Prediction method suitable for growth average weight of chicken flocks bred in large scale
By adopting the prediction method of standard growth curve and support vector machine model in chicken breeding, the problem of inaccurate prediction of chicken flock average weight in the existing technology is solved, and high-precision average weight prediction and chicken flock health management are achieved.
Patent Information
- Application Number
- CN202411304846.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-19
- Publication Date
- 2025-01-21
AI Technical Summary
In the prior art, when estimating the average weight of chickens, manual sampling and manual catching chickens are used, resulting in low estimation accuracy and affecting the health of chickens.
A prediction method suitable for large -scale breeding chicken group growth is designed. Through the standard growth curve and support vector machine model, it fits the average weight prediction curve in the future to achieve accurate predictions on the average of chickens.
By simulating the average weight prediction curve, the prediction accuracy can be controlled at ± 20g, which is much higher than artificial experience judgments, and automatically sampled through electronic chicken scale, reducing the stress of the chicken, realizing the entire process of chicken weight data data Tracking and management.
Smart Images
Figure FT_1 
Figure FT_2 
Figure FT_3
Abstract
Description
Technical Field
[0001] The invention belongs to the technical field of poultry breeding, and relates to a method for predicting the average growth weight of chickens in large-scale breeding. Background Art
[0002] In the prior art, when raising chickens on a large scale, weight estimation is generally required, that is, the average weight of the chickens must be estimated some time before the day of slaughter, in order to efficiently sell the chickens in batches. For example, if a chicken flock is to be slaughtered in 10 days, the average weight of the chicken flock must be estimated at present, so as to make a scheduled sale based on the weight of the chicken flock in 10 days.
[0003] However, the existing technology generally uses manual sampling and manual catching of chickens to estimate the average weight of a flock of chickens. Assuming that there are 100 chickens in a flock, about 20 are caught and weighed, which is used as a reference for the current 100 chickens. After the average weight is obtained, the average weight in 10 days is estimated based on daily experience, and the labels are marked for pre-sale. Obviously, this method is neither scientific nor accurate, and the estimation accuracy is low. The deviation between the estimated weight of this manual sampling and the actual settlement average is 100 to 150 grams. In addition, when the chickens are caught by manual sampling, they are frightened and stressed, resulting in special eating and drinking conditions for the chickens in the next few days, which affects the health of the chickens. Summary of the invention
[0004] In order to solve the technical problem of accurately predicting the future average weight of a flock of chickens, the present invention designs a prediction method for the average weight of growing chickens in large-scale breeding. The standard growth curve is used as a reference, and the average weight prediction curve for a period of time in the future is fitted by combining the actual average weight curve with a support vector machine model.
[0005] The technical solution adopted by the present invention is a prediction method for the average weight of growing chickens suitable for large-scale breeding. The steps of the prediction method are as follows:
[0006] a. Weigh the predicted flock every day to obtain the actual average weight curve;
[0007] b. The actual average weight curve is fitted with the standard growth curve with the same environmental parameters as the tested chicken flock with the help of the support vector machine model to obtain the average weight prediction curve of the tested chicken flock.
[0008] The steps of the method for generating the standard growth curve are as follows:
[0009] s1. Set the chickens with the same environmental parameters as a type of chickens, and divide such chickens 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 chickens are born on the same day;
[0010] s2. Record the daily average weight of each group of chickens;
[0011] s3. 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 environmental parameters;
[0012] s4. Change any one of the environmental parameters to form a new type of chicken flock, repeat steps s2 and s3, and obtain the standard growth curve of the average weight per day of different types of chicken flocks.
[0013] The daily average weight of the flock is obtained by placing an electronic chicken scale in the chicken house, obtaining data by having the chickens jump on and off the electronic chicken scale, and processing it through supporting software to obtain the daily average weight of the flock. The specific steps are as follows:
[0014] j1. Eliminate unreasonable data samples from all weighed data samples in the chicken flock to obtain preliminary denoised samples;
[0015] j2. 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;
[0016] j3. Filter the samples after preliminary denoising with the help of the minimum value and the maximum value, and obtain the data samples between the maximum value and the minimum value as the valid data samples;
[0017] j4. Average the valid data samples to obtain the initial weighted value;
[0018] j5. Set the weight coefficient to correct the initial average weight value to obtain the final average weight value.
[0019] The environmental parameters include breed, gender, region, chicken house control method, climate, and breeding mode.
[0020] The beneficial effect of the present invention is that through the simulated average weight prediction curve, the predicted average weight data on any day can be obtained. Combined with experimental verification, the prediction accuracy can be controlled within ±20g, which is much higher than manual experience judgment. At the same time, based on the sample data of the fully automatic and stress-free sampling of the electronic chicken scale, the measured value obtained by unsupervised machine learning and the actual average weight can basically realize the whole process tracking and management of the chicken group weight data, and finally realize the precise regulation of the growth of the chicken group. BRIEF DESCRIPTION OF THE DRAWINGS
[0021] Figure 1 This is a comparison chart of the forecast and historical curves of flock number 07003-45 of Lezhu Service Department.
[0022] Figure 2 This is a comparison chart of the forecast and historical curves of flock number 07120-45 of Lezhu Service Department.
[0023] Figure 3 This is a comparison chart of the forecast and historical curves of the flock number 07893-32 of Lezhu Service Department.
[0024] Figure 4 This is a comparison chart of the forecast and historical curves of the flock number 08360-30 of Lezhu Service Department.
[0025] Figure 5 This is a comparison chart of the forecast and historical curves of the flock number 1102266-13 of Lezhu Service Department.
[0026] Figure 6 This is a comparison chart of the forecast and historical curves of the flock number 1102291-15 of Lezhu Service Department.
[0027] Figure 7 This is a comparison chart of the forecast and historical curves of flock number 112237-15 of Lezhu Service Department.
[0028] Figure 8 This is a comparison chart of the forecast and historical curves of the flock number 112307-15 of the Lezhu Service Department.
[0029] Fig. 9 This is a comparison chart of the forecast and historical curves of flock number 112555-14 of Lezhu Service Department.
[0030] Fig.10 This is a comparison chart of the forecast and historical curves of the flock number 112731-13 of the Lezhu Service Department.
[0031] Fig.11 This is a comparison chart of the forecast and historical curves of the flock number 7606-22 of the Chegang Service Department.
[0032] Fig.12 This is a comparison chart of the forecast and historical curves of the flock number 7601-24 of the Chegang Service Department.
[0033] Fig.13 This is a comparison chart of the forecast and historical curves of the flock number 7639-22 of the Chegang Service Department.
[0034] Fig.14 This is a comparison chart of the forecast and historical curves of the flock number 7940-8 of the Chegang Service Department.
[0035] Fig.15 This is a comparison chart of the forecast and historical curves of the flock number 9982-80 of the Chegang Service Department. DETAILED DESCRIPTION
[0036] The core technical solution of the present invention is that the standard growth curve serves as a reference, but in reality the growth conditions of each flock of chickens with the same parameters are different. The actual average weight curve is used as a basis and the standard growth curve is used for guidance to finally fit the average weight prediction curve.
[0037] The steps for generating the standard growth curve are as follows:
[0038] s1. Set the chickens with the same environmental parameters as a type of chickens, and divide such chickens 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 chickens are born on the same day;
[0039] s2. Record the daily average weight of each group of chickens;
[0040] s3. 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 environmental parameters;
[0041] s4. Change any one of the environmental parameters to form a new type of chicken flock, repeat steps s2 and s3, and obtain the standard growth curve of the average weight per day of different types of chicken flocks.
[0042] The daily average weight data of the chickens is obtained by placing an electronic chicken scale in the chicken house, and obtaining data by having the chickens jump on and off the electronic chicken scale. The data is then processed through the supporting software to obtain the daily average weight of the chickens. The specific steps are as follows:
[0043] j1. Eliminate unreasonable data samples from all weighed data samples in the chicken flock to obtain preliminary denoised samples;
[0044] j2. 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;
[0045] j3. Filter the samples after preliminary denoising with the help of the minimum value and the maximum value, and obtain the data samples between the maximum value and the minimum value as the valid data samples;
[0046] j4. Average the valid data samples to obtain the initial weighted value;
[0047] j5. Set the weight coefficient to correct the initial average weight value to obtain the final average weight value.
[0048] In step j1, unreasonable data samples include null values, 0 values, and invalid data that is significantly less than the normal weight of the chicken. All weighed data samples include positive samples and negative samples. The positive samples are the incremental values of the chicken scale after the chicken jumps on the chicken scale, and the negative samples are the decrement values of the chicken scale after the chicken jumps off the chicken scale. In step j2, the minimum value is 0.5-0.8 times the average weight fuzzy value, and the maximum value is 1.3-1.6 times the average weight fuzzy value.
[0049] The weight coefficient is automatically generated by inputting parameters, wherein the input parameters include the type of chickens and the current age. After the user inputs the type of chickens and the current age, the weight coefficient is automatically generated, and the initial value of the average weight is corrected. Of course, the weight coefficient is directly input by the user based on experience.
[0050] In order to verify the effectiveness of the prediction method of the present invention, 10 chicken flocks from the farmers of the Lezhu Service Department and 5 chicken flocks from the farmers of the Chegang Service Department were selected for data prediction. The prediction results are shown in Table 1, where the actual average weight curves of all days for these 15 groups of chicken flocks are obtained. In order to verify the accuracy of the average weight prediction curve, we will go back 7 days and 14 days from the market date and make predictions respectively, that is, compare the generated 7-day and 14-day average weight prediction curves with the actual average weight curves of 7 days and 14 days to observe their consistency.
[0051] Table 1 Chicken flock information and prediction results of 15 farmers
[0052]
[0053] Take the example of farmer A from the Bamboo Service Department. Figure 1 All the average weight data from 94 days after the market release date, 14 days back and 7 days back were input into the software to let the software know the actual average weight of the flock in the past. Then the standard growth curve with the same environmental parameters as the flock was called up, and the curve was fitted with the support vector machine model of the software, and finally two average weight prediction curves were obtained. It should be noted here that the standard growth curve serves as a reference, but the actual growth of each flock of chickens with the same parameters is different. Based on the actual average weight curve and guided by the standard growth curve, the average weight prediction curve was finally fitted. Figure 1 It can be seen that the flock prediction accuracy of farmer A from Lezhu Service Department is 99.7%.
[0054] According to the above method, the prediction accuracy of the other 14 groups of chickens can also be obtained. The ideas and methods are the same as those of the case of farmer A of Lezhu Service Department. The specific actual and predicted comparison charts are shown in Figures 2 to 14 .
[0055] Through the above examples, it can be seen that according to the provided sample chickens, the comprehensive accuracy can reach 99.41%. Among them, the average weight prediction error of ±20g reached 73.65%, and the average weight prediction error of ±25g exceeded 80%. It can be said that this method can be fully used for the accurate prediction of the future average weight of the chickens. When the average weight prediction can be known, it is more convenient for farmers to regulate and manage the chickens.
Claims
1. A method for predicting the average weight of chickens in large-scale breeding, characterized by: The prediction method steps are as follows: a. Weigh the predicted flock every day to obtain the actual average weight curve; b. The actual average weight curve is fitted with the standard growth curve with the same environmental parameters as the tested chicken flock with the help of the support vector machine model to obtain the average weight prediction curve of the tested chicken flock.
2. The method for predicting the average weight of chickens in large-scale breeding according to claim 1, characterized in that: The steps of the method for generating the standard growth curve are as follows: s1. Set the chickens with the same environmental parameters as a type of chickens, and divide such chickens 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 chickens are born on the same day; s2. Record the daily average weight of each group of chickens; s3. 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 environmental parameters; s4. Change any one of the environmental parameters to form a new type of chicken flock, repeat steps s2 and s3, and obtain the standard growth curve of the average weight per day of different types of chicken flocks.
3. A method for predicting the average weight of chickens in large-scale breeding according to claim 1 or 2, characterized in that: The daily average weight of the flock is obtained by placing an electronic chicken scale in the chicken house, obtaining data by having the chickens jump on and off the electronic chicken scale, and processing it through supporting software to obtain the daily average weight of the flock. The specific steps are as follows: j1. Eliminate unreasonable data samples from all weighed data samples in the chicken flock to obtain preliminary denoised samples; j2. 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; j3. Filter the samples after preliminary denoising with the help of the minimum value and the maximum value, and obtain the data samples between the maximum value and the minimum value as the valid data samples; j4. Average the valid data samples to obtain the initial weighted value; j5. Set the weight coefficient to correct the initial average weight value to obtain the final average weight value.
4. A method for predicting the average weight of chickens in large-scale breeding according to claim 1 or 2, characterized in that: The environmental parameters include breed, gender, region, chicken house control method, climate, and breeding mode.