CN110956310A - Fish feed feeding amount prediction method and system based on feature selection and support vector - Google Patents
Fish feed feeding amount prediction method and system based on feature selection and support vector Download PDFInfo
- Publication number
- CN110956310A CN110956310A CN201911113222.3A CN201911113222A CN110956310A CN 110956310 A CN110956310 A CN 110956310A CN 201911113222 A CN201911113222 A CN 201911113222A CN 110956310 A CN110956310 A CN 110956310A
- Authority
- CN
- China
- Prior art keywords
- fish feed
- model
- data
- prediction
- feeding amount
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/02—Agriculture; Fishing; Mining
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A40/00—Adaptation technologies in agriculture, forestry, livestock or agroalimentary production
- Y02A40/80—Adaptation technologies in agriculture, forestry, livestock or agroalimentary production in fisheries management
- Y02A40/81—Aquaculture, e.g. of fish
Abstract
The invention discloses a fish feed feeding amount prediction method and system based on feature selection and support vectors, wherein a Lasso feature selection model is used for extracting a plurality of factors influencing large fish feed feeding amount, a gray prediction model of the extracted factors and the fish feed feeding amount is established for predicting the optimal fish feed feeding amount, the Lasso feature selection model is used for further screening the features, the requirements on the quality of the factors are further reduced, and a better effect is achieved.
Description
Technical Field
The disclosure relates to the field of aquaculture and the technical field of artificial intelligence, in particular to a fish feed input prediction method and system based on feature selection and support vectors.
Background
The bait is in a relatively important position in artificial culture and is a relatively large expense in aquaculture cost. In recent years, as the aquaculture industry is developed rapidly, particularly the scale of the fish culture industry is enlarged continuously, and some farmers cannot correctly master the feed feeding quantity, the unit yield is low, the diseases are more and the economic benefit is poor, so that the problem of feed waste in the fish culture process is more and more prominent, and the feed feeding becomes a key link in the aquaculture. Predicting and determining accurate feeding amount is a main measure for reducing feed waste, reducing feed coefficient and improving benefit. Only by correctly predicting and grasping the feeding quantity of the fish feed and scientifically feeding, the utilization rate of the feed can be improved, the occurrence of various fish diseases is reduced, and the yield and income increase of the fish pond are ensured.
In the prior art, a method based on a random forest regression model is used for predicting the feeding amount of fish feed. The method comprises the steps of constructing a sample data set, carrying out normalization pretreatment on the sample data, distributing the sample data set, constructing a random forest model, testing the model, evaluating the prediction precision and the like. However, the technique often increases the computational complexity, and the prediction result is less ideal due to lower prediction precision.
In the prior art, firstly, the annual feeding amount, the monthly feeding amount and the daily feeding amount are calculated by using the net weight gain multiple of fish and a feed coefficient, secondly, the daily feeding amount is determined according to the number of fish ponds and the ingestion condition of the fish, and generally, the feeding amount is determined by using a fixed arithmetic formula and the experience of some cultured fishes, and the two methods have the defects that the feeding amount is determined by neglecting the influence of various environmental factors, and the prediction is inaccurate due to the deviation of death and crime; the other technology is to use a random forest algorithm to predict the feeding amount, but the method has higher calculation complexity and the calculation result is often not accurate in imagination.
Disclosure of Invention
The fish feed feeding amount prediction method based on the Lasso characteristic selection and the support vector regression model is characterized in that data of various factors possibly influencing the fish feed feeding amount are collected, then the Lasso characteristic selection model is used for extracting a plurality of factors influencing the fish feed feeding amount to be larger, and a gray prediction model of the extracted factors and the fish feed feeding amount is established to predict the optimal fish feed feeding amount, so that the fish feed feeding amount can be predicted more accurately, and the fish feed feeding efficiency is effectively improved; in order to reduce the computational complexity, improve the prediction accuracy and provide a good environment for healthy and sustainable growth of fishes, the method is adopted to collect data of various factors possibly influencing the feeding amount of fish feed, then use a Lasso characteristic selection model to extract a plurality of factors influencing the feeding amount of the fish feed, and then establish a gray prediction model of the extracted factors and the feeding amount of the fish feed to predict the optimal feeding amount of the fish feed.
To achieve the above objects, according to one aspect of the present disclosure, there is provided a method for predicting fish feed dosage based on feature extraction and support vector, the method comprising the steps of:
step 1: collecting fish feed feeding data. The fish feed feeding data comprises: the protein content of the fish feed, the particle size of the fish feed, the weight, the water temperature, the dissolved oxygen amount, the pH value and the survival rate of the fish are respectively recorded as x1 *,x2 *,x3 *,x4 *,x5 *,x6 *,x7 *Recording the feeding amount of fish feed as y*Importing the data into an Excel table;
step 2: and (4) carrying out standardized processing on the fish feed feeding data. Data x for fish feed feeding1 *,x2 *,x3 *,x4 *,x5 *,x6 *,x7 *Feeding amount y of fish feed*Performing z-score standardization, also called standard deviation standardization, which gives the mean value and standard deviation of the original data to perform data standardization; the processed data are in accordance with the standard normal distribution, namely the mean value is 0, the standard deviation is 1, and the conversion function is as follows:
xi=(xi *-μ)/σ
where μ is the mean of all sample data, i is an integer between 1 and 7, and σ is the standard deviation of all sample data; finally obtaining x1,x2,x3,x4,x5,x6,x7,y。
And step 3: and analyzing the correlation of the data characteristics of the feeding amount of the fish feed. Using Pearson correlation coefficient ri(i-1, 2, …,7) is used to measure two features xiThe correlation (strength of linear correlation) between (i ═ 1,2, … 7) and y, respectively, is the simplest correlation coefficient, and has a value range of [ -1,1 [ - ]]. As shown in the formula:when 0 is present<r<When 1, it represents xi(i ═ 1,2, … 7) and y exhibit a positive correlation; when-1<r<At 0, represents xi(i ═ 1,2, … 7) and y exhibit a negative correlation. If r is 1, denotes xi(i ═ 1,2, … 7) and y are all positively correlated; if r is-1, denotes xi(i-1, 2, … 7) and y are all in a negative-positive correlation. The closer | r | is to 1, the more x is indicatediThe smaller the (i ═ 1,2, … 7) and y gap, the greater the correlation. According to the principle, the characteristic x with strong correlation with y is selectedi(i=1,2,…,n)(0<n is less than or equal to 7), and n is an integer between 0 and 7, and the characteristics need to be further screened because of the repetition of information among the characteristics.
And 4, step 4: and extracting key characteristics of the fish feed feeding amount prediction by using a Lasso characteristic selection model. Lasso is a contraction estimation method with the idea of reducing feature sets (price reduction). The Lasso method can compress the coefficients of the features and make some regression coefficients 0, thereby achieving the purpose of feature selection and being widely applied to model improvement and selection. The purpose of feature selection is realized by selecting a penalty function and borrowing Lasso thought and method. Model selection is a process of seeking sparse representation of the model, and this process can be accomplished by optimizing a "loss" + "penalty" function problem. The detailed implementation process is as follows:
respectively selecting the features x selected in step 3i(i=1,2,…,n)(0<n is less than or equal to 7) is substituted into the Lasso characteristic selection model to obtain the Lasso correlation coefficient valueAs shown in the formula:(0<n≤7s.t.i=1nβi<t. Where λ is a non-negative regular parameter, controls the complexity of the model, and t>0 is an adjustment parameter, and the compression of the overall regression coefficient can be realized by controlling the adjustment parameter t. The larger the lambda is, the greater the penalty degree of the linear model with more characteristics is, so as to finally obtain a model with less characteristics,referred to as penalty terms. And determining the parameter lambda by adopting a cross validation method, and selecting the lambda value with the minimum cross validation error. Finally, the model is re-fitted with all data according to the obtained lambda value. According to which the principle isThe characteristic of (a) is selected as a key characteristic for predicting the feeding amount of the fish feed.
And 5: the predicted values of the key features are calculated by a grey prediction model. The grey prediction is based on a grey model, here a grey prediction model (GM (1,1) model) is used. Suppose one of the features x extracted in step 4 is(0)={x(0)(i),i=1,2,…,m(0<m is less than or equal to n), establishing a gray prediction model as follows:
step 5.1, first, for x(0)Performing a first accumulation to obtain a first accumulation sequence x(1)={x(1)(k),k=0,1,2,…,m};
Step (ii) of5.2, for x(0)Establishing a first order linear differential equation of the formulaNamely, GM (1,1) model, where ∈ is the coefficient of progression and μ is the amount of gray effect;
Step 5.4, because the GM (1,1) model obtains the first accumulation amount, the data obtained by the GM (1,1) model is usedIs reduced intoI.e. x(0)The gray prediction model formula is
Finally, the predicted value of each key feature is obtained and recordedn is an integer between 0 and 7, and m is an integer between 0 and n.
Step 6: and predicting the fish feed feeding amount at the next moment according to the predicted value of each key characteristic through the SVR model.
And performing regression analysis on the data by using the SVR algorithm and adopting the idea of a support vector machine during fitting. Handle As a training set, among others, and R is a real number, training the prediction model, and updating the training sample in real time before predicting the next moment, namely adding the actual fish feed feeding amount and the selected principal component data at the previous moment and removing the most original data. For the sampleOutput f (c) according to the modeli) And true valueThe difference between them to calculate the loss, if and only ifThe loss is zero. Allowing f (c)i) Andwith a maximum deviation epsilon between them. Only whenThe loss is calculated. When in useConsidering the prediction to be accurate, finally obtaining the fish feed feeding amount at the next moment of the fish feed and recording as yn。
The present disclosure also provides a system for predicting fish feed dosage based on feature selection and support vector, the system comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, the processor executing the computer program to run in the units of the following system:
the data acquisition unit is used for acquiring fish feed feeding data;
the standardized processing unit is used for carrying out standardized processing on the fish feed feeding data;
the correlation analysis unit is used for analyzing the correlation of the fish feed feeding amount data characteristics;
the characteristic extraction unit is used for extracting key characteristics of the fish feed feeding amount prediction by using a Lasso characteristic selection model;
the predicted value calculating unit is used for calculating the predicted value of the key feature through the grey prediction model;
and the feeding amount prediction unit is used for predicting the feeding amount of the fish feed at the next moment according to the prediction values of the key characteristics through the SVR model.
The beneficial effect of this disclosure does: compared with the prior art, the method and the system for predicting the fish feed feeding amount based on the feature selection and the support vector can greatly improve the prediction precision of the fish feed feeding amount and predict the fish feed feeding amount in a shorter time. In the prior art, a random forest regression model is used for prediction, which is more convenient than the traditional technology, but when most of influencing factors are closely related to dependent variables, the key characteristics of fish feed prediction cannot be determined, so that the effect of reducing the calculation amount by the method cannot be achieved. The invention selects the model by using the Lasso characteristics, further screens the characteristics, further reduces the requirements on the quality of factors, achieves better effect, obtains the final predicted value by using the gray prediction model and the SVR algorithm, can obtain the fish feed feeding amount with high accuracy by the prediction model only by collecting less influence factor data sets, can help fishermen accurately put in the fish feed in real time, fully utilizes the feed, and achieves the effect of increasing the yield and increasing the efficiency and gain of the aquaculture industry.
Drawings
The foregoing and other features of the present disclosure will become more apparent from the detailed description of the embodiments shown in conjunction with the drawings in which like reference characters designate the same or similar elements throughout the several views, and it is apparent that the drawings in the following description are merely some examples of the present disclosure and that other drawings may be derived therefrom by those skilled in the art without the benefit of any inventive faculty, and in which:
FIG. 1 is a flow chart of a method for predicting fish feed dosage based on feature selection and support vector;
fig. 2 is a block diagram of a fish feed dosage prediction system based on feature selection and support vectors.
Detailed Description
The conception, specific structure and technical effects of the present disclosure will be clearly and completely described below in conjunction with the embodiments and the accompanying drawings to fully understand the objects, aspects and effects of the present disclosure. It should be noted that the embodiments and features of the embodiments in the present application may be combined with each other without conflict.
Referring to fig. 1, a flow chart of a method for predicting fish feed dosage based on feature extraction and support vector according to the present disclosure is shown, and the method for predicting fish feed dosage based on feature extraction and support vector according to an embodiment of the present disclosure is described below with reference to fig. 1.
The invention provides a fish feed dosage prediction method based on feature selection and support vectors, which specifically comprises the following steps:
step 1: collecting fish feed feeding data. The fish feed feeding data comprises: the protein content of the fish feed, the particle size of the fish feed, the weight of fish, the water temperature, the dissolved oxygen amount, the pH value and the survival rate of fish are respectively recorded as x1, x2, x3, x4, x5, x6 and x7, the feeding amount of the fish feed is recorded as y, and the data are imported into an Excel table;
step 2: and (4) carrying out standardized processing on the fish feed feeding data. Data x for fish feed feeding1 *,x2 *,x3 *,x4 *,x5 *,x6 *,x7 *Feeding amount y of fish feed*The z-score normalization, also called standard deviation normalization, is performed to give the raw data a uniform meanThe values (mean) and standard deviations (standard deviation) were used to normalize the data.
The processed data are in accordance with the standard normal distribution, namely the mean value is 0, the standard deviation is 1, and the conversion function is as follows:
xi=(xi *-μ)/σ
where μ is the mean of all sample data and σ is the standard deviation of all sample data. Finally obtaining x1,x2,x3,x4,x5,x6,x7,y。
And step 3: and analyzing the correlation of the data characteristics of the feeding amount of the fish feed. Using Pearson correlation coefficient ri(i-1, 2, …,7) is used to measure two features xiThe correlation (strength of linear correlation) between (i ═ 1,2, … 7) and y, respectively, is the simplest correlation coefficient, and has a value range of [ -1,1 [ - ]]. As shown in the formula:when 0 is present<r<When 1, it represents xi(i ═ 1,2, … 7) and y exhibit a positive correlation; when-1<r<At 0, represents xi(i ═ 1,2, … 7) and y exhibit a negative correlation. If r is 1, denotes xi(i ═ 1,2, … 7) and y are all positively correlated; if r is-1, denotes xi(i-1, 2, … 7) and y are all in a negative-positive correlation. The closer | r | is to 1, the more x is indicatediThe smaller the (i ═ 1,2, … 7) and y gap, the greater the correlation. According to the principle, the characteristic x with strong correlation with y is selectedi(i=1,2,…,n)(0<n is less than or equal to 7), and n is an integer between 0 and 7, and the characteristics need to be further screened because of the repetition of information among the characteristics.
And 4, step 4: and extracting key characteristics of the fish feed feeding amount prediction by using a Lasso characteristic selection model. Lasso is a contraction estimation method with the idea of reducing feature sets (price reduction). The Lasso method can compress the coefficients of the features and make some regression coefficients 0, thereby achieving the purpose of feature selection and being widely applied to model improvement and selection. The purpose of feature selection is realized by selecting a penalty function and borrowing Lasso thought and method. Model selection is a process of seeking sparse representation of the model, and this process can be accomplished by optimizing a "loss" + "penalty" function problem. The detailed implementation process is as follows:
respectively selecting the features x selected in step 3i(i=1,2,…,n)(0<n is less than or equal to 7) is substituted into the Lasso characteristic selection model to obtain the Lasso correlation coefficient valueAs shown in the formula:n≤7,i=1nβi<t. Where λ is a non-negative regular parameter, controls the complexity of the model, and t>0 is an adjustment parameter, and the compression of the overall regression coefficient can be realized by controlling the adjustment parameter t. The larger the lambda is, the greater the penalty degree of the linear model with more characteristics is, so as to finally obtain a model with less characteristics,referred to as penalty terms. And determining the parameter lambda by adopting a cross validation method, and selecting the lambda value with the minimum cross validation error. Finally, the model is re-fitted with all data according to the obtained lambda value. According to which the principle isThe characteristic of (a) is selected as a key characteristic for predicting the feeding amount of the fish feed.
And 5: the predicted values of the key features are calculated by a grey prediction model. The grey prediction is based on a grey model, here a grey prediction model (GM (1,1) model) is used. Suppose x, one of the features extracted in step 4(0)={x(0)(i),i=1,2,…,m(0<m is less than or equal to n), establishing a gray prediction model as follows:
step 5.1, first, for x(0)Performing a first accumulation to obtain a first accumulation sequence x(1)={x(1)(k),k=0,1,2,…,m};
Step 5.2, for x(0)Establishing a first order linear differential equation of the formulaNamely, GM (1,1) model, where ∈ is the coefficient of progression and μ is the amount of gray effect;
Step 5.4, because the GM (1,1) model obtains the first accumulation amount, the data obtained by the GM (1,1) model is usedIs reduced intoI.e. x(0)The gray prediction model formula is
Finally, the predicted value of each key feature is obtained and recordedn is an integer between 0 and 7, and m is an integer between 0 and n.
Step 6: and predicting the fish feed feeding amount at the next moment according to the predicted value of each key characteristic through the SVR model.
And performing regression analysis on the data by using the SVR algorithm and adopting the idea of a support vector machine during fitting. Handle As a training set, among others, training the prediction model, and updating the training sample in real time before predicting the next moment, namely adding the actual fish feed feeding amount and the selected main component data at the previous moment and removing the most original data. For the sampleOutput f (c) according to the modeli) And true valueThe difference between them to calculate the loss, if and only ifThe loss is zero. Allow forAndthere is at most an epsilon deviation between them. Only whenThe loss is calculated. When in useConsidering the prediction to be accurate, finally obtaining the fish feed feeding amount at the next moment of the fish feed and recording as yn。
Fig. 2 is a structural diagram of a fish feed dosage prediction system based on feature selection and a support vector according to an embodiment of the present disclosure, and the fish feed dosage prediction system based on feature selection and a support vector according to the embodiment includes: a processor, a memory and a computer program stored in and executable on the processor, the processor when executing the computer program implementing the steps in the above-described embodiment of a system for fish feed dosage prediction based on feature extraction and support vectors.
The system comprises: a memory, a processor, and a computer program stored in the memory and executable on the processor, the processor executing the computer program to run in the units of the following system:
the data acquisition unit is used for acquiring fish feed feeding data;
the standardized processing unit is used for carrying out standardized processing on the fish feed feeding data;
the correlation analysis unit is used for analyzing the correlation of the fish feed feeding amount data characteristics;
the characteristic extraction unit is used for extracting key characteristics of the fish feed feeding amount prediction by using a Lasso characteristic selection model;
the predicted value calculating unit is used for calculating the predicted value of the key feature through the grey prediction model;
and the feeding amount prediction unit is used for predicting the feeding amount of the fish feed at the next moment according to the prediction values of the key characteristics through the SVR model.
The fish feed feeding amount prediction system based on feature selection and support vectors can be operated in computing equipment such as desktop computers, notebooks, palm computers and cloud servers. The fish feed dosage prediction system based on feature extraction and support vector can be operated by a system comprising, but not limited to, a processor and a memory. It will be appreciated by those skilled in the art that the examples are merely illustrative of a feature selection and support vector based fish feed dosage prediction system and do not constitute a limitation of a feature selection and support vector based fish feed dosage prediction system, and may include more or less components than the features of the features selection and support vector based fish feed dosage prediction system, or some components in combination, or different components, for example, the feature selection and support vector based fish feed dosage prediction system may further include input and output devices, network access devices, buses, and the like.
The processor may be a Central Processing Unit (CPU), other general purpose processor, a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field Programmable Gate Array (FPGA) or other programmable logic device, discrete gate or transistor logic, discrete hardware components, etc. The general purpose processor may be a microprocessor or the processor may be any conventional processor or the like, said processor being the control centre of the characteristic selection and support vector based fish food dosage prediction system operating system, the various parts of the overall characteristic selection and support vector based fish food dosage prediction system operable system being connected by means of various interfaces and lines.
The memory may be used for storing the computer programs and/or modules, and the processor may be adapted to implement the various functions of the system for fish feed dosage prediction based on feature selection and support vectors by running or executing the computer programs and/or modules stored in the memory and by invoking data stored in the memory. The memory may mainly include a storage program area and a storage data area, wherein the storage program area may store an operating system, an application program required by at least one function (such as a sound playing function, an image playing function, etc.), and the like; the storage data area may store data (such as audio data, a phonebook, etc.) created according to the use of the cellular phone, and the like. In addition, the memory may include high speed random access memory, and may also include non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) card, a flash memory card (FlashCard), at least one disk storage device, a flash memory device, or other volatile solid state storage device.
While the present disclosure has been described in considerable detail and with particular reference to a few illustrative embodiments thereof, it is not intended to be limited to any such details or embodiments or any particular embodiments, but it is to be construed as effectively covering the intended scope of the disclosure by providing a broad, potential interpretation of such claims in view of the prior art with reference to the appended claims. Furthermore, the foregoing describes the disclosure in terms of embodiments foreseen by the inventor for which an enabling description was available, notwithstanding that insubstantial modifications of the disclosure, not presently foreseen, may nonetheless represent equivalent modifications thereto.
Claims (7)
1. A fish feed dosage prediction method based on feature selection and support vectors is characterized by comprising the following steps:
step 1: collecting fish feed feeding data;
step 2: carrying out standardized processing on fish feed feeding data;
and step 3: analyzing the correlation of the data characteristics of the feeding amount of the fish feed;
and 4, step 4: extracting key characteristics of fish feed feeding amount prediction by using a Lasso characteristic selection model;
and 5: calculating the predicted value of the key feature through a grey prediction model;
step 6: and predicting the fish feed feeding amount at the next moment according to the predicted value of each key characteristic through the SVR model.
2. The method for predicting fish feed dosage based on feature extraction and support vector as claimed in claim 1, wherein in step 2, the method for normalizing the fish feed feeding data comprises: data x for fish feed feeding1 *,x2 *,x3 *,x4 *,x5 *,x6 *,x7 *Feeding amount y of fish feed*Performing z-score standardization, also called standard deviation standardization, which gives the mean value and standard deviation of the original data to perform data standardization; the processed data are in accordance with the standard normal distribution, namely the mean value is 0, the standard deviation is 1, and the conversion function is as follows:
xi=(xi *-μ)/σ
where μ is of all sample dataMean, i is an integer between 1 and 7, σ is the standard deviation of all sample data; finally obtaining x1,x2,x3,x4,x5,x6,x7,y。
3. The method for fish feed dosage prediction based on feature extraction and support vector of claim 2, wherein in step 3, the method for analyzing the correlation of the features of the fish feed dosage data comprises: using Pearson correlation coefficient ri(i-1, 2, …,7) is used to measure two features xiThe correlation between (i ═ 1,2, … 7) and y, i.e. the strength of linear correlation, is in the range of [ -1,1]As shown in the formula:when 0 is present<r<When 1, it represents xi(i ═ 1,2, … 7) and y exhibit a positive correlation; when-1<r<At 0, represents xi(i ═ 1,2, … 7) and y exhibit a negative correlation; if r is 1, denotes xi(i ═ 1,2, … 7) and y are all positively correlated; if r is-1, denotes xi(i-1, 2, … 7) and y are all positively and negatively correlated, the closer | r | is to 1, indicating that x is positivei(i is 1,2, … 7) and the smaller the difference between y is, the more relevant is, and the characteristic x which is strongly relevant to y is selectedi(i ═ 1,2, …, n), n is an integer between 0 and 7.
4. The method for fish feed dosage prediction based on feature selection and support vector as claimed in claim 3, wherein in step 4, the method for extracting key features of fish feed dosage prediction by using Lasso feature selection model comprises: respectively combining the features xi(i=1,2,…,n)(0<n is less than or equal to 7) is substituted into the Lasso characteristic selection model to obtain the Lasso correlation coefficient valueAs shown in the formula: wherein, the lambda is a nonnegative regular parameter and controls the complexity of the model, the larger the lambda is, the larger the punishment intensity to the linear model with more characteristics is, thereby finally obtaining a model with less characteristics,the parameter lambda is determined by adopting a cross validation method, the lambda value with the minimum cross validation error is selected, and finally, the model is re-fitted by using all data according to the obtained lambda valueThe characteristic of (a) is selected as a key characteristic for predicting the feeding amount of the fish feed.
5. The method for predicting fish feed dosage based on feature extraction and support vector as claimed in claim 4, wherein in step 5, the method for calculating the predicted value of the key feature by using the grey prediction model comprises the following steps: the grey prediction is based on a grey model, here the GM (1,1) model is used, if one of the key features is: x is the number of(0)={x(0)(i),i=1,2,…,m}(0<m is less than or equal to n), establishing a gray prediction model as follows:
step 5.1, first, for x(0)Performing a first accumulation to obtain a first accumulation sequence x(1)={x(1)(k),k=0,1,2,…,m};
Step 5.2, for x(0)Establishing a first order linear differential equation of the formulaNamely, GM (1,1) model, where ∈ is the coefficient of progression and μ is the amount of gray effect;
Step 5.4, because the GM (1,1) model obtains the first accumulation amount, the data obtained by the GM (1,1) model is usedIs reduced intoI.e. x(0)The gray prediction model formula is
6. The method for predicting fish feed feeding amount based on feature extraction and support vector as claimed in claim 5, wherein in step 6, the method for predicting the fish feed feeding amount at the next moment according to the predicted value of each key feature by SVR model comprises: will be provided withAs a training set, among others,training the prediction model, and updating a training sample in real time before predicting at the next moment, namely adding the actual fish feed feeding amount and the selected main component data at the previous moment and removing the most original data; for the sampleOutput f (c) according to the modeli) And true valueThe difference between them to calculate the loss, if and only ifWhen the loss is zero; allowing f (c)i) Andthere is a maximum deviation of epsilon between; only whenCalculating the loss only when the loss is not calculated; when in useConsidering the prediction to be accurate, finally obtaining the fish feed feeding amount at the next moment of the fish feed and recording as yn。
7. Fish feed dosage prediction system based on feature selection and support vector, characterized in that the system comprises: a memory, a processor, and a computer program stored in the memory and executable on the processor, the processor executing the computer program to run in the units of the following system:
the data acquisition unit is used for acquiring fish feed feeding data;
the standardized processing unit is used for carrying out standardized processing on the fish feed feeding data;
the correlation analysis unit is used for analyzing the correlation of the fish feed feeding amount data characteristics;
the characteristic extraction unit is used for extracting key characteristics of the fish feed feeding amount prediction by using a Lasso characteristic selection model;
the predicted value calculating unit is used for calculating the predicted value of the key feature through the grey prediction model;
and the feeding amount prediction unit is used for predicting the feeding amount of the fish feed at the next moment according to the prediction values of the key characteristics through the SVR model.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201911113222.3A CN110956310B (en) | 2019-11-14 | 2019-11-14 | Fish feed dosage prediction method and system based on feature selection and support vector |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201911113222.3A CN110956310B (en) | 2019-11-14 | 2019-11-14 | Fish feed dosage prediction method and system based on feature selection and support vector |
Publications (2)
Publication Number | Publication Date |
---|---|
CN110956310A true CN110956310A (en) | 2020-04-03 |
CN110956310B CN110956310B (en) | 2023-04-28 |
Family
ID=69977253
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201911113222.3A Active CN110956310B (en) | 2019-11-14 | 2019-11-14 | Fish feed dosage prediction method and system based on feature selection and support vector |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110956310B (en) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112085926A (en) * | 2020-08-03 | 2020-12-15 | 佛山科学技术学院 | River water pollution early warning method and system |
CN112101658A (en) * | 2020-09-14 | 2020-12-18 | 安徽工业大学 | Fattening pig breeding feed consumption prediction method |
CN112819217A (en) * | 2021-01-27 | 2021-05-18 | 合肥综合性国家科学中心人工智能研究院(安徽省人工智能实验室) | Method, system and storage medium for predicting main influence factors of mobile source pollution emission |
CN113555004A (en) * | 2021-07-15 | 2021-10-26 | 复旦大学 | Voice depression state identification method based on feature selection and transfer learning |
CN113854221A (en) * | 2021-10-29 | 2021-12-31 | 广州市蓝得生命科技有限公司 | Intelligent feeding control system |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2000003586A2 (en) * | 1998-07-15 | 2000-01-27 | Eco-Fish Ltd. | Fish feeding control in aquaculture on the basis of sound emitted by fish |
CN105511346A (en) * | 2015-12-01 | 2016-04-20 | 中国水产科学研究院南海水产研究所 | Fish bait casting quantity control system for deep-water cage culture |
CN108805176A (en) * | 2018-05-21 | 2018-11-13 | 青岛农业大学 | A kind of fish meal feeding volume prediction technique returned based on random forest |
US20190021292A1 (en) * | 2017-07-21 | 2019-01-24 | RoboGardens LLC | System and method for adaptive aquatic feeding based on image processing |
CN109636007A (en) * | 2018-11-20 | 2019-04-16 | 佛山科学技术学院 | A kind of water demands forecasting method and device based on big data |
-
2019
- 2019-11-14 CN CN201911113222.3A patent/CN110956310B/en active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2000003586A2 (en) * | 1998-07-15 | 2000-01-27 | Eco-Fish Ltd. | Fish feeding control in aquaculture on the basis of sound emitted by fish |
CN105511346A (en) * | 2015-12-01 | 2016-04-20 | 中国水产科学研究院南海水产研究所 | Fish bait casting quantity control system for deep-water cage culture |
US20190021292A1 (en) * | 2017-07-21 | 2019-01-24 | RoboGardens LLC | System and method for adaptive aquatic feeding based on image processing |
CN108805176A (en) * | 2018-05-21 | 2018-11-13 | 青岛农业大学 | A kind of fish meal feeding volume prediction technique returned based on random forest |
CN109636007A (en) * | 2018-11-20 | 2019-04-16 | 佛山科学技术学院 | A kind of water demands forecasting method and device based on big data |
Non-Patent Citations (6)
Title |
---|
刘晓宁: "基于Lasso特征选择的方法比较", 《安徽电子信息职业技术学院学报》 * |
张启敏: "灰色预测模型", 《宁夏大学学报(自然科学版)》 * |
朱庆国等: "杂交鲍苗微囊饲料投喂量测算方法", 《福建农业学报》 * |
董雁萍: "支持向量机预测模型的构建及其应用", 《中国优秀博硕士学位论文全文数据库(硕士)》 * |
陈彩文: "基于计算机视觉的鱼群摄食行为分析研究", 《中国优秀博硕士学位论文全文数据库(硕士)》 * |
陈碧晗: "数据挖掘中的变量选择问题研究", 《中国优秀博硕士学位论文全文数据库(硕士)》 * |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112085926A (en) * | 2020-08-03 | 2020-12-15 | 佛山科学技术学院 | River water pollution early warning method and system |
CN112101658A (en) * | 2020-09-14 | 2020-12-18 | 安徽工业大学 | Fattening pig breeding feed consumption prediction method |
CN112819217A (en) * | 2021-01-27 | 2021-05-18 | 合肥综合性国家科学中心人工智能研究院(安徽省人工智能实验室) | Method, system and storage medium for predicting main influence factors of mobile source pollution emission |
CN113555004A (en) * | 2021-07-15 | 2021-10-26 | 复旦大学 | Voice depression state identification method based on feature selection and transfer learning |
CN113854221A (en) * | 2021-10-29 | 2021-12-31 | 广州市蓝得生命科技有限公司 | Intelligent feeding control system |
Also Published As
Publication number | Publication date |
---|---|
CN110956310B (en) | 2023-04-28 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110956310A (en) | Fish feed feeding amount prediction method and system based on feature selection and support vector | |
Klibisz et al. | Fast, simple calcium imaging segmentation with fully convolutional networks | |
Peng et al. | Automatic image analysis for gene expression patterns of fly embryos | |
CN111382616B (en) | Video classification method and device, storage medium and computer equipment | |
CN110264407B (en) | Image super-resolution model training and reconstruction method, device, equipment and storage medium | |
Chen et al. | Multi-threshold image segmentation of maize diseases based on elite comprehensive particle swarm optimization and otsu | |
CN111047073B (en) | Aquaculture water quality prediction method and system based on neural network | |
Yang et al. | Detection and classification of damaged wheat kernels based on progressive neural architecture search | |
Wang et al. | A lightweight CNN-based model for early warning in sow oestrus sound monitoring | |
Suo et al. | Casm-amfmnet: A network based on coordinate attention shuffle mechanism and asymmetric multi-scale fusion module for classification of grape leaf diseases | |
CN114708212A (en) | Heart image segmentation method based on SEA-Unet | |
Kiratiratanapruk et al. | Automatic detection of rice disease in images of various leaf sizes | |
Ullah et al. | Automatic diseases detection and classification in maize crop using convolution neural network | |
Klenovšek et al. | An ontogenetic perspective on the study of sexual dimorphism, phylogenetic variability, and allometry of the skull of European ground squirrel, Spermophilus citellus (Linnaeus, 1766) | |
Lee et al. | Prediction of average daily gain of swine based on machine learning | |
CN107480630B (en) | Method for zoning forest ecological function by using remote sensing technology | |
CN113011532A (en) | Classification model training method and device, computing equipment and storage medium | |
CN113066528A (en) | Protein classification method based on active semi-supervised graph neural network | |
Gong et al. | A superpixel segmentation algorithm based on differential evolution | |
CN114219051B (en) | Image classification method, classification model training method and device and electronic equipment | |
CN112085926B (en) | River water pollution early warning method and system | |
CN113011669B (en) | Method for predicting monthly stock quantity of live pigs | |
CN114897884A (en) | No-reference screen content image quality evaluation method based on multi-scale edge feature fusion | |
CN111738410B (en) | Beef individual growth curve acquisition method, device and storage medium | |
CN115358157B (en) | Prediction analysis method and device for litter size of individual litters and electronic equipment |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |