Water chemical oxygen demand quantification model construction method based on multi-source spectrum fusion

Abstract

The invention relates to the technical field of road sound insulation, and discloses a water chemical oxygen demand quantification model construction method based on multi-source spectrum fusion, which comprises a multi-source spectrum sensing layer and a multi-source fusion model, and specifically comprises the following steps: step 1, constructing a multi-source spectrum synchronous determination and effective fusion step; 2, constructing a dynamic interference compensation module; step 3, constructing a feature level fusion architecture; and 4, constructing a weighted decision fusion model. According to the invention, through the collaborative design of three optical paths, a coaxial light splitting probe is initiated, three types of spectrums are synchronously obtained through single sampling, and temperature / pH coupling correction coefficients alpha and beta are realized through a multi-stage fusion strategy, feature-level fusion (spectral feature extraction) + decision-level fusion (dynamic weight distribution) and drift real-time compensation, so that the spectrum drift error is reduced from 12% to 1.5%.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of highway sound insulation, more particularly to a water chemical oxygen demand quantification model construction method based on multi-source spectrum fusion. BACKGROUND

[0002] The prior art oxygen demand quantification model construction method has the following defects: 1. Single spectrum limitation: the sensitivity of ultraviolet absorption spectroscopy to low concentration water samples is low; near-infrared spectroscopy is easily disturbed by external interference, and the signal-to-noise ratio is low; fluorescence spectroscopy has unstable factors such as quenching and self-absorption; 2. Poor model universality: a single model cannot adapt to different water types (rivers / lakes / sewage), and the migration application error is more than 50%; 3. Lack of dynamic interference compensation: temperature and pH fluctuations cause spectral drift; At the same time, single spectrum technology cannot simultaneously meet the high-precision (error <5%), wide-range (1-500mg / L), and strong anti-interference COD detection requirements. Therefore, we propose a water chemical oxygen demand quantification model construction method based on multi-source spectrum fusion. SUMMARY

[0003] The present application relates to the technical field of highway sound insulation, more particularly to a water chemical oxygen demand quantification model construction method based on multi-source spectrum fusion.

[0004] The water chemical oxygen demand quantification model construction method based on multi-source spectrum fusion includes a multi-source spectrum perception layer and a multi-source fusion model, Specifically, the following steps are included: Step 1, construct multi-source spectrum synchronous determination and effective fusion Step 2, construct a dynamic interference compensation module; Step 3, construct a feature-level fusion architecture; Step 4, construct a weighted decision fusion model.

[0005] Further: the step 1 constructs multi-source spectrum synchronous determination and effective fusion, which includes: Deuterium halogen dual spectrum + 2048 pixel CCD, used for detecting ultraviolet-visible absorption light, with a detection range of 200-700nm, UV LED array + photomultiplier tube; used for detecting three-dimensional fluorescence, with a detection range of Ex: 250-550nm, Em: 300-650nm; InGaAs detector; used for detecting near-infrared diffuse reflection light, with a detection range of 900-1700nm.

[0006] Furthermore: Step 2 includes a dynamic interference compensation module, which feeds back to the spectral correction algorithm in real time by setting a built-in temperature / pH sensor; The specific algorithm is as follows: λcorrected=λraw×[1+α(T−25)+β(pH−7)] (α=0.002 / ℃, β=0.0015 / pH unit).

[0007] Furthermore: Step 3 includes: spectral feature extraction plus decision-level fusion.

[0008] Furthermore: the decision-level fusion includes a weighted decision fusion model, which includes the construction of sub-models, as detailed below: UV-Vis model: Partial Least Squares Regression (PLSR); EEMs ​​model: Support Vector Regression (SVR); NIR model: Random Forest (RF).

[0009] Furthermore, the decision-level fusion also includes dynamic weight allocation: the allocation logic is as follows: wi=1 / RMSEi2∑(1 / RMSEj2) (i=UV,EEMs,NIR) (RMSEi represents the error of each sub-model on the validation set). Beneficial effects

[0010] 1. Three-optical-path collaborative design The first coaxial spectrophotometer enables simultaneous acquisition of three types of spectra in a single sampling. 2. Multi-level fusion strategy Feature-level fusion (spectral feature extraction) + decision-level fusion (dynamic weight allocation) 3. Real-time drift compensation The temperature / pH coupling correction coefficients α and β reduced the spectral drift error from 12% to 1.5%.

[0011] This invention achieves the following through a three-tiered innovation: "multi-source spectral synchronous acquisition - real-time drift compensation - dynamic fusion modeling": Full spectrum coverage: ultraviolet (aromatic organic matter) + fluorescence (soluble microbial products) + near-infrared (suspended organic matter); Cross-scenario adaptation: Transfer learning enables the model to achieve an error of less than 10% in rivers / sewage / industrial water bodies. Attached Figure Description

[0012] Figure 1 This is a component diagram of the core module of the present invention; Detailed Implementation

[0013] In the description of this invention, it should be understood that the terms "center," "longitudinal," "lateral," "length," "width," "thickness," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," "outer," "clockwise," and "counterclockwise," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0014] In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0015] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "installed," "equipped with," "sleeved / connected," "connected," etc., should be interpreted broadly. For example, "connection" can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium; it can be a connection within two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances. Example

[0016] The core of this invention lies in two points: the multi-source spectral sensing layer and the multi-source fusion modeling layer. (I) Multi-source spectral sensing layer This includes simultaneous measurement and effective fusion of multi-source spectra. The detection ranges and core components corresponding to different spectral types are as follows: UV-Vis absorption, 200-700nm, deuterium-halogen dual spectrum + 2048 pixel CCD; Three-dimensional fluorescence (EEMs), Ex: 250-550nm, Em: 300-650nm, UV LED array + photomultiplier tube; Near-infrared diffuse reflection, 900-1700nm, InGaAs detector.

[0017] By combining a deuterium-halogen dual-spectrum detector with a 2048-pixel CCD, an ultraviolet LED array with a photomultiplier tube, and an InGaAs detector in a simple adaptable combination, different types of light can be detected through multi-modal combinations. Considering the interference effects between different detection units and different types of light, an interference compensation mechanism is designed as follows: Dynamic interference compensation module By incorporating a built-in temperature / pH sensor, real-time feedback is provided to the spectral correction algorithm. The specific algorithm is as follows: λcorrected=λraw×[1+α(T−25)+β(pH−7)] (α=0.002 / ℃, β=0.0015 / pH unit).

[0018] (II) Multi-source fusion modeling layer Feature-level fusion architecture: includes (spectral feature extraction) + decision-level fusion (dynamic weight allocation).

[0019] Weighted decision fusion model This includes the construction of sub-models, as detailed below: UV-Vis model: Partial Least Squares Regression (PLSR); EEMs ​​model: Support Vector Regression (SVR); NIR model: Random Forest (RF).

[0020] It also includes dynamic weight allocation: wi=1 / RMSEi2∑(1 / RMSEj2) (i=UV,EEMs,NIR) (RMSEi represents the error of each sub-model on the validation set).

[0021] The present invention, through the above design, can be used to validate data in multiple water bodies (measurements in different water bodies). This includes cleaning rivers, urban sewage, industrial wastewater, and agricultural drainage.

[0022] This invention achieves the following through a three-tiered innovation: "multi-source spectral synchronous acquisition - real-time drift compensation - dynamic fusion modeling": Full spectrum coverage: ultraviolet (aromatic organic matter) + fluorescence (soluble microbial products) + near-infrared (suspended organic matter); Cross-scenario adaptation: Transfer learning enables the model to achieve an error of less than 10% in rivers / sewage / industrial water bodies.

[0023] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely preferred examples and are not intended to limit the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for constructing a quantitative model of chemical oxygen demand (COD) in water based on multi-source spectral fusion, characterized in that: Including a multi-source spectral sensing layer and a multi-source fusion model, The specific steps are as follows: Step 1: Construct a system for simultaneous measurement and effective fusion of multi-source spectra; Step 2: Construct a dynamic interference compensation module; Step 3: Construct a feature-level fusion architecture; Step 4: Construct a weighted decision fusion model.

2. The method for constructing a quantitative model of chemical oxygen demand in water based on multi-source spectral fusion according to claim 1, characterized in that: Step 1, which involves constructing simultaneous multi-source spectral determination and effective fusion, includes: Deuterium-halogen dual-spectrum detector with a 2048-pixel CCD for detecting ultraviolet-visible absorption light, with a detection range of 200-700nm. Ultraviolet LED array + photomultiplier tube; used for detecting three-dimensional fluorescence, with a detection range of Ex: 250-550nm, Em: 300-650nm; InGaAs detectors are used to detect near-infrared diffuse reflected light, with a detection range of 900-1700nm.

3. The method for constructing a quantitative model of chemical oxygen demand in water based on multi-source spectral fusion according to claim 1, characterized in that: Step 2 includes a dynamic interference compensation module, which feeds back to the spectral correction algorithm in real time by setting a built-in temperature / pH sensor; The specific algorithm is as follows: λcorrected=λraw×[1+α(T−25)+β(pH−7)] (α=0.002 / ℃, β=0.0015 / pH unit).

4. The method for constructing a quantitative model of chemical oxygen demand in water based on multi-source spectral fusion according to claim 3, characterized in that: Step 3 includes: spectral feature extraction plus decision-level fusion.

5. The method for constructing a quantitative model of chemical oxygen demand in water based on multi-source spectral fusion according to claim 4, characterized in that: The decision-level fusion includes a weighted decision fusion model, which comprises the construction of sub-models, as detailed below: UV-Vis model: Partial Least Squares Regression (PLSR); EEMs ​​model: Support Vector Regression (SVR); NIR model: Random Forest (RF).

6. The method for constructing a quantitative model of chemical oxygen demand in water based on multi-source spectral fusion according to claim 5, characterized in that: The decision-level fusion also includes dynamic weight allocation: the allocation logic is as follows: wi=1 / RMSEi2∑(1 / RMSEj2) (i=UV,EEMs,NIR) (RMSEi represents the error of each sub-model on the validation set).

Images