Poultry multi-stage feed net energy in-vitro prediction method and device based on factorial screening

Abstract

The invention discloses a poultry multi-stage feed net energy in-vitro prediction method and device based on factorial screening, and relates to the technical field of animal husbandry. The method comprises the following steps: collecting and screening poultry feed raw material samples; analyzing the factors of the raw material components influencing the net energy value by adopting a factorization method to obtain a key component set; based on poultry feed raw material samples, determining the content of key components of each feed raw material and net energy values of the poultry feed raw material samples in different growth stages of poultry; based on the measured content of the key components of each feed raw material and the net energy values of the poultry feed raw material samples in different growth stages of poultry, performing principal component analysis to obtain predictive factors; constructing target net energy prediction equations of different growth stages of the poultry by using a linear regression analysis method based on the prediction factors; and performing net energy prediction on the poultry feed raw materials to be detected based on the target net energy prediction models of different growth stages of the poultry to obtain predicted net energy values of different growth stages of the poultry.

Description

Technical Field

[0001] The present application relates to the field of livestock technology, and particularly to a method and device for in vitro prediction of net energy of poultry multi-stage feed based on factorial screening. Background Art

[0002] Energy is the primary nutrient considered in feed formulation in poultry production. Metabolizable energy is the most widely used effective energy system in current feed formulation. Metabolizable energy refers to the remaining energy after subtracting fecal energy, urinary energy, and the energy of combustible gases in the digestive tract from the gross energy. Compared with the metabolizable energy system, the net energy system can better evaluate the effective energy value of poultry feed, which is the ultimate goal of poultry feed energy evaluation and also the future development direction. Net Energy (NE) refers to the energy remaining after subtracting the heat increment (HI) in the body from the energy obtained by animals from feed, which is the proportion of energy truly utilized by animals in feed and an important indicator for measuring the quality of nutritional supply of feed for animal growth and production. Precise evaluation of the net energy (NE) of feed raw materials is of great significance for formulating feed economically, reducing feed costs, and increasing breeding benefits. Evaluating the net energy of poultry feed is of great value for improving poultry production performance and feed utilization efficiency, can accurately reflect the actual energy requirements of poultry, and save feed costs in actual production. Clarifying the net energy of feed raw materials and designing precise diet formulations are also of great significance for reducing scarce feed resources such as corn and soybean meal and reducing feed costs.

[0003] Currently, the net energy assessment of laying hen feed raw materials usually determines the net energy value of feed raw materials through animal experiments and constructs a reference database for poultry net energy. However, factors such as the source and variety of feed raw materials have a great impact on net energy, and there may also be differences in net energy values between different batches of feed raw materials. The reference net energy values in the poultry net energy database do not consider the above influencing factors and batch differences. In addition, the reference net energy values in this database only consider one growth stage of poultry, that is, it is defaulted that the net energy values of all growth stages of poultry are the same, but there are significant differences in the net energy values of the same feed at different growth stages of poultry. Therefore, it is difficult for the existing technology to accurately predict the net energy value of feed raw materials and impossible to achieve precise formulation of poultry feed.

[0004] In addition, the method of obtaining the measured net energy value through animal experiments requires animal experiments, which are expensive, time-consuming, and cumbersome. And almost all of the existing prediction methods for the net energy of poultry feed raw materials are based on traditional conventional nutrient components. However, each conventional nutrient actually includes several nutrients or components, and their physicochemical properties vary greatly, and their contributions to net energy are different, resulting in insufficient prediction accuracy for net energy. Summary of the Invention

[0005] In view of this, the present application provides a method and device for in vitro prediction of net energy of poultry multi-stage feed based on factorial screening, with the main purpose of being able to reduce the detection cost, improve the prediction accuracy of net energy of poultry feed raw materials, and thus be able to achieve the economic and accurate formulation of poultry feed.

[0006] According to the first aspect of the present application, there is provided a method for in vitro prediction of net energy of poultry multi-stage feed based on factorial screening, the method comprising:

[0007] Collect and screen poultry feed raw material samples; wherein, the poultry feed raw material samples include feed raw materials from different origins, different times, different processing plants, and different batches;

[0008] Use the factorial method to factorize the raw material components affecting the net energy value to obtain a key component set, wherein the key component set includes key components of each feed raw material, and the key components of each feed raw material respectively belong to the components before factoring, primary key components, first-level key components, second-level key components, and third-level key components;

[0009] Based on the poultry feed raw material samples, determine the contents of the key components of each feed raw material, and the net energy values of the poultry feed raw material samples at different growth stages of poultry;

[0010] Based on the determined contents of the key components of each feed raw material and the net energy values of the poultry feed raw material samples at different growth stages of poultry, perform principal component analysis to obtain prediction factors;

[0011] Based on the prediction factors, use the linear regression analysis method to construct candidate net energy prediction equations for different growth stages of poultry;

[0012] Based on the determination coefficients and significances corresponding to the candidate net energy prediction equations for different growth stages of poultry, screen the net energy prediction equations for different growth stages of poultry;

[0013] Verify the prediction accuracy of the net energy prediction equations for different growth stages of poultry, and according to the verification results, determine the target net energy prediction models for different growth stages of poultry;

[0014] Based on the target net energy prediction models for different growth stages of poultry, perform net energy prediction on the to-be-tested poultry feed raw materials to obtain the predicted net energy values for different growth stages of poultry.

[0015] According to the second aspect of the present application, there is provided a device for in vitro prediction of net energy of poultry multi-stage feed based on factorial screening, the device comprising:

[0016] A collection unit for collecting and screening samples of poultry feed raw materials; wherein, the poultry feed raw material samples include feed raw materials from different origins, at different times, from different processing factories, and of different batches.

[0017] A factorial unit for factoring the raw material components affecting the net energy value by the factorial method to obtain a key component set, wherein the key component set includes the key components of each feed raw material, and the key components of each feed raw material respectively belong to the components before factoring, primary key components, first-level key components, second-level key components, and third-level key components.

[0018] A determination unit for determining the contents of the key components of each feed raw material and the net energy value of the poultry feed raw material samples at different growth stages of poultry based on the poultry feed raw material samples.

[0019] An analysis unit for performing principal component analysis based on the determined contents of the key components of each feed raw material and the net energy value of the poultry feed raw material samples at different growth stages of poultry to obtain prediction factors.

[0020] A construction unit for constructing candidate net energy prediction equations for different growth stages of poultry by using the linear regression analysis method based on the prediction factors.

[0021] A screening unit for screening the net energy prediction equations for different growth stages of poultry based on the determination coefficients and significance levels corresponding to the candidate net energy prediction equations for different growth stages of poultry.

[0022] A verification unit for verifying the prediction accuracy of the net energy prediction equations for different growth stages of poultry and determining the target net energy prediction models for different growth stages of poultry according to the verification results.

[0023] A prediction unit for predicting the net energy of the to-be-tested poultry feed raw materials based on the target net energy prediction models for different growth stages of poultry to obtain the predicted net energy values for different growth stages of poultry.

[0024] According to the third aspect of the present application, there is provided a storage medium on which a computer program is stored, and when the program is executed by a processor, the above-mentioned in vitro prediction method for the net energy of multi-stage poultry feed based on factorial screening is implemented.

[0025] According to the fourth aspect of the present application, there is provided an electronic device including a storage medium, a processor, and a computer program stored on the storage medium and executable on the processor, and when the processor executes the program, the above-mentioned in vitro prediction method for the net energy of multi-stage poultry feed based on factorial screening is implemented.

[0026] With the above technical solution, a method and device for in vitro prediction of net energy of poultry multi-stage feed based on factorial screening provided by the present application, compared with the prior art, by introducing feed raw materials of different products, different times, different processing plants and different batches when establishing the target net energy prediction model, the influence of factors such as the source and variety of feed raw materials on net energy can be fully considered, so that the variability of the modeling samples can be increased, and the prediction accuracy of the net energy of poultry feed raw materials can be improved. At the same time, by constructing the target net energy prediction model for different growth stages of poultry, the net energy value of feed raw materials at different growth stages of poultry can be accurately predicted. In addition, by analyzing the relationship between the key component set and net energy and determining the prediction factors, dimensionality reduction processing can be carried out, so that the number of detection indexes and the detection cost can be reduced. Further, by double-verifying the net energy prediction equation, the accuracy and reliability of the finally determined target net energy equation can be ensured. Further, by determining the pre-factorial components, primary key components, first-level key components, second-level key components and third-level key components, prediction factors with high correlation with net energy can be fully excavated, so that the prediction accuracy of the net energy of poultry feed raw materials can be further improved, and further a more reliable method can be provided for promoting the rapid in vitro determination of the net energy of feed raw materials, accurately formulating poultry diets in real-time dynamic manner, and reducing the substitution of corn and soybean meal.

[0027] The above description is only an overview of the technical solution of the present application. In order to be able to understand the technical means of the present application more clearly, it can be implemented according to the content of the specification. And in order to make the above and other objects, features and advantages of the present application more obvious and understandable, the specific embodiments of the present application are hereinafter specifically exemplified. Brief Description of the Drawings

[0028] The drawings described herein are used to provide a further understanding of the present application, and constitute a part of the present application. The schematic embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation to the present application. In the drawings:

[0029] Figure 1 A schematic flow chart of a method for in vitro prediction of net energy of poultry multi-stage feed based on factorial screening provided by an embodiment of the present application is shown;

[0030] Figure 2 A schematic flow chart of the determination of the factor loading matrix provided by an embodiment of the present application is shown;

[0031] Figure 3 A schematic diagram of the main component load of cottonseed meal provided by an embodiment of the present application is shown;

[0032] Figure 4 A schematic diagram of the verification result of the accuracy of the prediction equation of cottonseed meal at the peak egg production period provided by an embodiment of the present application is shown;

[0033] Figure 5 It shows a schematic diagram of the linear distribution of the net energy predicted value and the true value of cottonseed meal provided in the embodiment of the present application during the peak laying period;

[0034] Figure 6 It shows a schematic diagram of the main component load of rapeseed meal provided in the embodiment of the present application;

[0035] Figure 7 It shows a schematic diagram of the verification result of the accuracy rate of the prediction equation of rapeseed meal provided in the embodiment of the present application during the growth period of laying hens;

[0036] Figure 8 It shows a schematic diagram of the linear distribution of the net energy predicted value and the true value of rapeseed meal provided in the embodiment of the present application during the growth period of laying hens;

[0037] Figure 9 It shows a schematic diagram of the structure of an in vitro prediction device for the net energy of multi-stage poultry feed based on factorial screening provided in the embodiment of the present application. Detailed implementation manners

[0038] In the following, the present application will be described in detail with reference to the drawings and in combination with embodiments. It should be noted that, without conflict, the embodiments in the present application and the features in the embodiments can be combined with each other.

[0039] Since the reference net energy values in the poultry net energy database do not consider influencing factors such as the source and variety of feed raw materials and batch differences, and the reference net energy values for all growth stages of poultry in this database are defaulted to be the same without considering the net energy differences in different growth stages of poultry, it is difficult to accurately predict the net energy value of feed raw materials, and thus it is impossible to achieve the precise formulation of poultry feed.

[0040] To solve the above problems, Embodiment 1 of the present invention provides an in vitro prediction method for the net energy of multi-stage poultry feed based on factorial screening, as Figure 1 shown, this method includes:

[0041] Step 10: Collect and screen samples of poultry feed raw materials.

[0042] Among them, the poultry feed raw material samples include feed raw materials from different origins, different times, different processing factories, and different batches. The feed raw materials include at least one raw material that can supply energy in the poultry feed formula, such as grain processing by-products like cottonseed meal and rapeseed meal. Poultry includes at least one of chicken, duck, goose, turkey, ostrich, quail, and pigeon.

[0043] The embodiments of the present invention are mainly applicable to the in vitro prediction of the net energy of poultry feed raw materials. The execution subject of the embodiments of the present invention is a device or equipment capable of performing in vitro prediction of the net energy of poultry feed raw materials.

[0044] To ensure the prediction accuracy of the net energy prediction equation, the embodiments of the present invention will pre-collect feed raw materials from different origins, at different times, from different processing plants, and in different batches as samples, so as to enrich the modeling sample library. To further reduce the workload and costs, the embodiments of the present invention can also screen poultry feed raw materials. Specifically, the contents of several conventional components (such as gross energy, crude fat, starch, neutral detergent fiber) in the poultry feed raw material samples can be measured first, and then based on the contents of the above components, the correlation coefficients between the above components and the net energy are determined, as shown in Table 1. Then, according to the correlation coefficients, the conventional indicators with relatively high correlations are screened out from the above components, and based on the contents of the conventional indicators, a rough regression calculation of the net energy is performed, so as to select poultry feed raw material samples with large differences in composition and net energy according to the calculation results as modeling samples for later net energy determination.

[0045] It should be noted that if the user lacks the correlation coefficient data between the conventional components and the net energy, the correlation coefficient or regression relationship between the conventional components and the metabolic energy can be used for rough screening of the samples.

[0046] For example, first collect 30 cottonseed meal samples from different seasons, different products, and different batches, and then select 10 cottonseed meal raw material samples with large differences in composition and net energy as modeling samples for later net energy determination. The 10 selected cottonseed meal feed raw material samples are from Xinjiang (cottonseed meal 1), Shandong (cottonseed meal 2), Xinjiang (cottonseed meal 3), Shandong (cottonseed meal 4), Shandong (cottonseed meal 5), Henan (cottonseed meal 6), Xinjiang (cottonseed meal 7), Hebei (cottonseed meal 8), Xinjiang (cottonseed meal 9), and Shandong (cottonseed meal 10). Thus, by reducing the number of modeling samples for net energy determination, manpower, material resources, financial resources, and time can be saved.

[0047] It should be noted that in order to make the net energy prediction accuracy have a broader representativeness, within an acceptable range of costs, etc., the poultry feed raw material samples with net energy variation should be as many as possible, generally not less than 10 kinds.

[0048] Table 1 Contents of various components in the poultry cottonseed meal raw material samples and their correlation coefficients (r) with the net energy

[0049] Item Crude Protein Ether Extract Ash Crude Fiber Neutral Detergent Fiber Acid Detergent Fiber Gross Energy Net Energy Items CP EE Ash CF NDF ADF GE Net Energy Net Energy 0.292 -0.005 0.113 -0.091 <![CDATA[-0.936 ** > -0.262 <![CDATA[0.961 ** > 1.000

[0050] Among them, "**" indicates that the correlation is significant at a confidence level of 99%, and "*" indicates that the correlation is significant at a confidence level (two-sided) of 95%.

[0051] Step 20: Use the factorial method to factor the raw material components affecting the net energy value to obtain the key component set.

[0052] Among them, the key ingredient set includes key ingredients of each feed raw material, and the key ingredients of each feed raw material respectively belong to pre-factorial ingredients, primary key ingredients, first-level key ingredients, second-level key ingredients, and third-level key ingredients; the primary key ingredients are obtained by primary dissection of pre-factorial ingredients, the first-level key ingredients are obtained by first-level dissection of the primary key ingredients, the second-level key ingredients are obtained by second-level dissection of the first-level key ingredients according to different physical and chemical properties, and the third-level key ingredients are obtained by third-level dissection of the second-level key ingredients according to different physical and chemical properties. The pre-factorial ingredients include gross energy or metabolizable energy; the primary key ingredients specifically include moisture and dry matter; the first-level key ingredients specifically include crude protein, crude ash, nitrogen-free extract, crude fat, crude fiber, and various conventional anti-nutritional factors, etc.; the second-level key ingredients specifically include saturated fatty acids, unsaturated fatty acids, total starch, soluble sugars, insoluble sugars, calcium, phosphorus, globulin, gliadin, albumin, glutelin, neutral detergent fiber, acid detergent fiber, lignin, and anti-nutritional factors, etc.; the third-level key ingredients specifically include branched-chain amino acids, straight-chain amino acids, short-chain fatty acids, medium-chain fatty acids, long-chain fatty acids, oligosaccharides, and polysaccharides, etc. Specifically, in order to ensure the mining accuracy of prediction factors and expand the range of prediction factors in the embodiments of the present invention, the prediction factors can involve pre-factorial ingredients, primary key ingredients, first-level key ingredients, second-level key ingredients, and third-level key ingredients.

[0053] Step 30: Based on the poultry feed raw material samples, determine the contents of the key ingredients of each feed raw material, and the net energy values of the poultry feed raw material samples at different growth stages of poultry.

[0054] The embodiments of the present invention give the following methods for determining the contents of key ingredients of feed raw materials:

[0055] For albumin (Alb), globulin (Glob), gliadin (Gli), and glutelin (Glut), use a kit (Bradford, China) for determination.

[0056] For calcium, it is determined according to "Feed Analysis and Detection" edited by He Jianhua (2011).

[0057] For total starch, the soluble sugars in the sample can be separated from the starch with 80% ethanol first, and then the content of amylopectin is determined by double-wavelength colorimetry.

[0058] For amylose, the soluble sugars in the sample can be separated from the starch with 80% ethanol first, and then the amylose forms a complex with iodine and has an absorption peak at 620 nm. Then, the content of amylopectin is determined by colorimetry.

[0059] For amylopectin, it can be calculated based on the measurement results of total starch and amylose, that is, obtained by using the formula amylopectin = total starch - amylose.

[0060] It should be noted that the measurement methods of the key component contents of feed raw materials are not limited to the above, and the embodiments of the present invention do not make specific limitations thereto. The embodiments of the present invention give the measurement data of the contents of some key components, as shown in Table 2.

[0061] Table 2 Contents of Some Key Components Selected from Cottonseed Meal Raw Materials Based on the Factorial Method

[0062]

[0063]

[0064] It should be noted that since the starch content of cottonseed meal is extremely low, it is not further divided into secondary nutrient amylopectin and amylose. During the establishment of the regression equation between the predictor and net energy, the starch in the primary key components is directly combined with other secondary key components for regression analysis. Other raw materials can select the appropriate diversity level and key components according to the specific raw material situation.

[0065] Furthermore, when measuring the net energy values of poultry feed raw material samples at different growth stages of poultry, the following steps are mainly included: experimental animals and grouping, feed composition, feeding management, growth performance measurement, respiratory calorimetry data and excreta sample collection, and calculation of net energy.

[0066] It should be noted that in this embodiment, the indirect calorimetry method combined with the substitution algorithm is used for net energy measurement. In actual operation, those skilled in the art can understand that the net energy measurement can also be achieved by using the "comparative slaughter method", and modifying the net energy measurement method still belongs to the scope of application of the present invention.

[0067] For experimental animals and grouping, this embodiment adopts a completely randomized design, selects 144 healthy Hy-Line Brown laying hens at 33 weeks of age representing the peak laying period, with similar body weights and good health conditions. There are 6 replicates for each feed, and 2 laying hens in each replicate. The pre-feeding period is 7 days, and each group conducts respiratory calorimetry for 4 days (including 1 day of adaptation period and 3 days of formal test period) in a metabolic chamber equipped with a respiratory calorimetry device. The average value of the initial weight and final weight of the experimental hens during respiratory calorimetry is used as the body weight data, and the metabolic energy is measured during respiratory calorimetry.

[0068] For the feed composition, the feed used in the embodiments of the present invention includes 1 kind of corn-soybean meal-based basal feed and 10 kinds of cottonseed meal test feeds with a substitution ratio of 20%. The feed composition components are shown in Table 3, and the feed nutrient levels are shown in Table 4.

[0069] Table 3 Composition of Experimental Diets for Laying Hens at Peak Laying

[0070]

[0071]

[0072] Table 4 Nutritional levels of feed for laying hens at peak egg production

[0073]

[0074] It should be noted that the embodiment of the present invention is a feed formula designed for the net energy measurement of cottonseed meal for laying hens at the peak of egg production. For different poultry, different growth stages and different feed raw materials, the basic feed composition, nutritional level and raw material replacement ratio can be adjusted according to the characteristics of the animals and raw materials, so that the replaced feed can maintain a level without significant toxicity to the animals.

[0075] For feeding and management, the temperature during the experiment was (22±1)℃, the relative humidity was (72±10)%, the light duration was 16 h, including 8 h of darkness (lights turned off at 22:00 and turned on at 06:00 the next day), and the light intensity was 10 lx. The specific feeding and management parameters were to meet the physiological needs of the experimental animals and make them comfortable and stress-free.

[0076] For growth performance determination, the initial and final weights, feed intake and total excrement of laying hens were weighed and recorded at the beginning and end of the test period. Among them, the weight of the test chicken (BW) = (initial weight + final weight) / 2.

[0077] For the collection of respiratory calorimetry data and excrement samples, the operation of the respiratory calorimetry equipment was checked every 2 hours during the experimental lighting period, and feed and water were added at regular times every day (such as 09:00). The laying hens were allowed to eat and drink freely throughout the formal period. The heat production was calculated based on the O2 consumption and CO2 emission data of the laying hens in the respiratory calorimetry chamber collected by the sensor. At the same time, the excrement of the laying hens in the positive test period for 3 days was collected by the full feces collection method, and 10% dilute hydrochloric acid was sprayed to fix nitrogen. The feces samples collected daily were placed in a -20℃ refrigerator. After the test, the feces collected in 3 days were mixed together, placed in a 65℃ oven for 72 hours, and then rehydrated at room temperature for 24 hours. The samples were crushed until they all passed through a 40-mesh sieve and placed in a ziplock bag, and the samples were kept for later use. In addition, the feed intake, the weight of the laying hens before and after the test, the excrement and the total energy of the feed intake were measured.

[0078] For the calculation of net energy, the method includes: obtaining the basal diet and experimental diet of the poultry at different growth stages, where the experimental diet contains the poultry feed raw material sample; respectively determining the total energy intake of the basal diet and the total energy intake of the experimental diet; calculating the apparent metabolizable energy intake of the basal diet and the apparent metabolizable energy intake of the experimental diet based on the total energy intake of the basal diet and the total energy intake of the experimental diet; calculating the net energy of the basal diet and the net energy of the experimental diet according to the apparent metabolizable energy intake of the basal diet and the apparent metabolizable energy intake of the experimental diet; calculating the net energy value of the poultry feed raw material sample at different growth stages of the poultry based on the net energy of the experimental diet and the net energy of the basal diet. The specific steps are as follows:

[0079] 1. Feed intake and metabolic body weight: The feed intake (kg) is equal to the total feed intake during the respiration calorimetry test, the body weight (BW) is equal to the average of the body weights of the laying hens before and after the test, and the metabolic body weight is equal to BW 0.75 。

[0080] 2. Calculation of respiratory quotient:

[0081] Respiratory quotient (RQ) = VCO2 / VO2

[0082] where VO2 is the oxygen consumption (L) and VCO2 is the carbon dioxide consumption (L).

[0083] 3. Energy intake, excretion and apparent metabolizable energy intake of feed

[0084] Total energy intake (GEI, MJ) = Feed total energy (MJ / kg) × Intake feed amount (kg)

[0085] Total energy of excreta (GEO, MJ) = Excreta total energy (MJ / kg) × Excreta weight (kg)

[0086] Apparent metabolizable energy intake of feed (AMEI, MJ / kg) = [(GEI (MJ) - GEO (MJ)] / Feed intake (kg)

[0087] Nitrogen retention (RN) (g) = Nitrogen intake (NE) (g) - Nitrogen excretion (NE) (g)

[0088] Nitrogen-corrected apparent metabolizable energy (AMEn) (MJ / kg) = AME - RN (kg) × 34.39

[0089] where the total energy intake is the total energy of the feed ingested, and the feed total energy and the excreta total energy can be determined by drying the feed and excreta and using an oxygen bomb calorimeter.

[0090] 4. Heat increment and feed net energy

[0091] Total heat production (THP, kJ) = 16.18 × oxygen consumption (L) + 5.02 × carbon dioxide production (L)

[0092] Fasting heat production (FHP) of laying hens = 370 kJ / kg BW 0.75 ·d

[0093] Heat increment = Total heat production - Fasting heat production

[0094] Net energy of feed (Net energy, MJ / kg) = (AMEI - Heat increment) / Feed intake

[0095] Among them, d represents the number of days.

[0096] 5. Net energy of raw materials

[0097] Apparent metabolizable energy of raw materials (AME, MJ / kg) = (AME of test feed - AME of basal feed × a) / b

[0098] Net energy of raw materials (Net energy, MJ / kg) = (Net energy of test feed - Net energy of basal feed × a) / b.

[0099] Among them, a is the proportion (%) of the energy-providing raw material part of the basal feed in the test feed; b is the proportion (%) of cottonseed meal in the test feed; the net energy of the raw material is the net energy value of cottonseed meal for laying hens at the peak laying period finally calculated.

[0100] The production performance and respiratory data of different cottonseed meal raw materials in laying hens at the peak laying period are shown in Table 5. Based on the production performance and respiratory data of different cottonseed meal raw materials in laying hens at the peak laying period in Table 5, data such as the total heat production, heat increment, gross energy intake, apparent metabolizable energy intake, net energy intake, feed apparent metabolizable energy, and feed net energy of different cottonseed meal raw materials in Table 6 are obtained.

[0101] Table 5 Production performance and respiratory data of different cottonseed meal raw materials in laying hens at the peak laying period

[0102]

[0103] Table 6 Effects of different cottonseed meals on energy metabolism of laying hens at the peak laying period

[0104]

[0105]

[0106] Furthermore, based on the data in Table 6, the net energy values (measured values) of different cottonseed meal raw materials in laying hens at the peak laying period in Table 7 are obtained according to the above formula.

[0107] Table 7 Net energy values of different cottonseed meal raw materials during the peak laying period of laying hens (dry matter basis, MJ / kg DM)

[0108]

[0109] Step 40: Based on the measured contents of the key components of each feed raw material and the net energy values of the poultry feed raw material samples at different growth stages of poultry, perform principal component analysis to obtain prediction factors.

[0110] For the embodiments of the present invention, in order to perform dimensionality reduction processing on the measurement indexes (key components of each feed raw material), screen the set of key components of the raw materials with the greatest contribution to net energy, and use it to establish a prediction model, principal component analysis is performed between all the key components of the raw materials and net energy. That is, the measured values of all the key components of the raw materials and the measured values of net energy are input into a preset statistical analysis software for principal component analysis to obtain the set of key components of the raw materials with the greatest contribution to net energy, that is, the principal components. Among them, the preset statistical analysis software can specifically be SPASS statistical software, or other statistical analysis software, and the embodiments of the present invention do not make specific limitations on this.

[0111] During dimensionality reduction processing, first reduce the number of principal components, and then reduce the number of prediction factors in each principal component. For this process, as Figure 2 shown, it includes:

[0112] Step 41: Based on the measured contents of the key components of each feed raw material and the net energy values of the poultry feed raw material samples at different growth stages of poultry, determine a plurality of principal components.

[0113] For the embodiments of the present invention, the measured contents of the key components of each feed raw material and the net energy values are input into a preset statistical analysis software (such as SPASS) for statistical analysis to determine a plurality of principal components. When specifically inputting, all the key components of the raw materials can be packaged together, selectively package the key components across hierarchical combinations, or package them separately according to the classification of the key components.

[0114] Step 42: Based on the eigenvalue, variance interpretation rate, and cumulative contribution rate corresponding to each principal component among the plurality of principal components, determine candidate principal components from the plurality of principal components.

[0115] For the embodiments of the present invention, according to the actual business requirements, the eigenvalue threshold, the variance explanation rate threshold, and the cumulative contribution rate threshold are preset in advance. Then, according to the above thresholds, candidate principal components are screened out from multiple principal components, that is, the principal components with eigenvalues, and / or variance explanation rates, and / or cumulative contribution rates greater than the corresponding thresholds are selected as candidate principal components. In Table 8, the embodiments of the present invention perform principal component analysis on the key components of raw materials and traditional approximate nutrients (or conventional nutrients) respectively for comparison, so as to test the advantages of the factorial components of the embodiments of the present invention.

[0116] Table 8 Candidate Principal Components of Cottonseed Meal

[0117]

[0118] In this embodiment, candidate principal components (PCs) are extracted with the eigenvalue of each principal component being greater than 1, and the candidate PC with the highest cumulative contribution rate is obtained for subsequent prediction model establishment, that is, the content of Table 8, including: Item 1 performs principal component analysis with 5 conventional nutrients (CP, EE, Ash, CF, and Starch) as the analysis objects, obtaining two principal components, and the cumulative contribution rate reaches 75.16%. Among them, the cumulative contribution rate of principal component 1 is 51.28%, and the cumulative contribution rate of principal component 2 is 75.16%. Item 2 uses factorial component analysis, performs a more in-depth fitting with Alb, Glob, Gli, Glut, NDF, and ADF, and performs principal component analysis with 11 key components (Alb, Glob, Gli, Glut, EE, Ash, NDF, ADF, and Starch) as the analysis objects, screening 4 candidate principal components, and the highest cumulative contribution rate reaches 88.11%. Among them, the cumulative contribution rate of principal component 1 is 35.64%, the cumulative contribution rate of principal component 2 is 63.92%, the cumulative contribution rate of principal component 3 is 76.40%, and the cumulative contribution rate of principal component 4 is 88.11%. It can be seen that the method of factorial component analysis has a higher cumulative contribution degree to net energy for the principal components obtained than the method of traditional conventional nutrients (88.11% > 75.16%).

[0119] Step 43: Determine the number of extraction factors based on the number of the candidate principal components, and determine the factor loading matrix and / or loading diagram corresponding to the candidate principal components according to the number of extraction factors.

[0120] For the embodiments of the present invention, first calculate the correlation coefficient matrix, and the specific formula is as follows:

[0121]

[0122] where x ij represents the content of the j-th key component of the i-th cottonseed meal sample, represents the mean value of the j-th key component, and r jkrepresents the correlation coefficient between the j-th key component and the k-th key component.

[0123] After calculating the correlation coefficient matrix according to the above formula, eigenvalue decomposition is performed to obtain eigenvalues and corresponding eigenvectors. The eigenvalues represent the amount of information contained in the principal components, and the eigenvectors represent the directions of the principal components. The eigenvalues are arranged in descending order, and according to the number m of candidate principal components, the first m eigenvalues and their corresponding eigenvectors are selected. Suppose λ1, λ2, …, λ m are the first m eigenvalues, and V1, V2, …, V m are the corresponding unit eigenvectors. Then the j-th column of the factor loading matrix is:

[0124]

[0125] where a j is the j-th column of the factor loading matrix.

[0126] Thus, the factor loading matrix corresponding to each candidate principal component and the Figure 3 visualized multi-dimensional principal component loading diagram in can be determined in the above manner. The factor loading matrix of the candidate principal components of cottonseed meal is shown in Table 9, and the loading diagram of the candidate principal components of cottonseed meal is as shown in Figure 3 . From Table 9 and Figure 3-2 c, it can be seen that among the probabilistic nutrients, CP, Ash, Starch, and CF contribute more to principal component 1; EE contributes more to principal component 2. From Table 9 and Figure 3-2 d, it can be seen that among the key components, Ash, Starch, Gli, and ADF contribute more to principal component 1; NDF, Alb, Gli, Glut, and Glob contribute more to principal component 2; EE contributes more to principal component 3; NDF and Gli contribute more to principal component 4.

[0127] Table 9 Factor loading matrix of candidate principal components of cottonseed meal

[0128]

[0129]

[0130] Step 44: Based on the factor loading matrix and / or the loading diagram, analyze the correlation between the key components of the feed raw materials contained in the candidate principal components and the net energy value to determine the prediction factors in the candidate principal components.

[0131] For the embodiments of the present invention, after determining the factor loading matrix and / or loading graph corresponding to each candidate principal component in the above manner, based on the factor loading matrix of the principal component and the visualized multi-dimensional loading graph, factors with large weights (high loading values, that is, high correlation) and easy to detect are screened from each candidate principal component, so as to achieve the purpose of dimensionality reduction and reduce the subsequent chemical analysis and calculation workload.

[0132] When selecting prediction factors in the embodiments of the present invention, factors such as the absolute value of the factor loading, the synergistic and antagonistic effects between factors, raw material characteristics, detection convenience, and cost are comprehensively considered. In practical applications, users can set the threshold of the factor loading according to the raw material characteristics and actual situation. For example, factors with an absolute value of the loading higher than 0.5 are selected as prediction factors.

[0133] Step 50: Based on the prediction factors, use the linear regression analysis method to construct a candidate net energy prediction equation for different growth stages of poultry.

[0134] For the embodiments of the present invention, for any growth stage of poultry, after determining the prediction factors in each group of candidate principal components, based on the prediction factors in each group of candidate principal components, a candidate net energy prediction equation for poultry in this growth stage is constructed, as shown in Table 10.

[0135] Table 10 In-vivo and in-vitro candidate prediction equations for the net energy of cottonseed meal raw materials

[0136]

[0137]

[0138] Wherein, R 2 is the coefficient of determination, also known as R-squared; R 2 adj is the adjusted coefficient of determination, also known as adjusted R-squared; P is the significance. The same equation (A1 = B1) is obtained for both the in-vivo and in-vitro combined method and the in-vitro method using conventional components.

[0139] In-vivo indicators obtained through animal experiments, such as apparent metabolizable energy and nitrogen-corrected apparent metabolizable energy, generally have a direct relationship with energy utilization, often have a relatively high contribution rate to net energy, and the measurement process and cost of apparent metabolizable energy are much lower than the animal experiment cost of net energy measurement. Therefore, the embodiments of the present invention add equations containing animal test indicators, and through comparison, verify the reliability and feasibility of establishing a regression equation only through in-vitro indicators by the factorial component method.

[0140] As can be seen from Table 10, the key components of the raw materials are screened based on the eigenvalue, cumulative contribution rate, and principal component load of the above steps to obtain a set of factors for the candidate principal components, that is, the prediction factor set. By constructing a linear regression prediction equation for the net energy of cottonseed meal raw materials, three net energy regression equations for the in vitro and in vivo combined method and two net energy regression equations for the in vitro method can be obtained. The R 2 adj values of the three prediction equations containing animal test indicators are 0.670, 0.888, and 0.913 respectively. When the factorial component method of the embodiment of the present invention uses all pure in vitro indicators, the adjusted R-squared can reach 0.920 (R 2 adj not less than 0.670 and 0.913 of the in vivo indicators). It can be seen that through the factorial component method, the regression prediction equation constructed by the prediction factors after principal component analysis and dimension reduction can get rid of animal tests and achieve the same accuracy as the prediction equation for obtaining apparent metabolizable energy by combining animal experiments, thereby saving the cost of animal experiments. At the same time, the reliability and feasibility of the method of the embodiment of the present invention can be verified to be high.

[0141] Step 60: Based on the adjusted determination coefficient and significance corresponding to the candidate net energy prediction equations at different growth stages of the poultry, screen the net energy prediction equations at different growth stages of the poultry.

[0142] For the embodiment of the present invention, for the net energy prediction equation at any stage, a candidate net energy prediction equation with an adjusted determination coefficient (adjusted R-squared) and significance greater than a certain value can be screened out from multiple candidate net energy prediction equations as the net energy prediction equation at this growth stage.

[0143] For example, from Table 10, a candidate prediction equation with R 2 adj greater than 0.80 is screened out to obtain the above-mentioned 1 in vitro net energy prediction equation (B2). If selected according to R 2 adj greater than 0.9 or in descending order of priority, 1 optimal in vitro net energy prediction equation (B2) is also obtained. Since the prediction equation A2 obtained by the in vitro and in vivo combined method does not contain the metabolizable energy (apparent metabolizable energy or nitrogen-corrected apparent metabolizable energy) that requires essential animal tests, it can also be used for in vitro prediction. Although the R 2 adj of A3 is also very high (0.913), but it requires animal tests to obtain nitrogen-corrected apparent metabolizable energy, so it cannot be used as an in vitro prediction method. In addition, the R 2 adj of B2 is 0.920 > R 2 adj 0.888 of A2, so B2 is the optimal in vitro prediction equation. Therefore, taking laying hens at the peak laying period and cottonseed meal as an example, the following target net energy prediction equation is given.

[0144] When the predictors include prolamin and neutral detergent fiber, the target net energy prediction model is as follows:

[0145] Predicted net energy value = 11.893 + 11.908 * prolamin - 20.108 * neutral detergent fiber;

[0146] When the predictors include neutral detergent fiber, albumin, globulin, and gluten, the target net energy prediction model is as follows:

[0147] Predicted net energy value = 17.226 - 20.987 * neutral detergent fiber - 8.440 * albumin - 7.254 * globulin - 9.987 * gluten.

[0148] It should be noted specifically that in this embodiment, in order to verify the accuracy of the equation established by the in vitro method compared with the in vivo and in vitro combined method, the in vivo and in vitro combined method is set for comparison between the two. In the specific application of the present invention, there is no need to screen the equation for in vitro application through the in vivo and in vitro combined method, and only the in vitro method needs to be applied.

[0149] Step 70: Verify the prediction accuracy of the net energy prediction equations for different growth stages of the poultry, and determine the target net energy prediction models for different growth stages of the poultry according to the verification results.

[0150] For the embodiment of the present invention, in order to ensure the accuracy of the net energy prediction equation, it needs to be verified. For this process, the method includes: calculating the net energy value based on the net energy prediction equation for any one of the different growth stages, and comparing the calculated net energy value with the measured net energy value to determine the net energy deviation; performing a first verification on the net energy prediction equation for any one of the growth stages based on the net energy deviation, and screening out the net energy prediction equations that pass the first verification from the net energy prediction equations for any one of the growth stages; using a preset neural network model to continue performing a second verification on the net energy prediction equations that pass the first verification, and screening out the net energy prediction equations that pass the second verification from the net energy prediction equations that pass the first verification; determining the net energy prediction equations that pass the second verification as the target net energy prediction models for any one of the growth stages. Among them, the number of net energy prediction equations for any one of the growth stages is at least one.

[0151] Specifically, substitute the content corresponding to the predictor into the net energy prediction equation for any one of the growth stages to calculate the net energy. As shown in Table 11, the measured average value is 7.65 MJ / kg, and the relative deviation is between 1.339% and 3.195%. Among them, the relative deviation of Equation B2 is only 1.339%, so Equation B2 is the in vitro prediction equation with the highest accuracy.

[0152] Table 11 Comparison of predicted and measured net energy values of cottonseed meal raw materials

[0153]

[0154] After passing the above primary verification, the neural network algorithm is used to continue the secondary verification of the equations that have passed the primary verification. For example, for each of the 5 net energy prediction equations in Table 11, a neural network model is constructed according to the prediction factors of each prediction equation. The input of the neural network model is the content of the prediction factors, and the output is the range of net energy values. If the net energy value calculated using the prediction equation is within the range predicted by the neural network model, it means that the prediction equation passes the secondary verification. From Figure 4 and Figure 5 it can be seen that the accuracy rate of prediction equation B2 during the peak egg production period is 96.8%. The predicted net energy value and the measured value show a linear distribution. It can be seen that the reliability and accuracy of the prediction equation are relatively high, and it can be used as the final target net energy prediction model for in vitro prediction of the net energy of poultry feed ingredients.

[0155] Step 80: Based on the target net energy prediction model for different growth stages of the poultry, perform net energy prediction on the to-be-tested poultry feed ingredients to obtain the predicted net energy values for different growth stages of the poultry.

[0156] For the embodiments of the present invention, after obtaining the target net energy prediction model for different growth stages of the poultry, when wanting to predict the net energy of the poultry at any growth stage, select the target net energy prediction model for the corresponding growth stage, and input the content of the prediction factors in the to-be-tested poultry feed ingredients into the target net energy prediction model for net energy prediction to obtain the predicted net energy value of the poultry at this growth stage.

[0157] By comparing the prediction method provided by the embodiments of the present invention with the traditional in vivo determination method, the traditional in vivo determination method requires animal experiments, uses expensive respiration calorimetry devices, purchases and hires breeders to raise poultry, and measures the net energy based on animal experiments. Taking the determination of the net energy of 96 raw materials as an example, as shown in Table 12, the traditional net energy determination method costs about 576,000 yuan from the test preparation to the end of the test, takes 130 - 191 days, and requires farm lease fees, farm management fees, hiring fees for breeders and test personnel, and equipment purchase fees such as energy measuring instruments, resulting in serious consumption of manpower, financial resources and time. However, for the in vitro prediction method provided by the embodiments of the present invention, after constructing the target net energy prediction model, only 1 - 4 key prediction factors in cottonseed meal need to be measured in subsequent determinations, and the predicted value of the net energy can be directly calculated without animal experiments. Measuring the above 96 cottonseed meal samples only costs about 4,800 yuan, which is 0.83% of the traditional cost, and the measurement period is 1 - 4 days, which is 0.77% - 2% of the traditional measurement method period, greatly saving manpower, material resources, financial resources and time. In the future, if this method is combined with the near-infrared spectroscopy method (NIRS), the detection cost will be further reduced.

[0158] Table 12 Comparison between the prediction method of the embodiment of the present invention and the traditional method

[0159]

[0160] Compared with the traditional in vivo method, the in vitro determination method of the embodiment of the present invention does not require animal experiments, saves the rental fee of the farm, the cost of animal purchase and feeding, avoids the harm to animal welfare caused by force-feeding method, greatly improves the number of sample evaluations per unit time, saves the evaluation cycle, and thus greatly saves manpower, material resources, financial resources and time. In addition, due to the wide range of prediction factors screened in the embodiment of the present invention, the prediction accuracy of net energy is significantly improved.

[0161] In addition, the prediction method of the embodiment of the present invention is simple and easy to implement. Compared with the traditional Excel calculation method, after embedding the prediction equation into the net energy prediction applet for poultry raw materials, it can be widely promoted through WeChat applet or other mobile applications and computer software, and has high application value.

[0162] In some embodiments, if the content of the prediction factor in the to-be-detected poultry feed raw material exceeds the first threshold or the predicted net energy value exceeds the second threshold, it is determined that the to-be-detected poultry feed raw material is an abnormal sample, and an alarm message is sent. And based on the abnormal sample, the target net energy prediction model for different growth stages of the poultry is updated, that is, the above-mentioned process of constructing the prediction equation is repeated to realize the iterative upgrade of the method.

[0163] In order to make the technical solution of the embodiment of the present invention clearer, taking laying hens and rapeseed meal at other growth stages as an example, the construction and verification process of the target net energy prediction model is described again.

[0164] First, 30 rapeseed meal feed raw material samples in different seasons, different products and different batches are collected. Then, 10 rapeseed meal raw material samples with large differences in composition and net energy are selected as modeling samples for later net energy determination. The 10 selected rapeseed meal feed raw material samples are from Shanghai (rapeseed meal 1), Qinzhou, Guangxi (rapeseed meal 2), Shijiazhuang, Hebei (rapeseed meal 3), Nankou, Beijing (rapeseed meal 4), Zhuozhou, Hebei (rapeseed meal 5), Huanggang, Hubei (rapeseed meal 6), Chengdu, Sichuan (rapeseed meal 7), Maoming, Guangdong (rapeseed meal 8), Fuyu, Jilin (rapeseed meal 9), and Meishan, Sichuan (rapeseed meal 10).

[0165] After that, based on the determination method in step 30, the content of the key components selected for rapeseed meal by the factorial method is obtained, as shown in Table 13.

[0166] Table 13 Content of key components of rapeseed meal raw materials

[0167]

[0168]

[0169] Furthermore, net energy determination was performed according to the respiratory calorimetry in step 30 to obtain the respiratory data in Table 14 and the energy metabolism data in Table 15, including the net energy values ​​of 10 kinds of rapeseed meal raw materials on growing laying hens.

[0170] Table 14 Production performance and respiratory data of different rapeseed meal raw materials in growing laying hens

[0171]

[0172]

[0173] Table 15 Effects of different diets on energy metabolism and net energy value of growing laying hens

[0174]

[0175] Furthermore, based on the data in Table 14 and Table 15, principal component analysis was performed to obtain the eigenvalues ​​and contribution rates of the principal components in Table 16, and the principal component factor loading matrix and Figure 6 Visualized multidimensional principal component loading plot in .

[0176] Table 16 Principal component eigenvalues ​​and contribution rates

[0177]

[0178]

[0179] In the two methods of designing in vivo and in vitro combined prediction factors, the cumulative contribution rate of the principal component is much higher than that of conventional nutrients (84.31%>74.30%), and the cumulative contribution rate of the principal component in the in vitro method is also much higher than that of conventional nutrients (90.44%>77.93%). It can be seen that using the factorial method instead of the conventional nutrient method to mine prediction factors has a higher prediction accuracy.

[0180] In the factorial method, the highest cumulative contribution rate of the principal component of the in vitro and in vivo combined method was 84.31%, which was lower than the highest cumulative contribution rate of the principal component of the in vitro method, which was 90.44%. It can be seen that replacing the animal experiments involved in the in vivo method with the in vitro prediction method can greatly save costs while ensuring the accuracy of the net energy prediction. That is, the in vitro net energy prediction method based on the factorial method has a high application potential in replacing animal experiments.

[0181] Table 17 Principal component factor loading matrix

[0182]

[0183]

[0184] Based on the factor loading matrix in Table 17 and the multi-dimensional visualization loading diagram in Figure 6 , screening the prediction factors to establish a candidate net energy prediction equation, as shown in Table 18.

[0185] Table 18 Net energy prediction equation of rapeseed meal raw materials in laying hens during the growth period

[0186]

[0187]

[0188] Among them, R 2 is the coefficient of determination, also known as R-squared; R 2 adj is the adjusted coefficient of determination, also known as adjusted R-squared; P is the significance. The same equation (A1 = B1) was obtained for both the in vivo-in vitro combined method and the in vitro method using conventional components.

[0189] Table 18 shows that whether in the in vivo-in vitro combined method or the in vitro method, the R 2 adj of the prediction equations (A3 - A5, B2) obtained through factorial components is greater than that of the prediction equations (A1, A2, B1) obtained based on conventional nutrients; and it is shown in the in vivo-in vitro combined method that the R 2 adj of the equation (A3) that only factors fiber is lower than that of the equation (A4) that factors both fiber and protein at the same time. 2 adj 2 adj adj . It can be seen that factoring key components and obtaining prediction equations by factoring multiple key components have higher accuracy than prediction equations constructed using conventional nutrients.

[0190] Screening out the candidate prediction equations with R 2 adj greater than 0.80 from Table 18, the above 4 net energy prediction equations are obtained. Among them, although the R 2 adj of A5 is also very high (0.971), but the nitrogen-corrected apparent metabolizable energy needs to be obtained through animal experiments, so it cannot be used as an in vitro prediction method; the equations A3, A4, and B2 can achieve in vitro prediction of net energy. The R 2 adjReached 0.955, which is similar to the determination coefficients of A3 and A4 established by the in vitro and in vivo binding method (0.938, 0.966), with high accuracy. And no animal experiments are required during the equation construction process, which is simple and easy to implement. It can be used to establish an in vitro prediction method for net energy. The finally constructed equations A3, A4, and B2 in this embodiment can all achieve in vitro prediction of net energy. Therefore, in this embodiment, when the poultry is a laying hen in the growth period and the poultry feed raw material sample is rapeseed meal, when the prediction factors include neutral detergent fiber and acid detergent fiber, the target net energy prediction model is:

[0191] Predicted net energy value = 8.017 + 1.684 * neutral detergent fiber - 7.735 * acid detergent fiber;

[0192] When the prediction factors include acid detergent fiber, albumin, and prolamin, the target net energy prediction model is:

[0193] Predicted net energy value = 9.287 - 5.849 * acid detergent fiber - 3.572 * albumin - 19.287 * prolamin;

[0194] When the prediction factors include starch, neutral detergent fiber, and acid detergent fiber, the target net energy prediction model is:

[0195] Predicted net energy value == 7.617 + 36.728 * starch + 2.499 * neutral detergent fiber - 8.533 * acid detergent fiber.

[0196] It should be specifically noted that in this embodiment, in order to verify the accuracy of the equation established by the in vitro method compared with the in vitro and in vivo binding method, the in vitro and in vivo binding method is set for comparison between the two. In the specific application of the present invention, there is no need to screen the equation for in vitro application through the in vitro and in vivo binding method, and only the in vitro method needs to be applied.

[0197] Furthermore, based on the adjusted determination coefficient and significance, a net energy prediction equation is screened from the candidate net energy prediction equations. Then, the net energy prediction equation for rapeseed meal on laying hens in the growth period is verified.

[0198] Table 19 Comparison of predicted and measured values of net energy of rapeseed meal raw materials

[0199]

[0200] From the data in Table 19, it can be seen that the prediction equation A5 has the highest goodness of fit among all prediction equations, with a relative standard deviation of 0.834%. In the in vitro method, the relative deviation of equation B2 is the smallest, which is 0.983%. The deviation of equation B2 is close to that of the optimal equation (equation A5) in the in vitro-in vivo combined method, but it has the advantage of getting rid of animal experiments. Therefore, through the first verification, it can be known that the prediction equation B2 is the best candidate equation for in vitro prediction established directly through the in vitro method. Then, the neural network model was used to conduct a secondary verification on the prediction equation B2, and the accuracy verification results are as Figure 7 and Figure 8 shown. From Figure 7 and Figure 8 it can be known that the accuracy of the net energy prediction equation for rapeseed meal of laying hens in the growth period is 97.6%. The predicted value of net energy and the measured value are linearly distributed. The prediction equation B2 can be used as the target net energy prediction model and used as a method for in vitro prediction of the net energy of rapeseed meal for laying hens in the growth period.

[0201] It should be noted that the prediction method of the embodiment of the present invention can be used not only for the net energy prediction of cottonseed meal for laying hens at the peak laying period and the net energy prediction of rapeseed meal for laying hens in the growth period, but also for the net energy prediction of poultry at different stages and different energy-providing raw materials.

[0202] An in vitro prediction method for the net energy of multi-stage poultry feed based on factorial screening provided by the embodiment of the present invention can fully consider the influence of factors such as the source and variety of feed raw materials on the net energy by introducing feed raw materials of different products, different times, different processing plants and different batches when establishing the target net energy prediction model, so as to increase the variability of the modeling samples and improve the prediction accuracy of the net energy of poultry feed raw materials. At the same time, by constructing the target net energy prediction model for different growth stages of poultry, the embodiment of the present invention can accurately predict the net energy value of feed raw materials at different growth stages of poultry. In addition, by analyzing the relationship between the key component set and the net energy and determining the prediction factors, the embodiment of the present invention can perform dimensionality reduction processing, thereby reducing the number of detection indexes and the detection cost. Further, by conducting a double verification on the net energy prediction equation, the embodiment of the present invention can ensure the accuracy and reliability of the finally determined target net energy equation. Further, by determining the components before factoring, primary key components, first-level key components, second-level key components and third-level key components, the embodiment of the present invention can fully explore the prediction factors with high correlation with the net energy, thereby further improving the prediction accuracy of the net energy of poultry feed raw materials, and further providing a more reliable method for promoting the rapid in vitro determination of the net energy of feed raw materials, accurately formulating poultry diets in real-time dynamics, and reducing the substitution of corn and soybean meal.

[0203] Further, as Figure 1 and Figure 2For the specific implementation of the method described above, this embodiment provides a device for in vitro prediction of the net energy of poultry multi-stage feed based on factorial screening, as Figure 9 shown. The device includes: a collection unit 101, a factorial unit 102, a measurement unit 103, an analysis unit 104, a construction unit 105, a screening unit 106, a verification unit 107, and a prediction unit 108.

[0204] The collection unit 101 can be used to collect and screen samples of poultry feed raw materials; among them, the poultry feed raw material samples include feed raw materials from different origins, different times, different processing factories, and different batches.

[0205] The factorial unit 102 can be used to factorize the raw material components affecting the net energy value by the factorial method to obtain a key component set, where the key component set includes the key components of each feed raw material, and the key components of each feed raw material belong to the components before factoring, primary key components, first-level key components, second-level key components, and third-level key components respectively.

[0206] The measurement unit 103 can be used to measure the content of the key components of each feed raw material and the net energy value of the poultry feed raw material samples at different growth stages of poultry based on the poultry feed raw material samples.

[0207] The analysis unit 104 can be used to perform principal component analysis based on the measured content of the key components of each feed raw material and the net energy value of the poultry feed raw material samples at different growth stages of poultry to obtain prediction factors.

[0208] The construction unit 105 can be used to construct candidate net energy prediction equations for different growth stages of poultry by using the linear regression analysis method based on the prediction factors.

[0209] The screening unit 106 can be used to screen the net energy prediction equations for different growth stages of poultry based on the determination coefficients and significances corresponding to the candidate net energy prediction equations for different growth stages of poultry.

[0210] The verification unit 107 can be used to verify the prediction accuracy of the net energy prediction equations for different growth stages of poultry and determine the target net energy prediction models for different growth stages of poultry according to the verification results.

[0211] The prediction unit 108 can be used to perform net energy prediction on the to-be-tested poultry feed raw materials based on the target net energy prediction models for different growth stages of poultry to obtain the predicted net energy values for different growth stages of poultry.

[0212] In some embodiments, the primary key components are obtained by primary dissection of the pre-factorial components; the first-level key components are obtained by first-level dissection of the primary key components, the second-level key components are obtained by second-level dissection of the first-level key components according to different physical and chemical properties, and the third-level key components are obtained by third-level dissection of the second-level key components according to different physical and chemical properties; and / or the poultry feed raw material sample includes at least one raw material that can provide energy in the poultry feed formula; the poultry includes at least one of chicken, duck, goose, turkey, ostrich, quail, and pigeon; the different growth stages include the brooding stage, the growing stage, and the reproductive stage.

[0213] In some embodiments, the pre-factorial components include gross energy or metabolizable energy; the primary key components include moisture and dry matter; the first-level key components include crude protein, crude ash, nitrogen-free extract, crude fat, crude fiber, and various conventional anti-nutritional factors; the second-level key components include saturated fatty acids, unsaturated fatty acids, total starch, soluble sugars, insoluble sugars, calcium, phosphorus, globulin, prolamin, albumin, glutelin, neutral detergent fiber, acid detergent fiber, lignin, and anti-nutritional factors; the third-level key components include branched-chain amino acids, straight-chain amino acids, short-chain fatty acids, medium-chain fatty acids, long-chain fatty acids, oligosaccharides, and polysaccharides.

[0214] In some embodiments, the determination unit 103 can be specifically configured to obtain the basal feed and the test feed of the poultry at different growth stages, where the test feed contains the poultry feed raw material sample; respectively determine the gross energy intake of the basal feed and the gross energy intake of the test feed; based on the gross energy intake of the basal feed and the gross energy intake of the test feed, calculate the apparent metabolizable energy intake of the basal feed and the apparent metabolizable energy intake of the test feed respectively; according to the apparent metabolizable energy intake of the basal feed and the apparent metabolizable energy intake of the test feed, calculate the net energy of the basal feed and the net energy of the test feed respectively; based on the net energy of the test feed and the net energy of the basal feed, calculate the net energy value of the poultry feed raw material sample at different growth stages of the poultry.

[0215] In some embodiments, the analysis unit 104 can be specifically used to determine multiple principal components based on the measured content of each key component of the feed raw materials and the net energy value of the poultry feed raw material sample at different growth stages of poultry; determine candidate principal components from the multiple principal components based on the eigenvalue, variance explanation rate and cumulative contribution rate corresponding to each principal component in the multiple principal components; determine the number of extracted factors based on the number of candidate principal components, and determine the factor loading matrix and / or loading diagram corresponding to the candidate principal components based on the number of extracted factors; analyze the correlation between the key components of the feed raw materials contained in the candidate principal components and the net energy value based on the factor loading matrix and / or loading diagram to determine the prediction factors in the candidate principal components.

[0216] In some embodiments, the verification unit 107 can be specifically used to calculate the net energy value based on the net energy prediction equation of any growth stage in the different growth stages, and compare the calculated net energy value with the measured net energy value to determine the net energy deviation; verify the net energy prediction equation of any growth stage based on the net energy deviation, and screen out the net energy prediction equation that passes the primary verification from the net energy prediction equations of any growth stage; continue to perform a secondary verification on the net energy prediction equation that passes the primary verification using a preset neural network model, and screen out the net energy prediction equation that passes the secondary verification from the net energy prediction equation that passes the primary verification; and determine the net energy prediction equation that passes the secondary verification as the target net energy prediction model for any growth stage.

[0217] In some embodiments, the device further comprises: an alarm unit.

[0218] The alarm unit can be used to determine that the poultry feed raw material to be tested is an abnormal sample and send an alarm message if the content of the prediction factor in the poultry feed raw material to be tested exceeds a first threshold or the predicted net energy value exceeds a second threshold; based on the abnormal sample, the target net energy prediction model for the poultry at different growth stages is updated.

[0219] In some embodiments, when the poultry is a laying hen in the peak egg-laying period and the poultry feed raw material sample is cottonseed meal,

[0220] When the prediction factors include prolamin and neutral detergent fiber, the target net energy prediction model is:

[0221] Predicted net energy value = 11.893 + 11.908 * prolamin - 20.108 * neutral detergent fiber;

[0222] When the predictors include neutral detergent fiber, albumin, globulin, and glutelin, the target net energy prediction model is: Predicted net energy value = 17.226 - 20.987 * neutral detergent fiber - 8.440 * albumin - 7.254 * globulin - 9.987 * glutelin.

[0223] In some embodiments, when the poultry is a growing laying hen and the poultry feed raw material sample is rapeseed meal,

[0224] When the predictors include neutral detergent fiber and acid detergent fiber, the target net energy prediction model is:

[0225] Predicted net energy value = 8.017 + 1.684 * neutral detergent fiber - 7.735 * acid detergent fiber;

[0226] When the predictors include acid detergent fiber, albumin, and prolamin, the target net energy prediction model is:

[0227] Predicted net energy value = 9.287 - 5.849 * acid detergent fiber - 3.572 * albumin - 19.287 * prolamin;

[0228] When the predictors include starch, neutral detergent fiber, and acid detergent fiber, the target net energy prediction model is:

[0229] Predicted net energy value == 7.617 + 36.728 * starch + 2.499 * neutral detergent fiber - 8.533 * acid detergent fiber.

[0230] It should be noted that for other corresponding descriptions of each functional unit involved in the in vitro prediction device for net energy of multi-stage poultry feed based on factorial screening provided in the embodiments of the present invention, reference can be made to Figure 1 and Figure 2 the corresponding descriptions therein, which will not be elaborated here.

[0231] Based on the above methods as shown in Figure 1 and Figure 2 correspondingly, the present embodiment also provides a storage medium, on which a computer program is stored, and when the program is executed by a processor, it implements the in vitro prediction method for net energy of multi-stage poultry feed based on factorial screening as shown in Figure 1 and Figure 2 above.

[0232] Based on such understanding, the technical solution of the present application can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as a CD-ROM, a USB flash drive, a mobile hard disk, etc.), and includes several instructions for causing an electronic device (such as a personal computer, a server, or a network device, etc.) to execute the methods of various implementation scenarios of the present application.

[0233] Based on the above such Figure 1 and Figure 2 shown method, as well as Figure 9 the virtual device embodiment shown, in order to achieve the above object, the embodiments of the present application also provide an electronic device, which can specifically be a personal computer, a tablet computer, a server, or other network devices, etc. The device includes a storage medium and a processor; the storage medium is used to store a computer program; the processor is used to execute the computer program to implement the above such Figure 1 and Figure 2 shown in vitro prediction method of net energy of poultry multi-stage feed based on factorial screening.

[0234] Optionally, the above-mentioned physical device may further include a user interface, a network interface, a camera, a radio frequency (RF) circuit, sensors, an audio circuit, a WI-FI module, etc. The user interface may include a display screen (Display), an input unit such as a keyboard (Keyboard), etc. Optionally, the user interface may further include a USB interface, a card reader interface, etc. The network interface may optionally include a standard wired interface, a wireless interface (such as a WI-FI interface), etc.

[0235] Those skilled in the art can understand that the above-mentioned physical device structure provided in this embodiment does not limit the physical device, and it may include more or fewer components, or combine some components, or have different component arrangements. The storage medium may further include an operating system and a network communication module. The operating system is a program for managing the hardware and software resources of the above-mentioned physical device, and supports the operation of information processing programs and other software and / or programs. The network communication module is used to implement communication between components inside the storage medium, as well as communication between other hardware and software in the information processing physical device.

[0236] Through the description of the above embodiments, those skilled in the art can clearly understand that the present application can be implemented by means of software plus a necessary general hardware platform, or can also be implemented by hardware.

[0237] When establishing the target net energy prediction model in the embodiments of the present invention, by introducing feed raw materials of different products, different times, different processing plants and different batches, the influence of factors such as the source and variety of feed raw materials on net energy can be fully considered, thereby increasing the variability of the modeling samples and improving the prediction accuracy of the net energy of poultry feed raw materials. At the same time, in the embodiments of the present invention, by constructing the target net energy prediction model for different growth stages of poultry, the net energy value of feed raw materials at different growth stages of poultry can be accurately predicted. In addition, in the embodiments of the present invention, by analyzing the relationship between the key component set and net energy to determine the prediction factors, dimensionality reduction processing can be performed, thereby reducing the number of detection indicators and the detection cost. Further, in the embodiments of the present invention, by double-verifying the net energy prediction equation, the accuracy and reliability of the finally determined target net energy equation can be ensured. Further, in the embodiments of the present invention, by determining the pre-factorial components, primary key components, first-level key components, second-level key components and third-level key components, prediction factors with high correlation with net energy can be fully excavated, thereby further improving the prediction accuracy of the net energy of poultry feed raw materials, and further providing a more reliable method for promoting the rapid in vitro determination of the net energy of feed raw materials, accurately formulating poultry diets in real-time and dynamically, and reducing the substitution of corn and soybean meal.

[0238] Those skilled in the art can understand that the drawings are only schematic diagrams of a preferred implementation scenario, and the modules or processes in the drawings are not necessarily essential for implementing the present application. Those skilled in the art can understand that the modules in the device in the implementation scenario can be distributed in the device of the implementation scenario according to the description of the implementation scenario, or can be correspondingly changed and located in one or more devices different from the present implementation scenario. The modules in the above implementation scenario can be combined into one module, or further split into multiple sub-modules.

[0239] The above serial numbers of the present application are only for description and do not represent the advantages or disadvantages of the implementation scenarios. The above disclosure is only several specific implementation scenarios of the present application. However, the present application is not limited thereto, and any changes that can be thought of by those skilled in the art should fall within the protection scope of the present application.

Claims

1. An in vitro prediction method for the net energy of poultry multi-stage feed based on factorial screening, characterized in that, Comprising: Collecting and screening samples of poultry feed raw materials; wherein, the samples of poultry feed raw materials include feed raw materials from different origins, at different times, from different processing plants, and of different batches; Using the factorial method to factorize the raw material components affecting the net energy value to obtain a key component set, wherein the key component set includes key components of each feed raw material, and the key components of each feed raw material respectively belong to the components before factoring, primary key components, first-level key components, second-level key components, and third-level key components; Based on the samples of poultry feed raw materials, determining the contents of the key components of each feed raw material, and the net energy values of the samples of poultry feed raw materials at different growth stages of poultry; Based on the determined contents of the key components of each feed raw material and the net energy values of the samples of poultry feed raw materials at different growth stages of poultry, performing principal component analysis to obtain prediction factors; Based on the prediction factors, using the linear regression analysis method to construct candidate net energy prediction equations for different growth stages of poultry; Based on the adjusted determination coefficients and significances corresponding to the candidate net energy prediction equations for different growth stages of poultry, screening the net energy prediction equations for different growth stages of poultry; Verifying the prediction accuracy of the net energy prediction equations for different growth stages of poultry, and determining the target net energy prediction models for different growth stages of poultry according to the verification results; Based on the target net energy prediction models for different growth stages of poultry, predicting the net energy of the to-be-tested poultry feed raw materials to obtain the predicted net energy values at different growth stages of poultry.

2. The method according to claim 1, characterized in that The primary key components are obtained by primary dissection of the components before factoring; the first-level key components are obtained by first-level dissection of the primary key components, the second-level key components are obtained by second-level dissection of the first-level key components according to different physical and chemical properties, and the third-level key components are obtained by third-level dissection of the second-level key components according to different physical and chemical properties; and / or the samples of poultry feed raw materials include at least one raw material that can provide energy in the poultry feed formula; the poultry includes at least one of chicken, duck, goose, turkey, ostrich, quail, and pigeon; the different growth stages include the brooding period, the growth period, and the breeding period.

3. The method according to claim 1, wherein The components before factoring include gross energy or metabolizable energy; the primary key components include water and dry matter; the first-level key components include crude protein, crude ash, nitrogen-free extract, crude fat, crude fiber, and various conventional antinutritional factors; the second-level key components include saturated fatty acids, unsaturated fatty acids, total starch, soluble sugars, insoluble sugars, calcium, phosphorus, globulin, prolamin, albumin, glutelin, neutral detergent fiber, acid detergent fiber, lignin, and antinutritional factors; the third-level key components include branched-chain amino acids, straight-chain amino acids, short-chain fatty acids, medium-chain fatty acids, long-chain fatty acids, oligosaccharides, and polysaccharides.

4. The method according to claim 1, wherein Based on the samples of poultry feed raw materials, determining the net energy values of the samples of poultry feed raw materials at different growth stages of poultry, including: Obtaining the basal feed and test feed of the poultry at different growth stages, wherein the test feed contains the samples of poultry feed raw materials; Determine the gross energy intake of the basal diet and the gross energy intake of the test diet respectively; Based on the gross energy intake of the basal diet and the gross energy intake of the test diet, calculate the apparent metabolizable energy intake of the basal diet and the apparent metabolizable energy intake of the test diet respectively; According to the apparent metabolizable energy intake of the basal diet and the apparent metabolizable energy intake of the test diet, calculate the net energy of the basal diet and the net energy of the test diet respectively; Based on the net energy of the test diet and the net energy of the basal diet, calculate the net energy value of the poultry feed raw material sample at different growth stages of poultry.

5. The method according to claim 1, wherein Based on the measured contents of the key components of each feed raw material, and the net energy values of the poultry feed raw material sample at different growth stages of poultry, perform principal component analysis to obtain prediction factors, including: Based on the measured contents of the key components of each feed raw material, and the net energy values of the poultry feed raw material sample at different growth stages of poultry, determine multiple principal components; Based on the eigenvalue, variance interpretation rate and cumulative contribution rate corresponding to each principal component in the multiple principal components, determine candidate principal components from the multiple principal components; Based on the number of candidate principal components, determine the number of extraction factors, and according to the number of extraction factors, determine the factor loading matrix and / or loading plot corresponding to the candidate principal components; Based on the factor loading matrix and / or loading plot, analyze the correlation between the key components of the feed raw materials included in the candidate principal components and the net energy value to determine the prediction factors in the candidate principal components.

6. The method according to claim 1, wherein Verify the prediction accuracy of the net energy prediction equation for different growth stages of poultry, and according to the verification result, determine the target net energy prediction model for different growth stages of poultry, including: Based on the net energy prediction equation for any one of the different growth stages, calculate the net energy value and compare the calculated net energy value with the measured net energy value to determine the net energy deviation; Based on the net energy deviation, perform a first verification on the net energy prediction equation for any one of the growth stages, and screen out the net energy prediction equations that pass the first verification from the net energy prediction equations for any one of the growth stages; Use a preset neural network model to continue to perform a second verification on the net energy prediction equations that pass the first verification, and screen out the net energy prediction equations that pass the second verification from the net energy prediction equations that pass the first verification; Determine the net energy prediction equation that passes the second verification as the target net energy prediction model for any one of the growth stages; and / or the method further includes: If the content of the prediction factor in the to-be-tested poultry feed raw material exceeds the first threshold or the predicted net energy value exceeds the second threshold, determine that the to-be-tested poultry feed raw material is an abnormal sample and send an alarm message; Based on the abnormal sample, update the target net energy prediction model for different growth stages of poultry.

7. The method according to any one of claims 1-6, characterized in that, In the case where the poultry is a laying hen at the peak laying period and the poultry feed raw material sample is cottonseed meal, When the prediction factors include prolamin and neutral detergent fiber, the target net energy prediction model is: Predicted net energy value = 11.893 + 11.908 * prolamin - 20.108 * neutral detergent fiber; When the prediction factors include neutral detergent fiber, albumin, globulin, and gluten, the target net energy prediction model is: predicted net energy value = 17.226-20.987*neutral detergent fiber-8.440*albumin-7.254*globulin-9.987*gluten; and / or In the case where the poultry is a laying hen in the growing period and the poultry feed raw material sample is rapeseed meal, When the prediction factors include neutral detergent fiber and acid detergent fiber, the target net energy prediction model is: Predicted net energy value = 8.017 + 1.684 * neutral detergent fiber - 7.735 * acid detergent fiber; When the prediction factors include acid detergent fiber, albumin, and prolamin, the target net energy prediction model is: Predicted net energy value = 9.287-5.849*acid detergent fiber-3.572*albumin-19.287*gliadin; When the prediction factors include starch, neutral detergent fiber, and acid detergent fiber, the target net energy prediction model is: Predicted net energy value = 7.617 + 36.728*starch + 2.499*neutral detergent fiber - 8.533*acid detergent fiber.

8. An in vitro prediction device for the net energy of poultry multi-stage feed based on factorial screening, characterized in that, include: A collection unit, used for collecting and screening poultry feed raw material samples; wherein the poultry feed raw material samples include feed raw materials from different origins, different times, different processing factories, and different batches; A factorial analysis unit is used to factor the raw material components that affect the net energy value by a factorial analysis method to obtain a key component set, wherein the key component set includes each key component of the feed raw material, and each key component of the feed raw material belongs to a pre-factorial analysis component, a primary key component, a primary key component, a secondary key component and a tertiary key component respectively; A determination unit, used for determining the content of each key component of the feed raw material and the net energy value of the poultry feed raw material sample at different growth stages of poultry based on the poultry feed raw material sample; An analysis unit, for performing a main component analysis based on the determined contents of the key components of each feed raw material and the net energy value of the poultry feed raw material sample at different growth stages of poultry to obtain a prediction factor; A construction unit is used to construct candidate net energy prediction equations for different growth stages of poultry based on the prediction factors using a linear regression analysis method; A screening unit, for screening net energy prediction equations for poultry at different growth stages based on the determination coefficients and significance corresponding to the candidate net energy prediction equations for poultry at different growth stages; A verification unit, used to verify the prediction accuracy of the net energy prediction equation for the poultry at different growth stages, and determine the target net energy prediction model for the poultry at different growth stages according to the verification result; The prediction unit is used to perform net energy prediction on the poultry feed raw materials to be tested based on the target net energy prediction model for the poultry at different growth stages, so as to obtain the predicted net energy values ​​of the poultry at different growth stages.

9. A storage medium, on which a computer program is stored, characterized in that, When the computer program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

10. An electronic device, comprising a storage medium, a processor, and a computer program stored on the storage medium and executable on the processor, characterized in that, When the processor executes the computer program, the method described in any one of claims 1 to 7 is implemented.

Images