Turbidity detection method and device based on intelligent sensing technology

Abstract

The invention discloses a turbidity detection method and device based on an intelligent sensing technology, and the method comprises the following steps: step 1, original signal collection and noise preprocessing, step 2, automatic range switching judgment and hardware path switching, step 3, intelligent multi-dimensional error compensation, and step 4, data linearization conversion. The method comprises the following steps: carrying out automatic range switching judgment and hardware channel switching, carrying out hysteresis threshold judgment, then controlling an analog switch / light path to switch a corresponding detection module through a microcontroller, and synchronously adjusting AD sampling gain, so that measurement switching in various environments can be realized, and the turbidity detection accuracy is improved; temperature compensation, light source drift compensation and bubble interference compensation can be carried out on the acquired signals through intelligent multi-dimensional error compensation, and the accuracy of signal detection is improved.

Description

Technical Field

[0001] This invention relates to the field of turbidity measurement technology, and specifically to a turbidity detection method and device based on intelligent sensing technology. Background Technology

[0002] Turbidity refers to the degree to which a solution impedes the passage of light, including the scattering of light by suspended matter and the absorption of light by solute molecules. The turbidity of water is related not only to the content of suspended matter but also to their size, shape, and refractive index. It is generally applicable to the determination of turbidity in natural water, drinking water, and some industrial water. Water samples for turbidity determination should be tested as soon as possible, or must be refrigerated at 4°C and tested within 24 hours. Before testing, the water sample should be vigorously shaken and allowed to return to room temperature.

[0003] According to patent number CN111624176B-Turbidity Measurement Method and Turbidity Meter, it describes "a turbidity measurement method and turbidity meter that can calculate the turbidity of the test liquid in a manner corresponding to the turbidity of the test liquid measured using other light sources in existing turbidity meters". It can be seen that the turbidity meter uses the scattered light method for measurement, but it does not effectively filter out interference in the light signal, which leads to a deviation in the accuracy of turbidity measurement. Moreover, existing turbidity meters or turbidity detection devices have limited application scenarios and cannot meet the needs of multiple scenarios, which further limits the practicality of the turbidity detection device.

[0004] In summary, a turbidity detection method and device based on intelligent sensing technology were designed. Summary of the Invention

[0005] To overcome the above-mentioned shortcomings, the present invention provides a turbidity detection method and device based on intelligent sensing technology.

[0006] The present invention achieves the above objectives through the following technical solutions: A turbidity detection method based on intelligent sensing technology includes the following steps: Step 1: Raw signal acquisition and noise preprocessing. Three optical channels are set up for inputting scattered light intensity signal, transmitted light intensity signal, and light source calibration channel signal, respectively. A temperature sensor is also set up to acquire temperature signals. After noise preprocessing of the four signals, the following signals are output: scattered light intensity signal, transmitted light intensity signal, calibration light intensity signal, and real-time temperature signal. The scattered light intensity signal is... The transmitted light intensity signal is The calibration light intensity signal is and real-time temperature signal ; Step 2: Automatic range switching judgment and hardware path switching. After the system is powered on, the scattering method is used by default for low range. The hysteresis threshold is judged based on the scattered light intensity signal, transmitted light intensity signal, calibration light intensity signal and real-time temperature signal output in Step 1. Then, the microcontroller controls the analog switch / optical path to switch the corresponding detection module and adjusts the AD sampling gain synchronously. Step 3: Intelligent multi-dimensional error compensation. The scattered light intensity signal, transmitted light intensity signal, calibration light intensity signal and real-time temperature signal output in Step 2 are subjected to temperature compensation, light source drift compensation and bubble interference compensation, and the compensated scattered light intensity signal, transmitted light intensity signal, calibration light intensity signal and real-time temperature signal are output. Step 4: Data linearization conversion. Perform linearization conversion on the compensated scattered light intensity signal, transmitted light intensity signal, calibration light intensity signal and real-time temperature signal, and output the linearized standard turbidity value.

[0007] Preferably, the specific steps of noise preprocessing in step one are as follows: S11. First, calculate the signal variance of 10 consecutive sampling points. The larger the variance, the stronger the noise in the current scene; S12. Set the trigger threshold: At that time, among them Based on the current average light intensity, this is determined to be a strong industrial electromagnetic interference scene, so the filter window is increased to 10. When the scene is identified as a weak noise scene in the field, the window is reduced to 3. The intermediate value is adjusted by linear interpolation to adjust the window size. The noise intensity is quantified by variance to achieve dynamic adaptation of the filter window. This ensures both the strong interference suppression capability of the industrial scene and the response speed of the turbidity change of the dynamic water sample in the field, thus solving the problem of "paying attention to one thing but losing another" with a fixed window. S13. Corresponding filtering processing is performed on the three optical channel signals. Accurate noise reduction is achieved through noise feature matching to avoid over-filtering of some channels resulting in the loss of effective details or insufficient filtering leaving residual interference. The scattered light channel uses low-pass FIR filtering. The main noise in the scattered light channel is low-frequency interference from particle distribution fluctuations. Low-pass FIR filtering (cutoff frequency 0.5Hz) is used to filter out high-frequency noise while retaining the turbidity change trend. The transmitted light channel uses band-pass filtering. The main noise in the transmitted light channel is visible light interference caused by water sample color. Band-pass filtering (only retaining the frequency signal corresponding to the 860nm infrared band) is used to specifically eliminate color interference. The calibration channel uses adaptive recursive filtering. The main noise in the calibration channel is weak random fluctuations of the light source. Adaptive recursive filtering (the recursion coefficient is dynamically adjusted according to the stability of the calibration light intensity) is used to improve the smoothness of the reference signal.

[0008] Preferably, in step two, The specific steps for determining the hysteresis threshold are as follows: First, calculate the maximum fluctuation value of turbidity within 30 consecutive seconds. If the scenario is identified as a high-fluctuation scenario (such as industrial wastewater or rivers after heavy rain), the hysteresis bandwidth is set to 30 NTU (the trigger switching threshold is 170 NTU / 230 NTU). In stable scenarios (such as waterworks), the hysteresis bandwidth is set to 10 NTU (trigger threshold 190 NTU / 200 NTU). In intermediate fluctuating scenarios, the bandwidth is adjusted by linear interpolation to dynamically adapt to the fluctuation characteristics of the water sample, avoid frequent switching and hardware wear in high-fluctuation scenarios, and ensure the timeliness of range switching in stable scenarios, thus solving the problem of the universality of fixed thresholds. The specific steps for hardware path switching are as follows: when the hysteresis threshold is judged to indicate that a range switch is about to occur, a pre-switching command is sent one sampling cycle in advance: first, the AD sampling gain is adjusted to the appropriate level, and at the same time, the light intensity value after the switch is predicted by the linear fitting model of the first 5 sampling points. The sampling value at the moment of switching is replaced by the predicted value, and the subsequent sampling values ​​gradually transition to the actual measurement value. Through hardware pre-adjustment and signal transition compensation, the signal jump caused by optical path / gain switching is completely eliminated, and a smooth "imperceptible" transition of range switching is achieved.

[0009] Preferably, in step three, the specific steps of temperature compensation are as follows: a visible light sensor is built into the temperature sensor, and the visible light sensor collects the ratio of the light intensity of the water sample at 550nm (visible light) to 860nm (infrared). , The water sample was identified as a high-transmittance sediment sample, and a linear function compensation model was used. The water sample was identified as having high scattering due to algae, and a quadratic function compensation model was used. The intermediate value uses linear interpolation to fuse the two models, distinguishes the suspended matter type by spectral characteristics, and specifically matches the temperature compensation law to solve the problem of large compensation error of the general model under complex water quality.

[0010] Preferably, the specific steps for light source drift compensation in step three are as follows: A quartz reference plate with known reflectivity is placed in the calibration channel, and the calibration light intensity is synchronously collected during each measurement. Light intensity compared to the reference film ; Dynamic calculation of optical path loss correction factor ,in, The original compensation coefficients are updated to the initial reference light intensity. Then, the measured light intensity is compensated, no longer relying on the initial calibration of the optical path consistency. The optical path loss difference is calibrated in real time through the reference plate, which solves the compensation failure problem caused by lens contamination and fiber aging during long-term use.

[0011] Preferably, the specific steps for bubble interference compensation in step three are as follows: Calculate the light intensity abrupt change rate between the current sampling point and the previous sampling point. ,in, This represents the change in light intensity between the current sampling point and the previous sampling point. This represents the time difference between the current sampling point and the previous sampling point. like Further determine the duration of the mutation (the number of sampling points that continuously exceed the threshold). Each sampling period It was determined to be an air bubble.

[0012] Preferably, the bubble interference compensation includes frequency feature verification. Through fast Fourier transform analysis, the energy of the bubble signal is concentrated above 10Hz, while the energy of the suspended matter change signal is concentrated below 1Hz, further filtering out misjudgments. Bubbles are identified from multiple dimensions of "amplitude + time + frequency," avoiding misjudgments based on a single change rate threshold and solving the problems of bubble signals being masked under high turbidity and misjudging suspended matter changes under low turbidity.

[0013] Preferably, the specific steps of step four are as follows: Low-range linearization of scattering method: using a pre-calibrated linear model. for Fitting parameters, goodness of fit , Expressed as turbidity value; High-range linearization of transmission method: Based on the Lambert-Beer law, it is converted into a linear relationship. , The initial incident light intensity, These are the fitting parameters for the high range.

[0014] A turbidity detection device based on intelligent sensing technology as described above, wherein the device body is provided with a controller, and the controller is loaded with the method described above.

[0015] The beneficial effects of this invention are: in this turbidity detection method and device based on intelligent sensing technology, In step two, automatic range switching and hardware path switching are performed, hysteresis threshold is determined, and then the corresponding detection module is switched by analog switch / optical path through microcontroller. The AD sampling gain is adjusted synchronously, which enables measurement switching in various environments and improves the accuracy of turbidity detection. Intelligent multi-dimensional error compensation can compensate for temperature, light source drift, and bubble interference in the acquired signal, thereby improving the accuracy of signal detection. Attached Figure Description

[0016] The present invention will be described by way of example and with reference to the accompanying drawings, wherein: Figure 1 This is a flowchart illustrating the steps of the present invention. Detailed Implementation

[0017] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention, and therefore only show the components relevant to the invention.

[0018] like Figure 1 As shown, a turbidity detection method based on intelligent sensing technology includes the following steps: Step 1: Raw signal acquisition and noise preprocessing. Three optical channels are set up for inputting scattered light intensity signal, transmitted light intensity signal, and light source calibration channel signal, respectively. A temperature sensor is also set up to acquire temperature signals. After noise preprocessing of the four signals, the following signals are output: scattered light intensity signal, transmitted light intensity signal, calibration light intensity signal, and real-time temperature signal. The scattered light intensity signal is... The transmitted light intensity signal is The calibration light intensity signal is and real-time temperature signal ; Step 2: Automatic range switching judgment and hardware path switching. After the system is powered on, the scattering method is used by default for low range. The hysteresis threshold is judged based on the scattered light intensity signal, transmitted light intensity signal, calibration light intensity signal and real-time temperature signal output in Step 1. Then, the microcontroller controls the analog switch / optical path to switch the corresponding detection module and adjusts the AD sampling gain synchronously. Step 3: Intelligent multi-dimensional error compensation. The scattered light intensity signal, transmitted light intensity signal, calibration light intensity signal and real-time temperature signal output in Step 2 are subjected to temperature compensation, light source drift compensation and bubble interference compensation, and the compensated scattered light intensity signal, transmitted light intensity signal, calibration light intensity signal and real-time temperature signal are output. Step 4: Data linearization conversion. Perform linearization conversion on the compensated scattered light intensity signal, transmitted light intensity signal, calibration light intensity signal and real-time temperature signal, and output the linearized standard turbidity value.

[0019] Specifically, the noise preprocessing steps in step one are as follows: S11. First, calculate the signal variance of 10 consecutive sampling points. The larger the variance, the stronger the noise in the current scene; S12. Set the trigger threshold: At that time, among them Based on the current average light intensity, this is determined to be a strong industrial electromagnetic interference scene, so the filter window is increased to 10. When the scene is identified as a weak noise scene in the field, the window is reduced to 3. The intermediate value is adjusted by linear interpolation to adjust the window size. The noise intensity is quantified by variance to achieve dynamic adaptation of the filter window. This ensures both the strong interference suppression capability of the industrial scene and the response speed of the turbidity change of the dynamic water sample in the field, thus solving the problem of "paying attention to one thing but losing another" with a fixed window. S13. Corresponding filtering processing is performed on the three optical channel signals. Accurate noise reduction is achieved through noise feature matching to avoid over-filtering of some channels resulting in the loss of effective details or insufficient filtering leaving residual interference. The scattered light channel uses low-pass FIR filtering. The main noise in the scattered light channel is low-frequency interference from particle distribution fluctuations. Low-pass FIR filtering (cutoff frequency 0.5Hz) is used to filter out high-frequency noise while retaining the turbidity change trend. The transmitted light channel uses band-pass filtering. The main noise in the transmitted light channel is visible light interference caused by water sample color. Band-pass filtering (only retaining the frequency signal corresponding to the 860nm infrared band) is used to specifically eliminate color interference. The calibration channel uses adaptive recursive filtering. The main noise in the calibration channel is weak random fluctuations of the light source. Adaptive recursive filtering (the recursion coefficient is dynamically adjusted according to the stability of the calibration light intensity) is used to improve the smoothness of the reference signal.

[0020] Specifically, in step two, The specific steps for determining the hysteresis threshold are as follows: First, calculate the maximum fluctuation value of turbidity within 30 consecutive seconds. If the scenario is identified as a high-fluctuation scenario (such as industrial wastewater or rivers after heavy rain), the hysteresis bandwidth is set to 30 NTU (the trigger switching threshold is 170 NTU / 230 NTU). In stable scenarios (such as waterworks), the hysteresis bandwidth is set to 10 NTU (trigger threshold 190 NTU / 200 NTU). In intermediate fluctuating scenarios, the bandwidth is adjusted by linear interpolation to dynamically adapt to the fluctuation characteristics of the water sample, avoid frequent switching and hardware wear in high-fluctuation scenarios, and ensure the timeliness of range switching in stable scenarios, thus solving the problem of the universality of fixed thresholds. The specific steps for hardware path switching are as follows: When the hysteresis threshold is judged to indicate that a range switch is about to occur, a pre-switching command is sent one sampling cycle in advance: First, the AD sampling gain is adjusted to the appropriate level. At the same time, the light intensity value after the switch is predicted using the linear fitting model of the first 5 sampling points. The sampling value at the moment of switching is replaced by the predicted value, and the subsequent sampling values ​​gradually transition to the actual measured value. For example, the first switching sampling point uses 100% of the predicted value, the second uses 70% of the predicted value + 30% of the actual value, and the third uses the actual value completely. Through hardware pre-adjustment and signal transition compensation, the signal jump caused by optical path / gain switching is completely eliminated, and a smooth "imperceptible" transition of range switching is achieved.

[0021] Specifically, in step three, the temperature compensation process involves the following steps: a visible light sensor is integrated into the temperature sensor, and the visible light sensor collects the ratio of the light intensity of the water sample at 550nm (visible light) to 860nm (infrared). , The water sample was identified as a high-transmittance sediment sample, and a linear function compensation model was used. The water sample was identified as having high scattering due to algae, and a quadratic function compensation model was used. The intermediate value uses linear interpolation to fuse the two models, distinguishes the suspended matter type by spectral characteristics, and specifically matches the temperature compensation law to solve the problem of large compensation error of the general model under complex water quality.

[0022] Specifically, the steps for light source drift compensation in step three are as follows: A quartz reference plate with known reflectivity is placed in the calibration channel, and the calibration light intensity is synchronously collected during each measurement. Light intensity compared to the reference film ; Dynamic calculation of optical path loss correction factor ,in, The original compensation coefficients are updated to the initial reference light intensity. Then, the measured light intensity is compensated, no longer relying on the initial calibration of the optical path consistency. The optical path loss difference is calibrated in real time through the reference plate, which solves the compensation failure problem caused by lens contamination and fiber aging during long-term use.

[0023] Specifically, the steps for bubble interference compensation in step three are as follows: Calculate the light intensity abrupt change rate between the current sampling point and the previous sampling point. ,in, This represents the change in light intensity between the current sampling point and the previous sampling point. This represents the time difference between the current sampling point and the previous sampling point. like Further determine the duration of the mutation (the number of sampling points that continuously exceed the threshold). Each sampling period It was determined to be an air bubble.

[0024] Specifically, the bubble interference compensation includes frequency feature verification. Through fast Fourier transform analysis, the energy of the bubble signal is concentrated above 10Hz, while the energy of the suspended matter change signal is concentrated below 1Hz, further filtering out false positives. Bubbles are identified from multiple dimensions of "amplitude + time + frequency," avoiding false positives based on a single mutation rate threshold and solving the problems of bubble signals being masked under high turbidity and false positives of suspended matter changes under low turbidity.

[0025] Specifically, the steps in step four are as follows: Low-range linearization of scattering method: using a pre-calibrated linear model. for Fitting parameters, goodness of fit , Expressed as turbidity value; High-range linearization of transmission method: Based on the Lambert-Beer law, it is converted into a linear relationship. , The initial incident light intensity, These are the fitting parameters for the high range.

[0026] A turbidity detection device based on intelligent sensing technology as described above, wherein the device body is provided with a controller, and the controller is loaded with the method described above.

[0027] Specific Implementation Case 1: Online Monitoring of Turbidity in Water Leaving a Water Treatment Plant The application scenario is characterized by long-term stable turbidity of the water sample. The environment has weak electromagnetic interference and a narrow temperature fluctuation range (15-25℃). The measurement accuracy required for long-term online operation is ±0.01NTU. It is necessary to avoid the cumulative error caused by light source drift and lens contamination.

[0028] Detailed implementation process Step 1: Raw signal acquisition and noise preprocessing The three channels of scattered light, transmitted light, and calibration light are simultaneously activated, and a temperature sensor is used with a sampling frequency of 1Hz.

[0029] Calculate the signal variance over 10 consecutive sampling points, considering the scene stability variance. Current average light intensity The scene was determined to be a low-noise outdoor environment, and the filter window was reduced to 3.

[0030] The scattered light channel uses a low-pass FIR filter (cutoff frequency 0.5Hz), the transmitted light channel uses a band-pass filter (only retaining the 860nm infrared band signal), and the calibration channel uses an adaptive recursive filter, outputting four types of pre-processed signals.

[0031] Step 2: Automatic range switching judgment and hardware path switching By default, the system uses the low-range scattering method upon power-up to calculate the maximum turbidity fluctuation value within 30 consecutive seconds. The scenario was determined to be stable, and the hysteresis bandwidth was set to 10 NTU (switching threshold 190 NTU / 200 NTU). Because the turbidity of the water sample was much lower than the threshold, the scattering method was kept at a low range, and the AD sampling gain was fixed at 16 times.

[0032] Step 3: Intelligent Multi-Dimensional Error Compensation Temperature compensation: Collect the light intensity ratio of 550nm and 860nm. The water sample was identified as high-transmittance clear water and a linear function compensation model was used. (Base temperature is 20℃).

[0033] Light source drift compensation: Simultaneous acquisition and calibration of light intensity for each measurement. Light intensity reflected from quartz reference plate Dynamically calculate optical path loss correction coefficient After updating the compensation coefficient, the intensity of the scattered light is corrected.

[0034] Bubble interference compensation: After the water sample is allowed to settle, it enters the monitoring tank, and the light intensity abrupt change rate is considered. The system was determined to have no bubble interference, and no additional compensation was required.

[0035] Step 4: Data linearization transformation The linear model is pre-calibrated using a low-range scattering method: Output standard turbidity value. Implementation effect

[0036] After 30 days of continuous online operation, the deviation between the turbidity measurement value and the laboratory standard method is less than 0.01 NTU, the light source drift error is reduced by 95%, and there is no hardware loss caused by frequent range switching, which fully meets the high-precision monitoring requirements of water leaving the water treatment plant. Specific Implementation Case 2: Turbidity Monitoring of Industrial Chemical Wastewater Discharge Outlets The application scenario is characterized by drastic fluctuations in water turbidity. The workshop experiences strong electromagnetic interference (from motors and frequency converters), and the wastewater temperature fluctuates greatly. The flow process easily generates bubbles, requiring the equipment to have high turbidity range adaptability, strong anti-interference ability, and measurement error. .

[0037] Detailed implementation process Step 1: Raw signal acquisition and noise preprocessing Enable the three optical channels and the temperature sensor, and set the sampling frequency to 2Hz.

[0038] Calculate the signal variance of 10 consecutive sampling points. The scenario was determined to be a strong industrial electromagnetic interference scene, and the filter window was increased to 10.

[0039] Signals are processed according to the corresponding filtering rules of the channels: scattered light low-pass FIR filtering, transmitted light band-pass filtering, calibration channel adaptive recursive filtering, and four types of signals are output after filtering electromagnetic noise.

[0040] Step 2: Automatic range switching judgment and hardware path switching The system defaults to the low range of the scattering method, calculating the maximum turbidity fluctuation value within 30 consecutive seconds. The scenario was identified as high-fluctuation, and the hysteresis bandwidth was set to 30 NTU (the switching threshold). ).

[0041] When the turbidity corresponding to the scattered light intensity reaches 230 NTU, a pre-switching command is sent one sampling cycle in advance: first, the AD sampling gain is adjusted to 1x, and the transmitted light intensity value is predicted using the linear fitting model of the first 5 sampling points. The predicted value is then used to replace the actual value at the moment of switching. The second sampling point uses... During the transition, the third sampling point is completely switched to the true value, achieving seamless range switching.

[0042] Step 3: Intelligent Multi-Dimensional Error Compensation Temperature compensation: Collect the light intensity ratio of 550nm and 860nm. For water samples identified as containing mixed suspended solids, a linear interpolation model combining primary and secondary compensation methods was used. .

[0043] Light source drift compensation: Synchronously acquire calibration light intensity and reference film reflected light intensity, and calculate correction coefficient. After updating the compensation coefficient, the transmitted light intensity is corrected to offset the loss caused by lens contamination and fiber optic aging.

[0044] Bubble interference compensation: Calculation of light intensity abrupt change rate and duration After sampling for one period, FFT analysis was performed to verify that the signal energy was concentrated at 12Hz. The signal was identified as bubble interference, and the data segment was removed and the data was completed by linear interpolation of the sampling points before and after sampling.

[0045] Step 4: Data linearization transformation High-range model using transmission method: Output standard turbidity value. Implementation effect

[0046] In environments with strong electromagnetic interference, the measurement error is reduced from ±5 NTU to ±1 NTU by traditional methods, there is no signal jump phenomenon when switching ranges, and the bubble interference rejection rate reaches 98%, which fully meets the compliance monitoring requirements for industrial wastewater discharge.

[0047] Specific Implementation Case 3: Mobile Turbidity Monitoring of Watershed Environment in the Field (Portable Equipment) The application scenario is characterized by a large range of water sample turbidity. In the wild, there is weak electromagnetic interference, drastic temperature fluctuations (5-35℃), water flow disturbances that easily generate bubbles, and mixed types of suspended matter (alternating between silt and algae), requiring the equipment to be portable and adaptable to different scenarios.

[0048] Detailed implementation process Step 1: Raw signal acquisition and noise preprocessing The portable device simultaneously activates three types of optical channels and a temperature sensor, with the sampling frequency set to 1.5Hz.

[0049] Calculate the signal variance of 10 consecutive sampling points. The scene was determined to be a medium noise scene. The filter window was adjusted to 6 by linear interpolation to balance noise suppression and response speed to turbidity changes.

[0050] After filtering is completed according to the channel rules, the preprocessed four types of signals are output.

[0051] Step 2: Automatic range switching judgment and hardware path switching Upper reaches of the river Turbidity fluctuations for 30 consecutive seconds By using linear interpolation, the hysteresis bandwidth is set to 15 NTU to maintain a low range for the scattering method. Downstream of the river Turbidity fluctuations for 30 consecutive seconds Set the hysteresis bandwidth to 25 NTU. When the turbidity reaches 225 NTU, smoothly switch to the high range of the transmission method and adjust the AD sampling gain to 2 times.

[0052] Step 3: Intelligent Multi-Dimensional Error Compensation Temperature compensation: Collect the light intensity ratio of 550nm and 860nm. The water sample was identified as having high scattering due to algae, and a quadratic function compensation model was used: .

[0053] Light source drift compensation: Due to frequent power-on and power-off cycles, the reflected light intensity of the quartz reference plate is calibrated and a correction factor is calculated before each measurement. This compensates for the light intensity signal to counteract temporary light source drift.

[0054] Bubble disturbance compensation: Water flow disturbance causes abrupt changes in light intensity. After two consecutive sampling cycles, the signal energy was concentrated at 11Hz according to FFT analysis, which was determined to be bubble interference. This data point was removed and the previous sampled value was used to complete the signal.

[0055] Step 4: Data linearization transformation The upstream low-turbidity section uses a scattering method model: ; The downstream high-turbidity section uses a transmission method model: Output standard turbidity value. Implementation effect

[0056] Measurement deviations in different river sections are less than ±0.5 NTU, temperature compensation errors are reduced by 80%, the influence of bubble interference on measurements is completely eliminated, and the equipment can adapt to complex water environments in the field, meeting the flexible needs of mobile watershed monitoring.

[0057] In summary, this turbidity detection method based on intelligent sensing technology can suppress noise in the original signal through raw signal acquisition and noise preprocessing, thereby improving the accuracy of signal acquisition. Through automatic range switching judgment and hardware path switching, it can reliably switch ranges and hardware paths in different situations, meeting the requirements of reliable detection in various scenarios. Intelligent multi-dimensional error compensation can compensate for multi-dimensional errors in the data, improving the accuracy of turbidity detection.

[0058] Based on the above description, those skilled in the art can make various changes and modifications without departing from the technical concept of this invention. The technical scope of this invention is not limited to the contents of the specification, but must be determined according to the scope of the claims.

Claims

1. A turbidity detection method based on intelligent sensing technology, characterized in that: Includes the following steps: Step 1: Raw signal acquisition and noise preprocessing. Three optical channels are set up for inputting scattered light intensity signal, transmitted light intensity signal, and light source calibration channel signal, respectively. A temperature sensor is also set up to acquire temperature signals. After noise preprocessing of the four signals, the following signals are output: scattered light intensity signal, transmitted light intensity signal, calibration light intensity signal, and real-time temperature signal. The scattered light intensity signal is... The transmitted light intensity signal is The calibration light intensity signal is and real-time temperature signal ; Step 2: Automatic range switching judgment and hardware path switching. After the system is powered on, the scattering method is used by default for low range. The hysteresis threshold is judged based on the scattered light intensity signal, transmitted light intensity signal, calibration light intensity signal and real-time temperature signal output in Step 1. Then, the microcontroller controls the analog switch / optical path to switch the corresponding detection module and adjusts the AD sampling gain synchronously. Step 3: Intelligent multi-dimensional error compensation. The scattered light intensity signal, transmitted light intensity signal, calibration light intensity signal and real-time temperature signal output in Step 2 are subjected to temperature compensation, light source drift compensation and bubble interference compensation, and the compensated scattered light intensity signal, transmitted light intensity signal, calibration light intensity signal and real-time temperature signal are output. Step 4: Data linearization conversion. Perform linearization conversion on the compensated scattered light intensity signal, transmitted light intensity signal, calibration light intensity signal and real-time temperature signal, and output the linearized standard turbidity value.

2. The turbidity detection method based on intelligent sensing technology according to claim 1, characterized in that: The specific steps of noise preprocessing in step one are as follows: S11. First, calculate the signal variance of 10 consecutive sampling points. ; S12. Set the trigger threshold: At that time, among them Based on the current average light intensity, this is determined to be a strong industrial electromagnetic interference scene, so the filter window is increased to 10. When the scene is identified as a low-noise outdoor environment, the window size is reduced to 3; for intermediate values, linear interpolation is used to adjust the window size. S13. Perform corresponding filtering processing on the three optical channel signals: low-pass FIR filtering is used for the scattered light channel, band-pass filtering is used for the transmitted light channel, and adaptive recursive filtering is used for the calibration channel.

3. The turbidity detection method based on intelligent sensing technology according to claim 1, characterized in that: In step two, The specific steps for determining the hysteresis threshold are as follows: First, calculate the maximum fluctuation value of turbidity within 30 consecutive seconds. Determined to be a high-fluctuation scenario, the hysteresis bandwidth is set to... ; Determined to be a stable scenario, hysteresis bandwidth is set to... In scenarios with intermediate fluctuations, linear interpolation is used to adjust the bandwidth. The specific steps for hardware path switching are as follows: when the hysteresis threshold is judged to predict that a range switch is about to occur, a pre-switching command is sent one sampling cycle in advance: first, the AD sampling gain is adjusted to the appropriate level, and at the same time, the light intensity value after switching is predicted by the linear fitting model of the first 5 sampling points. The sampled value at the moment of switching is replaced by the predicted value, and the subsequent sampled values ​​gradually transition to the actual measured value.

4. The turbidity detection method based on intelligent sensing technology according to claim 1, characterized in that: In step three, the specific steps for temperature compensation are as follows: a visible light sensor is built into the temperature sensor, and the visible light sensor collects the ratio of the light intensity of the water sample at 550nm and 860nm. , The water sample was identified as a high-transmittance sediment sample, and a linear function compensation model was used. ; The water sample was identified as having high scattering due to algae, and a quadratic function compensation model was used. The intermediate value is obtained by fusing the two models using linear interpolation.

5. The turbidity detection method based on intelligent sensing technology according to claim 1, characterized in that: In step three, the specific steps for light source drift compensation are as follows: A quartz reference plate with known reflectivity is placed in the calibration channel, and the calibration light intensity is synchronously collected during each measurement. Light intensity compared to the reference film ; Dynamic calculation of optical path loss correction factor ,in, The original compensation coefficients are updated to the initial reference light intensity. Then, the measured light intensity is compensated.

6. The turbidity detection method based on intelligent sensing technology according to claim 1, characterized in that: In step three, the specific steps for bubble interference compensation are as follows: Calculate the light intensity abrupt change rate between the current sampling point and the previous sampling point. ,in, This represents the change in light intensity between the current sampling point and the previous sampling point. This represents the time difference between the current sampling point and the previous sampling point. like Further determine the duration of the mutation. Each sampling period was identified as a bubble.

7. The turbidity detection method based on intelligent sensing technology according to claim 6, characterized in that: The bubble interference compensation includes frequency feature verification. Through fast Fourier transform analysis, the energy of the bubble signal is concentrated above 10Hz, while the energy of the suspended matter change signal is concentrated below 1Hz, further filtering out misjudgments.

8. The turbidity detection method based on intelligent sensing technology according to claim 1, characterized in that: The specific steps of step four are as follows: Low-range linearization of scattering method: using a pre-calibrated linear model. ; High-range linearization of transmission method: Based on the Lambert-Beer law, it is converted into a linear relationship. .

9. A turbidity detection device based on intelligent sensing technology according to any one of claims 1-8, characterized in that: The device body includes a controller, which loads the method described in claims 1-8.

Images