradient index G_T is defined to characterize the deviation degree of the local hot spot relative to the overall average temperature, specifically as:

[0081] G_T = T_max - T_avg

[0082] When G_T is larger, it indicates that there is an obvious local overheating phenomenon or the temperature non-uniformity in the monitoring area is aggravated, and there is a safety risk.

[0083] In order to give a clear warning trigger condition, the temperature gradient safety threshold τ_G is calibrated in advance according to the maximum allowable temperature rise of the equipment and historical operation data. When it satisfies: G_T > τ_G, it is determined that the current spatial temperature gradient exceeds the safety threshold. The controller triggers the linkage control: reduce the peristaltic pump PWM and increase the atomizing sheet PWM to reduce the reaction heat and local temperature rise; when G_T ≤ τ_G, it is considered to be in a safe state, and the current dropping speed and atomizing power remain unchanged. τ_G can be set to 1 - 5°C according to different equipment and working conditions, and more preferably 2 - 3°C.

[0084] The uniformity of the mixing between wastewater and reagents is diagnosed based on the color change rate dC / dt. The speed of the magnetic stirrer is adjusted in real time to ensure that the reagents and wastewater are mixed quickly and evenly, thereby improving the water treatment effect.

[0085] The calculation method for dC / dt is as follows: The chromaticity C of the visible spectrum sensor is collected every Δt seconds, and dC / dt is calculated as follows: - ) / Δt.

[0086] Uniformity criterion: When |dC / dt|> This indicates incomplete mixing, when |dC / dt| < This indicates that the mixture has been thoroughly mixed.

[0087] Speed ​​adjustment rule: If |dC / dt|> Then increase the rotational speed Δnrpm; if |dC / dt| < 0 for T consecutive seconds To save energy, the rotation speed is reduced. , These are all set values ​​for the rate of color change, set according to the requirements of the actual reaction conditions, typically... It can be set to 1.5 to 3.0 Abs / min, more preferably 2.0 Abs / min; It can be set to 0.1 to 0.5 Abs / min, more preferably 0.2 Abs / min.

[0088] Specifically, in step S4, the fusion strategy is as follows: when the pH sensor detection value falls within the target pH range for water treatment input in step S1, and the target pollutant concentration C (i.e., C_polluant) calculated by the visible spectrum sensor using the Lambert-Beer model reaches the preset wastewater discharge standard, the first microcontroller confirms that the water treatment reaction has ended.

[0089] The expression for the Lambert-Beer model is: ,in, A Absorbance A = , I is the incident light intensity of pure water, and I is the transmitted light intensity of wastewater collected by the visible spectrum sensor. K Absorption coefficient is an inherent property of pollutants. The absorbance at a specific wavelength needs to be pre-calibrated. K The value is 0.05 to 0.5. Preferably 0.1 ; l Optical path length is the distance light travels through water (such as the thickness of a cuvette or the distance between probes, which is usually fixed, such as 1 cm).

[0090] The above description is a preferred embodiment of the present invention, used to explain the technical solution of the present invention, and is not intended to limit the present invention. Those skilled in the art can make conventional modifications, equivalent substitutions and improvements within the spirit and principles of the present invention, all of which are still included within the protection scope of the present invention.

Claims

1. A smart multimodal fusion water treatment closed-loop control system, characterized in that: It includes a chemical working platform adapted for water treatment scenarios, a dual-core intelligent control module, a multimodal data acquisition module, and a cloud server; The chemical working platform includes a layered experimental chamber, which is divided into an electronic control layer, a fluid control layer, and a chemical experimental layer from top to bottom by a corrosion-resistant partition. A dual-core intelligent control module is installed on the electronic control layer. The fluid control layer is equipped with a peristaltic pump, a reagent storage tank, and an ultrasonic atomizing plate drive board. The chemical experimental layer is equipped with a water treatment reactor. The reagent storage tank is connected to the water treatment reactor through the peristaltic pump. A magnetic stirrer is installed on the water treatment reactor. An ultrasonic atomizing plate is installed inside the water treatment reactor, and the ultrasonic atomizing plate is electrically connected to the ultrasonic atomizing plate drive board. The dual-core intelligent control module includes a first microcontroller and a second microcontroller. The first microcontroller is connected to the multimodal data acquisition module via an I2C bus for real-time data processing, environmental adaptive calibration, and low-level real-time control. The second microcontroller is connected to the first microcontroller via UART / SPI and communicates with the cloud server for encrypted data storage, remote transmission, and optimized parameter reception, thereby realizing intelligent decision-making and control of the water treatment process. The peristaltic pump, magnetic stirrer, and ultrasonic atomizing plate drive board are respectively controlled by the first microcontroller to achieve precise dripping of water treatment agents, uniform stirring of wastewater and agents, and emergency cooling. The multimodal data acquisition module includes a pH sensor, a TDS sensor, a thermal imaging lens, a visible spectrum sensor, and a ToF ranging module. The pH sensor, TDS sensor, thermal imaging lens, and visible spectrum sensor are all installed in the chemical experiment layer and are used to collect data on pH value, TDS change, spatial temperature gradient of the reaction system, and solution color in the water treatment reactor during the water treatment process, respectively. The ToF ranging module is installed in the fluid control layer and is used to collect data on liquid level changes in the reagent storage tank. The thermal imaging lens, visible spectrum sensor, and ToF ranging module are all electrically connected to the first microcontroller, and the pH sensor and TDS sensor are both electrically connected to the first microcontroller through an analog-to-digital converter. The cloud server is used to receive encrypted data, call a pre-trained large language model to perform comprehensive evaluation and expert-level diagnosis of the data, generate an expert-level intelligent diagnostic report for the water treatment process, and output optimized control parameters.

2. The intelligent multimodal fusion water treatment closed-loop control system according to claim 1, characterized in that: The electronic control layer of the layered experimental box is also equipped with a human-computer interaction central control screen, which is electrically connected to the first microcontroller.

3. The intelligent multimodal fusion water treatment closed-loop control system according to claim 1, characterized in that: The multimodal data acquisition module also includes an environmental sensing unit, which includes an ambient light sensor and a temperature, humidity and pressure sensor, used to collect data on the light, temperature, humidity and pressure of the environment around the water treatment reactor. The ambient light sensor and the temperature, humidity and pressure sensor are all electrically connected to the first microcontroller.

4. The intelligent multimodal fusion water treatment closed-loop control system according to claim 1, characterized in that: It also includes an external storage device, which is used to record local logs and historical operating data to facilitate iterative updates of control data. The external storage device is electrically connected to the second microcontroller.

5. The intelligent multimodal fusion water treatment closed-loop control system according to claim 1, characterized in that: The pH sensor is installed above the water treatment reactor by suspension, with the probe submerged below the wastewater surface and kept at a safe distance from the magnetic stir bar placed inside the water treatment reactor; the TDS sensor is made of corrosion-resistant material and is immersed in the wastewater solution, and is set at an interval from the pH sensor; the thermal imaging lens and the visible spectrum sensor are installed in a non-contact manner through a transparent window, covering the key areas of the water treatment reactor, and are installed 3-8 cm away from the liquid surface.

6. A smart multimodal fusion water treatment closed-loop control method, implemented based on the smart multimodal fusion water treatment closed-loop control system described in any one of claims 1-5, characterized in that, Includes the following steps: S1: Input the target pH range, target pollutant concentration standard, TDS prediction threshold and safe temperature gradient threshold parameters for water treatment through the human-computer interaction central control screen; S2: Start the system. The peristaltic pump adds the treatment agent to the wastewater at the initial speed. The magnetic stirrer starts. The multimodal data acquisition module simultaneously collects wastewater pH value, TDS change data, reaction system spatial temperature gradient, wastewater color, agent storage tank liquid level change, and ambient temperature, humidity / light / pressure data, and transmits them to the first microcontroller. S3: The first microcontroller performs environmental adaptive calibration, and calculates the TDS change slope d(TDS) / dt and the spatial temperature gradient G in real time. T And the color change rate dC / dt, based on the above data, implement reagent minimization titration control and safety-uniformity adaptive adjustment of the water treatment process; S4: The first microcontroller determines the endpoint of the water treatment reaction based on the fusion strategy of pH and target pollutant concentration. If the endpoint is not reached, it returns to step S3 to continue closed-loop control. If the endpoint is reached, it stops the peristaltic pump and magnetic stirrer. S5: After the entire water treatment data is processed by the first microcontroller, it is transmitted to the second microcontroller via UART / SPI, stored in an external storage device and encrypted, and then uploaded to the cloud server via Wi-Fi; The cloud server calls a pre-trained large language model to perform comprehensive evaluation and expert-level diagnosis of water treatment data, generate an expert-level intelligent diagnostic report for the water treatment process, and output optimized control parameters. The optimized control parameters are encrypted and transmitted to the second microcontroller, and then transmitted to the first microcontroller to complete the parameter update.

7. The intelligent multimodal fusion water treatment closed-loop control method according to claim 6, characterized in that, In step S3, the environmental adaptive calibration specifically involves the first microcontroller calling the environmental temperature, humidity, and light data collected by the multimodal data acquisition module to compensate and calibrate the raw water treatment-related readings collected by the pH sensor and the visible spectrum sensor, thereby offsetting the interference of environmental factors on the sensor data.

8. The intelligent multimodal fusion water treatment closed-loop control method according to claim 6, characterized in that, In step S3, the reagent minimization titration control specifically involves: when d(TDS) / dt approaches the preset TDS prediction threshold, the first microcontroller controls the peristaltic pump to enter the nonlinear deceleration microdroplet mode; at the same time, the ToF ranging module monitors the liquid level change ΔH in the reagent storage tank, calculates the actual reagent consumption volume ΔV, and performs real-time calibration of the flow coefficient K of the peristaltic pump based on ΔV.

9. The intelligent multimodal fusion water treatment closed-loop control method according to claim 6, characterized in that, In step S3, the safety-uniformity adaptive adjustment specifically involves: if the spatial temperature gradient G... T If the safe temperature gradient threshold is exceeded, the first microcontroller immediately controls the peristaltic pump to stop dripping the agent and increases the driving power of the ultrasonic atomizing plate to cool down and dissipate heat, avoiding local thermal runaway caused by strong exothermic reaction during water treatment; the uniformity of mixing between wastewater and agent is diagnosed based on the color change rate dC / dt, and the speed of the magnetic stirrer is adjusted in real time to ensure that the agent and wastewater are mixed quickly and evenly.

10. The intelligent multimodal fusion water treatment closed-loop control method according to claim 6, characterized in that, In step S4, the fusion strategy is as follows: when the pH sensor detection value falls within the target pH range for water treatment input in step S1, and the target pollutant concentration C calculated by the visible spectrum sensor using the Lambert-Beer model reaches the preset wastewater discharge standard, the first microcontroller confirms that the water treatment reaction has ended.

Images