Long-acting low-maintenance turbidity sensor with intelligent compensation function and calibration method thereof

Abstract

The present application relates to water quality monitoring sensor technical field, specifically to a kind of long-acting low-maintenance turbidity sensor with intelligent compensation function's calibration method, comprising the following steps: accurate calibration of temperature sensor;Establish the mathematical model of the current I of sensor internal detection photodiode o With the mathematical model of the current I of sensor internal detection photodiode pd ; Reference calibration measurement;Dynamic differential measurement;Dynamic differential multiple regression compensation, complete turbidity compensation calibration.The present application also provides a kind of long-acting low-maintenance turbidity sensor with intelligent compensation function, applies the calibration method described above.The present application provides a kind of long-acting low-maintenance turbidity sensor with intelligent compensation function and its calibration method, through dynamic dynamic differential multiple regression calibration algorithm, the ability of environmental interference and the long-acting nature of measurement accuracy are significantly improved;Through intelligent self-cleaning system, the influence of precision due to biological and non-biological attachment is significantly reduced, and the maintenance frequency of sensor is reduced.

Description

Technical Field

[0001] This invention relates to the field of water quality monitoring sensor technology, and in particular to a long-lasting, low-maintenance turbidity sensor with intelligent compensation function and its calibration method. Background Technology

[0002] Turbidity is a key indicator of water cleanliness, and its detection is widely used in drinking water safety, wastewater treatment, industrial process control, and environmental monitoring. Currently, most mainstream online turbidity sensors employ the principle of optical scattering, which involves emitting a light beam towards a water sample and detecting the intensity of scattered light caused by suspended particles in the water, thereby calculating the turbidity value. However, existing turbidity sensors generally face the following technical bottlenecks in practical applications, and relevant existing technology cases also confirm the prevalence and difficulty in solving these problems:

[0003] Light source stability and measurement reference drift: Traditional sensors widely use light-emitting diodes (LEDs) as the light source. LEDs are inexpensive, but their emission spectrum is broad (typically tens of nanometers), making them susceptible to absorption interference from specific colored substances in the water sample. More importantly, the output light intensity of LEDs drifts significantly with changes in ambient temperature and their own aging. Early models of products such as the Hach 1720D turbidimeter from the United States required frequent (e.g., monthly or quarterly) manual calibration in practical use to compensate for reference errors caused by light source attenuation, resulting in high maintenance costs and an inability to achieve long-term continuous and stable measurements.

[0004] Measurement errors introduced by ambient light interference: Especially for sensors operating outdoors or in open pools, stray ambient light is a major source of error. While existing technologies employ physical light shields (such as the Turb 430T sensor from WTW, Germany) or modulation / demodulation techniques (such as some schemes using pulsed LEDs and lock-in amplifier circuits) to suppress interference, the former has limited effectiveness under complex lighting conditions, and the latter involves complex circuitry and is expensive. For example, in some low-cost sensor designs, relying solely on simple photodetectors to receive scattered light is almost insufficient to eliminate the severe impact of sunlight fluctuations on low turbidity measurements, leading to significant discrepancies between nighttime and daytime readings.

[0005] The Systemic Impact of Temperature Drift on Optoelectronic Detection Systems: The performance of photodetectors and related signal conditioning circuits is temperature-sensitive. Temperature variations in photodiode responsivity and operational amplifier gain are significant sources of measurement error. Although some high-end products (such as some models from Toray Industries, Inc.) attempt to incorporate thermistors for coarse compensation, these methods are mostly applied to single components and lack comprehensive temperature modeling and real-time closed-loop compensation for the entire opto-mechanical-electrical system. Therefore, when experiencing drastic seasonal temperature changes, measured values ​​still exhibit non-negligible drift.

[0006] Signal attenuation and failure due to optical window contamination: This is the most critical issue limiting the long-term reliability of all immersion optical sensors. Whether it's the online turbidimeter cited in the US EPA standard method or products from companies like Systea in Italy, biofouling and inorganic scaling are extremely common after prolonged contact with wastewater. Existing solutions mainly rely on periodic manual disassembly and cleaning, or automatic rinsing with chemical cleaning agents. However, manual maintenance is time-consuming and labor-intensive, and chemical cleaning can lead to secondary pollution and system complexity.

[0007] The mismatch between static calibration and dynamic operation, and the lack of adaptive capability: Most existing sensors are calibrated statically only once at the factory using a standard solution (such as formalin). However, in actual operation, the light source intensity, detector sensitivity, and optical window transmittance all dynamically decline. For example, maintenance records of a certain brand of turbidimeter widely used in municipal wastewater treatment plants show that after six months of operation, even with a relatively clean optical window, its readings may deviate by more than 10% compared to the original calibration curve. This is because its algorithm cannot distinguish and compensate for the aforementioned dynamic factors in real time, resulting in a linear decrease in measurement accuracy over time.

[0008] Existing turbidity sensing technologies have significant shortcomings in key areas such as light source stability, anti-interference capability, temperature adaptability, anti-fouling and self-cleaning, and intelligent adaptive calibration, making it difficult to meet the long-term online monitoring requirements for high precision, high stability, and low maintenance in complex and harsh water quality environments. Therefore, there is an urgent need for an innovative, comprehensive solution that integrates light source, detection, compensation, and maintenance. Summary of the Invention

[0009] This invention provides a long-lasting, low-maintenance turbidity sensor with intelligent compensation function and its calibration method, in order to solve the technical problems of low long-term online measurement accuracy, poor stability, and frequent maintenance caused by unstable light source, ambient light interference, temperature drift, optical window contamination, and lack of real-time dynamic compensation in existing turbidity sensors.

[0010] This invention provides a calibration method for a long-lasting, low-maintenance turbidity sensor with intelligent compensation function, comprising the following steps: Step 1: Precise calibration of the temperature sensor; Step 2: Establish optical power P o The sensor internally detects the photodiode current I. pd Mathematical model; Step 3: Benchmark calibration measurement; Step 4: Dynamic differential measurement; Step 5: Dynamic differential multivariate regression compensation to complete turbidity compensation calibration.

[0011] As a preferred technical solution, the precise calibration process of the temperature sensor in step one is as follows: S1: Acquire multiple sample values ​​of the original temperature signal and perform ADC conversion on each sample value; then use the mean method for smoothing and noise reduction, and output the stable value under this state; The formula for calculating the stable value is as follows:

[0012] x i This is the raw ADC reading obtained from the i-th sample; m is the size of the data window; X is the calculated average value, which serves as the final stable reading under this condition; S2, collect standard temperature data; Multiple temperature measurements from the water bath temperature control system are collected. Different weights are assigned to the multiple temperature measurements using a variable weight Gaussian function, and a weighted average is calculated for temperature compensation. Variable weight Gaussian function:

[0013] in: W(i) is the initial weight value of the i-th data point; σ is the standard deviation of the variable-weight Gaussian function, which controls the degree of concentration of the weight distribution; the larger the value of σ, the flatter the weight distribution. c is the center position of the weight distribution; i is the index of the data point within the window; Weight normalization:

[0014] in: ω (i) It is the normalized weight of the i-th data point; m is the size of the data window; The initial weights W for all data points within the window. (j) The sum; Gaussian weighted mean calculation: T(i)

[0015] Wherein: T filtered This is the filtered standard temperature value; T (i) It is the i-th original temperature measurement value within the window; ω (i) It is the normalized weight corresponding to the i-th data point; m is the total number of data points involved in the calculation; S3. Establish a mathematical model and perform polynomial fitting using least squares on the stable values ​​in S1 and the standard temperature data in S2. Establish sensor stability values Compared with the standard temperature T after variable weight Gaussian filtering filtered A precise mathematical model between them was established, and the mathematical model was fitted using polynomials, with the least squares method used in the fitting process. The polynomial form is:

[0016] Where a0, a1, ..., a n denoted as the coefficients of the polynomial to be determined; n is the number of terms in the polynomial.

[0017] As a preferred technical solution, the mathematical model in step two is:

[0018] Among them, P o The actual laser output power measured by an external standard optical power meter; I pd The sensor internally detects the current of the photodiode; The coefficients of the polynomial are and ; This is for internal detection of photodiode current.

[0019] As a preferred technical solution, the benchmark calibration measurement process in step three is as follows: In a dark room under conditions free from ambient light interference, the sensor was immersed in a series of standard solutions with different known turbidities; For each standard solution, the laser source was alternately turned on and off, and the total scattered signal value V measured by the photodiode was recorded when the source was on. on_std The ambient light background signal value V measured when the light source is off. off_std ; Calculate the effective scattering signal difference ΔV at this turbidity. std =V on_std -V off_std V off_std ≈0; Record multiple sets of data at different temperatures: ; Establish an effective signal V std The precise mathematical model Q between the known turbidity value and the known turbidity value T (k, V std ), that is, the characteristic curve of the entire photodiode under different turbidity and light intensity.

[0020] As a preferred technical solution, the dynamic differential measurement process in step four is as follows: The laser source is controlled to alternately turn on and off, and the total scattered signal V is measured in real time when the source is on. on_measure and the ambient light background signal V when the light source is off off_base Calculate the current effective signal difference ΔV measure =V on_measure -V off_base The effective signal difference ΔV measure Substitute it into the mathematical model Q T (k, V std ) to perform the calculation; Its functional relationship is expressed as: NTU measure = Q T (k,ΔV measure ).

[0021] As a preferred technical solution, in step five, compensation for temperature T and light source power Po is added based on the dynamic differential measurement to construct a dynamic differential multiple regression model:

[0022] in, k1 and k2 are powers; P o The output optical power of the laser diode; T represents the ambient temperature.

[0023] This invention also provides a long-lasting, low-maintenance turbidity sensor with intelligent compensation function. Using the above method, the sensor includes a cylindrical body, with a flow-guiding protective cover at the head of the cylindrical body. The flow-guiding protective cover has multiple flow-guiding holes. Inside the flow-guiding protective cover, a copper alloy flow-guiding block is provided, with multiple flow-guiding channels on the copper alloy flow-guiding block. Inside the flow-guiding protective cover, an intelligent self-cleaning system is provided, comprising a sensing module, a decision-making module, and an execution module. The sensing module includes a temperature sensor and a turbidity sensor, which are used to monitor the ambient temperature T and the real-time turbidity reading and its changing trend in real time, respectively. The execution module includes ultraviolet LED beads and a high-frequency linear motor; The decision-making module includes a microcontroller.

[0024] As a preferred technical solution, the decision module generates cleaning instructions, including: Antibacterial control of UV LED beads: When the decision module determines that the risk of biological contamination has reached a preset threshold, the UV LED beads are activated; Vibration control of high-frequency linear motor: When the decision module determines that the risk of particulate matter pollution has reached a preset threshold, the high-frequency linear motor is activated.

[0025] As a preferred technical solution, the decision module determines whether to trigger cleaning and the cleaning intensity based on a cleaning determination coefficient, and the steps are as follows: Step 21, Data Synchronization and Feature Calculation: Data acquisition: The microcontroller unit of the single-chip microcomputer synchronously acquires the temperature sensor reading T at a fixed period, as well as the accurate turbidity value N after dynamic differential processing and multivariate regression compensation; Change rate calculation: Based on the turbidity value N in the time series, the turbidity change rate dN / dt per unit time is calculated in real time through differential or linear fitting methods to characterize the instantaneous speed of pollutant adhesion; Step 22, Quantitative assessment of cleaning needs: The real-time data (T-T0), N, and dN / dt are input into the function F to calculate the cleanliness determination coefficient s at the current moment. Step 23, Hierarchical decision-making and execution based on s: The calculated s value is compared with the preset multi-level thresholds s1, s2, and s3, where s1 is the low-risk threshold, s2 is the medium-risk threshold, and s3 is the high-risk threshold. When s < s1: the condition is determined to be clean or extremely low risk, no cleaning action is initiated, only continuous monitoring is performed; When s1 ≤ s < s2: it is judged as mild pollution or risk accumulation; If (T-T0) contributes a large proportion, it is determined that the risk is mainly microbial, and preventive ultraviolet LED irradiation is initiated. The irradiation duration is proportional to (s - s1). If the contribution of dN / dt is large, it is determined that particulate matter has begun to adhere, and low-frequency or short-term high-frequency linear motor (10) vibration is started for early intervention. When s ≥ s3: it is determined to be severe or complex pollution, and the ultraviolet LED and high-frequency linear motor are immediately activated.

[0026] As a preferred technical solution, in step 22, the formula for calculating the cleanliness determination coefficient is as follows:

[0027] in: s: Cleaning decision coefficient, used to determine whether cleaning is triggered and the intensity of cleaning; T: Real-time measured water temperature; T0: Baseline temperature for microbial reproduction. When T>T0, the risk of biocontamination increases significantly. N: Turbidity value measured in real time; dN / dt: Turbidity change rate, reflecting the rate of pollution accumulation.

[0028] This invention provides a long-lasting, low-maintenance turbidity sensor with intelligent compensation function and its calibration method. Through a dynamic differential multivariate regression calibration algorithm, it significantly improves the ability to withstand environmental interference and the long-term effectiveness of measurement accuracy. Through an intelligent self-cleaning system, it significantly reduces the impact of biological and non-biological adhesion on accuracy and reduces the maintenance frequency of the sensor. Attached Figure Description

[0029] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0030] Figure 1 This is a schematic diagram of the structure of the long-lasting, low-maintenance turbidity sensor with intelligent compensation function of the present invention; In the picture: 1-Flow guide protective cover; 2-Copper alloy flow guide block; 3-Narrow bandpass filter; 4-Quartz glass plate; 5-Optical path structure mounting base; 6-Photodiode; 7-UV LED bead; 8-Laser diode; 9-Temperature sensor; 10-High frequency linear motor; 11-Fixing component; 12-Sealing ring; 13-Circuit board; 14-Cylinder body; 15-Tail end cap; 16-Watertight connector. Detailed Implementation

[0031] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0032] Furthermore, the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. Additionally, the character " / " in this article, unless otherwise specified, generally indicates that the preceding and following related objects have an "or" relationship.

[0033] The embodiments of the present invention will now be described in further detail with reference to the accompanying drawings.

[0034] Example 1: This invention provides a calibration method for a long-lasting, low-maintenance turbidity sensor with intelligent compensation function. This method introduces a dynamic differential correction mechanism and considers multiple factors that affect water turbidity, such as temperature and light source aging rate (excitation light intensity), to achieve calibration measurement of the sensor.

[0035] Includes the following steps: Step 1: Precise calibration of the temperature sensor. This includes: S1 acquires multiple sample values ​​of the original temperature signal and performs ADC conversion on each sample value; then, it uses the mean method for smoothing and noise reduction to output a stable value under this condition.

[0036] The formula for calculating the stable value is as follows:

[0037] x i This is the raw ADC reading obtained from the i-th sample; m is the size of the data window; X is the calculated average value, which serves as the final stable reading for this state.

[0038] The core function of smoothing and noise reduction using the mean method is to suppress random errors generated during temperature acquisition by smoothing the data, thereby obtaining a stable value that represents the true output level of the sensor at a specific temperature point.

[0039] This step is crucial because the raw temperature signal directly read from the sensor during calibration may contain random fluctuations and noise. Therefore, precise calibration of the temperature sensor is necessary. Simultaneously, this step ensures that the sensor acquires accurate ambient temperature data, which is fundamental for all subsequent compensation.

[0040] S2, collect standard temperature data.

[0041] Multiple temperature measurements from the water bath temperature control system are collected, and different weights are assigned to the multiple temperature measurements using a variable-weight Gaussian function. The weighted average value is then calculated for temperature compensation.

[0042] Variable weight Gaussian function:

[0043] in: W(i) is the initial weight value of the i-th data point.

[0044] σ is the standard deviation of the variable-weight Gaussian function, which controls the degree of concentration of the weight distribution (the larger the value of σ, the flatter the weight distribution).

[0045] c is the center position of the weight distribution, which is usually set in the center of the current data window.

[0046] i is the index of the data point within the window.

[0047] Weight normalization:

[0048] in: ω (i) It is the normalized weight of the i-th data point.

[0049] m is the size of the data window.

[0050] The initial weights W for all data points within the window. (j) The sum of .

[0051] Gaussian weighted mean calculation: T(i)

[0052] Wherein: T filtered This is the filtered standard temperature value.

[0053] T (i) It is the i-th original temperature measurement value within the window.

[0054] ω (i) It is the normalized weight corresponding to the i-th data point.

[0055] m is the total number of data points (window size) involved in the calculation.

[0056] In step S2, during temperature calibration, it is necessary to acquire the corresponding standard temperature signal. However, current water bath temperature control systems typically have high thermal inertia and slow response, inevitably leading to fluctuations during use, which can affect calibration accuracy. Therefore, a variable-weight Gaussian filtering algorithm can be used. This algorithm assigns different weights to a series of temperature measurements using a Gaussian function, and then calculates a weighted average, aiming to effectively suppress random noise and improve data stability and accuracy.

[0057] S3. Establish a mathematical model and perform polynomial fitting using least squares on the stable values ​​in S1 and the standard temperature data in S2.

[0058] Establish sensor stability values Compared with the standard temperature T after variable weight Gaussian filtering filtered A precise mathematical model is needed. The mathematical model mainly uses polynomial fitting, and the fitting process usually employs the least squares method.

[0059] The polynomial form is:

[0060] Where a0, a1, ..., an denoted as the coefficients of the polynomial to be determined; n is the number of terms in the polynomial.

[0061] Step 2: Establish optical power P o The sensor internally detects the photodiode current I. pd The mathematical model.

[0062] The mathematical model is as follows:

[0063] Among them, P o The actual laser output power measured by an external standard optical power meter; I pd The sensor internally detects the current of the photodiode; The coefficients of the polynomial are and ; This is for internal detection of photodiode current.

[0064] Under constant current drive, the laser output optical power P o The degradation directly reflects its aging state. The key lies in establishing P... o With internal monitoring of photodiode current I pd The relationship between the two key data points is as follows: during the accelerated aging experiment, two key data points are measured simultaneously and periodically: the actual laser output optical power P measured using an external standard optical power meter. o The laser internally monitors the current I generated by the PD. pd .

[0065] To quickly and accurately establish the compensation relationship between the laser output intensity and the signal of the internal monitoring photodiode, an accelerated aging test method must be employed. The core of this method lies in accelerating the heat-related degradation process within the laser by applying environmental stress (primarily increased temperature) above normal levels, thereby simulating the aging effects under long-term use within a shorter experimental period. This solves the problem that, due to the extremely long lifespan of laser diodes under normal sensor operating conditions, the natural degradation process of their performance is too slow, making it difficult to obtain aging rate data through direct measurement within an acceptable timeframe.

[0066] The accelerated aging test method uses the Arrhenius equation to describe the exponential effect of temperature on chemical reaction rates, including aging rates:

[0067] in: AF: Acceleration factor, which means that spending 1 hour under accelerated conditions is equivalent to spending AF hours under normal conditions.

[0068] E aActivation energy represents the sensitivity of a laser material to temperature during a specific aging process, and is measured in electron volts (eV). For semiconductor lasers, this value is typically around 0.7 eV.

[0069] Boltzmann constant k: 8.617 × 10 -5 eV / K is a physical constant.

[0070] T use Normal rated operating temperature (unit: Kelvin, K).

[0071] T stress : The experimental temperature for accelerated aging (unit: Kelvin, K).

[0072] Step 3: Benchmark calibration measurement.

[0073] In a dark room under conditions free from ambient light interference, the sensor was immersed in a series of standard solutions with different known turbidities.

[0074] For each standard solution, the laser source was alternately turned on and off, and the total scattered signal value V measured by the photodiode was recorded when the source was on. on_std The ambient light background signal value V measured when the light source is off. off_std .

[0075] Calculate the effective scattering signal difference ΔV at this turbidity. std =V on_std -V off_std Because it is under darkroom conditions, therefore V off_std ≈0, which can be ignored.

[0076] Record multiple sets of data at different temperatures: .

[0077] Establish an effective signal V std The precise mathematical model Q between the known turbidity value and the known turbidity value T (k, V std ), that is, the characteristic curve of the entire photodiode under different turbidity and light intensity.

[0078] Step 4: Dynamic differential measurement.

[0079] In a real-world measurement environment, for the water body being measured, the laser light source is also controlled to alternately turn on and off, and the total scattered signal V is measured in real time when the light source is on. on_measure and the ambient light background signal V when the light source is off off_base Calculate the current effective signal difference ΔV measure =V on_measure -V off_base and the effective signal difference ΔV measureInput the mathematical model Q T( k,V off_measure The calculation is performed within the NTU (Network Turbine Unit) to obtain the real-time turbidity value. The functional relationship is: NTU measure = Q T (k,(V on_measure ))- Q T (k,(V off_base This can be optimized to represent as: NTU measure = Q T (k,ΔV measure ).

[0080] Step 5: Dynamic differential multivariate regression compensation to complete turbidity compensation calibration.

[0081] Based on the above dynamic differential measurement, compensation for temperature T and light source power Po is added to construct a dynamic differential multiple regression model:

[0082] in, k1 and k2 are powers; P o The output optical power of the laser diode is measured by the diode current I through an internal optical power detection sensor. pd get.

[0083] T represents the ambient temperature.

[0084] Dynamic differential multivariate regression compensation can effectively improve the long-term stability and accuracy of turbidity measurement under complex working conditions such as temperature variation, light source aging, and ambient light changes.

[0085] Example 2: See Figure 1 The present invention also provides a long-lasting, low-maintenance turbidity sensor with intelligent compensation function, comprising a cylindrical body 14, wherein a flow guide protective cover 1 is provided at the head (measuring end) of the cylindrical body 14, and the flow guide protective cover 1 has multiple flow guide holes; a copper alloy flow guide block 2 is provided inside the flow guide protective cover 1, and the copper alloy flow guide block 2 has multiple flow guide channels. The liquid to be detected enters the copper alloy flow guide block 2 through the multiple flow guide holes on the flow guide cover 1.

[0086] The flow guide cover 1 is equipped with an intelligent self-cleaning system. The intelligent self-cleaning system senses the characteristics of the water environment where the sensor in the above embodiment is located, and adaptively triggers and adjusts the cleaning strategy to combat the contamination of the optical window and ensure the stability and accuracy of long-term measurements.

[0087] The intelligent self-cleaning system includes a sensing module, a decision-making module, and an execution module.

[0088] The sensing module includes a temperature sensor and a turbidity sensor, which are used to monitor the ambient temperature T and the real-time turbidity reading and its changing trend in real time, respectively.

[0089] The execution module includes ultraviolet LED beads 7 and a high-frequency linear motor.

[0090] The decision-making module includes a microcontroller unit (MCU) storing a decision-making program. This module analyzes the ambient temperature (T) and turbidity signals input from the sensing module and determines whether cleaning should be performed. Specifically, the decision-making module intelligently judges the state of pollutants (type, accumulation rate, and severity) and then precisely controls the operating mode, activation timing, duration, and frequency of the execution module, achieving a shift from "periodic cleaning" to "on-demand cleaning."

[0091] Furthermore, the decision-making module determines the pollution status by analyzing the following parameters: Temperature monitoring and biofouling assessment: The decision-making module continuously reads temperature sensor data. When the temperature is within a suitable range for microbial reproduction (e.g., 10°C to 30°C), the risk of biofilm (such as algae and bacteria) adhesion is determined to be significantly increased. The higher the temperature and the longer the duration, the higher the assessed biofouling risk level, thus providing a basis for initiating ultraviolet antibacterial treatment.

[0092] Turbidity monitoring and particulate matter pollution assessment: The decision module analyzes the readings and rate of change of the turbidity sensor in real time. When the turbidity value remains high or an abnormal jump occurs, it is determined that the probability of suspended particulate matter in the water adhering to the optical window has increased. The higher the turbidity value and the faster the rate of increase in the reading, the higher the assessed risk level of particulate matter pollution, thus providing a signal to initiate mechanical cleaning.

[0093] The decision-making module outputs cleaning instructions to the execution module, and the execution module performs corresponding actions based on the cleaning instructions. The specific cleaning instructions generated by the decision-making module include: Antibacterial control of UV LED bead 7: When the decision module determines that the risk of biological contamination reaches a preset threshold, the UV LED bead 7 is activated. Its irradiation duration and frequency are not fixed but adaptively adjusted according to the contamination risk level. For example, during hot seasons when microbial reproduction is active, an irradiation strategy of "low power always on" or "high frequency short pulse" may be adopted; while during cold seasons, the operating frequency can be reduced or a dormant mode can be entered, thereby achieving a balance between efficient antibacterial control and energy consumption control.

[0094] Vibration control of the high-frequency linear motor: When the decision module determines that the risk of particulate matter contamination has reached a preset threshold, the high-frequency linear motor is activated. Its vibration intensity, duration, and operating mode (such as intermittent vibration or continuous strong vibration) are also dynamically adjusted by the algorithm according to the severity of the contamination. For slight contamination trends, a brief, low-intensity vibration is triggered for preventative cleaning; for severe blockage signals, a high-intensity, sustained vibration may be activated to ensure thorough removal of adhering substances.

[0095] The decision module determines whether to trigger cleaning and the cleaning intensity based on a cleaning determination coefficient. The specific steps are as follows: Step 21: Data synchronous acquisition and feature calculation.

[0096] Data acquisition: The microcontroller unit of the single-chip microcomputer synchronously acquires the temperature sensor reading T at a fixed period, as well as the accurate turbidity value N after dynamic differential processing and multivariate regression compensation.

[0097] Change rate calculation: Based on the turbidity value N in the time series, the turbidity change rate dN / dt per unit time is calculated in real time through differential or linear fitting methods to characterize the instantaneous velocity of pollutant adhesion.

[0098] Step 22, quantitative assessment of cleaning needs.

[0099] The real-time data (T-T0), N, and dN / dt are input into the function F to calculate the cleanliness determination coefficient s at the current moment.

[0100] The formula for calculating the cleanliness judgment coefficient is as follows:

[0101] in: s: Cleaning decision coefficient (output), used to determine whether to trigger cleaning and the cleaning intensity.

[0102] T: Real-time measured water temperature.

[0103] T0: The baseline temperature for microbial reproduction (e.g., 20°C). When T>T0, the risk of biocontamination increases significantly.

[0104] N: Turbidity value measured in real time (NTU).

[0105] dN / dt: Turbidity change rate, reflecting the rate of pollution accumulation. A positive and large value indicates that pollutants are rapidly adhering.

[0106] The function F makes: The (T-T0) item quantifies the risk of biological contamination; the higher the temperature, the faster the microorganisms multiply, and the greater the contribution value.

[0107] The N-item quantifies the current pollution load; the higher the absolute value of turbidity, the more severe the pollution.

[0108] The dN / dt term quantifies the accelerating trend of pollution; the greater the rate of change, the faster the pollution is accumulating, requiring urgent intervention.

[0109] Step 23: Hierarchical decision-making and execution based on the s-value.

[0110] The calculated s value is compared with preset multi-level thresholds (s1, s2, s3), where s1 is the low-risk threshold, s2 is the medium-risk threshold, and s3 is the high-risk threshold. When s < s1 (low-risk threshold): the status is determined to be clean or extremely low risk, no cleaning action is initiated, only continuous monitoring is performed.

[0111] When s1 ≤ s < s2 (medium risk threshold): it is judged as mild pollution or risk accumulation.

[0112] If (T-T0) contributes a large proportion, it is determined that the risk is mainly microbial, and preventive ultraviolet LED irradiation is initiated. The irradiation duration is proportional to (s-s1).

[0113] If the contribution of dN / dt is large, it is determined that particulate matter has begun to adhere, and a low-frequency or short-term high-frequency linear motor 10 is started to vibrate for early intervention.

[0114] When s ≥ s3 (high-risk threshold): it is judged as severe or compound pollution.

[0115] Instantly activate the high-intensity collaborative cleaning mode: simultaneously activate the UV LED 7 and the high-frequency linear motor 10, with both working intensity and duration increasing as the S value exceeds the standard.

[0116] See Figure 1 In terms of structural connection, the two parts of the cylinder 14 are respectively provided with a fixing member 11 and a tail end cap 15. The fixing member 11 and the tail end cap 15 are equipped with sealing rings 12 (grooves are provided on the tail end cap 15 and the fixing member 11, and the grooves cooperate with the sealing rings to achieve a seal), so that the inner cavity of the cylinder 14 forms a sealed structure. The tail end cap 15 is provided with a watertight connector 16 to ensure the overall protection level.

[0117] The end of the fixing member 11 is fitted with a flow guide protective cover 1. The liquid to be detected enters the copper alloy flow guide block 2 through multiple flow guide holes on the flow guide cover 1, and contacts the optical path structure mounting base 5 through the flow guide channel on the copper alloy flow guide block 2.

[0118] The circuit board 13 is located in the inner cavity of the cylinder 14, and the optical path structure mounting base 5 is located in the flow guide protective cover 1 with its bottom in contact with the liquid being tested.

[0119] The optical path structure mounting base 5 is equipped with a laser emission optical path and a signal receiving optical path. The laser emission optical path and the signal receiving optical path are at 90° and perpendicular to each other. The intersection point of the optical paths is arranged in the water sample outside the optical window, thereby forming a fixed turbidity detection area to ensure the consistency and accuracy of the measurement.

[0120] The optical window includes a narrow bandpass filter 3 and two quartz glass plates 4.

[0121] The laser emission path includes a laser diode 8 and a quartz glass plate 4, with the laser diode 8 serving as the emission source. The highly collimated monochromatic laser beam emitted by the laser diode 8 passes through the quartz glass plate 4 and enters the liquid to be tested. The laser diode 8 integrates a monitoring photodiode for real-time, in-situ monitoring of the laser emission power, providing a first feedback signal for light source fluctuation compensation.

[0122] The signal receiving optical path includes a photodiode 6 and a narrowband bandpass filter 3, the center wavelength of which is 860 nm. The light signal scattered by suspended particles in the water passes through the narrowband filter 3 and is then received by the photodiode 6. The center wavelength of the narrowband bandpass filter 3 matches the emission wavelength of the laser diode 6 to efficiently suppress ambient stray light interference. Subsequently, the photodiode 6 converts the optical signal into an electrical signal.

[0123] Specifically, the photodiode 6, ultraviolet LED 7, laser diode 8, and high-frequency linear motor 10 are all mounted on the optical path structure mounting base 5. A narrow-band bandpass filter 3 is provided at the front end of the photodiode 6 (on the side closer to the liquid being measured); a quartz glass plate 4 is provided at the front end of the laser diode 8; and a quartz glass plate 4 is also provided at the front end of the ultraviolet LED 7.

[0124] A temperature sensor 9 is also provided on the optical path structure mounting base 5. The temperature sensor 9 is located near the optical window. The temperature sensing end of the temperature sensor 9 directly contacts the medium being measured, acquiring the medium temperature in real time and providing a second feedback signal for temperature compensation of the turbidity sensor.

[0125] Both the ultraviolet LED beads 7 and the high-frequency linear motor 10 are mounted on the optical path structure mounting base 5.

[0126] The microcontroller on the circuit board 13 integrates a dynamic differential multivariate regression calibration model. This dynamic differential multivariate regression calibration model can simultaneously acquire the monitoring photodiode signal in the laser diode 8, the signal from the temperature sensor 9, and the signal from the photodiode 6. Through the compensation algorithm of the dynamic differential multivariate regression calibration model, it performs real-time digital correction on the fluctuation of the light source itself and the temperature drift of the system, and finally outputs a high-precision turbidity value.

[0127] System hardware configuration:

[0128] Laser source: A VCSEL laser diode with a wavelength of 860nm is used, which has a built-in monitoring photodiode (PD) to monitor laser power fluctuations in real time. The laser beam is collimated into parallel light after passing through a collimating lens and is perpendicularly incident on the water body to be tested through a quartz glass window.

[0129] Temperature sensor: PT100 is sealed and installed near the optical path to measure ambient temperature T.

[0130] Optical path design: The transmitting optical path and the receiving optical path are at a 90° angle. The scattered light is collected by the receiving lens and then filtered out by a narrow bandpass filter (3) with a center wavelength of 860nm and a bandwidth of ±20nm. Finally, it is converted into an electrical signal by a silicon photodiode.

[0131] Optical path calibration: The optical path focus is adjusted to be 5mm outside the quartz glass window by using precision optical path structural components to avoid interference from reflections on the window surface.

[0132] The turbidity sensor of this invention has a triple protection mechanism of "prevention-inhibition-removal": First, a passive antibacterial layer: a copper alloy guide block 2 is embedded in the flow guide protective cover 1 in front of the optical window, which uses the continuous release of copper ions to inhibit the adhesion of microorganisms. Second, an active disinfection layer: ultraviolet LED beads 7 are integrated in the optical path structure mounting base 5, and their irradiation frequency and duration are controlled by the circuit board 13 to directly kill bacteria. Third, a mechanical removal layer: a high-frequency linear motor 10 drives the optical path structure mounting base 5 to perform high-frequency short-amplitude vibration for physical removal of adhering contaminants.

[0133] This invention provides a long-lasting, low-maintenance turbidity sensor with intelligent compensation function and its calibration method. Through a dynamic differential multivariate regression calibration algorithm, it significantly improves the ability to withstand environmental interference and the long-term effectiveness of measurement accuracy. Through an intelligent self-cleaning system, it significantly reduces the impact of biological and non-biological adhesion on accuracy and reduces the maintenance frequency of the sensor.

[0134] Of course, the present invention may have other various embodiments. Without departing from the spirit and essence of the present invention, those skilled in the art can make various corresponding changes and modifications according to the present invention, but these corresponding changes and modifications should all fall within the protection scope of the appended claims.

Claims

1. A calibration method for a long-lasting, low-maintenance turbidity sensor with intelligent compensation function, characterized in that, Includes the following steps: Step 1: Precise calibration of the temperature sensor; Step 2: Establish optical power P o The sensor internally detects the photodiode current I. pd Mathematical model; Step 3: Benchmark calibration measurement; Step 4: Dynamic differential measurement; Step 5: Dynamic differential multivariate regression compensation to complete turbidity compensation calibration.

2. The calibration method for a long-lasting, low-maintenance turbidity sensor with intelligent compensation function according to claim 1, characterized in that, In step one, the precise calibration process of the temperature sensor is as follows: S1: Acquire multiple sample values ​​of the original temperature signal and perform ADC conversion on each sample value; then use the mean method for smoothing and noise reduction, and output the stable value under this state; The formula for calculating the stable value is as follows: x i This is the raw ADC reading obtained from the i-th sample; m is the size of the data window; X is the calculated average value, which serves as the final stable reading under this condition; S2, collect standard temperature data; Multiple temperature measurements from the water bath temperature control system are collected. Different weights are assigned to the multiple temperature measurements using a variable weight Gaussian function, and a weighted average is calculated for temperature compensation. Variable weight Gaussian function: in: W(i) is the initial weight value of the i-th data point; σ is the standard deviation of the variable-weight Gaussian function, which controls the degree of concentration of the weight distribution; the larger the value of σ, the flatter the weight distribution. c is the center position of the weight distribution; i is the index of the data point within the window; Weight normalization: in: ω (i) It is the normalized weight of the i-th data point; m is the size of the data window; The initial weights W for all data points within the window. (j) The sum; Gaussian weighted mean calculation: T(i) Wherein: T filtered This is the filtered standard temperature value; T (i) It is the i-th original temperature measurement value within the window; ω (i) It is the normalized weight corresponding to the i-th data point; m is the total number of data points involved in the calculation; S3. Establish a mathematical model and perform polynomial fitting on the stable values ​​in S1 and the standard temperature data in S2 using least squares. Establish sensor stability value Compared with the standard temperature T after variable weight Gaussian filtering filtered A precise mathematical model between them was established, and the mathematical model was fitted using polynomials, with the least squares method used in the fitting process. The polynomial form is: Where a0, a1, ..., a n denoted as the coefficients of the polynomial to be determined; n is the number of terms in the polynomial.

3. The calibration method for a long-lasting, low-maintenance turbidity sensor with intelligent compensation function according to claim 2, characterized in that, In step two, the mathematical model is as follows: Among them, P o The actual laser output power measured by an external standard optical power meter; I pd The sensor internally detects the current of the photodiode; The coefficients of the polynomial are and ; This is for internal detection of photodiode current.

4. The calibration method for a long-lasting, low-maintenance turbidity sensor with intelligent compensation function according to claim 3, characterized in that, In step three, the benchmark calibration measurement process is as follows: In a dark room under conditions free from ambient light interference, the sensor was immersed in a series of standard solutions with different known turbidities; For each standard solution, the laser source was alternately turned on and off, and the total scattered signal value V measured by the photodiode was recorded when the source was on. on_std The ambient light background signal value V measured when the light source is off. off_std ; Calculate the effective scattering signal difference ΔV at this turbidity. std =V on_std -V off_std V off_std ≈0; Record multiple sets of data at different temperatures: ; Establish an effective signal V std The precise mathematical model Q between the known turbidity value and the known turbidity value T (k, V std ), that is, the characteristic curve of the entire photodiode under different turbidity and light intensity.

5. The calibration method for a long-lasting, low-maintenance turbidity sensor with intelligent compensation function according to claim 4, characterized in that, In step four, the dynamic differential measurement process is as follows: The laser source is controlled to alternately turn on and off, and the total scattered signal V is measured in real time when the source is on. on_measure and the ambient light background signal V when the light source is off off_base Calculate the current effective signal difference ΔV measure =V on_measure -V off_base The effective signal difference ΔV measure Substitute it into the mathematical model Q T (k, V std ) to perform the calculation; Its functional relationship is expressed as: NTU measure = Q T (k,ΔV measure ).

6. The calibration method for a long-lasting, low-maintenance turbidity sensor with intelligent compensation function according to claim 5, characterized in that, In step five, based on the dynamic differential measurement, compensation for temperature T and light source power Po is added to construct a dynamic differential multiple regression model: in, k1 and k2 are powers, respectively; P o The output optical power of the laser diode; T represents the ambient temperature.

7. A long-lasting, low-maintenance turbidity sensor with intelligent compensation function, applied to the method described in any one of claims 1-6, characterized in that, The device includes a cylindrical body (14), the head of which is provided with a flow guide protective cover (1), and the flow guide protective cover (1) is provided with multiple flow guide holes; a copper alloy flow guide block (2) is provided inside the flow guide protective cover (1), and the copper alloy flow guide block (2) is provided with multiple flow guide channels; an intelligent self-cleaning system is provided inside the flow guide protective cover (1), and the intelligent self-cleaning system includes a sensing module, a decision-making module and an execution module; The sensing module includes a temperature sensor and a turbidity sensor, which are used to monitor the ambient temperature T and the real-time turbidity reading and its changing trend in real time, respectively. The execution module includes ultraviolet LED beads (7) and a high-frequency linear motor (10). The decision-making module includes a microcontroller.

8. A long-lasting, low-maintenance turbidity sensor with intelligent compensation function according to claim 7, characterized in that, The decision module generates cleaning instructions, including: Antibacterial regulation of ultraviolet LED beads (7): When the decision module determines that the risk of biological contamination reaches the preset threshold, the ultraviolet LED beads (7) are activated. Vibration control of high-frequency linear motor (10): When the decision module determines that the risk of particulate matter pollution reaches the preset threshold, the high-frequency linear motor (10) is started.

9. A long-lasting, low-maintenance turbidity sensor with intelligent compensation function according to claim 7, characterized in that, The decision module determines whether to trigger cleaning and the cleaning intensity based on a cleaning determination coefficient, as follows: Step 21, Data Synchronization and Feature Calculation: Data acquisition: The microcontroller unit of the single-chip microcomputer synchronously acquires the temperature sensor reading T at a fixed period, as well as the accurate turbidity value N after dynamic differential processing and multivariate regression compensation; Change rate calculation: Based on the turbidity value N in the time series, the turbidity change rate dN / dt per unit time is calculated in real time through differential or linear fitting methods to characterize the instantaneous speed of pollutant adhesion; Step 22, Quantitative assessment of cleaning needs: The real-time data (T-T0), N, and dN / dt are input into the function F to calculate the cleanliness determination coefficient s at the current moment. Step 23, Hierarchical decision-making and execution based on s: The calculated s value is compared with the preset multi-level thresholds s1, s2, and s3, where s1 is the low-risk threshold, s2 is the medium-risk threshold, and s3 is the high-risk threshold. When s < s1: the condition is determined to be clean or extremely low risk, no cleaning action is initiated, only continuous monitoring is performed; When s1 ≤ s < s2: it is judged as mild pollution or risk accumulation; If (T-T0) contributes a large proportion, it is determined that the risk is mainly microbial, and preventive ultraviolet LED irradiation is initiated. The irradiation duration is proportional to (s - s1). If the contribution of dN / dt is large, it is determined that particulate matter has begun to adhere, and low-frequency or short-term high-frequency linear motor (10) vibration is started for early intervention. When s ≥ s3: it is determined to be severe or complex pollution, and the ultraviolet LED and high-frequency linear motor are immediately activated.

10. A long-lasting, low-maintenance turbidity sensor with intelligent compensation function according to claim 9, characterized in that, In step 22, the formula for calculating the cleanliness judgment coefficient is as follows: in: s: Cleaning decision coefficient, used to determine whether cleaning is triggered and the intensity of cleaning; T: Real-time measured water temperature; T0: Baseline temperature for microbial reproduction. When T>T0, the risk of biocontamination increases significantly. N: Turbidity value measured in real time; dN / dt: Turbidity change rate, reflecting the rate of pollution accumulation.

Images