Optical dissolved oxygen sensor calibration method based on nonlinear temperature compensation

Abstract

The invention discloses an optical dissolved oxygen sensor calibration method based on nonlinear temperature compensation, and belongs to the field of sensor calibration, and the method comprises the following steps: S1, building a calibration platform; s2, data acquisition: under a plurality of temperature gradients, generating a dissolved oxygen saturated water sample through nitrogen-oxygen mixed gas, and synchronously acquiring fluorescence lifetime and influence factors; s3, establishing a nonlinear compensation model of the fluorescence lifetime and the temperature, and optimizing model parameters through a Levenberg-Marquardt algorithm; s4, dynamically allocating a weight coefficient; and S5, predicting the drift distance of the optical dissolved oxygen sensor to be calibrated, and realizing closed-loop self-maintenance calibration. By adopting the optical dissolved oxygen sensor calibration method based on nonlinear temperature compensation, not only is the influence of temperature on the optical dissolved oxygen sensor considered, but also multiple environmental factors such as salinity and pressure are incorporated, and the calibration accuracy and comprehensiveness are improved through multi-parameter collaborative correction.

Description

Technical Field

[0001] The present invention relates to the technical field of sensor calibration, and in particular to a calibration method for an optical dissolved oxygen sensor based on nonlinear temperature compensation. Background Art

[0002] An optical dissolved oxygen sensor is a precision device used to measure dissolved oxygen concentration in water. It detects oxygen levels through a photochemical reaction. This type of sensor typically consists of a driver light source (LED) and a photodetector. When blue light of a specific wavelength strikes a fluorescent substance at the sensor's front end, the substance absorbs the energy and emits red light. If oxygen molecules are present in the surrounding environment, they collide with the excited fluorescent substance, causing its luminescence intensity to decrease or cease. The resulting signal change can be used to calculate the dissolved oxygen concentration.

[0003] The traditional optical dissolved oxygen sensor calibration method has the following technical defects: 1. Oversimplification of temperature compensation models: Conventional methods often use linear or second-order polynomials to fit the relationship between temperature and fluorescence lifetime. However, these methods struggle to describe nonlinear thermodynamic processes (such as changes in the activation energy of fluorescence quenching) over a wide temperature range (0-40°C), leading to cumulative errors exceeding ±2.5% in high-temperature regions. For example, existing techniques use iodine titration to obtain a dissolved oxygen baseline. While this method can eliminate some temperature interference, it requires manual preparation of chemical reagents and lacks in-situ calibration. This makes the process complex (taking ≥30 minutes per test) and reduces accuracy by up to 35% at salinity levels above 20 PSU.

[0004] 2. Lack of coordinated correction of multiple environmental parameters: Interference factors such as salinity and pressure are not included in the calibration system. Existing technologies mostly use a step-by-step correction strategy (such as temperature compensation first and then pressure correction), which leads to an error superposition effect.

[0005] 3. Inadequate dynamic drift suppression: Traditional light source control relies on a constant current drive module without a closed-loop feedback mechanism. This results in the inability to correct for LED intensity drift (±3%) and baseline drift caused by optical window contamination (>0.8mg / L / week) in real time. Some methods use static calibration coefficients to compensate for temperature effects but lack the integration of dynamic algorithms (such as LSTM prediction models), resulting in poor long-term stability (drift >1.2mg / L over 30 days). Summary of the Invention

[0006] The purpose of the present invention is to provide an optical dissolved oxygen sensor calibration method based on nonlinear temperature compensation to solve the above technical problems.

[0007] To achieve the above object, the present invention provides a calibration method for an optical dissolved oxygen sensor based on nonlinear temperature compensation, comprising the following steps: S1. Build a calibration platform: Fill the multi-temperature zone reaction chamber with water and immerse the optical dissolved oxygen sensor, temperature sensor, salinity sensor, and pressure sensor to be calibrated in the water; S2. Data acquisition: Generate dissolved oxygen-saturated water samples using nitrogen and oxygen mixed gases at multiple temperature gradients, and simultaneously collect fluorescence lifetime and influencing factors; S3. Model construction: Based on the joint fitting of the Arrhenius equation and the fourth-order polynomial, a nonlinear compensation model of fluorescence lifetime and temperature was established, and the model parameters were optimized using the Levenberg-Marquardt algorithm; S4. The improved Weiss equation is used to calculate the theoretical dissolved oxygen saturation concentration, and the weight coefficient is dynamically assigned based on the measurement standard deviation of the salinity sensor and the pressure sensor to achieve multi-parameter coordinated correction; S5. Use the PID controller to adjust the intensity fluctuation of the driving light source of the optical dissolved oxygen sensor to be calibrated, and use the LSTM network to predict the drift of the optical dissolved oxygen sensor to be calibrated, so as to achieve closed-loop self-maintenance calibration.

[0008] Preferably, the multiple temperature gradients described in step S2 include 0°C, 10°C, 20°C, 30°C, and 40°C; the influencing factors include dissolved oxygen concentration ,salinity and pressure .

[0009] Preferably, step S3 specifically includes the following steps: S31. Use the Arrhenius equation to describe the effect of temperature on the activation energy of fluorescence quenching: (1); Where, Indicates water temperature The time constant of the fluorescence quenching process under represents the activation energy of the fluorescence quenching process, in eV; Represents the theoretical fluorescence lifetime at absolute zero, in μs; is the Boltzmann constant, and ; S32. Compensation of nonlinear optical drift caused by temperature using a fourth-order polynomial: (2); Where, represents the predicted fluorescence lifetime; 、 、 、 and They are baseline drift compensation constant term, linear drift compensation constant term, thermal expansion effect compensation constant term, nonlinear optical response constant term and high-order thermodynamic effect constant term; S33. Combining formula (1) and formula (2) to obtain the nonlinear compensation model of fluorescence lifetime and temperature: (3); Where, Indicates water temperature The final fluorescence lifetime; S34, optimize model parameters by Levenberg-Marquardt algorithm; S341, initialization parameters: , , initial damping factor , initial step size scaling factor , and estimated by the least squares method ; S342, iterative optimization: S3421, calculate the residual sum of squares : (4); Where, Indicates the amount of data collected for each temperature gradient; Indicates water temperature Fluorescence lifetime measured under ; S3422, adjust the damping factor , update the parameter vector ;‌ S3423, iterating steps S3421-S3422 until the set number of iterations is reached, and outputting the optimized nonlinear compensation model of fluorescence lifetime and temperature; S343. Use goodness of fit, residual control, and temperature sensitivity to verify the optimized nonlinear compensation model of fluorescence lifetime and temperature, and output the optimized nonlinear compensation model of fluorescence lifetime and temperature after verification.

[0010] Preferably, step S4 specifically includes the following steps: S41. Construct the Weiss standard equation: (5); Where, Indicates water temperature Saturated vapor pressure under 、 and are empirical coefficients, and ; Indicates standard atmospheric pressure; S42. Construct comprehensive saturation concentration formula: (6); Where, represents the comprehensive saturation concentration; represents the theoretical saturation concentration considering only pressure; represents the theoretical saturation concentration considering only salinity; in, (7); (8); Where, Indicates salinity in PSU; 、 and are salinity correction factors‌, and ; Indicates real-time pressure in kPa; represents the pressure nonlinear correction coefficient‌, and ; S43. Dynamically allocate weight coefficients: (9); Where, Indicates the corrected dissolved oxygen value; Represents the salinity sensor weight coefficient; Indicates the raw dissolved oxygen measurement value in mg / L; Represents the pressure sensor weight coefficient; in, (10); Where, Indicates the standard deviation of salinity sensor measurement; Indicates the pressure sensor measurement standard deviation.

[0011] Preferably, step S5 specifically includes the following steps: S51, data acquisition: synchronously collect the current value of the driving light source every 1 second , raw dissolved oxygen measurement value , water temperature and salinity And pre-processing; S52, PID adjusts the light source intensity; S521, real-time acquisition of current fluctuation signals of driving light sources and set the value Generate for benchmark Error signal at time ; S522, using incremental PID algorithm to calculate the control quantity of the driving light source intensity : (11); Where, 、 and Represents the proportional gain coefficient, integral gain coefficient and differential gain coefficient; represents the initial moment; Indicates the sampling period; express Error signal at the moment; S523, suppressing the current fluctuation of the driving light source within the set threshold range by outputting the control voltage through PWM modulation; S53, using the LSTM network to predict the drift of the optical dissolved oxygen sensor to be calibrated; S531, construct an LSTM network structure, which includes an input layer, two hidden layers and an output layer arranged in sequence, wherein the input variable of the input layer is the water temperature ,salinity , Accumulated working time of driving light source ; Output layer output Predicted value of momentary drift , represents the Sigmoid function, represents the weight matrix, represents the state of the hidden layer, represents the bias term; S532, set the current drift prediction value Dynamic compensation is performed when it is greater than the set value: (12); Where, express Time-calibrated dissolved oxygen value; express Original dissolved oxygen value; Indicates the drift change rate suppression factor; S533, parameter self-update: (13); Where, Represents the updated PID controller parameters; Indicates the PID parameter learning rate; Indicates control performance indicators.

[0012] Preferably, the pre-processing in step S51 includes removing abnormal current values ​​with fluctuation ranges greater than ±5% of the set value, and performing the pre-processing on the original dissolved oxygen measurement value. Perform sliding window mean filtering.

[0013] Preferably, the set threshold range in step S523 is ±0.3 mA.

[0014] Therefore, the present invention adopts the above-mentioned optical dissolved oxygen sensor calibration method based on nonlinear temperature compensation, which has the following beneficial effects: 1. Improved calibration accuracy: Through nonlinear temperature compensation model and multi-parameter coordinated correction, the calibration accuracy of optical dissolved oxygen sensors is significantly improved, especially in wide temperature ranges and high salinity conditions; 2. Enhanced environmental adaptability: The impact of multiple environmental factors such as temperature, salinity and pressure on the sensor is comprehensively considered, enhancing the adaptability and stability of the sensor in different environments; 3. Achieve closed-loop self-maintenance calibration: By combining a PID controller and an LSTM network, real-time monitoring and dynamic compensation of light source intensity fluctuations and sensor drift are achieved, reducing maintenance costs and improving the long-term stability of the sensor. 4. Simplified operation process: No need to prepare chemical reagents manually, and in-situ calibration can be achieved, which simplifies the operation process and improves calibration efficiency; 5. Improve sensor performance: By optimizing model parameters and dynamic correction strategies, the sensor's baseline drift and light intensity drift are effectively suppressed, improving the sensor's measurement accuracy and stability.

[0015] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 This is a flow chart of a calibration method for an optical dissolved oxygen sensor based on nonlinear temperature compensation according to the present invention. DETAILED DESCRIPTION

[0017] In order to make the purposes, technical solutions and advantages disclosed in the embodiments of the present invention clearer, the embodiments of the present invention are further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the embodiments of the present invention and are not intended to limit the embodiments of the present invention. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application. Examples of the embodiments are shown in the accompanying drawings, where the same or similar numbers throughout represent the same or similar elements or elements with the same or similar functions.

[0018] It should be noted that the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or server that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or are inherent to these processes, methods, products or devices.

[0019] The embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0020] like Figure 1 As shown, a calibration method for an optical dissolved oxygen sensor based on nonlinear temperature compensation includes the following steps: S1. Build a calibration platform: Fill the multi-temperature zone reaction chamber with water and immerse the optical dissolved oxygen sensor, temperature sensor, salinity sensor, and pressure sensor to be calibrated in the water; S2. Data acquisition: Generate dissolved oxygen-saturated water samples using nitrogen and oxygen mixed gases at multiple temperature gradients, and simultaneously collect fluorescence lifetime and influencing factors; The multiple temperature gradients described in step S2 include 0°C, 10°C, 20°C, 30°C, and 40°C; the influencing factors include dissolved oxygen concentration ,salinity and pressure .

[0021] S3. Model construction: Based on the joint fitting of the Arrhenius equation and the fourth-order polynomial, a nonlinear compensation model of fluorescence lifetime and temperature was established, and the model parameters were optimized using the Levenberg-Marquardt algorithm; Step S3 specifically includes the following steps: S31. Use the Arrhenius equation to describe the effect of temperature on the activation energy of fluorescence quenching: (1); Where, Indicates water temperature The time constant of the fluorescence quenching process under represents the activation energy of the fluorescence quenching process, in eV; Represents the theoretical fluorescence lifetime at absolute zero, in μs; is the Boltzmann constant, and ; S32. Compensation of nonlinear optical drift caused by temperature using a fourth-order polynomial: (2); Where, represents the predicted fluorescence lifetime; 、 、 、 and They are baseline drift compensation constant term, linear drift compensation constant term, thermal expansion effect compensation constant term, nonlinear optical response constant term and high-order thermodynamic effect constant term; S33. Combining formula (1) and formula (2) to obtain the nonlinear compensation model of fluorescence lifetime and temperature: (3); Where, Indicates water temperature The final fluorescence lifetime; S34, optimize model parameters by Levenberg-Marquardt algorithm; S341, initialization parameters: , , initial damping factor , initial step size scaling factor , and estimated by the least squares method ; S342, iterative optimization: S3421, calculate the residual sum of squares : (4); Where, Indicates the amount of data collected for each temperature gradient; Indicates water temperature Fluorescence lifetime measured under ; S3422, adjust the damping factor , update the parameter vector ;‌ S3423, iterating steps S3421-S3422 until the set number of iterations is reached, and outputting the optimized nonlinear compensation model of fluorescence lifetime and temperature; S343. Use goodness of fit, residual control, and temperature sensitivity to verify the optimized nonlinear compensation model of fluorescence lifetime and temperature, and output the optimized nonlinear compensation model of fluorescence lifetime and temperature after verification.

[0022] S4. The improved Weiss equation is used to calculate the theoretical dissolved oxygen saturation concentration, and the weight coefficient is dynamically assigned based on the measurement standard deviation of the salinity sensor and the pressure sensor to achieve multi-parameter coordinated correction; Step S4 specifically includes the following steps: S41. Construct the Weiss standard equation: (5); Where, Indicates water temperature Saturated vapor pressure under 、 and are empirical coefficients, and ; Indicates standard atmospheric pressure; S42. Construct comprehensive saturation concentration formula: (6); Where, represents the comprehensive saturation concentration; represents the theoretical saturation concentration considering only pressure; represents the theoretical saturation concentration considering only salinity; in, (7); (8); Where, Indicates salinity in PSU; 、 and are salinity correction factors‌, and ; Indicates real-time pressure in kPa; represents the pressure nonlinear correction coefficient‌, and ; S43. Dynamically allocate weight coefficients: (9); Where, Indicates the corrected dissolved oxygen value; Represents the salinity sensor weight coefficient; Indicates the raw dissolved oxygen measurement value in mg / L; Represents the pressure sensor weight coefficient; in, (10); Where, Indicates the standard deviation of salinity sensor measurement; Indicates the pressure sensor measurement standard deviation.

[0023] S5. Use the PID controller to adjust the intensity fluctuation of the driving light source of the optical dissolved oxygen sensor to be calibrated, and use the LSTM network (long short-term memory network) to predict the drift of the optical dissolved oxygen sensor to be calibrated, thereby achieving closed-loop self-maintenance calibration.

[0024] Step S5 specifically includes the following steps: S51, data acquisition: synchronously collect the current value of the driving light source every 1 second , raw dissolved oxygen measurement value , water temperature and salinity And pre-processing; Preferably, the pre-processing in step S51 includes removing abnormal current values ​​with fluctuation ranges greater than ±5% of the set value, and performing the pre-processing on the original dissolved oxygen measurement value. Perform sliding window mean filtering.

[0025] S52, PID adjusts the light source intensity; S521, real-time acquisition of current fluctuation signals of driving light sources and set the value mA is used as the reference to generate Error signal at time ; S522, using incremental PID algorithm to calculate the control quantity of the driving light source intensity : (11); Where, 、 and Represents the proportional gain coefficient, integral gain coefficient and differential gain coefficient; represents the initial moment; Indicates the sampling period; express Error signal at the moment; S523, outputting a control voltage through PWM modulation (pulse width modulation) to suppress fluctuations in the current driving the light source within a set threshold range; Preferably, the set threshold range in step S523 is ±0.3 mA.

[0026] S53, using the LSTM network to predict the drift of the optical dissolved oxygen sensor to be calibrated; S531, construct an LSTM network structure, which includes an input layer, two hidden layers and an output layer arranged in sequence, wherein the input variable of the input layer is the water temperature ,salinity , Accumulated working time of driving light source ; Output layer output Predicted value of momentary drift , represents the Sigmoid function, represents the weight matrix, represents the state of the hidden layer, represents the bias term; S532, set the current drift prediction value Dynamic compensation is performed when it is greater than the set value: (12); Where, express Time-calibrated dissolved oxygen value; express Original dissolved oxygen value; Indicates the drift change rate suppression factor; S533, parameter self-update: (13); Where, Represents the updated PID (proportional-integral-derivative) controller parameters; Indicates the PID parameter learning rate; Indicates control performance indicators.

[0027] Simulation experiment: 1. Experimental Preparation: Build the calibration platform: Fill the multi-temperature reaction chamber with water and immerse the optical dissolved oxygen sensor, temperature sensor, salinity sensor, and pressure sensor to be calibrated. Set the temperature gradient: Select 0°C, 10°C, 20°C, 30°C, and 40°C as the experimental temperature gradient. Prepare a dissolved oxygen-saturated water sample: Generate a dissolved oxygen-saturated water sample using a nitrogen and oxygen gas mixture.

[0028] 2. Data acquisition: At each temperature gradient, fluorescence lifetime, dissolved oxygen concentration, salinity, and pressure data are collected simultaneously.

[0029] 3. Data preprocessing: Abnormal current values ​​with fluctuation amplitude greater than ±5% are eliminated, and the original dissolved oxygen measurement values ​​are filtered using a sliding window mean.

[0030] 4. Model construction and optimization: Nonlinear compensation model: Based on the joint fitting of the Arrhenius equation and the fourth-order polynomial, a nonlinear compensation model of fluorescence lifetime and temperature is established.

[0031] Parameter Optimization: The Levenberg-Marquardt algorithm is used to optimize the model parameters. The residual sum of squares is calculated iteratively, the damping factor is adjusted, and the parameter vector is updated until the set number of iterations is reached.

[0032] 5. Multi-parameter coordinated correction: The theoretical dissolved oxygen saturation concentration is calculated using the improved Weiss equation. Combined with the standard deviation of the salinity and pressure sensors, weight coefficients are dynamically assigned to achieve multi-parameter coordinated correction.

[0033] 6. Closed-loop self-maintenance calibration: PID adjustment of light source intensity: Real-time acquisition of current fluctuation signals driving the light source, calculation of the control quantity through incremental PID algorithm, and output control voltage through PWM modulation to suppress light source current fluctuations within the range of ±0.3mA.

[0034] LSTM network drift prediction: Build an LSTM network structure to predict sensor drift. When the predicted value is greater than the set value, dynamic compensation is performed and the PID controller parameters are automatically updated.

[0035] The simulation results are as follows: In a wide temperature range (0-40°C), the calibration accuracy of the optical dissolved oxygen sensor is significantly improved through a nonlinear temperature compensation model. In particular, the cumulative error in the high temperature range (40°C) is reduced to within ±1.5%, an improvement of more than 30% compared with traditional methods.

[0036] Under the conditions of salinity fluctuation of ±10‰ and pressure change of ±50kPa, multi-parameter coordinated correction was achieved through a dynamic weight allocation strategy, with an overall measurement error of ≤2%, which is significantly better than the step-by-step correction strategy.

[0037] Fluctuations in light source intensity were effectively suppressed, reducing light intensity drift to within ±1.5% and baseline drift to 1 / 8 of its original value, improving the sensor's long-term stability. Furthermore, the LSTM network achieved over 90% accuracy in predicting sensor drift. After dynamic compensation, the sensor's long-term drift was kept within ±0.5 mg / L, demonstrating the effectiveness of this invention.

[0038] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit the same. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that they can still modify or replace the technical solutions of the present invention with equivalents, and these modifications or equivalent replacements cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A calibration method for an optical dissolved oxygen sensor based on nonlinear temperature compensation, characterized in that: The following steps are involved: S1. Build a calibration platform: fill the multi-temperature zone reaction chamber with water, and immerse the optical dissolved oxygen sensor, temperature sensor, salinity sensor and pressure sensor to be calibrated in the water; S2. Data acquisition: Generate dissolved oxygen saturated water samples through nitrogen and oxygen mixed gas under multiple temperature gradients, and simultaneously collect fluorescence lifetime and influencing factors; S3, model construction: Based on the joint fitting of the Arrhenius equation and the fourth-order polynomial, a nonlinear compensation model of fluorescence lifetime and temperature was established, and the model parameters were optimized by the Levenberg-Marquardt algorithm; S4, using the improved Weiss equation to calculate the theoretical dissolved oxygen saturation concentration, combining the measurement standard deviation of the salinity sensor and the pressure sensor to dynamically allocate the weight coefficient, and realize multi-parameter coordinated correction; S5. The intensity fluctuation of the driving light source of the optical dissolved oxygen sensor to be calibrated is adjusted by the PID controller, and the drift of the optical dissolved oxygen sensor to be calibrated is predicted by the LSTM network to realize closed-loop self-maintenance calibration.

2. The optical dissolved oxygen sensor calibration method based on nonlinear temperature compensation according to claim 1, characterized in that: The multiple temperature gradients described in step S2 include 0°C, 10°C, 20°C, 30°C, and 40°C; the influencing factors include dissolved oxygen concentration ,salinity and pressure .

3. The optical dissolved oxygen sensor calibration method based on nonlinear temperature compensation according to claim 2, characterized in that: Step S3 specifically includes the following steps: S31. Use the Arrhenius equation to describe the effect of temperature on the activation energy of fluorescence quenching: （1）； In the formula, Indicates water temperature The time constant of the fluorescence quenching process under represents the activation energy of fluorescence quenching process, in eV; It represents the theoretical fluorescence lifetime at absolute zero, in μs; is the Boltzmann constant, and ; S32. Compensation of nonlinear optical drift caused by temperature using a fourth-order polynomial: （2）； In the formula, represents the predicted fluorescence lifetime; , , , and They are baseline drift compensation constant term, linear drift compensation constant term, thermal expansion effect compensation constant term, nonlinear optical response constant term and high-order thermodynamic effect constant term; S33. Combining formula (1) and formula (2) to obtain the nonlinear compensation model of fluorescence lifetime and temperature: （3）； In the formula, Indicates water temperature The final fluorescence lifetime; S34, optimize model parameters by Levenberg-Marquardt algorithm; S341, initialization parameters: , , initial damping factor , initial step size scaling factor , and estimated by the least squares method ; S342, Iterative Optimization: S3421, calculate the residual sum of squares : （4）； In the formula, Indicates the amount of data collected for each temperature gradient; Indicates water temperature The fluorescence lifetime measured under S3422, adjust the damping factor , update the parameter vector ; S3423, iterating steps S3421-S3422 until the set number of iterations is reached, and outputting the optimized nonlinear compensation model of fluorescence lifetime and temperature; S343. Use goodness of fit, residual control and temperature sensitivity to verify the optimized nonlinear compensation model of fluorescence lifetime and temperature, and output the optimized nonlinear compensation model of fluorescence lifetime and temperature after verification.

4. The optical dissolved oxygen sensor calibration method based on nonlinear temperature compensation according to claim 3 is characterized in that: Step S4 The specific steps include: S41. Construct the Weiss standard equation: （5）； In the formula, Indicates water temperature Saturated vapor pressure under , and are empirical coefficients, and ; Indicates standard atmospheric pressure; S42. Construct comprehensive saturation concentration formula: （6）； In the formula, represents the comprehensive saturation concentration; represents the theoretical saturation concentration considering only pressure; represents the theoretical saturation concentration considering only salinity; in, （7）； （8）； In the formula, Indicates salinity in PSU; , and are salinity correction factors‌, and ; Indicates the real-time pressure in kPa; represents the pressure nonlinear correction coefficient‌, and ; S43, Dynamically allocate weight coefficient: （9）； In the formula, Indicates the corrected dissolved oxygen value; Represents the weight coefficient of the salinity sensor; Indicates the original dissolved oxygen measurement value in mg / L; Represents the weight coefficient of the pressure sensor; in, （10）； In the formula, Represents the standard deviation of salinity sensor measurement; Indicates the pressure sensor measurement standard deviation.

5. The optical dissolved oxygen sensor calibration method based on nonlinear temperature compensation according to claim 4, characterized in that: Step S5 specifically includes the following steps: S51, data collection: synchronously collect the current value of the driving light source every 1 second , Raw dissolved oxygen measurements , water temperature and salinity And pre-processing; S52, PID adjusts the light source intensity; S521, real-time acquisition of current fluctuation signal of driving light source , and set the value Generate for benchmark Error signal at time ; S522, using incremental PID algorithm to calculate the control quantity of the intensity of the driving light source : （11）； In the formula, , and Represents the proportional gain coefficient, integral gain coefficient and differential gain coefficient; represents the initial moment; Indicates the sampling period; express The error signal at the moment; S523, suppressing the current fluctuation of the driving light source within a set threshold range by outputting a control voltage through PWM modulation; S53, using the LSTM network to predict the drift of the optical dissolved oxygen sensor to be calibrated; S531, construct an LSTM network structure, which includes an input layer, two hidden layers and an output layer arranged in sequence, wherein the input variable of the input layer is the water temperature ,salinity , Accumulated working time of driving light source ; Output layer output Prediction value of drift at the moment , represents the Sigmoid function, represents the weight matrix, represents the state of the hidden layer, represents the bias term; S532, set the drift prediction value Dynamic compensation is performed when it is greater than the set value: （12）； In the formula, express Dissolved oxygen value after calibration at all times; express Original dissolved oxygen value; It represents the drift change rate suppression factor; S533, parameter self-update: （13）； In the formula, Represents the updated PID controller parameters; Indicates the PID parameter learning rate; Represents control performance indicators.

6. The optical dissolved oxygen sensor calibration method based on nonlinear temperature compensation according to claim 5, characterized in that: The pre-processing in step S51 includes removing abnormal current values ​​with fluctuation ranges greater than ±5% of the set value and processing the original dissolved oxygen measurement value. Perform sliding window mean filtering.

7. The optical dissolved oxygen sensor calibration method based on nonlinear temperature compensation according to claim 5, characterized in that: The setting threshold range described in step S523 is ±0.3 mA.

Images