Dairy cow cold stress multi-level fuzzy comprehensive evaluation method

Abstract

The application discloses a kind of dairy cow cold stress multistage fuzzy comprehensive evaluation method, comprising the following steps: step (1): using analytic hierarchy process, from warm environment (temperature, relative humidity, wind speed and illumination), physiological factors (respiratory rate, body surface temperature) and air quality (NH3, CO2, PM 10 ) 3 dimensions construct dairy cow cold stress evaluation index system;Step (2): establish the judgment matrix of each index layer, and make consistency check, if meet the consistency requirement, then determine the weight of each index layer;Step (3): if it does not meet the consistency check, use genetic algorithm for testing and correcting the judgment matrix, the weight of each index layer is calculated by crossing, mutation and other operations;Step (4): the degree of cold stress of dairy cow is obtained by multistage fuzzy comprehensive evaluation;The application combines environmental parameters with dairy cow physiological parameters, and can realize comprehensive and accurate evaluation of dairy cow cold stress.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of cold stress evaluation of dairy cows, and particularly relates to a multi-level fuzzy comprehensive evaluation method for cold stress of dairy cows.

[0002] BACKGROUND AND TECHNICAL FIELD

[0003] Cold stress refers to the functional disorders and defense reactions exhibited by animals exposed to cold environments. Although dairy cows are a kind of cold-tolerant and heat-averse ruminants, they need to produce more metabolic heat to maintain core temperature through body regulation when facing extreme low temperature climate. The generated heat is diffused to the outside world through evaporation, conduction, convection and radiation. When the heat production and dissipation of the cow reach a relative balance, the core temperature of the cow's body is relatively constant, which means that its health and production performance are normal. However, when the external environment is too cold, the heat dissipation is greater than the heat production, which will destroy this heat balance, cause changes in hormones in the cow's body, and eventually cause physiological dysfunction and cold stress.

[0004] Dairy cows in a cold stress state have decreased production performance and decreased disease resistance, and can even directly cause diseases. Therefore, evaluating the cold stress condition of dairy cows is the key to scientific management of dairy cow production.

[0005] Currently, the evaluation methods for cold stress of dairy cows include warm and hot environment evaluation index, wind and cold temperature, and comprehensive climate index. However, the above indexes only consider the warm and hot environment parameters, including temperature, humidity, wind speed and solar radiation, and cannot comprehensively reflect the cold stress condition of dairy cows. With the development of modern information technology, various physiological parameters of dairy cows can be accurately obtained through wearable or non-contact monitoring technology, and it is found that cold stress of dairy cows is significantly related to their physiological responses. Among them, respiratory rate and body surface temperature are the core indicators for evaluating whether the balance of ruminant body is achieved, and can most directly reflect the health status of dairy cows. This is because dairy cows in a cold stress state will reduce respiratory rate and increase respiratory depth to achieve the purpose of reducing body heat dissipation; and the skin, as the interface between the body and the environment, can reflect the changes made by the body to adapt to the environment. In addition, in winter, in order to prevent cold and keep warm, the breeding farm chooses to raise dairy cows in the shed all year round, and high-density breeding has become a common form of intensive production. Due to the increase of breeding density, the input of feed per unit area and the output of manure are also increased, which has a considerable impact on the air quality in the breeding shed. Manure storage, surplus feed and individual dairy cows will produce various air pollutants, including gaseous pollutants carbon dioxide (CO2), ammonia (NH3), hydrogen sulfide (H2S) and methane (CH4), and solid pollutants particulate matter (PM). Among them, CO2, NH3 and PM 10 have the highest concentration content and have the greatest impact on the production performance and health status of dairy cows, further deepening the cold stress degree of dairy cows in winter. SUMMARY

[0006] The present application aims to overcome the shortcomings of the current method for evaluating cold stress of dairy cows, and provides a multi-level fuzzy comprehensive evaluation method for cold stress of dairy cows, which can comprehensively and accurately evaluate the degree of cold stress of dairy cows, so as to better guide the cold prevention work in the process of dairy cow production and breeding.

[0007] The object of the present application can be achieved by the following technical solutions:

[0008] A multi-level fuzzy comprehensive evaluation method for cold stress of dairy cows comprises the following steps:

[0009] Step (1): According to literature investigation and expert consultation, factors related to cold stress of dairy cows are classified by characteristics, and a multi-level analysis method is used to construct a comprehensive evaluation index system for cold stress of dairy cows;

[0010] Step (2): Establish the judgment matrix of each index layer and conduct consistency test, if the consistency requirement is met, determine the weight of each index layer, otherwise, use genetic algorithm to optimize the weight of each index layer;

[0011] Step (3): Genetic algorithm is used to test and correct the judgment matrix, and the weight of each index layer is calculated through operations such as crossover and mutation, and finally the optimal feasible solution that meets the consistency of the judgment matrix is obtained;

[0012] Step (4): From the aspects of cold stress degree and applicability, set up the comment set, and establish the fuzzy evaluation matrix model of membership degree, and comprehensively evaluate the multi-level index factors, and finally determine the degree of cold stress of dairy cows according to the principle of maximum membership degree.

[0013] Further, in the step (1), the evaluation index system is divided into four levels, including target layer, constraint layer, criterion layer and scheme layer; the target layer is the comprehensive evaluation result of cold stress of dairy cows, the constraint layer is the first level index layer, including three dimensions of warm environment, physiological factors and air quality, the criterion layer is the second level index, further reflecting the content of the constraint layer, including temperature, relative humidity, wind speed, light, respiratory rate, body temperature, CO2, NH3 and PM 10 9 indexes, and the scheme layer is divided into five cold stress evaluation grades of none, mild, moderate, high and extreme.

[0014] Further, the step (2) comprises the following steps:

[0015] (2-1) Construct the judgment matrix: when there are factors affecting the evaluation object, then the set U={u1, u2, , un} called factor set, u i (i=1, 2, ,n) is an influencing factor; group each factor, set as U={U1,U2, ,U n}, wherein U i has m influencing factors, namely U i ={u i1 , u i2 , , u im} is a first-level index factor, u im is a second-level index factor, and satisfies the following two conditions:

[0016]

[0017] By using 1-9 scale method, experts quantize the relative importance between factors in each index layer, and construct a judgment matrix of pairwise comparison from the scale of quantization results; perform single ordering calculation on the judgment matrix to construct a judgment matrix , as follows:

[0018]

[0019] In formula (2), is the importance degree of element compared with , and >0, =1, =1 / ;

[0020] The 1-9 scale method represents that, compared with u i and u j , 1 represents that the two elements are equally important, 3 represents that u i is slightly more important than u j , 5 represents that u i is obviously more important than u j , 7 represents that u i is strongly more important than u j , 9 represents that u i is extremely more important than u j , and 2, 4, 6 and 8 are intermediate values of the above judgments.

[0021] (2-2) Calculate the weight index: adopt square root method to determine the weight of each index:

[0022] First, calculate the product of each row of the judgment matrix, that is:

[0023]

[0024] Secondly, the nth root of M i , that is:

[0025]

[0026] Finally, the vector is normalized, that is:

[0027]

[0028] According to formulas (3)-(5), the weight vector of the first-level evaluation index U is denoted as W=(w1, w2, , i w i ), 0≤w i ≤1; the weight vector of the second-level evaluation index U i is denoted as W i1 =(w i2 , w im ), 0≤w im ≤1;

[0029] (2-3) Consistency check of the judgment matrix: first, calculate the consistency index (CI) of the judgment matrix:

[0030]

[0031] In formula (6), n is the rank of the judgment matrix, n>1; is the maximum eigenvalue of the judgment matrix;

[0032] Secondly, calculate the consistency ratio (CR) of the judgment matrix:

[0033]

[0034] In formula (7), RI is the random consistency index of the judgment matrix. When 0≤CR<0.1, it indicates that the judgment matrix meets the consistency requirement, and the vector w is taken as the weight vector solution. However, when CR≥0.1, it indicates that the consistency check fails, and the weight of the obtained judgment matrix needs to be optimized.

[0035] Further, the step (3) includes the following steps:

[0036] (3-1) First, take the judgment matrix of the first-level evaluation index U={U1, U2, , n U } as an example, and the weight vector is W=(w1, w2, , i w

[0037]

[0038]

[0039] In formula (8), CIF(n) is a consistency index function, w k is the optimized weight vector; the closer CIF(n) is to 0, the better the corresponding weight value is to judge the matrix;

[0040] (3-2) Secondly, the population is randomly initialized, and the initial population is randomly set to generate a set of feasible solutions, wherein the individual coding method is selected as real number coding, and each individual is a real number string; the population size is n, the evolution times are 200, the crossover probability is 0.3, and the mutation probability is 0.2;

[0041] And the fitness function is constructed: the sum of the absolute values of the errors between the predicted output and the expected output of the objective function CIF(n) is taken as the individual fitness value F:

[0042]

[0043] In the formula, n is the number of network output nodes, Y i is the expected output of the i-th node of the objective function CIF(n), X i is the predicted output of the i-th node, and k is a coefficient;

[0044] At the same time, the selection operation is performed: the roulette method is selected as the selection strategy of the fitness proportion, and the selection probability of each individual i is p i :

[0045]

[0046]

[0047] In formulas (11) and (12), F i is the fitness value of individual i, and since the smaller the fitness value is, the better, the reciprocal of the fitness value is calculated before the individual is selected, and n is the number of population individuals;

[0048] (3-3) Then, the crossover operation and the mutation operation are completed: the real number crossover method is used for the crossover operation, and the j-th gene of the k-th chromosome c k and the j-th gene of the l-th chromosome c l are crossed:

[0049]

[0050] The j-th gene c ij of the i-th individual is selected for mutation operation, and the method is as follows:

[0051]

[0052] In the formula (13)~(14), b is the cross probability, c max is the last generation of gene c ij , c min is the next generation of gene c ij , r2 is a random number between [0,1], r2=0.2, g is the current iteration number, G max is the maximum evolution number, and r1 is the mutation probability.

[0053] (3-4) Finally, it is determined whether the updated weight value satisfies the constraint condition of the formula (9), if it satisfies, the iteration is terminated, and the corresponding optimal feasible solution is the best weight value; otherwise, the selection operation is performed again.

[0054] Further, the step (4) comprises the following steps:

[0055] (4-1) First, according to the dairy cow cold stress evaluation index system, the evaluation index factor set and the comment set are set: wherein the first evaluation index set U={(U1,U2,U3)}, the second evaluation index set U1=(u 11 ,u 12 ,u 13 ,u 14 ), U2=(u 21 ,u 22 ), U3=(u 31 ,u 32 ,u 33 ), the comment set V j =(V1,V2,V3,V4,V5), and V1~V5 in the comment set respectively represent no, mild, moderate, high and extreme 5 cold stress evaluation levels;

[0056] (4-2) Then, the division of the cold stress degree of 9 indexes is completed, the quantitative analysis is performed on the temperature, relative humidity, wind speed, respiratory rate, body temperature, CO2, NH3 and PM 10 , the trapezoidal distribution and the semi-trapezoidal distribution are selected to construct the membership function, each distribution is divided into small, medium and large, and the formula is as follows:

[0057]

[0058]

[0059]

[0060] In the formula (15)~(17), , , These are the membership functions for the smaller, intermediate, and larger types, respectively, x i The numerical value corresponding to the indicator. These are the boundary values ​​for each fuzzy set;

[0061] In addition, based on three aspects—lighting duration, artificial light intensity, and uniformity of light distribution—experts conducted a qualitative analysis of the cold stress degree of light indicators, drawing on their own experience.

[0062] (4-3) Finally, a multi-level fuzzy comprehensive evaluation model is established, mainly including the comprehensive evaluation of secondary indicators and the comprehensive evaluation of primary indicators; in the comprehensive evaluation of secondary indicators, the secondary indicator factor set U is evaluated. i ={u i1 ,u i2 , u im The uth im Each factor is evaluated individually to obtain a single-factor evaluation matrix, which determines u. im Comment V j The membership degree is used to obtain the single-factor evaluation results:

[0063]

[0064] Single-factor evaluation is actually a fuzzy mapping from the set of indicator factors to the set of comments. i By evaluating each factor in the equation, we can obtain U. i Fuzzy evaluation matrix R i :

[0065]

[0066] Secondary indicator factor U i The weight is W i =(w i1 ,w i2 ,⋯w im The comprehensive judgment results of the secondary indicators are as follows:

[0067]

[0068] In the comprehensive evaluation of the primary indicators, the results of the secondary evaluation are used as factor values ​​for the primary comprehensive evaluation, thus forming the secondary fuzzy evaluation set B. i =(B i1 B i2 B ij ) T The fuzzy relation matrix R, used as the primary comprehensive evaluation, is:

[0069]

[0070] The first-level evaluation index U weight vector is W=(w1, w2, …, w i ), so that the first-level index comprehensive judgment vector is:

[0071]

[0072] The synthesis operator of the fuzzy matrix is represented by b j The membership degree of the evaluation object to each comment is obtained, and finally the cold stress degree of the dairy cow is obtained according to the maximum membership degree principle.

[0073] The application proposes a dairy cow cold stress multi-level fuzzy comprehensive evaluation method, first adopts the analytic hierarchy process to construct a dairy cow cold stress comprehensive evaluation index system; secondly, the index weight is optimized and established in combination with a genetic algorithm, and the evaluation index factors are quantitatively and qualitatively analyzed; then, the comment set and membership function are established based on the fuzzy set theory, the dairy cow cold stress degree is divided into five levels of none, mild, moderate, high and extreme; finally, the dairy cow cold stress degree is obtained through multi-level fuzzy comprehensive evaluation. Compared with the prior art, the following advantages are obtained:

[0074] (1) The fuzzy comprehensive evaluation method is adopted in the application, which not only comprehensively considers the environmental indexes such as temperature, relative humidity, wind speed, illumination, NH3, CO2 and PM 10 in the cowshed, but also considers the physiological characteristics of the dairy cow such as respiratory rate and body surface temperature in the evaluation method, so that the dairy cow cold stress can be comprehensively evaluated.

[0075] (2) The knowledge structure and cognitive level of each expert are different, sometimes the constructed judgment matrix cannot meet the consistency requirement, the genetic algorithm can convert the judgment matrix consistency test into a nonlinear constraint optimization problem, the weight vector of the judgment matrix can be fine-tuned under the premise of retaining the original judgment information to the maximum extent, so that the judgment matrix meets the consistency, and the weights of the indexes at all levels are determined more objectively.

[0076] (3) The dairy cow cold stress comprehensive evaluation method not only overcomes the subjectivity of the expert experience which is not easy to be quantified, but also avoids the neglect of the index weight in the fuzzy evaluation process, is superior to the "simple exponential" model which only relies on the fitting of environmental parameters, and can accurately and effectively reflect the cold stress condition of the dairy cow. DETAILED DESCRIPTION

[0077] Figure 1 The dairy cow cold stress multi-level fuzzy comprehensive evaluation flowchart used in the application example

[0078] Figure 2 The dairy cow cold stress comprehensive evaluation index system diagram used in the application example

[0079] Figure 3A membership degree function diagram used for the example of the present application

[0080] Figure 4 A weight diagram of each level evaluation index used for the example of the present application

[0081] Figure 5 A diagram of the average daily cold stress duration of a dairy cow used for the example of the present application DETAILED DESCRIPTION

[0082] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the present application will be further described below in combination with the drawings of the specification.

[0083] As shown in Figure 1 : the dairy cow cold stress comprehensive evaluation method proposed by the present application includes the following four steps:

[0084] Step (1): according to literature investigation and expert consultation, factors related to dairy cow cold stress are classified by characteristics, and a dairy cow cold stress comprehensive evaluation index system is constructed by using a multi-level analysis method;

[0085] Step (2): a judgment matrix of each index layer is established, and consistency test is performed, if the consistency requirement is met, the weight of each index layer is determined, otherwise, the weight of each index layer is optimized by using a genetic algorithm.

[0086] Step (3): the genetic algorithm is mainly used for testing and correcting the judgment matrix, the weight of each index layer is calculated through operations such as crossover and mutation, and finally the optimal feasible solution that meets the consistency of the judgment matrix is obtained;

[0087] Step (4): on the basis of the above, from the aspects of cold stress degree and applicability, a comment set is set, and a membership degree fuzzy evaluation matrix model is established, the multi-level index factors are comprehensively evaluated, and finally the degree of dairy cow cold stress is determined according to the maximum membership degree principle.

[0088] In the above step (1), a dairy cow cold stress evaluation index system is established:

[0089] As shown in Figure 2 , the evaluation index system is divided into four levels, including the target layer, the constraint layer, the criterion layer and the scheme layer. The target layer is the core problem of the present application, that is, the construction of the dairy cow cold stress comprehensive evaluation index system; the constraint layer is the first level index, including three dimensions of warm environment, physiological factors and air quality; the criterion layer is the second level index, which further reflects the main content of the constraint layer, including nine indexes (temperature, relative humidity, wind speed, light, respiratory rate, body temperature, CO2, NH3 and PM 10 ); finally, the scheme layer is divided into five cold stress evaluation grades of none, mild, moderate, high and extreme.

[0090] In the above step (2), the judgment matrix is constructed, the weight index is calculated, and the consistency of the judgment matrix is tested. The specific steps are as follows:

[0091] (2-1) Construct the judgment matrix: when the factors affecting the evaluation object are , then the set U={u1, u2, , u n} is called the factor set, and u i (i=1, 2, ,n) is the influencing factor; group each factor, and set it as U={U1,U2, ,U n}, where U i has m influencing factors, i.e. U i ={u i1 , u i2 , , u im} is the first-level index factor, and u im is the second-level index factor, and satisfies the following two conditions:

[0092]

[0093] The relative importance between the factors in each index layer is quantified by experts using the 1-9 scale method, and the judgment matrix of pairwise comparison is constructed from the scale of the quantification result; the single ordering calculation of the judgment matrix is performed to construct the judgment matrix , as follows:

[0094]

[0095] In formula (2), is the importance of element compared with , and >0, =1, =1 / ;

[0096] The 1-9 scale method represents that u i and u j are compared, 1 represents that the two elements are equally important, 3 indicates that u i is slightly more important than u j , 5 indicates that u i is obviously more important than u j , 7 indicates that u i is strongly more important than u j , and 9 indicates that u i is more important than u jVery important, 2, 4, 6, 8 are the intermediate values of the above judgment.

[0097] (2-2) Calculate the weight index: the weight of each index is determined by the square root method:

[0098] First, calculate the product of each row of the judgment matrix, that is:

[0099]

[0100] Second, calculate the nth root of M i , that is:

[0101]

[0102] Finally, normalize the vector , that is:

[0103]

[0104] According to formulas (3) ~ (5), the weight vector of the first-level evaluation index U is denoted as W=(w1,w2, ,w i ), 0≤ w i ≤1; the weight vector of the second-level evaluation index U i is W i =(w i1 ,w i2 , ,w im ), 0 ≤ w im ≤1;

[0105] (2-3) Consistency test of the judgment matrix: first, calculate the consistency index (CI) of the judgment matrix:

[0106]

[0107] In formula (6), n is the rank of the judgment matrix, n>1; is the largest eigenvalue of the judgment matrix, (Aw) i represents the ith component of Aw;

[0108] Second, calculate the consistency ratio (CR) of the judgment matrix:

[0109]

[0110] In formula (7), RI is the random consistency index of the judgment matrix. When 0≤CR<0.1, it indicates that the judgment matrix meets the consistency requirement, and the vector w is taken as the weight vector solution. However, when CR≥0.1, it indicates that the consistency test fails, and the weight of the obtained judgment matrix needs to be optimized.

[0111] In the above step (3), the genetic algorithm is used to optimize the consistency check problem in the judgment matrix, so that the judgment matrix meets the consistency, and the optimization model is as follows:

[0112] (3-1) First, take the judgment matrix of the first-level evaluation index U={U1, U2, , U n} as an example, and the weight vector is W=(w1, w2, , w i ), and the optimization objective function is constructed:

[0113]

[0114]

[0115] In formula (8), CIF(n) is the consistency index function, w k is the optimized weight vector; the closer CIF(n) is to 0, the better the corresponding weight value is the best weight value of the judgment matrix;

[0116] (3-2) Secondly, randomly initialize the population, randomly set the initial population, generate a group of feasible solutions, and the individual coding method is selected as real number coding, each individual is a real number string, the population size is n, the evolution times are 200, the crossover probability is 0.3, and the mutation probability is 0.2;

[0117] And build the fitness function: take the sum of the absolute value of the error between the predicted output of the objective function CIF(n) and the expected output as the individual fitness value F:

[0118]

[0119] In the formula, n is the number of network output nodes, Y i is the expected output of the i-th node of the objective function CIF(n), X i is the predicted output of the i-th node, and k is the coefficient;

[0120] At the same time, selection operation is carried out: roulette wheel selection method is selected as the selection strategy of the fitness ratio, and the selection probability of each individual i is p i :

[0121]

[0122]

[0123] In formulas (11)-(12), F i is the fitness value of individual i, and since the smaller the fitness value is, the better, the fitness value is inverted before individual selection, and n is the number of population individuals;

[0124] (3-3) Then perform the crossover and mutation operations: The crossover method uses the real number crossover method, and the k-th chromosome c k and the lth chromosome c l Cross at position j:

[0125]

[0126] Select the j-th gene c of the i-th individual ij The mutation operation is performed as follows:

[0127]

[0128] In formulas (13) to (14), b is the crossover probability, and c max For gene c ij The previous session, c min For gene c ij The next iteration, r2 is a random number between [0,1], r2=0.2, g is the current iteration number, G max r1 represents the maximum number of evolutions and r1 represents the mutation probability.

[0129] (3-4) Finally, determine whether the updated weights satisfy the constraints of formula (9). If they do, terminate the iteration. The corresponding optimal feasible solution is the best weight value. Otherwise, perform the selection operation again.

[0130] In step (4) above, since the evaluation values ​​of each indicator are different, different levels are often formed. The set composed of various different judgments is called the comment set. This invention is based on the dairy cow cold stress evaluation index system ( Figure 1 The specific steps for setting the evaluation indicator factor set and the comment set are as follows:

[0131] (4-1) First, based on the aforementioned evaluation index system for cold stress in dairy cows, set the evaluation index factor set and the comment set: where the first-level evaluation index set U={(U1,U2,U3)}, and the second-level evaluation index set U1=(u 11 ,u 12 ,u 13 ,u 14 ), U2=(u 21 ,u 22 ), U3=(u 31 ,u 32 ,u 33 ), Comments Collection V j =(V1,V2,V3,V4,V5), where V1~V5 in the evaluation set represent five levels of cold stress: none, mild, moderate, high, and extreme, respectively;

[0132] (4-2) The application divides the cold stress degree of 9 indexes according to the environmental quality standard and carries out quantitative and qualitative analysis. In order to ensure the quality of livestock products, the northern large-scale dairy farm needs to follow the prescribed environmental quality standard, i.e. "dairy farm shed area, field area, buffer zone environmental quality standard (DB11 / T 426-2007)" in the actual breeding process. The standard is used for environmental quality control, monitoring and environmental management of the dairy farm. Among them, the range and threshold of temperature, relative humidity, wind speed, light, CO2, NH3 and PM 10 are shown in Table 1:

[0133] Table 1 Environmental quality standard of northern large-scale dairy farm

[0134] The units of CO2, NH3 and H2S specified in the standard are mg / m 3 , and the data unit collected by the environmental parameter multifunctional measuring instrument used in the application is ppm. Therefore, the application converts mg / m 3 into ppm as the data unit. The conversion formula between ppm and mg / m 3 is: volume concentration (ppm) = 24.5 x mass concentration (mg / m 3 ) / molecular weight.

[0135] The basic idea of the fuzzy mathematics model construction is the idea of membership degree, and the membership degree of each evaluation grade of dairy cow cold stress can be converted through the corresponding fuzzy mathematics language. The range of membership degree is [0, 1], and the greater the value, the more it belongs to the set. The application selects trapezoidal distribution and half trapezoidal distribution to construct the membership function, and each distribution is divided into small type, large type and intermediate type. Eight characteristic indexes (temperature, relative humidity, wind speed, respiratory rate, body temperature, CO2, NH3 and PM 10 ) are quantitatively analyzed using their respective membership functions, and the formulas are shown in (15)~(17).

[0136]

[0137]

[0138]

[0139] In formulas (15)~(17), , , respectively, the small type, intermediate type and large type membership degree functions, x i is the value corresponding to the index, is the boundary value of each fuzzy set;

[0140] Meanwhile, this invention uses the average daily milk yield of dairy cows of the same lactation age in spring as a benchmark, and classifies the degree of cold stress in respiratory rate and abdominal surface temperature according to the degree of milk yield reduction. This invention employs a generalized linear mixture model to conduct mathematical statistical analysis on the relationship between respiratory rate, abdominal surface temperature, and average daily milk yield. The analysis results show that all fixed-effects models in the generalized linear mixture model are statistically significant (F=1270.122, P=0.021<0.01; F=2319.157, P=0.014<0.01), with intercepts of -25.54 and -92.20 respectively (P<0.001, Table 5). The main effect of average daily milk yield is significant (P<0.001), with coefficients of 1.43 and 3.41 respectively, indicating that for every 1 kg decrease in milk yield, the respiratory rate decreases by 1.43 breaths / min and the body surface temperature decreases by 3.41℃. This invention categorizes respiratory rate and body surface temperature into five levels based on a 5% reduction in dairy production. Figure 3 Five membership function graphs for each of the eight indicators are given.

[0141] Furthermore, the uniformity of light distribution was judged by the hanging method, distance, and illumination range of the lights in the cowshed, as well as the presence of alternating light and dark phenomena in the cowshed. In this invention, winter-collected data and cowshed monitoring videos were sent to experts who scored the light indicators. The average score from 43 experts was taken as the final score for the light indicators. Tables 2 and 3 show the evaluation content and scoring criteria. The indicator scores correspond to the degree of cold stress in dairy cows; higher scores indicate a more positive impact of the light indicators on dairy cows. In this invention, the average scores given by experts for the light indicators were 76 and 58, respectively, indicating that the dairy cows on the north and south sides were experiencing "mild cold stress" and "moderate cold stress," respectively.

[0142] Table 2 Specific Evaluation Contents of Illumination Indicators

[0143]

[0144] Table 3. Correspondence between qualitative index evaluation scores and cold stress levels

[0145]

[0146] (4-3) Finally, a multi-level fuzzy comprehensive evaluation model is established, mainly including the comprehensive evaluation of secondary indicators and the comprehensive evaluation of primary indicators; in the comprehensive evaluation of secondary indicators, the secondary indicator factor set U is evaluated. i ={u i1 ,u i2 , u im The uth im Each factor is evaluated individually to obtain a single-factor evaluation matrix, which determines u.im Comment V j The membership degree is used to obtain the single-factor evaluation results:

[0147]

[0148] Single-factor evaluation is actually a fuzzy mapping from the set of indicator factors to the set of comments. i By evaluating each factor in the equation, we can obtain U. i Fuzzy evaluation matrix R i :

[0149]

[0150] Secondary indicator factor U i The weight is W i =(w i1 ,w i2 ,⋯w im The comprehensive judgment results of the secondary indicators are as follows:

[0151]

[0152] In the comprehensive evaluation of the primary indicators, the results of the secondary evaluation are used as factor values ​​for the primary comprehensive evaluation. Therefore, the secondary fuzzy evaluation set B... i =(B i1 B i2 B ij ) T The fuzzy relation matrix R, used as the primary comprehensive evaluation, is:

[0153]

[0154] The weight vector of the primary evaluation index U is W=(w1,w2,⋯,w i Thus, the comprehensive judgment vector of the first-level indicators is obtained as follows:

[0155]

[0156] b represents the composition operator of fuzzy matrices; j To determine the degree of membership of the evaluation object to each comment, the degree of cold stress experienced by the dairy cow is finally determined based on the principle of maximum membership.

[0157] The present application is evaluated by 30 experts in the field of dairy cattle breeding (including 12 experts in dairy cattle nutrition and feed science, 13 experts in dairy cattle production, and 5 experts in intelligent dairy cattle breeding) and 13 dairy cattle breeders using a 1-9 scale method to score the importance of each group of indicators, and 43 group weight vector values are obtained and mean value processing is performed. Table 4 only lists the scoring of the importance of each level of indicators by one expert and one breeder and the calculation process of the weight vector. Figure 4 To evaluate the maximum weight value of each index in the index system.

[0158] Table 4 Calculation of weight of each level of indicators of dairy cattle cold stress according to expert scoring

[0159]

[0160] Taking the data of any one dairy cow as an example: {(temperature: -7℃), (humidity: 84℃), (wind speed: 0.23m / s), (light: 76 minutes), (respiratory rate: 21.3 times / minute), (body temperature: 24.2℃), (CO2: 2639ppm), (NH3: 5.3ppm), (PM 10 : 230ug / m 3 )}, first, according to the single factor fuzzy evaluation of the secondary indicators u 11 , u 12 , u 13 , u 14 , the fuzzy evaluation matrix R1 of the warm environment U1 is obtained:

[0161]

[0162] According to Figure 4 , the weight of the secondary indicators of U1 is W1=[0.25,0.13,0.19,0.04], and the comprehensive determination result of the warm environment is as follows: B1=W1∘R1=[0.076,0.404,0.13,0,0]. Similarly, the comprehensive determination results of physiological factors and air quality indicators are as follows:

[0163]

[0164]

[0165] Secondly, the comprehensive evaluation of the primary indicators is carried out, and the secondary fuzzy evaluation set B i =(B1,B2,B3) T is used as the fuzzy relationship matrix of the primary comprehensive evaluation, and the weight of the primary indicators is W=[0.59,0.34.0.07]. The comprehensive determination vector B is calculated, and after normalization processing, the following is obtained:

[0166]

[0167] It can be seen that the evaluation result b2=0.65 is the maximum, according to the maximum membership principle, at this time the comprehensive evaluation of the cold stress of the dairy cow belongs to the second level V2 in the evaluation set V, and the comprehensive evaluation result is "mild cold stress".

[0168] Figure 5 The average daily cold stress duration of the south and north sides is mainly mild cold stress, but the duration is long. Among them, the average total duration of mild cold stress is 605.3h (25.22d), 725.5h (30.23d), the average duration of moderate cold stress is 67.2h (2.8d), 96h (4.0d), which mainly occurs in December, January and February. The cold stress duration of the south side is less than that of the north side by 120.2h, 28.8h.

[0169] The generalized linear mixed model is used for statistical data analysis of the daily average cold stress duration and the behavior characteristics (including daily average milk yield, daily average forage intake, daily average lying duration and daily average activity steps). The fixed effect parameter estimation results are shown in Table 5. The correlation between the daily average cold stress duration of the dairy cow and the behavior characteristics is significant (P<0.001), which verifies the feasibility of the proposed method. The cold stress comprehensive evaluation method proposed in the application can reflect the cold stress condition of the dairy cow in northern China in winter, and provides a new idea for evaluating the cold stress degree of the dairy cow in winter.

[0170] Table 5 Fixed effect parameter estimation results

[0171]

[0172] 1 Fixed effect parameter estimation results of daily average cold stress duration and daily average milk yield and behavior characteristics of the south side

[0173] 2 Fixed effect parameter estimation results of daily average cold stress duration and daily average milk yield and behavior characteristics of the north side

[0174] It should be noted that the purpose of publishing the embodiments is to help further understand the application, but those skilled in the art can understand that various substitutions and modifications are possible without departing from the spirit and scope of the application and the appended claims. Therefore, the application should not be limited to the disclosed content of the embodiments, and the scope of protection claimed by the application is defined by the scope of the claims.

Claims

1. A multi-level fuzzy comprehensive evaluation method for cold stress in dairy cows, characterized in that: Includes the following steps: Step (1): Based on literature review and expert consultation, the factors related to cold stress in dairy cows are classified by characteristics, and a comprehensive evaluation index system for cold stress in dairy cows is constructed using multi-level analysis. Step (2): Establish the judgment matrix for each indicator layer and perform a consistency check. If the consistency requirement is met, determine the weight of each indicator layer; otherwise, use a genetic algorithm to optimize the weight of each indicator layer. Step (3) uses a genetic algorithm to test and correct the judgment matrix. The weights of each index layer are calculated through crossover and mutation operations, and finally the optimal feasible solution that satisfies the consistency of the judgment matrix is ​​obtained. Step (4): From the perspective of cold stress degree and applicability, set up a set of comments and establish a membership degree fuzzy evaluation matrix model to comprehensively evaluate the multi-level index factors. Finally, based on the principle of maximum membership degree, determine the degree of cold stress in dairy cows. Step (4) includes the following steps: (4-1) First, based on the aforementioned comprehensive evaluation index system for cold stress in dairy cows, set the evaluation index factor set and the comment set: where the first-level evaluation index set U={(U1,U2,U3)}, and the second-level evaluation index set U1=(u 11 ,u 12 ,u 13 ,u 14 ), U2=(u 21 ,u 22 ), U3=(u 31 ,u 32 ,u 33 ), Comments Collection V j =(V1,V2,V3,V4,V5), where V1~V5 in the comment set represent five cold stress rating levels: none, mild, moderate, high, and extreme, respectively. (4-2) Then, classify the degree of cold stress based on nine indicators, including temperature, relative humidity, wind speed, respiratory rate, perceived temperature, CO2, NH3, and PM2.

5. 10 Quantitative analysis was conducted, and membership functions were constructed using trapezoidal and semi-trapezoidal distributions. Each distribution was categorized into small-scale, intermediate-scale, and large-scale distributions, as shown in the following formulas: In formulas (15)~(17), , , These are the membership functions for the smaller, intermediate, and larger types, respectively, x i The numerical value corresponding to the indicator. These are the boundary values ​​for each fuzzy set; In addition, based on three aspects—lighting duration, artificial light intensity, and uniformity of light distribution—experts conducted a qualitative analysis of the cold stress degree of light indicators, drawing on their own experience. (4-3) Finally, a multi-level fuzzy comprehensive evaluation model is established, mainly including the comprehensive evaluation of secondary indicators and the comprehensive evaluation of primary indicators; in the comprehensive evaluation of secondary indicators, the secondary indicator factor set U is evaluated. i ={u i1 ,u i2 , u im The uth im Each factor is evaluated individually to obtain a single-factor evaluation matrix, which determines u. im Comment V j The membership degree is used to obtain the single-factor evaluation results: Single-factor evaluation is actually a fuzzy mapping from the set of indicator factors to the set of comments. i By evaluating each factor in the equation, we can obtain U. i Fuzzy evaluation matrix R i : Secondary indicator factor U i The weight is W i =(w i1 ,w i2 ,⋯w im The comprehensive judgment results of the secondary indicators are as follows: In the comprehensive evaluation of the primary indicators, the results of the secondary evaluation are used as factor values ​​for the primary comprehensive evaluation, thus forming the secondary fuzzy evaluation set B. i =(B i1 B i2 B ij ) T The fuzzy relation matrix R, used as the primary comprehensive evaluation, is: The weight vector of the primary evaluation index U is W=(w1,w2,⋯,w i Thus, the comprehensive judgment vector of the first-level indicators is obtained as follows: b represents the composition operator of fuzzy matrices. j To determine the degree of membership of the evaluation object to each comment, the degree of cold stress experienced by the dairy cow is finally determined based on the principle of maximum membership.

2. The multi-level fuzzy comprehensive evaluation method for cold stress in dairy cows according to claim 1, characterized in that: In step (1), the evaluation index system is divided into four levels: target layer, constraint layer, criterion layer, and scheme layer. The target layer is the comprehensive evaluation result of cold stress in dairy cows. The constraint layer is a first-level index layer, including three dimensions: thermal environment, physiological factors, and air quality. The criterion layer is a second-level index layer, which further reflects the content of the constraint layer, including temperature, relative humidity, wind speed, light intensity, respiratory rate, perceived temperature, CO2, NH3, and PM2.

5. 10 There are a total of 9 indicators, and the proposed scheme is divided into 5 levels of cold stress evaluation: none, mild, moderate, high, and extreme.

3. The multi-level fuzzy comprehensive evaluation method for cold stress in dairy cows according to claim 1, characterized in that: Step (2) includes the following steps: (2-1) Constructing the judgment matrix: When the factors affecting the evaluation object are If there are , then the set U = {u1, u2, ...} , u n } is called the factor set, u i (i=1, 2, Let n be the influencing factors; group each factor into groups, denoted as U={U1,U2,...} U n }, where U i There are m influencing factors, i.e., U i ={u i1 , u i2 , , u im } is a primary indicator factor, u im It is a secondary indicator factor and meets the following two conditions: Experts used a 1-9 scale to quantify the relative importance of factors in each indicator layer. Based on the quantification results, pairwise comparison judgment matrices were constructed. A single-order calculation was then performed on these judgment matrices to construct the final judgment matrix. ,as follows: In formula (2), It is an element and The relative importance, and >0, =1, =1 / ; (2-2) Calculate the weighting indicators: The weight of each indicator is determined using the square root method: First, calculate the product of each row of the judgment matrix, that is: Next, calculate M. i The nth root, that is: Finally, the vector Normalization is performed, that is: Based on formulas (3) to (5), the weight vector of the first-level evaluation index U is obtained and denoted as W=(w1,w2, ,w i ), 0≤ w i ≤1, secondary evaluation index U i The weight vector is W i =(w i1 ,w i2 , ,w im ), 0 ≤ w im ≤1; (2-3) Consistency test of the judgment matrix: First, calculate the consistency index (CI) of the judgment matrix: In formula (6), n is the rank of the judgment matrix, and n > 1; It is the largest eigenvalue of the judgment matrix; Next, calculate the consistency ratio (CR) of the judgment matrix: In formula (7), RI is the random consistency index of the judgment matrix. When 0≤CR<0.1, it means that the judgment matrix meets the consistency requirements and the vector w is used as the solution of the weight vector. However, when CR≥0.1, it means that the consistency test has not passed and the weight of the obtained judgment matrix needs to be optimized.

4. The multi-level fuzzy comprehensive evaluation method for cold stress in dairy cows according to claim 1, characterized in that: Step (3) includes the following steps: (3-1) First, take the first-level evaluation index U={U1,U2, U n Taking the judgment matrix of} as an example, its weight vector is W=(w1,w2, ,w i Construct the optimization objective function: In formula (8), CIF(n) is the consistency index function, w k It is the optimized weight vector; the closer CIF(n) is to 0, the better the corresponding weight value is for the judgment matrix. (3-2) Next, the population is randomly initialized. The initial population is randomly set to generate a set of feasible solutions. The individual encoding method is real number encoding, each individual is a real number string, the population size is n, the number of evolutions is 200, the crossover probability is 0.3, and the mutation probability is 0.

2. And construct a fitness function: the sum of the absolute values ​​of the errors between the predicted output and the expected output of the objective function CIF(n) is used as the individual fitness value F: In the formula, n is the number of network output nodes, Y i Let X be the expected output of the i-th node of the objective function CIF(n). i Let k be the predicted output of the i-th node, and k be the coefficient. Simultaneously, a selection operation is performed: roulette wheel selection is chosen as the fitness-proportional selection strategy, and the selection probability for each individual i is p. i : In equations (11) to (12), F i Let be the fitness value of individual i. Since a smaller fitness value is better, the reciprocal of the fitness value is taken before individual selection. n is the number of individuals in the population. (3-3) Then perform the crossover and mutation operations: The crossover method uses the real number crossover method, and the k-th chromosome c k and the lth chromosome c l Cross at position j: Select the j-th gene c of the i-th individual ij The mutation operation is performed as follows: In formulas (13) to (14), b is the crossover probability, and c max For gene c ij The previous session, c min For gene c ij The next iteration, r2 is a random number between [0,1], r2=0.2, g is the current iteration number, G max r1 represents the maximum number of evolutions and r1 represents the mutation probability. (3-4) Finally, determine whether the updated weights satisfy the constraints of formula (9). If they do, terminate the iteration. The corresponding optimal feasible solution is the best weight value. Otherwise, perform the selection operation again.

5. The multi-level fuzzy comprehensive evaluation method for cold stress in dairy cows according to claim 3, characterized in that, The 1-9 scale method described above represents u i with u j When comparing two elements, 1 represents that both elements are equally important, and 3 represents u. i Than u j Slightly more important, 5 represents u i Than u j Clearly important, 7 represents u i Than u j Strongly important, 9 represents u i Than u j Extremely important, 2, 4, 6, and 8 are the intermediate values ​​of the above judgment.

Images