Intelligent diagnosis method for aquatic diseases

Abstract

The invention discloses an intelligent diagnosis method for aquatic diseases, and relates to the technical field of aquatic disease diagnos.The method comprises the steps that aquaculture information is collected and stored through a data monitoring module, a preliminary diagnosis model is built through a multi-source fusion module, and the biological health state and the aquaculture risk state are evaluated; therefore, disease risks are evaluated in time to realize preliminary diagnosis and prediction, and then a deep learning model is constructed through a central processing unit to perform refined judgment on aquatic diseases, so that an aquatic disease diagnosis and prediction result based on environmental characterization is generated, and refined diagnosis and prediction on the aquatic diseases are realized. The defect of insufficient diagnosis accuracy caused by aquatic product disease characteristic differences in different environment states is overcome, the generalization ability of scheme application is improved for auxiliary reference of farmers, the deep learning model is adjusted and corrected through the optimization design module, and self-growth optimization of the deep learning model is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of aquatic disease diagnosis, and in particular to an intelligent aquatic disease diagnosis method. Background Art

[0002] In recent years, aquaculture, as an important part of agriculture, has achieved remarkable development worldwide. The scale of aquaculture has been continuously expanded and the types of aquaculture have become increasingly rich. However, with the expansion of aquaculture scale, the problem of aquaculture diseases has become increasingly prominent, becoming a key factor restricting the sustainable and healthy development of the industry. There are many types of diseases, including bacterial diseases, viral diseases, fungal diseases and parasitic diseases, which not only lead to high mortality rates and increase aquaculture costs, but also may pollute water quality and the ecological environment. Disease diagnosis is an important part of aquaculture disease prevention and control. Accurate diagnosis can timely understand the disease type, infection degree and transmission mode, and provide a basis for formulating scientific and effective prevention and control measures. Traditional diagnostic methods, such as monitoring the physical properties of aquatic organisms, may have problems such as misdiagnosis and missed diagnosis. Factors such as water temperature, dissolved oxygen, and pH value in different breeding environments may affect the manifestation of aquatic diseases, resulting in differences in the growth, diet and activity of aquatic organisms, and differences in the types and corresponding numbers of microorganisms, which in turn affect the feeding, medication and density management of aquaculture; Therefore, the existing intelligent diagnosis technology for aquatic diseases has the defects of insufficient diagnostic efficiency and generalization ability. Due to the differences in the characteristics of the same aquatic disease under different environmental conditions, the diagnosis efficiency and accuracy of aquatic diseases are insufficient, and the generalization ability of the solution application is insufficient. In view of the above technical defects, a solution is now proposed. Summary of the invention

[0003] The purpose of the present invention is to solve the defects of the prior art in terms of insufficient diagnostic efficiency, accuracy and generalization ability, and to achieve preliminary diagnosis and prediction and refined diagnosis and prediction of aquatic diseases by analyzing the differences in characteristics of aquatic diseases under different environmental conditions, thereby improving the generalization ability of the scheme application, and adjusting and correcting the model to achieve self-growth optimization of the model for auxiliary reference by farmers.

[0004] In order to achieve the above object, the present invention adopts the following technical solutions: An intelligent diagnosis method for aquatic diseases comprises the following steps: S1, the data monitoring module collects and stores aquaculture information: the data monitoring period Ts is set to regularly monitor the aquaculture information; the aquaculture information includes biological aquaculture data and aquaculture environment data; wherein the biological aquaculture data includes aquatic biological parameters and aquaculture parameters; the aquaculture environment data includes aquaculture water quality parameters and environmental bottom quality parameters; S2, the multi-source fusion module builds a preliminary diagnosis model to analyze and process biological aquaculture data: through edge computing, biological aquaculture data analysis is performed, aquatic biological parameters and aquaculture parameters are pre-processed in turn, and the biological health status and aquaculture risk status are evaluated respectively, so as to obtain disease feature data sets to evaluate disease risks and achieve preliminary diagnosis and prediction; S3, the CPU builds a deep learning model to make a refined judgment on aquatic diseases: a correlation function is built between the aquaculture environment data and the disease feature data set to generate a diagnosis and prediction result of aquatic diseases, and make a refined diagnosis and prediction of aquatic diseases; S4, the optimization design module adjusts and corrects the deep learning model: the actual diagnosis results of aquatic diseases are obtained through an expert team, and compared with the diagnosis prediction results, so as to determine the accuracy of the deep learning model, and provide feedback for targeted adjustments and corrections to achieve self-growth optimization of the deep learning model.

[0005] Furthermore, the specific process of the preliminary diagnosis model is as follows: Sa1, obtain microbial coefficient and aquatic biological coefficient through preliminary analysis of aquatic biological parameters, and then comprehensively obtain the biological status assessment index to evaluate the biological health status; Aquatic biological parameters include microbial vector group and aquatic biological vector group; Sa1-1, obtain the microbial coefficient by analyzing the microbial vector group, which includes the types and quantities of microorganisms, including viruses, bacteria and parasites; Sa1-2, aquatic biological coefficients are obtained by analyzing the aquatic biological vector group, which includes the growth value, diet value and activity value of aquaculture organisms; Sa1-3, integrate the microbial coefficient and the aquatic biological coefficient into a biological factor set, and obtain the biological status evaluation index through the biological factor set to evaluate the biological health status; Sa2, through preliminary analysis of aquaculture parameters, obtain the feeding management coefficient, drug management coefficient and density management coefficient, and then comprehensively obtain the aquaculture risk assessment index to evaluate the aquaculture risk status; Aquaculture parameters include feeding management vector group, drug management vector group and density management vector group; Sa2-1, obtain the feeding management coefficient through the feeding management vector group, the feeding management vector group includes the feeding amount, feeding frequency and bait type; Sa2-2, obtains the medication management coefficient through the medication management vector group, and the medication management vector group includes drug type, medication dosage and medication frequency; Sa2-3, obtain the density management coefficient through the density management vector group, the density management vector group includes the breeding density and the pond division time; Sa2-4, integrate the feeding management coefficient, drug management coefficient and density management coefficient into aquaculture risk factors, and obtain the aquaculture risk assessment index through the aquaculture risk factor data set to evaluate the aquaculture risk status; Sa3 integrates the biological factor data set, biological status assessment index, aquaculture risk factor data set and aquaculture risk assessment index into a disease characteristic data set, and evaluates the risk level of aquatic diseases through the disease characteristic data set to achieve preliminary diagnosis and prediction.

[0006] Furthermore, the specific process of the deep learning model is: Sb1, obtain water quality assessment coefficient through preliminary analysis of aquaculture water quality parameters; Aquaculture water quality parameters include water temperature, pH value and the content of water environmental compounds; Sb2, obtain the sediment assessment coefficient through preliminary analysis of environmental sediment parameters; Environmental sediment parameters include sediment thickness, organic matter content, and sulfide content; Sb3, constructs the correlation function between the water quality assessment coefficient, the bottom quality assessment coefficient and the disease characteristic data set, and generates the diagnosis and prediction results of aquatic diseases based on environmental characteristics, and performs refined diagnosis and prediction of aquatic diseases through the diagnosis and prediction results.

[0007] Furthermore, the specific process of preliminary analysis of aquatic biological parameters is as follows: Sa1-1, microbial coefficients were obtained by analyzing the microbial vector group; Label the microbial vector group as Vmic<ZLm0,SLm0> , where ZLm0 refers to the set of types of m0 microorganisms, and SLm0 refers to the set of numbers of m0 microorganisms; The type of any microorganism is marked as Zmic, and the number of microorganism Zmic is marked as Lmic; The microbial coefficient Mic is obtained comprehensively through the types Zmic and the number Lmic of m0 kinds of microorganisms; Sa1-2, obtain the aquatic biological coefficient by analyzing the aquatic biological vector group; Sampling monitoring is performed on the same batch of aquaculture organisms. It is preset to select m1 aquaculture organisms, mark any sampled aquaculture organism as H, monitor the weight Hg1, feed amount Hg2, swimming speed Hv and swimming time Ht of the aquaculture organism H, so as to obtain the growth value EA, diet value EB and activity value EC; Label the aquatic biological vector group as Vaqu<EAm1,EBm1,ECm1> , where EAm1 refers to the set of growth values ​​EA of m1 aquaculture organisms; EBm1 refers to the set of diet values ​​EB of m1 aquaculture organisms; ECm1 refers to the set of activity values ​​EC of m1 aquaculture organisms; The aquatic biological coefficient Aqu is obtained comprehensively through the aquatic biological vector group of m1 aquaculture organisms; Sa1-3, by combining the microbial coefficient Mic and the aquatic biological coefficient Aqu, the biological status assessment index ORG is obtained to evaluate the biological health status.

[0008] Furthermore, the specific process of assessing the risk status of aquaculture is as follows: Sa2-1, obtain the feeding management coefficient through the feeding management vector group; Mark the bait management vector group as Wfee<ZRm2,LRm2,FRm2> , where ZRm2 refers to the set of m2 bait types, LRm2 refers to the set of m2 bait feeding amounts, and FRm2 refers to the set of m2 bait feeding frequencies; The type of any bait is marked as Rfee, and the amount and frequency of bait Rfee are marked as Lfee and Ffee respectively; The feeding management coefficient Fee is obtained comprehensively through the feeding amount Lfee and feeding frequency Ffee of m2 kinds of bait; Sa2-2, obtains the medication management coefficient through the medication management vector group, and the medication management vector group includes drug type, medication dosage and medication frequency; Label the medication management vector group as Wmed<ZDm3,LDm3,FDm3> , where ZDm3 refers to the set of m3 drug types, LDm3 refers to the set of m3 drug dosages, and FDm3 refers to the set of m3 drug usage frequencies; The type of any drug is marked as Dmed, and the dosage and frequency of drug Dmed are marked as Lmed and Fmed respectively; The medication management coefficient Med is obtained comprehensively through the dosage Lmed and medication frequency Fmed of m3 drugs; Sa2-3, obtain the density management coefficient through the density management vector group, and mark the density management vector group as Wden<NSσ,NSt> , where NSσ and NSt refer to the stocking density and pond division time, respectively; Set up a parameter evaluation model, input parameter i and its standard interval [Ig, If], and obtain the evaluation value Ui of parameter i by combining parameter i and its standard interval [Ig, If]; By combining the stocking density NSσ and the pond division time NSt, the stocking density NSσ and the pond division time NSt are substituted into the parameter evaluation model in turn to obtain the density management coefficient Den; Sa2-4, by combining the feeding management coefficient Fee, the medication management coefficient Med and the density management coefficient Den, the aquaculture risk assessment index RISK is obtained to evaluate the aquaculture risk status.

[0009] Furthermore, the specific process of preliminary analysis of aquaculture water quality parameters is as follows: Aquaculture water quality parameters include water temperature wd, pH value and the content of water environmental compounds; Set the standard intervals of water temperature wd, pH value and water environment compounds, mark the number of water environment compounds as N0, and mark any water environment compound as Hj; The water temperature wd, pH value, water environmental compounds and their corresponding standard ranges are substituted into the parameter evaluation model to output the evaluation value Uwd of the water temperature wd, the evaluation value UpH of the pH value, the evaluation value UHj of the water environmental compound Hj, and then comprehensively obtain the water quality evaluation coefficient WQ.

[0010] Furthermore, the specific process of preliminary analysis of environmental sediment parameters is as follows: Set the standard ranges of sediment thickness hd, organic matter content and sulfide content, mark the number of organic matter types as N1, and mark any organic matter as OM; mark the number of sulfide types as N2, and mark any sulfide as SF; The standard ranges of sediment thickness hd, organic matter content and sulfide content are substituted into the parameter evaluation model to output the evaluation value Uhd of sediment thickness hd, the evaluation value Uom of organic matter OM and the evaluation value Usf of sulfide SF respectively, and then comprehensively obtain the bottom sediment evaluation coefficient BQ.

[0011] Furthermore, the specific process of generating the diagnosis and prediction results of aquatic diseases is as follows: Mark any aquatic disease category as Q, and fit the correlation function F0 of the aquatic disease category Q through the water quality assessment coefficient WQ, the bottom quality assessment coefficient BQ and the disease characteristic data set; The risk interval of the correlation function F0 is set, and the risk level of the category Q of aquatic diseases is evaluated by interval comparison.

[0012] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are: The present invention collects and stores aquaculture information through a data monitoring module, and constructs a preliminary diagnosis model through a multi-source fusion module to evaluate the biological health status and the aquaculture risk status, so as to timely evaluate the disease risk to achieve preliminary diagnosis and prediction, and then constructs a deep learning model through a central processing unit to make a refined judgment on aquatic diseases, thereby generating a diagnosis and prediction result of aquatic diseases based on environmental characteristics, and realizing refined diagnosis and prediction of aquatic diseases, which solves the defect of insufficient diagnostic accuracy caused by differences in aquatic disease characteristics under different environmental conditions, improves the generalization ability of the solution application, and provides auxiliary reference for farmers, and adjusts and corrects the deep learning model through the optimization design module to achieve self-growth optimization of the deep learning model. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 A schematic diagram showing the steps of the overall solution process of the present invention is shown; Figure 2 A schematic diagram of the process flow of the preliminary diagnosis model of the present invention is shown; Figure 3 A schematic diagram of the process flow of the deep learning model of the present invention is shown; Figure 4 A schematic diagram of module connection of the application device of the present invention is shown. DETAILED DESCRIPTION

[0014] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0015] Embodiment 1: like Figure 1-Figure 4 As shown, an intelligent diagnosis method for aquatic diseases comprises the following steps: S1, the data monitoring module collects and stores aquaculture information: the data monitoring period Ts is set to regularly monitor the aquaculture information; the aquaculture information includes biological aquaculture data and aquaculture environment data; wherein the biological aquaculture data includes aquatic biological parameters and aquaculture parameters; the aquaculture environment data includes aquaculture water quality parameters and environmental bottom quality parameters; The aquatic biological parameters include a microbial vector group and an aquatic biological vector group; the information of the aquatic biological parameters is collected through a biological detection sensor; S1-1, microbial vector group includes the types and quantities of microorganisms, including viruses, bacteria and parasites; Label the microbial vector group as Vmic<ZLm0,SLm0> , where ZLm0 refers to the set of types of m0 microorganisms, and SLm0 refers to the set of numbers of m0 microorganisms; The type of any microorganism is marked as Zmic, and the number of microorganism Zmic is marked as Lmic; S1-2, aquatic organism vector group includes growth values, diet values, and activity values ​​of aquaculture organisms; Sampling monitoring is performed on the same batch of aquaculture organisms. It is preset to select m1 aquaculture organisms, mark any sampled aquaculture organism as H, monitor the weight Hg1, feed amount Hg2, swimming speed Hv and swimming time Ht of the aquaculture organism H, so as to obtain the growth value EA, diet value EB and activity value EC; Label the aquatic biological vector group as Vaqu<EAm1,EBm1,ECm1> , where EAm1 refers to the set of growth values ​​EA of m1 aquaculture organisms; EBm1 refers to the set of diet values ​​EB of m1 aquaculture organisms; ECm1 refers to the set of activity values ​​EC of m1 aquaculture organisms; S1-3, aquaculture parameters include feeding management vector group, drug management vector group and density management vector group; aquaculture parameters are obtained by accessing the log of the aquaculture center, and the log is recorded and obtained by the aquaculture management equipment or management personnel; The feeding management vector group includes feeding amount, feeding frequency and feed type; the medication management vector group includes medication type, medication dosage and medication frequency; the density management vector group includes breeding density and pond division time; S1-4, aquaculture water quality parameters include water temperature, pH value and the content of water environment compounds; water environment compounds include dissolved oxygen, ammonia nitrogen compounds, nitrite and nitrate; Environmental sediment parameters include sediment thickness, organic matter content, and sulfide content; Bottom sediment thickness refers to the thickness of the bottom sediment in the aquaculture waters. When the bottom sediment is too high, it means that the impurity content in the water is high and aquatic metabolites accumulate and precipitate a lot, which may cause poor water environment quality and affect the health of aquatic organisms. Organic matter refers to organic matter containing a certain amount of carbon, nitrogen, phosphorus, sulfur and other elements. It can provide nutrients for the growth of microorganisms in the water, promote the reproduction and growth of plankton in the water body, and maintain water quality stability. However, excessive accumulation of organic matter will lead to the increase of harmful bacteria, trigger the outbreak of farmed animal diseases, and may produce toxic and harmful substances or gases, such as ammonia nitrogen, hypochlorite, methane, etc., which will harm the aquatic ecological environment. Sulfides mainly refer to sulfur-containing compounds, especially hydrogen sulfide (H2S) and its related forms, such as HS - 、S2- Sulfide is a compound of ions. Sulfide mainly comes from the external environment and biological metabolism. Among them, a large amount of sulfate contained in the external environment such as soil and rock layers, underground springs, etc. enters the aquaculture water body and is decomposed by sulfate-reducing bacteria under anaerobic conditions. Sulfur-containing amino acids and proteins in organic matter such as biological metabolites, leftover bait, and rotten aquatic plants are decomposed and utilized by microorganisms during the process of sedimentation. Sulfide is highly toxic to aquatic animals. S2, the multi-source fusion module builds a preliminary diagnosis model to analyze and process biological aquaculture data: through edge computing, biological aquaculture data analysis is performed, aquatic biological parameters and aquaculture parameters are pre-processed in turn, and the biological health status and aquaculture risk status are evaluated respectively, so as to obtain disease feature data sets to evaluate disease risks and achieve preliminary diagnosis and prediction; The specific process of the preliminary diagnosis model is: Sa1, obtain microbial coefficient and aquatic biological coefficient through preliminary analysis of aquatic biological parameters, and then comprehensively obtain the biological status assessment index to evaluate the biological health status; Sa1-1, obtain the microbial coefficient by analyzing the microbial vector group, and obtain the microbial coefficient Mic comprehensively through the type Zmic and quantity Lmic of m0 kinds of microorganisms: ; in, is the proportional coefficient of the number of microorganisms Zmic and Lmic Greater than 0; the proportional coefficient is preset and obtained after calculation through a large amount of experimental data; when the number Lmic of m0 microorganisms Zmic is higher, the microbial coefficient Mic is higher, and the microbial content of the water plant is evaluated to be higher; Sa1-2, obtain the aquatic biological coefficient by analyzing the aquatic biological vector group, and obtain the aquatic biological coefficient Aqu comprehensively through the aquatic biological vector group of m1 aquaculture organisms: ; It can be deduced that, ; in, , and are the conversion coefficients of the weight Hg1, feed amount Hg2 and swimming speed Hv of the aquaculture organism H. The conversion coefficient refers to the constant value of the aquaculture organism H vector group converted into the weight Hg1, feed amount Hg2 and swimming speed Hv of the aquaculture organism H. , and All are greater than 0; and are the growth rates of the weight Hg1 of aquaculture organisms H and the amount of feed Hg2, namely, It refers to the difference in weight Hg1 corresponding to two adjacent data monitoring periods; It refers to the difference in bait quantity Hg2 corresponding to two adjacent data monitoring periods; , , are the weight coefficients of growth value EA, diet value EB and activity value EC, respectively, and , , All are greater than 0. When the growth value EA, diet value EB and activity value EC are higher, the aquatic organism coefficient Aqu is higher, and the state of aquatic organisms is better. Sa1-3, integrate the microbial coefficient and the aquatic biological coefficient into a biological factor set, and obtain the biological status evaluation index through the biological factor set to evaluate the biological health status; By combining the microbial coefficient Mic and the aquatic biological coefficient Aqu, the biological status assessment index ORG is obtained: ; in, is the adjustment factor of the biological status assessment index ORG, and the adjustment factor To ensure A preset constant value that is always greater than 1; when the microbial coefficient Mic is lower and the aquatic biological coefficient Aqu is higher, the biological status assessment index ORG is higher; The evaluation interval of the biological status evaluation index ORG is set, and the biological health status is evaluated by interval comparison: when the biological status evaluation index ORG is higher, the biological health status is better; Sa2, through preliminary analysis of aquaculture parameters, obtain the feeding management coefficient, drug management coefficient and density management coefficient, and then comprehensively obtain the aquaculture risk assessment index to evaluate the aquaculture risk status; Sa2-1, obtain the feeding management coefficient through the feeding management vector group, and mark the feeding management vector group as Wfee<ZRm2,LRm2,FRm2> , where ZRm2 refers to the set of m2 bait types, LRm2 refers to the set of m2 bait feeding amounts, and FRm2 refers to the set of m2 bait feeding frequencies; The type of any bait is marked as Rfee, and the amount and frequency of bait Rfee are marked as Lfee and Ffee respectively; The feeding management coefficient Fee is obtained comprehensively through the feeding amount Lfee and feeding frequency Ffee of m2 kinds of bait: ; in, and are the weight coefficients of the bait quantity Lfee and the bait frequency Ffee, respectively, and and All are greater than 0; when the bait type m2, bait amount Lfee and bait frequency Ffee are higher, the bait management coefficient Fee is higher, and the effect of bait management evaluation is better; Sa2-2, obtain the medication management coefficient through the medication management vector group, and mark the medication management vector group as Wmed<ZDm3,LDm3,FDm3> , where ZDm3 refers to the set of m3 drug types, LDm3 refers to the set of m3 drug dosages, and FDm3 refers to the set of m3 drug usage frequencies; The type of any drug is marked as Dmed, and the dosage and frequency of drug Dmed are marked as Lmed and Fmed respectively; Through the dosage Lmed and frequency Fmed of m3 drugs, the medication management coefficient Med is obtained comprehensively: ; in, and are the weight coefficients of medication dose Lmed and medication frequency Fmed, respectively, and and All are greater than 0; when the type of drug m3, the dosage Lmed and the frequency of medication Fmed are higher, the medication management coefficient Med is lower, and the effect of evaluating medication management is worse; Sa2-3, obtain the density management coefficient through the density management vector group, and mark the density management vector group as Wden<NSσ,NSt> , where NSσ and NSt refer to the stocking density and pond division time, respectively; Set up the parameter evaluation model, input parameter i and its standard interval [Ig, If], and obtain the evaluation value Ui of parameter i by combining parameter i and its standard interval [Ig, If]: ; When parameter i is in the standard interval [Ig, If], the evaluation value Ui of parameter i is ≥ 0; when parameter i is not in the standard interval [Ig, If], the evaluation value Ui of parameter i is < 0, and the higher the deviation of parameter i from the standard interval [Ig, If], the smaller the evaluation value Ui of parameter i is; By combining the aquaculture density NSσ and the pond division time NSt, the aquaculture density NSσ and the pond division time NSt are substituted into the parameter evaluation model in turn to obtain the density management coefficient Den: ; Among them, [σg, σf] is preset as the standard interval of the breeding density NSσ, and [tg, tf] is preset as the standard interval of the pond division time NSt. When the breeding density NSσ is in the standard interval [σg, σf] and the pond division time NSt is in the standard interval [tg, tf], the density management coefficient Den is greater than or equal to 0, and the effect of the density management is good; on the contrary, when the breeding density NSσ is not in the standard interval [σg, σf] and the pond division time NSt is not in the standard interval [tg, tf], the density management coefficient Den is less than 0, and the greater the deviation of the breeding density NSσ and the pond division time NSt from the corresponding standard interval, the lower the density management coefficient Den, and the worse the effect of the density management is. Sa2-4, integrate the feeding management coefficient, drug management coefficient and density management coefficient into aquaculture risk factors, and obtain the aquaculture risk assessment index through the aquaculture risk factor data set to evaluate the aquaculture risk status; By combining the feeding management coefficient Fee, the medication management coefficient Med and the density management coefficient Den, the aquaculture risk assessment index RISK is obtained: ; Among them, e1, e2 and e3 are the weight coefficients of the feeding management coefficient Fee, the medication management coefficient Med and the density management coefficient Den, respectively, and e1, e2 and e3 are all greater than 0; is the adjustment coefficient of the aquaculture risk assessment index RISK, and the adjustment coefficient To ensure A preset constant value that is always greater than 1; when the feeding management coefficient Fee, the medication management coefficient Med, and the density management coefficient Den are higher, the aquaculture risk assessment index RISK is lower; Set the evaluation interval of the breeding risk assessment index RISK, and evaluate the breeding risk status through interval comparison: the lower the breeding risk assessment index RISK, the lower the breeding risk level; Sa3, integrates the biological factor data set, biological status assessment index, aquaculture risk factor data set and aquaculture risk assessment index into a disease characteristic data set, evaluates the risk level of aquatic diseases through the disease characteristic data set, and realizes preliminary diagnosis and prediction; Based on aquaculture experience and a large number of experimental measurements, the risk thresholds of each element of the disease characteristic data set are preset, and the risk status of each element is evaluated by threshold comparison. When the element data value is lower than its risk threshold, the element is determined to be a risk element. Risk elements include biological factors and aquaculture risk factors that are lower than the risk threshold. The more risk elements there are and the higher the amplitude, the higher the risk level of aquatic diseases. By comprehensively monitoring biological breeding data, we can timely discover disease risks and take corresponding prevention and control measures to ensure the healthy development of aquaculture; S3, the CPU builds a deep learning model to make a refined judgment on aquatic diseases: a correlation function is built between the aquaculture environment data and the disease feature data set to generate a diagnosis and prediction result of aquatic diseases based on environmental characteristics, and a refined diagnosis and prediction of aquatic diseases is made for reference by farmers; The specific process of the deep learning model is: Sb1, obtain water quality assessment coefficient through preliminary analysis of aquaculture water quality parameters; Set the standard intervals of water temperature wd, pH value and water environment compounds, mark the number of water environment compounds as N0, and mark any water environment compound as Hj; By substituting the water temperature wd, pH value, water environment compounds and their corresponding standard intervals into the parameter evaluation model, the evaluation value Uwd of the water temperature wd, the evaluation value UpH of the pH value, and the evaluation value UHj of the water environment compound Hj are output respectively; The water quality assessment coefficient WQ is obtained comprehensively through the assessment value Uwd of water temperature wd, the assessment value UpH of pH value, and the assessment value UHj of N0 water environment compounds: ; in, , , are the weight coefficients of the water temperature assessment value Uwd, pH assessment value UpH, and water environment compound assessment value UHj, respectively, and , , All are greater than 0; when the water temperature assessment value Uwd, pH assessment value UpH, and water environment compound assessment value UHj are less than 0 and the lower they are, the lower the water quality assessment coefficient WQ is, and the worse the water quality status is; Sb2, obtain the sediment assessment coefficient through preliminary analysis of environmental sediment parameters; Set the standard ranges of sediment thickness hd, organic matter content and sulfide content, mark the number of organic matter types as N1, and mark any organic matter as OM; mark the number of sulfide types as N2, and mark any sulfide as SF; Substitute the standard ranges of sediment thickness hd, organic matter content and sulfide content into the parameter evaluation model, thereby outputting the evaluation value Uhd of sediment thickness hd, the evaluation value Uom of organic matter OM and the evaluation value Usf of sulfide SF respectively; The bottom sediment assessment coefficient BQ is obtained comprehensively through the assessment value Uhd of the bottom sediment thickness hd, the assessment value Uom of the N1 type organic matter OM and the assessment value Usf of the N2 type sulfide SF: ; in, , and are the weight coefficients of the sediment thickness assessment value Uhd, the organic matter assessment value Uom and the sulfide assessment value Usf, respectively, and , and are all greater than 0; when the sediment thickness assessment value Uhd, the organic matter assessment value Uom and the sulfide assessment value Usf are less than 0 and the lower they are, the lower the bottom quality assessment coefficient BQ is, and the worse the assessed bottom quality state is; Sb3, construct the correlation function between the water quality assessment coefficient, the bottom quality assessment coefficient and the disease characteristic data set, and generate the diagnosis and prediction results of aquatic diseases based on environmental characteristics, and perform refined diagnosis and prediction of aquatic diseases through the diagnosis and prediction results; Mark any aquatic disease category as Q, and fit the correlation function F0 of the aquatic disease category Q through the water quality assessment coefficient WQ, the bottom quality assessment coefficient BQ and the disease characteristic data set: ; in, and are the proportional coefficients of water quality assessment coefficient WQ and bottom quality assessment coefficient BQ respectively, and and are greater than 0, is the adjustment coefficient of the correlation function F0, and the adjustment coefficient To ensure The preset constant value is always greater than 0; when the water quality assessment coefficient WQ and the bottom quality assessment coefficient BQ are higher, the correlation function F0 is higher, the risk of aquatic disease Q is lower, and the aquaculture status of aquatic organisms is better. Thus, the risk of aquatic disease Q can be extended through the water quality assessment coefficient WQ and the bottom quality assessment coefficient BQ of the aquaculture environment, and the proportional coefficients corresponding to different types of aquatic diseases are and There are differences, thus achieving the generation of diagnostic and prediction results of aquatic diseases based on environmental characteristics; Set the risk interval of the correlation function F0, and evaluate the risk level of the category Q of aquatic diseases by comparing the intervals; The diagnosis and prediction results include the categories of aquatic diseases and their corresponding risk levels; if there are Nq categories of aquatic diseases, then Nq correlation functions are constructed to analyze the functional relationship between the environmental characteristics of Nq aquatic diseases and the disease characteristic data sets, and output the measured data values ​​of the corresponding correlation functions under the same environmental characteristics, sort the Nq correlation function data values ​​in descending order, extract the top 3 aquatic disease categories and their risk levels, and integrate them into the diagnosis and prediction results; Through the diagnosis and prediction results, the aquatic diseases can be diagnosed and predicted in detail to provide reference for farmers; S4, the optimization design module adjusts and corrects the deep learning model: the expert team obtains the actual diagnosis results of aquatic diseases and compares them with the diagnosis prediction results to determine the accuracy of the deep learning model, and provides feedback for targeted adjustment and correction to achieve self-growth optimization of the deep learning model; The actual diagnosis results of aquatic diseases are obtained through an expert team and compared with the diagnosis prediction results to determine the accuracy of the deep learning model ACC: ; Among them, Ya, Yb, and Yc are the comparison values ​​of the top three aquatic disease categories, respectively. When the aquatic disease category in the diagnosis prediction result is consistent with the actual diagnosis result, the comparison value of the aquatic disease category is assigned a value of 1; when the aquatic disease category in the diagnosis prediction result is inconsistent with the actual diagnosis result, the comparison value of the aquatic disease category is assigned a value of 0; A deviation threshold X0 of the accuracy ACC of the deep learning model is set. When the accuracy ACC of the deep learning model is lower than the deviation threshold X0, it is determined that the accuracy of the deep learning model is insufficient. The information is fed back to the deep learning model for targeted adjustment and correction, thus realizing the self-growth optimization of the deep learning model. The specific process is as follows: through in-depth research on the actual diagnosis results of aquatic diseases and the disease characteristics corresponding to the actual diagnosis results, the proportion coefficient of aquatic diseases of this category is calculated. and Parameter adjustment is performed. The specific range of parameter adjustment needs to be combined with the actual situation. For example, when the first ranked aquatic disease category V1 is inconsistent with the actual diagnosis result V0, it means that the correlation function data value of the aquatic disease category V1 is too high. Therefore, the proportional coefficient in the correlation function of the aquatic disease category V1 is adjusted. and It is reduced, thereby reducing the correlation function data value of the aquatic disease category V1, and further reducing the ranking of the aquatic disease category V1. Similarly, for the aquatic disease categories that are inconsistent with the actual diagnosis result V0 and their corresponding correlation functions, parameters are adjusted to finally achieve that the aquatic disease category ranked first is consistent with the actual diagnosis result. Therefore, under the guidance of the same disease feature data set, adaptive predictive diagnosis of aquatic diseases can be achieved by modifying the deep learning model.

[0016] The above-mentioned intelligent diagnosis method for aquatic diseases is applied to an intelligent diagnosis device for aquatic diseases, which includes a data monitoring module, a multi-source fusion module, a central processing unit and an optimization design module, wherein the data monitoring module, the multi-source fusion module, the central processing unit and the optimization design module are communicatively connected.

[0017] In summary, the present invention collects and stores aquaculture information through a data monitoring module, and constructs a preliminary diagnosis model through a multi-source fusion module to evaluate the biological health status and aquaculture risk status, so as to timely evaluate the disease risk to achieve preliminary diagnosis and prediction, and then constructs a deep learning model through a central processing unit to make fine judgments on aquatic diseases, thereby generating diagnostic prediction results of aquatic diseases based on environmental characteristics, and achieving fine diagnosis and prediction of aquatic diseases for auxiliary reference by farmers, and adjusting and correcting the deep learning model through an optimization design module to achieve self-growth optimization of the deep learning model, thereby solving the defect of insufficient accuracy in disease category judgment under different environmental conditions and improving the generalization ability of the solution application.

[0018] The size of the interval and threshold is set to facilitate comparison. The size of the threshold depends on the amount of sample data and the number of bases set by technical personnel in this field for each group of sample data; as long as it does not affect the proportional relationship between the parameter and the quantized value.

[0019] The above formulas are all dimensionless and numerical calculations. The formula is a formula obtained by collecting a large amount of data and performing software simulation to obtain the most recent real situation. The preset parameters in the formula are set by technicians in this field according to actual conditions. The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. An intelligent diagnosis method for aquatic diseases, characterized by: The following steps are involved: S1, the data monitoring module collects and stores aquaculture information: the data monitoring period Ts is set to regularly monitor the aquaculture information; the aquaculture information includes biological aquaculture data and aquaculture environment data; wherein the biological aquaculture data includes aquatic biological parameters and aquaculture parameters; the aquaculture environment data includes aquaculture water quality parameters and environmental bottom quality parameters; S2, the multi-source fusion module builds a preliminary diagnosis model to analyze and process biological aquaculture data: through edge computing, biological aquaculture data analysis is performed, aquatic biological parameters and aquaculture parameters are pre-processed in turn, and the biological health status and aquaculture risk status are evaluated respectively, so as to obtain disease feature data sets to evaluate disease risks and achieve preliminary diagnosis and prediction; S3, the CPU builds a deep learning model to make a refined judgment on aquatic diseases: a correlation function is built between the aquaculture environment data and the disease feature data set to generate a diagnosis and prediction result of aquatic diseases, and make a refined diagnosis and prediction of aquatic diseases; S4, the optimization design module adjusts and corrects the deep learning model: the actual diagnosis results of aquatic diseases are obtained through an expert team, and compared with the diagnosis prediction results, so as to determine the accuracy of the deep learning model, and provide feedback for targeted adjustments and corrections to achieve self-growth optimization of the deep learning model.

2. The intelligent diagnosis method for aquatic diseases according to claim 1, characterized in that: The specific process of the preliminary diagnosis model is: Sa1, obtain microbial coefficient and aquatic biological coefficient through preliminary analysis of aquatic biological parameters, and then comprehensively obtain the biological status assessment index to evaluate the biological health status; Aquatic biological parameters include microbial vector group and aquatic biological vector group; Sa1-1, obtain the microbial coefficient by analyzing the microbial vector group, which includes the types and quantities of microorganisms, including viruses, bacteria and parasites; Sa1-2, aquatic biological coefficients are obtained by analyzing the aquatic biological vector group, which includes the growth value, diet value and activity value of aquaculture organisms; Sa1-3, integrate the microbial coefficient and the aquatic biological coefficient into a biological factor set, and obtain the biological status evaluation index through the biological factor set to evaluate the biological health status; Sa2, through preliminary analysis of aquaculture parameters, obtain the feeding management coefficient, drug management coefficient and density management coefficient, and then comprehensively obtain the aquaculture risk assessment index to evaluate the aquaculture risk status; Aquaculture parameters include feeding management vector group, drug management vector group and density management vector group; Sa2-1, obtain the feeding management coefficient through the feeding management vector group, the feeding management vector group includes the feeding amount, feeding frequency and bait type; Sa2-2, obtains the medication management coefficient through the medication management vector group, and the medication management vector group includes drug type, medication dosage and medication frequency; Sa2-3, obtain the density management coefficient through the density management vector group, the density management vector group includes the breeding density and the pond division time; Sa2-4, integrate the feeding management coefficient, drug management coefficient and density management coefficient into aquaculture risk factors, and obtain the aquaculture risk assessment index through the aquaculture risk factor data set to evaluate the aquaculture risk status; Sa3 integrates the biological factor data set, biological status assessment index, aquaculture risk factor data set and aquaculture risk assessment index into a disease characteristic data set, and evaluates the risk level of aquatic diseases through the disease characteristic data set to achieve preliminary diagnosis and prediction.

3. The intelligent diagnosis method for aquatic diseases according to claim 2 is characterized in that: The specific process of the deep learning model is: Sb1, obtain water quality assessment coefficient through preliminary analysis of aquaculture water quality parameters; Aquaculture water quality parameters include water temperature, pH value and the content of water environmental compounds; Sb2, obtain the sediment assessment coefficient through preliminary analysis of environmental sediment parameters; Environmental sediment parameters include sediment thickness, organic matter content, and sulfide content; Sb3, constructs the correlation function between the water quality assessment coefficient, the bottom quality assessment coefficient and the disease characteristic data set, and generates the diagnosis and prediction results of aquatic diseases based on environmental characteristics, and performs refined diagnosis and prediction of aquatic diseases through the diagnosis and prediction results.

4. The intelligent diagnosis method for aquatic diseases according to claim 3 is characterized in that: The specific process of analyzing aquatic biological parameters is: Sa1-1, microbial coefficients were obtained by analyzing the microbial vector group; Label the microbial vector group as Vmic<ZLm0,SLm0> , where ZLm0 refers to the set of types of m0 microorganisms, and SLm0 refers to the set of numbers of m0 microorganisms; The type of any microorganism is marked as Zmic, and the number of microorganism Zmic is marked as Lmic; The microbial coefficient Mic is obtained comprehensively through the types Zmic and the number Lmic of m0 kinds of microorganisms; Sa1-2, obtain the aquatic biological coefficient by analyzing the aquatic biological vector group; Sampling monitoring is performed on the same batch of aquaculture organisms. It is preset to select m1 aquaculture organisms, mark any sampled aquaculture organism as H, monitor the weight Hg1, feed amount Hg2, swimming speed Hv and swimming time Ht of the aquaculture organism H, so as to obtain the growth value EA, diet value EB and activity value EC; Label the aquatic biological vector group as Vaqu<EAm1,EBm1,ECm1> , where EAm1 refers to the set of growth values ​​EA of m1 aquaculture organisms; EBm1 refers to the set of diet values ​​EB of m1 aquaculture organisms; ECm1 refers to the set of activity values ​​EC of m1 aquaculture organisms; The aquatic biological coefficient Aqu is obtained comprehensively through the aquatic biological vector group of m1 aquaculture organisms; Sa1-3, by combining the microbial coefficient Mic and the aquatic biological coefficient Aqu, the biological status assessment index ORG is obtained to evaluate the biological health status.

5. The intelligent diagnosis method for aquatic diseases according to claim 4 is characterized in that: The specific process of assessing the risk status of aquaculture is as follows: Sa2-1, obtain the feeding management coefficient through the feeding management vector group; Mark the bait management vector group as Wfee<ZRm2,LRm2,FRm2> , where ZRm2 refers to the set of m2 bait types, LRm2 refers to the set of m2 bait feeding amounts, and FRm2 refers to the set of m2 bait feeding frequencies; The type of any bait is marked as Rfee, and the amount and frequency of bait Rfee are marked as Lfee and Ffee respectively; The feeding management coefficient Fee is obtained comprehensively through the feeding amount Lfee and feeding frequency Ffee of m2 kinds of bait; Sa2-2, obtains the medication management coefficient through the medication management vector group, and the medication management vector group includes drug type, medication dosage and medication frequency; Label the medication management vector group as Wmed<ZDm3,LDm3,FDm3> , where ZDm3 refers to the set of m3 drug types, LDm3 refers to the set of m3 drug dosages, and FDm3 refers to the set of m3 drug usage frequencies; The type of any drug is marked as Dmed, and the dosage and frequency of drug Dmed are marked as Lmed and Fmed respectively; The medication management coefficient Med is obtained comprehensively through the dosage Lmed and medication frequency Fmed of m3 drugs; Sa2-3, obtain the density management coefficient through the density management vector group, and mark the density management vector group as Wden<NSσ,NSt> , where NSσ and NSt refer to the stocking density and pond division time, respectively; Set up a parameter evaluation model, input parameter i and its standard interval [Ig, If], and obtain the evaluation value Ui of parameter i by combining parameter i and its standard interval [Ig, If]; By combining the stocking density NSσ and the pond division time NSt, the stocking density NSσ and the pond division time NSt are substituted into the parameter evaluation model in turn to obtain the density management coefficient Den; Sa2-4, by combining the feeding management coefficient Fee, the medication management coefficient Med and the density management coefficient Den, the aquaculture risk assessment index RISK is obtained to evaluate the aquaculture risk status.

6. The intelligent diagnosis method for aquatic diseases according to claim 5, characterized in that: The specific process of preliminary analysis of aquaculture water quality parameters is as follows: Aquaculture water quality parameters include water temperature wd, pH value and the content of water environmental compounds; Set the standard intervals of water temperature wd, pH value and water environment compounds, mark the number of water environment compounds as N0, and mark any water environment compound as Hj; The water temperature wd, pH value, water environmental compounds and their corresponding standard ranges are substituted into the parameter evaluation model to output the evaluation value Uwd of the water temperature wd, the evaluation value UpH of the pH value, the evaluation value UHj of the water environmental compound Hj, and then comprehensively obtain the water quality evaluation coefficient WQ.

7. The intelligent diagnosis method for aquatic diseases according to claim 6 is characterized in that: The specific process of analyzing environmental sediment parameters is as follows: Set the standard ranges of sediment thickness hd, organic matter content and sulfide content, mark the number of organic matter types as N1, and mark any organic matter as OM; mark the number of sulfide types as N2, and mark any sulfide as SF; The standard ranges of sediment thickness hd, organic matter content and sulfide content are substituted into the parameter evaluation model to output the evaluation value Uhd of sediment thickness hd, the evaluation value Uom of organic matter OM and the evaluation value Usf of sulfide SF respectively, and then comprehensively obtain the bottom sediment evaluation coefficient BQ.

8. The intelligent diagnosis method for aquatic diseases according to claim 7 is characterized by: The specific process of generating the diagnosis and prediction results of aquatic diseases is as follows: Mark any aquatic disease category as Q, and fit the correlation function F0 of the aquatic disease category Q through the water quality assessment coefficient WQ, the bottom quality assessment coefficient BQ and the disease characteristic data set; The risk interval of the correlation function F0 is set, and the risk level of the category Q of aquatic diseases is evaluated by interval comparison.

Images