CN-122020136-A - Biological influence assessment method and system for marine micro ecological system

CN122020136ACN 122020136 ACN122020136 ACN 122020136ACN-122020136-A

Abstract

The invention discloses a biological influence assessment method and system for a marine micro ecological system, which are characterized by acquiring environmental parameters, shooting images, obtaining morphological feature vectors through an image CNN model, calculating an environmental parameter change function, acquiring spectral tin, obtaining spectral feature vectors through the spectral CNN model, calculating a biological response function, and calculating a quantization index BI of environmental parameters on biological influence according to the environmental parameter change function and the biological response function, so that the biological influence degree can be assessed according to the BI index. The invention can dynamically capture the instantaneous change of the environment and the organism, synchronously extract the biological form, behavior and biochemical parameters, and quantify the influence degree of the environmental factors on the organism through the established mathematical model.

Inventors

YU DELIANG
WANG SEN
Qiao Tengsheng
ZHANG RUIHAO
WANG HAILIANG
WANG WEIMIN
SONG XIAOJIN
PAN KEHOU

Assignees

崂山国家实验室

Dates

Publication Date: 20260512
Application Date: 20260415

Claims (10)

1. A biological influence evaluation method of a marine micro-ecological system is characterized in that, Acquiring an environmental parameter x i of the marine micro-ecological system, and calculating a difference delta x i between the environmental parameter x i and the baseline environmental parameter; Computing environmental parameter variation functions Wherein omega i is the weight of the environmental parameter, N is the number of environmental parameters, The value range of the environmental parameter is obtained; Shooting an image of the marine micro ecological system, and inputting an image CNN model to obtain a morphological feature vector V I [ community density rho (p/mm < 2 >), plankton ingestion rate F (times/h), and morphological distortion rate D (%) ]; The spectrum of the marine micro-ecological system is obtained through a multispectral camera, and a spectrum CNN model is input to obtain a spectrum characteristic vector V λ [ chlorophyll (Chl, mug/L) and carotenoid (Car, mug/L) ]; calculating a biological response function Wherein V I (t) is an image feature vector extracted by a t-moment image CNN model, V I (0) is an image feature vector extracted by a base line state image CNN model, V λ (t) is a spectrum feature vector extracted by a t-moment spectrum CNN model, and V λ (0) is a spectrum feature vector extracted by a base line state spectrum CNN model; Calculating a quantization index BI of the environmental parameters on biological effects, Wherein ω E is an environmental parameter weight, ω B is a biological response weight, ω E +ω B =1, and ε is an error term.
2. The method for evaluating biological effects of a marine micro-ecological system according to claim 1, wherein the image CNN model and the spectrum CNN model are trained ,LOSS 1 =α×MSE(ρ pred ,ρ true )+β×MAE(V λpred ,V λrue )+γ×(1–cos(V I ,V Iaug )); through a LOSS function LOSS 1 , wherein MSE is a mean square error, MAE is an average absolute error, ρ pred is a predicted value of community density, ρ true is an actual measured value of community density, V λpred is a predicted value of biochemical characteristics, V λrue is an actual measured value of biochemical characteristics, V I is a feature vector of an original image, V Iaug is a feature vector of an enhanced image, α, β, γ are task weights, and the training process is dynamically adjusted according to an error of a verification set.
3. The biological influence assessment method of the marine micro ecological system according to claim 2 is characterized in that the extraction precision of CNN characteristics is verified, verification indexes are required to meet that ① determination coefficients R2 are more than or equal to 0.96 (the fitting goodness of a predicted value and a true value), ② chlorophyll concentration MAE is less than or equal to 0.5 mug/L, ③ carotenoid content MAE is less than or equal to 0.2 mug/L, and ④ characteristic vector cosine similarity is more than or equal to 0.95.
4. The method for evaluating biological influence of a marine micro-ecological system according to claim 2, wherein the image enhancement method is that the image is rotated by 0-180 degrees, scaled by 0.8-1.2 times and adjusted by brightness +/-10%.
5. The marine micro-ecological system biological impact assessment method according to any one of claims 1 to 4, wherein ω i 、ω B and ω E are optimized by LOSS function LOSS 2 =MSE(BI pred ,BI true )+λ×(1-r(BI pred ,BI true ) under the constraint condition that ω E ≥0、ω B is equal to or greater than 0 and ω E +ω B =1, ω i is equal to or greater than 0 and Σω i =1, where r is a BI pred and BI true pearson correlation coefficient, λ=0.5, BI pred is a model calculated biological impact index prediction value, BI true is a true value obtained by actually measuring a biological index, training set iteration update weights are adopted in the optimization process, verification set assessment convergence is performed, and convergence criterion is verification set pearson correlation coefficient r is equal to or greater than 0.85 and MSE is equal to or less than 0.02.
6. A marine micro-ecological system biological impact assessment system, the system comprising: The box body comprises a transparent partition board, and light sources are arranged around the box body; A cultivation water tank positioned in the box body; The cameras are used for acquiring images and spectrums in the cultivation water tank; the sensor is used for collecting environmental parameters in the cultivation water tank; A data processing module for performing marine organism impact assessment on the marine micro-ecosystem in the cultivation tank according to the assessment method of any one of claims 1-5.
7. The marine micro-ecosystem biological impact assessment system of claim 6, wherein the cameras comprise a microscopic imaging camera, a high-speed dynamic camera, and a multispectral camera.
8. The marine micro-ecological system biological impact assessment system according to claim 6, wherein the system comprises water inlet and outlet pipes, and water circulation and nutrient salt supplementation are achieved.
9. The marine micro-ecological system biological impact assessment system according to claim 8, wherein the system comprises a water control valve for controlling the water flow rate and the replacement frequency, and simulating the water flow environment in the sea area.
10. The marine micro-ecological system biological impact assessment system according to claim 6, wherein the system comprises an observation window providing an observation channel for the camera, the observation window comprising a high light transmission glass.

Description

Biological influence assessment method and system for marine micro ecological system Technical Field The invention belongs to the technical field of marine ecological observation and data analysis, and particularly relates to a biological influence assessment method and system for a marine micro-ecological system. Background In marine ecological research (especially in the islands-in-the-sea field), in-situ, real-time, long-term ecological observations are the core requirement for resolving bioelectrochemical processes (e.g., nutrient salt circulation, plankton community succession). The traditional research relies on field sample collection, a large amount of manpower and material resources (such as scientific investigation ship leasing, sample transportation and pretreatment) are required, the problems of high cost, long time consumption, short observation period (the complete life cycle of organisms cannot be covered) and the like exist, meanwhile, the ecological systems of islands and surrounding sea areas are complex, the field environment fluctuation is large, and the single variable is difficult to accurately control so as to explore the causal relationship of environmental parameter-biological response. The existing research is based on qualitative description (such as 'algae density reduction'), a quantitative model of 'environmental parameter change-biological response-influence degree' is not established, and an ecological stress threshold cannot be accurately judged. The above information disclosed in this background section is only for enhancement of understanding of the background section of the application and therefore it may not form the prior art that is already known to those of ordinary skill in the art. Disclosure of Invention The invention provides a biological influence evaluation method and system for a marine micro-ecological system, which are used for solving the technical problem that the ecological stress threshold cannot be accurately judged according to environmental parameters in the prior art. In order to achieve the above-mentioned invention/design purpose, the invention adopts the following technical scheme to realize: The biological influence evaluation method of the marine micro-ecological system comprises the steps of obtaining an environmental parameter x i of the marine micro-ecological system, and calculating a difference delta x i between the environmental parameter x i and a baseline environmental parameter; Computing environmental parameter variation functions Wherein omega i is the weight of the environmental parameter,N is the number of environmental parameters,The value range of the environmental parameter is obtained; Shooting an image of the marine micro ecological system, and inputting an image CNN model to obtain a morphological feature vector V I [ community density rho (p/mm < 2 >), plankton ingestion rate F (times/h), and morphological distortion rate D (%) ]; The spectrum of the marine micro-ecological system is obtained through a multispectral camera, and a spectrum CNN model is input to obtain a spectrum characteristic vector V λ [ chlorophyll (Chl, mug/L) and carotenoid (Car, mug/L) ]; calculating a biological response function Wherein V I (t) is an image feature vector extracted by a t-moment image CNN model, V I (0) is an image feature vector extracted by a base line state image CNN model, V λ (t) is a spectrum feature vector extracted by a t-moment spectrum CNN model, and V λ (0) is a spectrum feature vector extracted by a base line state spectrum CNN model; Calculating a quantization index BI of the environmental parameters on biological effects, Wherein ω E is an environmental parameter weight, ω B is a biological response weight, ω E+ωB =1, and ε is an error term. According to the biological influence assessment method for the marine micro-ecological system, the image CNN model and the spectrum CNN model are trained ,LOSS1=α×MSE(ρpred,ρtrue)+β×MAE(Vλpred,Vλrue)+γ×(1–cos(VI,VIaug)); through the LOSS function LOSS 1, wherein MSE is mean square error, MAE is mean absolute error, rho pred is a predicted value of community density, rho true is an actual measurement value of community density, V λpred is a predicted value of biochemical characteristics, V λrue is an actual measurement value of biochemical characteristics, V I is a characteristic vector of an original image, V Iaug is a characteristic vector of an enhanced image, alpha, beta and gamma are task weights, and the training process is dynamically adjusted according to the error of a verification set. According to the biological influence assessment method for the marine micro-ecological system, CNN characteristic extraction accuracy is verified, verification indexes are required to meet the requirements that ① determination coefficients R2 are more than or equal to 0.96 (the fitting goodness of predicted values and true values), ② chlorophyll concentration MAE is less than or equal to 0.5 m