Method for generating standard growth curve for chicken flock breeding

Abstract

A method for generating a standard growth curve for chicken flock breeding comprises the following specific steps that a, the same kind of chicken flock is divided into N groups to be bred, and it is guaranteed that all chickens in each group of chicken flock are born on the same day; b, recording the daily average weight of each group of chicken flocks; c, taking the daily average weight of all groups of chicken flocks as a training set, learning in a learning model, averaging, and finally generating a daily age average weight standard growth curve of the same type of chicken flocks by means of a Gompertz fitting model; and d, changing any condition of variety, gender, region, henhouse control mode, climate and breeding mode to form a new type of chicken flocks, and repeating the steps b and c to obtain daily age average weight standard growth curves of different types of chicken flocks. The daily age average weight standard growth curve obtained by the method is more consistent with the actual weight of the chicken flocks, the average weight prediction of the chicken flocks is more scientific and accurate, and meanwhile, the problem of physical health of the chickens caused by a mode of catching the chickens by sampling is avoided.

Description

Technical Field

[0001] The present invention belongs to the technical field of poultry farming, and relates to a method for generating a standard growth curve for chicken flock farming. Background Art

[0002] Monitoring the body weight of chicken flocks is a key indicator in the process of large-scale chicken farming. Generally, farmers also manage the chicken flocks by regularly monitoring the body weight of the chicken flocks and their changing trends, and the aspects involved include health management, feeding management, etc.

[0003] In the prior art, farmers mostly refer to experience. After obtaining the average weight of the current chicken flock and the historical average weight, they can know the data situation of the current chicken flock through the change of the average weight, and then judge the health status of the current chicken flock, whether the nutrition is good, how to feed in the next few days, and estimate the average weight of the chicken flock before it is about to be sold through experience. This method relies too much on the richness of personal experience. For example, when two people with different experiences analyze the same chicken flock, the conclusions obtained are very likely to be two different conclusions. Even if a farmer has little experience, it is completely impossible to carry out good monitoring and management. It can be seen that the traditional method is very unscientific and there is no unified standardized management. Moreover, when estimating the future weight of the chicken flock, only personal experience can be relied on for valuation, and there is no reference data. Especially in large-scale chicken flock farming, it is impossible for one person to supervise all chicken flocks. Therefore, it is necessary to formulate certain standard data for the growth management of large-scale chicken flock farming. Summary of the Invention

[0004] In order to solve the technical problem of realizing standardized management in large-scale chicken flock farming, the present invention designs a method for generating a standard growth curve for chicken flock farming. Through the training of a large amount of historical data, a standard growth curve of average daily weight is finally obtained. Through the standard growth curve, the average weight of the chicken flock at a certain age under the same conditions can be predicted.

[0005] The technical solution adopted by the present invention is a method for generating a standard growth curve for chicken flock farming, and the steps of the generating method are as follows:

[0006] a. Set the same breed, the same gender, the same region, the same chicken house control method, the same climate, and the same breeding mode as a group of chicken flocks, and divide this group of chicken flocks into N groups for breeding, where N is a positive integer greater than or equal to 2, and ensure that all chickens in each group of chicken flocks are born on the same day;

[0007] b. Record the average weight of each group of chicken flocks every day;

[0008] c. The daily average weight of all groups of chickens is used as a training set, and the learning model is used to learn and average. Finally, the Gompertz fitting model is used to generate a standard growth curve of average weight per day under the same breed, same gender, same region, same chicken house control method, same climate, and same breeding mode;

[0009] d. Change any one of the conditions including breed, gender, region, chicken house control method, climate, and breeding mode to form a new type of chicken flock, repeat steps b and c, and obtain the standard growth curve of average weight per day for different types of chicken flocks.

[0010] The regions are divided according to the average annual temperature or according to the geographical location.

[0011] The chicken house control method includes an environmentally controlled chicken house and a non-environmentally controlled chicken house.

[0012] The breeding modes include flat breeding and cage breeding.

[0013] The number of chickens in each group shall not be less than 30.

[0014] In the step b, the daily average weight of the chickens is obtained by placing an electronic chicken scale in the chicken house, obtaining data by the chickens jumping on and off the electronic chicken scale, and processing the data through supporting software to obtain the daily average weight of the chickens. The specific steps are as follows:

[0015] b1. Eliminate unreasonable data samples from all weighed data samples in the chicken flock to obtain preliminary denoised samples;

[0016] b2. Use the k-means algorithm to find the average weight fuzzy value in the chicken group, and determine the maximum and minimum values ​​of the secondary screening through the average weight fuzzy value;

[0017] b3. Filter the samples after preliminary denoising with the help of the minimum and maximum values, and obtain the data samples between the maximum and minimum values ​​as valid data samples;

[0018] b4. Average the valid data samples to obtain the initial weighted value;

[0019] b5. Set the weight coefficient to correct the initial average weight value to obtain the final average weight value.

[0020] The beneficial effect of the present invention is that the present invention obtains a standard growth curve by summarizing a large amount of historical data, which can be used for a series of matters such as breeding monitoring management and growth prediction, and provides a standard for breeders, thereby avoiding the problem of management ability differences caused by differences in personal experience.

[0021] In terms of obtaining the average weight data, since manual sampling in the prior art can cause stress to the chicken flock, weighing cannot be carried out daily. However, with the electronic chicken scale of the present invention, a large amount of historical data can be obtained daily. The average of the data of multiple groups of average weights under the same conditions can be calculated, and after learning and training, a standard growth curve can be generated. Through the standard growth curve, the distribution of the average weight of each day in the same kind of chicken flock can be obtained, thus avoiding the problem of inaccurate average weight data caused by sampling in the prior art. If one chicken scale is configured for every 2,000 chickens, the number of samples collected daily for each chicken flock is at least ≥1,000, which is more representative.

[0022] Secondly, the traditional manual sampling and weighing method will also cause stress to the chicken flock. After the chickens are stressed, their recent eating and drinking will be abnormal, which is time-consuming and laborious and also affects the healthy growth of the chicken flock. The method of the present invention does not require grasping and weighing the chickens, and prediction can be achieved through the standard growth curve, avoiding the stress reaction of the chickens and ensuring the healthy growth of the chicken flock.

[0023] Moreover, due to the low frequency of sample collection in the prior art, the data update frequency is also low, and the time period for machine learning is too long. However, the standard growth curve obtained by the method of the present invention can achieve daily weighing, so a large amount of new sample data can be obtained within one month. Then, data iteration is carried out with the previously generated standard growth curve. The iteration frequency is high, which helps to quickly update the learning data to obtain a more accurate standard growth curve. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 It is a diagram showing the relationship between the actual scenario requirements of the standard growth curve.

[0025] Figure 2 It is a schematic diagram of an embodiment of the standard growth curve of Lezhu Service Department.

[0026] Figure 3 It is a schematic diagram comparing the standard growth curve and the production curve of Lezhu Service Department. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0027] When the present invention is specifically implemented, obtaining the standard growth curve of the average weight at different ages requires a large amount of historical data, and the time for obtaining these data is relatively long. In order to make the standard growth curve more accurate, 2 years are used to obtain and summarize the breeding data.

[0028] During specific breeding, chicken flocks under the same conditions are classified into one category. For example, in spring in a southern city of China, a batch of yellow jade chickens raised in a flat and environmentally controlled manner are divided into 20 groups according to female and male respectively, and the number of chickens in each group is set to 50.

[0029] During daily weighing, farmers place intelligent electronic chicken scales in the chicken coop, constantly obtaining the weight data of chickens jumping onto the scale pans, and performing simple debouncing at the terminals in the electronic scales, then reporting the actual weights to the cloud. The method for obtaining the daily average weight is divided into four steps:

[0030] 1) Eliminate invalid data based on experience values

[0031] Based on experience values, the age at which a chicken can jump onto the scale, and the weight of a single chicken ≥ 50g. Therefore, all data < 50g will be eliminated as invalid data, and the resulting sample is called the original sample. Taking a certain chicken scale in a certain farm as an example, a total of 363 sample data were obtained on the same day. The valid samples are 325, and the average weight is 2.3 catties. In this step, 17 data such as null values, 0 values, and data significantly less than 50g were eliminated by removing unreasonable data samples. The above unreasonable data samples are related to the time when the chicken jumps onto the scale and stays. In extremely short time or other environmental factors, it may cause the reading of the scale to be a null value or 0 value.

[0032] 2) Use k-means to find the range where the average weight distribution of the chicken flock is relatively large

[0033] Denoising is performed through the k-means unsupervised machine learning algorithm. The principle of denoising is as follows: First, randomly generate k clustering centers in the data, then calculate the distance from each sample in the data to these k clustering centers, and assign the corresponding sample to the cluster corresponding to the clustering center with the smallest distance. After all samples are classified, recalculate the clustering center of each of the k clusters, that is, the centroid of all samples in each cluster, and repeat the above operations until k-means traverses the data. The centroid corresponds to the weight approximately equal to the average weight of the chicken flock, which is temporarily called the average weight fuzzy value. Suppose the average weight fuzzy value obtained by clustering through the k-means algorithm in this example is 2.1 catties.

[0034] 3) Determine the minimum and maximum values of the valid samples for secondary screening based on the average weight fuzzy value

[0035] Generally, the weight distribution of the chicken flock follows a normal distribution. Suppose we take Min as 0.7 times and Max as 1.4 times, and multiply the average weight fuzzy value obtained by k-means by 0.7 and 1.4 respectively, then the minimum weight is 1.47 catties and 2.94 catties. Secondary screening is performed within this interval to obtain valid data samples. In this example, through secondary screening and filtering, another 21 data samples were filtered out, and finally 325 valid data samples were obtained.

[0036] 4) Correct the arithmetic mean of the valid samples through the weight coefficient

[0037] After the above three steps, effective data samples are obtained. By taking the arithmetic mean of the effective samples, the initial average weight and the variance can be obtained. The variance is used to describe the weight distribution of the sample chickens.

[0038] In the pilot stage of the electronic chicken scale, the actual average weight obtained when the chicken flock is sold is often 2-3% higher than the initial average weight obtained by arithmetic mean. After research, the following two influencing factors are found:

[0039] First, the electronic chicken scale can read the weight readings of the chickens that jump onto the scale, but it cannot ensure that each chicken jumps onto the scale only once. That is to say, the 363 samples in the embodiment do not mean that there are 363 chickens in the chicken coop, and it cannot ensure that all chickens have the same number of times of getting on the scale. In fact, it is possible that some chickens jump onto the scale multiple times, and some chickens may not have jumped onto the scale at all.

[0040] Second, as is well known to those skilled in the art, according to the growth characteristics of the chicken flock, when the age is short, the relatively active chickens have a relatively large weight. These are the relatively healthy chickens in terms of growth. Therefore, the chickens with a weight exceeding the average weight are more active, and their weight in the sample is higher than the actual weight. If a simple arithmetic mean is taken, the obtained average weight will be greater than the actual average weight. On the contrary, as the age increases, some overweight chickens have reduced activity due to excessive weight, and the weight of the chickens exceeding the average weight of the chicken flock in the sample will be less than the actual weight. If a simple arithmetic evaluation is carried out, the obtained average weight of the chicken flock will be lower than the actual average weight.

[0041] Based on the above two factors, it is necessary to perform weighted average correction on the initial average weight according to the empirical value. Specifically, according to the growth cycle of the actual chicken flock, the correction idea is that at an early age, the average weight obtained by arithmetic mean needs to be compensated by a weight coefficient <1, and at a large age, the average weight of the chicken flock obtained by arithmetic mean needs to be compensated by a weight coefficient >1. Only in this way can the result after the weight coefficient correction be closer to the actual situation. Because the actual average weight of the chicken flock can only be obtained after it is sold, the average weight obtained by arithmetic mean from the samples obtained by the electronic chicken scale before slaughter is very likely to be 2%-3% lighter than the actual average weight. For other ages, since the actual average weight of the chicken flock cannot be obtained, it is very difficult to obtain the empirical value. The only available reference is the experimental data of the experimental farm. For example, the actual average weight of the pilot chicken flock is obtained by full sampling and weighing every week. If curve fitting is carried out and then introduced into the algorithm, it will increase the complexity of the algorithm. In addition, since this influencing factor is within 3%, it will not have too much impact on the production process management.

[0042] The average weight of all female and male yellow - feather chickens per day is obtained above. Of course, the chickens in each group should be born on the same day. For example, if they are raised for 60 days before being sold, the average weight data for each of the 60 days can be obtained. Of course, the average weight data for a period of time after generation actually has little reference significance for sales and is mainly used for predicting the growth and health status, which is helpful for daily feeding management. However, the data later is of great significance for batch sales.

[0043] Next, these historical data need to be imported into the learning model for learning and averaging. Finally, two sets of average - weight growth curves for different ages of this category are obtained. In other times of the first year, or the same period in the second year, or when other climatic conditions are similar, multiple sets of curves can be obtained and then averaged to ensure the accuracy of the standard growth curve.

[0044] In this way, the average - weight growth curve of this type of chicken group is obtained, which can be used to predict the average weight of the chicken group to be predicted in the future. Of course, after obtaining the actual data of the future chicken group, it can be put back into the learning model for learning. Through continuous data learning, the average - weight growth curve will be more accurate and standard.

[0045] Similarly, changing any of the other conditions will generate the average - weight growth curves of different categories of chicken groups. This is because under these conditions, there are significant differences in the average weight. For example, when the temperature is suitable, the chicken group grows relatively fast, while in relatively cold or hot areas, the chicken group grows relatively slowly. In addition, different breeding methods and chicken - house control methods will also result in differences in the muscle content of chickens. Therefore, six conditions are set, namely breed, gender, region, chicken - house control method, climate, and breeding mode. Especially for chicken breeds, different breeds have significant differences in chicken weight, and accurate prediction cannot be made even under other equal conditions.

[0046] The verification process of this method is also carried out in the present invention, which is specifically as follows:

[0047] First of all, in order to verify the universality and popularizability of the standard growth curve of the chicken group, we included all chicken groups that have been equipped with electronic chicken scales since 2022, have completed settlement, and have first - batch sales data and relevant financial indicators in the data set. The financial indicators include settlement average weight, feed - to - meat ratio, market average weight, market rate, settlement gross profit, etc.

[0048] Secondly, considering that the main application scenario of the standard growth curve is to provide reference for front-line production and a baseline for easy comparison, it is not very appropriate if the sample chicken flocks for generating the standard growth curve span several years. Additionally, in many cases, in order to guide front-line farmers to improve production performance, the standard growth curve is not the average value of all sample chicken flocks that meet the preliminary conditions in this region. Instead, it is often slightly higher than the average value of the set. That is to say, when screening sample chicken flocks, those chicken flocks with slightly better production performance will be selected. Therefore, each standard growth curve provides the key indicators such as feed-to-meat ratio, marketable rate, and settlement gross profit for screening and filtering, and finally determines the sample chicken flocks that ultimately participate in the calibration.

[0049] Refer to Table 1. In the development environment, all sample chicken flocks with electronic chicken scales put into use since 2022 were imported into the dataset, and the selected standard growth curves were obtained, with an additional screening condition of effective sample chicken flock number ≥ 30.

[0050] Table 1 Statistics of Standard Growth Curves in UAT Environment

[0051]

[0052] The supporting software architecture and algorithm logic of the method of the present invention are as follows:

[0053] Based on the Linux environment, it is developed using Python language and NumPy library, and supports Docker deployment. The details are as follows:

[0054] a. Select and set up a system development and operation environment for the chicken body weight standard algorithm. First, a Dockerfile is created in the Docker container to build a Python environment containing relevant dependency libraries (such as Scipy, NumPy, etc.).

[0055] b. Based on the conditions configured according to the rules of the standard growth curve, preprocess the selected sample chicken flocks, and eliminate those that do not meet the requirements in terms of business (such as chicken flocks with too low gross profit margin or too high feed-to-meat ratio).

[0056] c. Using the Python language, based on the SciPy, NumPy libraries and the Gompertz model, an algorithm for the chicken body weight standard of non-linear least squares multivariate differential gradient descent is implemented. And the JarqueBera test and Ljung-Box test are used to test the data distribution characteristics and time series correlation.

[0057] d. To ensure that the standard growth curve can truly reflect the growth trend of the chicken flock, there is a constraint in the "rule configuration" that the number of sample chicken flocks ≥ 30 (the default is 30, recommended based on statistical experience). After obtaining the standard growth curve of the chicken flock, the results will be screened based on this constraint to obtain the final subset of the standard growth curve of the chicken flock. For example, in the test environment, only 19 standard growth curves are actually output.

[0058] e. In addition, to analyze the fitted standard growth curve, this project also calculates the arithmetic mean based on the selected sample chicken flock data according to the age in days, and obtains a reference broken line that matches the actual situation of the sample chicken flock weight. At the same time, I compare the mean value of the fitted standard growth curve at this age with the mean value of the reference broken line, and use the degree of coincidence to represent the deviation between the two.

[0059] The specific verification steps are as follows:

[0060] First, we analyzed the application scenarios, data flow, etc. of the standard growth curve, as Figure 1 shown. The following three aspects need to be considered:

[0061] Problem - In traditional farmer management, the reference curve provided by the service department is static and dead data; to facilitate farmers' benchmarking, it is recommended to use the actual data on the production front line based on the enterprise's self-defined production department, such as a subsidiary company in a certain region. Specifically for Wenshi, it is recommended to the fourth-level company, such as Chegang Branch, and use Gompertz to fit a standard growth curve of the chicken flock weight. In addition, the average value can be calculated based on the sample chicken flock for specific ages to obtain a reference broken line of the actual average weight. The broken line is not smooth but serrated.

[0062] Goal - In the first ten days of each month, based on the weight data of the settled chicken flock, according to the rule configuration of the standard growth curve, timely update the standard growth curve of the enterprise-specified area, including four dimensions: breed / gender / feeding mode / whether environmental control. It should be noted that the standard growth curve that is not updated in real time in the existing technology only considers year-on-year comparison. Incorporating the data of the previous month for fitting here is equivalent to adding month-on-month comparison.

[0063] Application - First, for any specific chicken flock, the comparison between the reference broken line of the actual average weight and the standard growth curve can be provided to facilitate the process management and early warning prompts of the service department. Second, because the standard growth curve basically represents the actual average weight of this production department, the standard growth curves of different production departments can be compared horizontally, and the breeding performance can be compared horizontally for a certain breed between different fourth-level units, which is convenient for discovering problems and making timely adjustments and optimizations.

[0064] Secondly, for the calculation of the fitting degree with the retrograde standard growth curve, in order to test the rationality and scientificity of the standard growth curve, we introduced the fitting degree index. The fitting degree is an index that describes the deviation degree between the standard growth curve and the reference curve measured actually from the samples. First, calculate the difference by comparing the arithmetic mean of the average weight corresponding to a certain age after fitting with the actual average weight of the sample chicken flock at a certain age; secondly, calculate the geometric mean of the differences in the age range involved in the standard growth curve. Here, the geometric mean is selected instead of the arithmetic mean; finally, calculate a percentage with this geometric mean and the average weight of the first batch sale of the chicken flock. This percentage is the fitting degree. Note: Since the average weight of the first batch sale is around 2 kg, so 0.01 is approximately a deviation of ±20 g.

[0065] In the specific verification, in the first step, we need to standardize the error index. This can be achieved by dividing by the theoretically maximum error, which is usually the difference between the possible maximum and minimum values. Assuming the range of ytrue is known, we can use this range to standardize the error.

[0066] For the mean absolute error (MAE):

[0067]

[0068] For the root mean square error (RMSE):

[0069]

[0070] In the second step, to calculate the fitting degree, we can define the fitting degree as 1 minus the standardized error. The formula is as follows:

[0071] For the mean absolute error:

[0072] Agreement MAE = 1 - MAE normalized

[0073] For the root mean square error:

[0074] Agreement RMSE = 1 - RMSE normalized

[0075] In the third step, we convert the fitting degree into a percentage form:

[0076] Agreement MAE (%) = Agreement MAF × 100%

[0077] Agreement RMSE (%) = AgreementRMSE ×100%

[0078] In this way, the percentage representations of the two degrees of coincidence are obtained, and the closer they are to 100%, the better the prediction fits the actual data.

[0079] The following gives a specific embodiment, the verification of the standard growth curve of the Lezhu Service Department.

[0080] During the verification of the standard growth curve results, taking 13 chicken flock samples output by the Lezhu Service Department as an example, see Table 2.

[0081] Table 2 13 groups of chicken flock samples of the Lezhu Service Department

[0082]

[0083] Among them, the total number of sample chicken flocks involved is 1183, and the number of valid sample chicken flocks is 981; the maximum settlement age is 121 days. Note: At present, the starting age and settlement age of the standard growth curve are respectively the minimum age and the maximum age of the sample chicken flock. The breeds included are 10 breeds such as Datu No. 2, Tianlu Qingma Chicken No. 2, Tianlu Grass Chicken Castrated, Tianlu Black Chicken No. 5, Shankeng Phoenix, Yupin Chicken, Yupin Chicken Castrated, Yao Chicken, Aijiao Huang Chicken A, and Bamboo Silk Chicken No. 6; Gender, 13 standard growth curves, of which 4 are pure male and 9 are pure female; Other instructions, the sample chicken flocks are all in the free-range mode, not in the environmentally controlled house.

[0084] In addition, in order to more intuitively display the comparison between our latest standard growth curve and the production curve currently used by the front-line production department, on the display page of each standard growth curve, the actual production curve and the standard growth curve are loaded at the same time, see Figure 2 .

[0085] Taking the 13 standard growth curves of the Lezhu Service Department as the experimental data set, the functions and performances of the algorithms of these 13 chicken flock standard growth curves are evaluated, including the following indicators:

[0086] a. Verification of the degree of coincidence between the standard growth curve and the actual measured values of the sample chicken flock, see Table 3.

[0087] Table 3 Result statistics of 13 standard growth curves of the Lezhu Service Department

[0088]

[0089] 1) Degree of coincidence - represents the degree of deviation between the average weight corresponding to a certain age of the fitted standard growth curve and the arithmetic mean of the average weights corresponding to that age of the sample chicken flock. Since the average weight corresponding to the age of the sample chicken flock is a reference broken line, the difference in the degree of coincidence here is not as convergent as the growth prediction curve.

[0090] 2) Analysis of sample chicken flocks - A total of 1,183 sample chicken flocks were involved, among which 981 were valid sample chicken flocks, accounting for 82.92%, which is relatively high.

[0091] Note: The average degree of coincidence of the standard growth curve of Lezhu Service Department is 4.3%. This is mainly because the age - average weight of the sample chicken flock is originally a reference broken line, while the standard growth curve fitted based on the Gompertz curve model is a smooth curve. It is normal for there to be a deviation between the two. In addition, considering the application scenario of the standard growth curve, this degree of coincidence does not affect its use in specific business scenarios.

[0092] b. For the visual comparison between the standard growth curve and the production curve, see Figure 3 。

[0093] 1) Definition of the production curve - This is a standard growth curve manually produced by the third - level company based on empirical data and has been distributed to the front - line production departments and is currently being implemented. In order to better lead the front - line to move upward, most of the time the production curve is slightly higher than the average value of the actual chicken flocks in the region.

[0094] 2) Comparison between the production curve and the standard growth curve - Taking the experimental results of Lezhu Service Department as an example, for the standard growth curve numbered 1818100716325105666 (variety: Datu 2; gender: female; environmental control environment: non - environmental control; feeding mode: free - range; sample chicken flock: 98; age range: 35 - 97), after 39 days of age, the average weight of the chicken flock corresponding to the age of the standard growth curve starts to be lower than that of the production curve. By 90 days of age, the standard growth curve is 158.22 g lower than the production curve. This precisely verifies that the standard growth curve currently distributed to the front - line production departments is slightly higher than the average value of the actual measured data of chicken weights. In view of this, a series of screening conditions including feed - to - meat ratio and settlement gross profit are reserved for the business department in the rule configuration of the standard growth curve during this experimental process, aiming to facilitate the business department to screen sample chicken flocks according to actual needs and obtain the standard growth curve they expect.

[0095] Based on the results and analysis of the verification of the standard growth curve algorithm in this time, we draw the following conclusions:

[0096] a. The standard growth curve algorithm already has the conditions for large - scale promotion - Based on the big data of chicken weights collected by the electronic chicken scale and combined with the provided rule configuration of the standard growth curve, it can already meet the efficient customization of the standard growth curve with finer granularity by the business department. It is recommended to select relevant companies that have already installed electronic chicken scales to start small - scale application and gradually expand to other companies.

[0097] b. Iterative upgrade direction of the standard growth curve algorithm - First, it is necessary to promote the unity of consciousness of relevant personnel through standardization and normalization. During this verification process, it was found that the screening conditions for the sample chicken flock can be flexibly defined, which directly determines the proportion of effective sample chicken flocks. At present, the standards of different business departments are inconsistent. It is recommended to reach a consensus as much as possible through organizational exchanges to create more opportunities for value mining of standard growth curves. If the consciousness is not similar and the standards are not unified, there will be obstacles to horizontal comparison of the third-level and fourth-level companies at the group level. Secondly, through brainstorming, we will continue to explore the application scenarios of the standard growth curve. As long as we can find the value realization point and discover the needs, we will naturally continue to iterate and upgrade, continuously optimize, and achieve sustainable development.

[0098] c. The final conclusion is: if the degree of fit is higher, it means the deviation is smaller; if the degree of fit is 100%, it means the standard growth curve is completely consistent with the actual measurement data.

Claims

1. A method for generating a standard growth curve for chicken farming, characterized in that: The steps of the generation method are as follows: a. Chickens of the same breed, gender, region, house control method, climate, and breeding mode are classified into one group, and these chickens are divided into N groups for breeding, where N is a positive integer greater than or equal to 2, and it is ensured that all chickens in each group are born on the same day; b. Record the average daily weight of each group of chickens; c. The daily average weight of all groups of chickens is used as a training set, and the learning model is used to learn and average. Finally, the Gompertz fitting model is used to generate a standard growth curve of average weight per day under the same breed, same gender, same region, same chicken house control method, same climate, and same breeding mode; d. Change any one of the conditions including breed, gender, region, chicken house control method, climate, and breeding mode to form a new type of chicken flock, repeat steps b and c, and obtain the standard growth curve of average weight per day for different types of chicken flocks.

2. The method for generating a standard growth curve for chicken farming according to claim 1, characterized in that: The regions are divided according to the average annual temperature or according to the geographical location.

3. The method for generating a standard growth curve for chicken farming according to claim 1, characterized in that: The chicken house control method includes an environmentally controlled chicken house and a non-environmentally controlled chicken house.

4. The method for generating a standard growth curve for chicken farming according to claim 1, characterized in that: The breeding modes include flat breeding and cage breeding.

5. The method for generating a standard growth curve for chicken farming according to claim 1, characterized in that: The number of chickens in each group shall not be less than 30.

6. The method for generating a standard growth curve for chicken farming according to claim 1, characterized in that: In the step b, the daily average weight of the chickens is obtained by placing an electronic chicken scale in the chicken house, obtaining data by the chickens jumping on and off the electronic chicken scale, and processing the data through supporting software to obtain the daily average weight of the chickens. The specific steps are as follows: b1. Eliminate unreasonable data samples from all weighed data samples in the chicken flock to obtain preliminary denoised samples; b2. Use the k-means algorithm to find the average weight fuzzy value in the chicken group, and determine the maximum and minimum values ​​of the secondary screening through the average weight fuzzy value; b3. Filter the samples after preliminary denoising with the help of the minimum and maximum values, and obtain the data samples between the maximum and minimum values ​​as valid data samples; b4. Average the valid data samples to obtain the initial weighted value; b5. Set the weight coefficient to correct the initial average weight value to obtain the final average weight value.

Images