Feed production data optimization acquisition method and system

Abstract

The invention relates to the field of data processing, in particular to a feed production data optimization and collection method and system, and the method comprises the steps: collecting humidity data during feed production, selecting an adjacent data point set of missing data points, and obtaining a feed production data set based on the numerical difference between each adjacent data in the adjacent data point set and other adjacent data points; calculating the prominence degree of the adjacent data points, analyzing the numerical value change characteristics of the adjacent data points, and optimizing the numerical value prominence degree of each adjacent data point to obtain the noise degree of each adjacent data point of the missing data points; and performing weighted summation by taking the noise degree of the adjacent data points of the missing data points as the weight of the missing data points to obtain an interpolation result of the missing data points, and filling the missing data points in the humidity data according to the interpolation result to obtain complete humidity data. According to the invention, weighting adjustment is carried out through the noise characteristics of multiple groups of adjacent data point sets, and the accuracy of an interpolation result is improved.

Description

Technical Field

[0001] The present invention relates to the field of data processing, and in particular to a feed production data optimization collection method and system. Background Art

[0002] Feed production is an important supporting link in animal husbandry and breeding industry, which is directly related to the health, growth rate, reproductive capacity and quality of the final product of the animal. The feed production process involves multiple links. Humidity is an important control parameter in feed production, which directly affects the quality, nutritional content and stability of the finished product. Humidity data in the feed production process often has data missing due to environmental interference and other reasons. Data missing not only affects the precise control of the production process, but may also lead to unqualified production or waste of resources.

[0003] The existing Chinese patent application document with publication number CN116993527A discloses a method for optimizing the collection and monitoring of pig feed production data, which relates to the field of feed production technology. It includes grain raw material quality analysis, grain raw material grinding uniformity analysis, feed mixing uniformity analysis, finished feed quality analysis and abnormal information early warning. Through real-time monitoring and analysis of the pig feed production process, the operator needs to conduct further inspection and confirmation for the abnormal process, so as to find problems and optimize and adjust them in time, and analyze the quality of the finished feed, so as to improve production efficiency and quality.

[0004] This application document uses intelligent data collection and monitoring technology to achieve comprehensive monitoring and analysis of the production process. The current data interpolation technology uses the KNN algorithm to make up for data missing. However, when collecting humidity data, there are usually a large number of electrical equipment in the industrial environment. These devices may generate electromagnetic interference and affect the normal operation of the humidity sensor, resulting in measurement errors or data fluctuations, and generating noise data. Noise data will affect the accuracy of data interpolation. Summary of the invention

[0005] In order to solve the problem that electromagnetic interference affects the operation of the humidity sensor when collecting data, generates noise data, and thus affects the accuracy of interpolation of missing data, the present invention provides solutions in the following aspects.

[0006] In a first aspect, a method for optimizing the collection of feed production data includes: collecting humidity data during feed production, obtaining a set of neighboring data points of missing data points in the humidity data, and obtaining the prominence of each neighboring data point based on the numerical difference between each neighboring data point in the set of neighboring data points and all other neighboring data points; analyzing the numerical change characteristics of all neighboring data points of the missing data point, and optimizing the numerical prominence of each neighboring data point to obtain the noise level of each neighboring data point of the missing data point; taking the noise level of the neighboring data points of the missing data point as the weight of the missing data point for weighted summation to obtain the interpolation result of the missing data point, and filling the missing data points in the humidity data according to the interpolation result to obtain complete humidity data.

[0007] The optimized collection and processing of humidity data in the feed production process effectively fills in the missing data points and significantly improves the integrity and accuracy of the data. By analyzing the numerical differences and change characteristics of adjacent data points, the noise level of each data point is quantified, and weighted summation is performed based on this, so as to obtain a more reliable interpolation result. It not only reduces the impact of noise data on the interpolation results, but also enhances the reliability of the data. By obtaining accurate humidity data in the feed production process, it is helpful to provide reliable data support for intelligent analysis and decision-making in feed production, which helps to optimize the production process, improve product quality, and reduce production costs.

[0008] Preferably, the set of adjacent data points includes: Taking the missing data point as the center, a preset number of adjacent data points are evenly selected on both sides to construct an adjacent data point set, wherein the adjacent data point set cannot include the missing point. In response to the insufficient number of data points on one side of the missing point, the missing point is supplemented on the other side.

[0009] By taking the missing data point as the center and evenly selecting a preset number of adjacent data points on both sides to construct a set, the representativeness and integrity of the data are ensured. Even when the number of data points is insufficient, the integrity of the adjacent data point set can be ensured by supplementing it on the other side, thereby improving the accuracy and stability of the interpolation.

[0010] Preferably, obtaining the prominence of each adjacent data point includes: Take any missing data point as a marked missing point, obtain a set of adjacent data points of the marked missing point, take any adjacent point in the set of adjacent data points as a marked adjacent point, traverse and calculate the absolute value of the numerical difference between the marked adjacent data point and other adjacent data points, sum up the absolute values ​​of all numerical differences, and obtain the sum of the numerical differences of the marked adjacent data points; The absolute difference between the maximum and minimum values ​​in the adjacent point set and the maximum and minimum values ​​in the adjacent data point set excluding the marked adjacent data point are calculated respectively, and the absolute difference value plus the product of the sum of the hyperparameters and the sum of the numerical differences is taken as the numerical prominence of the marked adjacent point.

[0011] By calculating the sum of the absolute values ​​of the numerical differences between each adjacent data point and other data points, and considering the difference between the maximum and minimum values, the numerical prominence of each adjacent data point is accurately analyzed, thereby effectively identifying possible noise data points. This not only enhances the reliability of the interpolation results, but also improves the adaptability and robustness of data processing by introducing hyperparameter adjustment.

[0012] Preferably, obtaining the prominence of each adjacent data point further includes: Take any missing data point as a marked missing point, obtain a set of adjacent data points of the marked missing point, take any adjacent point in the set of adjacent data points as a marked adjacent point, traverse and calculate the absolute value of the numerical difference between the marked adjacent data point and other adjacent data points, sum up the absolute values ​​of all numerical differences, and obtain the sum of the numerical differences of the marked adjacent data points; The variance of the values ​​in the adjacent data set of the marked missing point is taken as the overall discrete degree, the variance of the values ​​of the remaining data points after removing the marked adjacent points is taken as the local discrete degree, the difference between the overall discrete degree and the local discrete degree is normalized, and the product of the normalized difference and the sum of the numerical differences is taken as the numerical prominence of the marked adjacent points.

[0013] By calculating the sum of the numerical differences of each adjacent data point and the difference between the overall and local discreteness, the numerical prominence of each adjacent data point is further quantified. Not only the direct difference between data points is considered, but also the overall and local variance changes are combined to more comprehensively identify possible noise data points.

[0014] Preferably, the noise level includes: Calculate the local mean of the variation between each adjacent data point of each missing data point and the data points corresponding to the sampling moments on both sides of the adjacent data point, and calculate the overall mean of the variation between all two adjacent data points of each missing data point; The absolute difference between the local mean change and the overall mean change is calculated, and the product of the absolute difference and the sum of the hyperparameters and the numerical prominence is taken as the noise level of each adjacent data point.

[0015] Preferably, the noise level also includes: Count the number of times that the positive and negative attributes of the numerical difference between each adjacent data point of each missing point and the previous sampling moment are opposite to the positive and negative attributes of the numerical difference between all other adjacent adjacent data points; Calculate the local mean of the variation between each adjacent data point of each missing data point and the corresponding data points at the sampling moments on both sides of the adjacent data point, and calculate the overall mean of the variation between all two adjacent data points of each missing data point; calculate the absolute difference between the local mean of variation and the overall mean of variation, and take the product of the sum of the absolute difference and the hyperparameter, the number of opposites, and the degree of numerical prominence as the noise level of each adjacent data point.

[0016] By counting the number of times the positive and negative attributes of the numerical difference of each adjacent data point are opposite, and calculating the difference between the local and overall change means, the noise level of each adjacent data point is comprehensively evaluated. Not only the numerical difference of the data point is considered, but also the change trend and consistency in the time series are analyzed.

[0017] Preferably, the interpolation result satisfies the following relationship: ; In the formula, Indicates The interpolation result of missing data points is Indicates different preset quantities Group a set of adjacent data points, Indicates of the missing data points The mean humidity value of a set of adjacent data points. Indicates of the missing data points The noise level of the group of adjacent data points, Indicates An exponential function with base .

[0018] Preferably, the step of obtaining complete humidity data further includes: The integrity of the filled humidity data is verified to check whether there are new missing values ​​or outliers, and the numerical differences between the filled data and adjacent data points are compared to ensure that the filled data points maintain reasonable consistency in numerical values ​​with the surrounding data points.

[0019] Preferably, comparing the value difference between the filled data and adjacent data points also includes: The error value after filling is calculated by using mean square error, and the accuracy of the filled data points is determined based on the error value and the preset error threshold; In response to the error value being greater than a preset threshold, it is considered that the accuracy of the filled data point does not meet the requirement, and the difference calculation is performed again. Otherwise, it is considered that the accuracy requirement of the data point is met.

[0020] In a second aspect, a feed production data optimization collection system includes: a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the above-mentioned feed production data optimization collection method is implemented.

[0021] The present invention has the following effects: 1. The present invention can identify and quantify the noise level of each group of data points by selecting a set of data points adjacent to the missing data points on the time axis and analyzing the numerical characteristics of the adjacent data points. The interpolation result is corrected based on the noise level of each set of adjacent data points as the weight of each missing data point, so that the interpolation result can more accurately reflect the actual humidity situation, avoid the interpolation error caused by the influence of noise data, and perform weighted adjustment according to the noise characteristics of multiple sets of adjacent data points, so that the adjacent data points with less noise contribute more to the final interpolation result, ensuring the accuracy of the interpolation result.

[0022] 2. The present invention uses the optimized KNN algorithm to interpolate the missing data points in the humidity data. The interpolated humidity data is more accurate, which effectively improves the quality of humidity data collection and processing, thereby improving the data analysis accuracy and production efficiency in the feed production process. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 It is a method flow chart of steps S1 to S3 in a method for optimizing and collecting feed production data according to an embodiment of the present invention.

[0024] Figure 2 The present invention is a structural block diagram of a feed production data optimization collection system according to an embodiment of the present invention. DETAILED DESCRIPTION

[0025] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments.

[0026] Reference Figure 1 A feed production data optimization collection method includes steps S1 to S3, which are as follows: S1: Collect moisture data during feed production, obtain a set of adjacent data points of missing data points in the moisture data, and obtain the prominence of each adjacent data point based on the numerical difference between each adjacent data point in the set of adjacent data points and all other adjacent data points.

[0027] Exemplarily, this solution uses a humidity sensor to collect humidity data during feed production, with a collection frequency of once every 2 seconds and a collection time of 2 hours; It should be noted that when performing interpolation processing, it is necessary to select adjacent data points related to the missing data points as references. The number and distribution of adjacent data points will affect the accuracy of the interpolation results. Therefore, it is necessary to ensure that the selected data points can reasonably reflect the actual situation of the missing data points. The specific steps are as follows: Taking the missing data point as the center, a preset number of adjacent data points are evenly selected on both sides to construct an adjacent data point set, wherein the adjacent data point set cannot include the missing point. In response to the insufficient number of data points on one side of the missing point, the missing point is supplemented on the other side.

[0028] Exemplarily, the preset number may be 6, 12, 18, 24, or 30 data points. That is, multiple sets of adjacent data points may be obtained according to the preset number, and each set of adjacent data points needs to be analyzed as follows: In this embodiment, the analysis is performed based on the largest set of adjacent data points corresponding to the missing data point being analyzed, wherein the largest set of adjacent data points refers to a set of adjacent data points consisting of 30 data points. Since this set of adjacent data points already includes all adjacent data points of the missing data point being analyzed, it can provide the most comprehensive local features that may affect the missing data point.

[0029] Get the prominence of each nearby data point, including: Take any missing data point as a marked missing point, obtain a set of adjacent data points of the marked missing point, take any adjacent point in the set of adjacent data points as a marked adjacent point, traverse and calculate the absolute value of the numerical difference between the marked adjacent data point and other adjacent data points, sum up the absolute values ​​of all numerical differences, and obtain the sum of the numerical differences of the marked adjacent data points; The absolute difference between the maximum and minimum values ​​in the adjacent point set and the maximum and minimum values ​​in the adjacent data point set excluding the marked adjacent data point are calculated respectively, and the absolute difference value plus the product of the sum of the hyperparameters and the sum of the numerical differences is taken as the numerical prominence of the marked adjacent point.

[0030] Specifically, the numerical prominence degree satisfies the following relationship: ; In the formula, Indicates of the missing data points The numerical prominence of adjacent data points, Indicates of the missing data points The sum of the numerical differences between a neighboring data point and all other neighboring data points, Indicates The range of the values ​​in the set of adjacent data points of the missing data point, Indicates Remove the missing data point from the set of all neighboring data points The range after the adjacent data point, is a hyperparameter.

[0031] That is to say, The larger the of the missing data points The greater the difference between the value of a neighboring data point and all other neighboring data points except itself, the more prominent the corresponding value is. However, since there may be multiple consecutive data points with values ​​close to or consistent with the data point, this neighboring data point may be a data point with a more prominent value caused by normal humidity changes, rather than a noise point; Reflects the The contribution of a neighboring data point to the range value of all neighboring data points is Very big, indicating The value of the adjacent data point is significantly different from the value range of other data points in the set, that is, The more isolated the value of an adjacent data point is within the set of adjacent data points, the more likely it is a noise data point, which will affect subsequent interpolation calculations and data analysis. By analyzing the range difference, we can identify those adjacent data points that deviate significantly from other data points in value. These data points may be noise data points caused by external factors such as electromagnetic interference. , which aims to prevent situation occurs.

[0032] In addition, another embodiment further includes: Take any missing data point as a marked missing point, obtain a set of adjacent data points of the marked missing point, take any adjacent point in the set of adjacent data points as a marked adjacent point, traverse and calculate the absolute value of the numerical difference between the marked adjacent data point and other adjacent data points, sum up the absolute values ​​of all numerical differences, and obtain the sum of the numerical differences of the marked adjacent data points; The variance of the values ​​in the adjacent data set of the marked missing point is taken as the overall discrete degree, the variance of the values ​​of the remaining data points after removing the marked adjacent points is taken as the local discrete degree, the difference between the overall discrete degree and the local discrete degree is normalized, and the product of the normalized difference and the sum of the numerical differences is taken as the numerical prominence of the marked adjacent points.

[0033] Specifically, the numerical prominence degree satisfies the following relationship: ; In the formula, Indicates of the missing data points The numerical prominence of adjacent data points, Indicates of the missing data points The sum of the numerical differences between a neighboring data point and all other neighboring data points, Indicates The variance of the values ​​in the set of neighboring data points of the missing data points, Indicates Remove the missing data point from the set of all neighboring data points The variance of the values ​​of the remaining data points after the adjacent data points, Represents the normalization function.

[0034] It should be noted that the numerical prominence cannot accurately represent the possibility that a data point belongs to noise to a certain extent. Since the humidity data may increase or decrease steadily over time, the numerical values ​​of the data points at the two end points of the data segment composed of all adjacent data points may be more prominent or even close to the numerical values ​​of the noise data points, but such data points are normal humidity data points rather than noise data points. Among them, the difference between humidity data points and noise data points is that the change amount of the real data point is more stable compared with the data points around its sampling time, and the positive and negative attributes of the change amount are also similar to other data points in the data segment composed of all adjacent data points. Therefore, the numerical change characteristics of all adjacent data points of each missing data point are further analyzed, and the numerical prominence of each adjacent data point of each missing data point is optimized to obtain the noise level of each adjacent data point of each missing data point.

[0035] S2: Analyze the numerical variation characteristics of all neighboring data points of the missing data point, and optimize the numerical prominence of each neighboring data point to obtain the noise level of each neighboring data point of the missing data point.

[0036] Get the noise level, including: Calculate the local mean of the variation between each adjacent data point of each missing data point and the data points corresponding to the sampling moments on both sides of the adjacent data point, and calculate the overall mean of the variation between all two adjacent data points of each missing data point; The absolute difference between the local mean change and the overall mean change is calculated, and the product of the absolute difference and the sum of the hyperparameters and the numerical prominence is taken as the noise level of each adjacent data point.

[0037] Specifically, the noise level satisfies the following relationship: ; In the formula, Indicates of the missing data points The noise level of adjacent data points is Indicates of the missing data points The numerical prominence of adjacent data points, Indicates of the missing data points The mean of the changes between the adjacent data points and the corresponding data points at the two sampling moments before and after them, Indicates The mean of the changes between all two adjacent data points of the missing data point, is a hyperparameter.

[0038] That is to say, The larger the The greater the relative size of the mean difference between the value of the adjacent data point and the data points on both sides of it and the mean difference between the value of all adjacent data points, the greater the relative size of the mean difference between the value of the adjacent data point and the data points on both sides of it. The greater the possibility that there are abnormal humidity fluctuations at adjacent data points, the greater the possibility that they are noise data points. is a hyperparameter, , which exists to prevent situation occurs.

[0039] By analyzing the size of the mean change in the numerical difference between each adjacent data point and the data points on both sides of it, the greater the difference between the size and the mean change between all two adjacent data points, the greater the possibility that this adjacent data point is a noise data point, and the greater the corresponding noise level will be.

[0040] In addition, another embodiment further includes: Count the number of times that the positive and negative attributes of the numerical difference between each adjacent data point of each missing point and the previous sampling moment are opposite to the positive and negative attributes of the numerical difference between all other adjacent adjacent data points; Calculate the local mean of the variation between each adjacent data point of each missing data point and the corresponding data points at the sampling moments on both sides of the adjacent data point, and calculate the overall mean of the variation between all two adjacent data points of each missing data point; calculate the absolute difference between the local mean of variation and the overall mean of variation, and take the product of the sum of the absolute difference and the hyperparameter, the number of opposites, and the degree of numerical prominence as the noise level of each adjacent data point.

[0041] Specifically, the noise level satisfies the following relationship: ; In the formula, Indicates of the missing data points The noise level of adjacent data points is Indicates of the missing data points The numerical prominence of adjacent data points, Indicates The positive and negative properties of the numerical differences between all two adjacent data points of the missing data point are the same as the The number of times the positive and negative attributes of the numerical difference between the adjacent data points and the previous sampling moment are opposite, Indicates of the missing data points The mean of the changes between the adjacent data points and the corresponding data points at the two sampling moments before and after them, Indicates The mean of the changes between all two adjacent data points of the missing data point, is a hyperparameter.

[0042] That is to say, The larger the The more inconsistent the positive and negative attributes of the numerical difference between the adjacent data point and the previous sampling moment and the positive and negative attributes of the numerical difference between the next moment and the previous moment of all adjacent data points are, the more The greater the possibility that there are abnormal humidity fluctuations at adjacent data points, the greater the possibility that they are noise data points.

[0043] Through statistics The number of times the positive and negative attributes of the numerical difference between a neighboring data point and the previous sampling moment are opposite to the positive and negative attributes of the numerical difference between all other neighboring data points can effectively identify data points with abnormal numerical fluctuations in the time series. And abnormal fluctuations may be caused by noise or abnormal events; at the same time, by considering The value of can give different weights to each adjacent data point. The weight of the noise data point will be reduced, while the weight of the normal data point will be increased, which makes the interpolation result more stable and enhances the robustness of data processing.

[0044] S3: The noise levels of the adjacent data points of the missing data points are taken as the weights of the missing data points for weighted summation to obtain the interpolation results of the missing data points, and the missing data points in the humidity data are filled according to the interpolation results to obtain complete humidity data.

[0045] Specifically, the interpolation result satisfies the following relationship: ; In the formula, Indicates The interpolation result of missing data points is Indicates different preset quantities Group a set of adjacent data points, Indicates of the missing data points The mean humidity value of a set of adjacent data points. Indicates of the missing data points The noise level of the group of adjacent data points, Indicates An exponential function with base .

[0046] That is to say, Indicates of the missing data points The weighted average calculation of the group of adjacent data points The weight when there are missing data points.

[0047] The integrity of the filled humidity data is verified to check whether there are new missing values ​​or outliers, and the numerical differences between the filled data and adjacent data points are compared to ensure that the filled data points maintain reasonable consistency in numerical values ​​with the surrounding data points.

[0048] The error value of the filled data point is calculated using the mean square error, and the accuracy of the filled data point is determined based on the error value and the preset error threshold; In response to the error value being greater than a preset threshold, it is considered that the accuracy of the filled data point does not meet the requirement, and the difference calculation is performed again. Otherwise, it is considered that the accuracy requirement of the data point is met.

[0049] The present invention also provides a feed production data optimization collection system. Figure 2 As shown, the system includes a processor and a memory, the memory stores computer program instructions, and when the computer program instructions are executed by the processor, a feed production data optimization collection method according to the first aspect of the present invention is implemented. The system also includes other components well known to those skilled in the art, such as a communication bus and a communication interface, whose settings and functions are known in the art, so they are not described here in detail.

[0050] It should be noted that, for those skilled in the art, several modifications and improvements can be made without departing from the concept of the present invention, and these modifications and improvements all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.

Claims

1. A feed production data optimization collection method, characterized in that: include: Collecting humidity data during feed production, obtaining a set of adjacent data points of missing data points in the humidity data, and obtaining the prominence of each adjacent data point based on the numerical difference between each adjacent data point in the set of adjacent data points and all other adjacent data points; Analyze the numerical variation characteristics of all adjacent data points of the missing data point, and optimize the numerical prominence of each adjacent data point to obtain the noise level of each adjacent data point of the missing data point; The noise levels of the adjacent data points of the missing data points are taken as the weights of the missing data points for weighted summation to obtain the interpolation results of the missing data points. The missing data points in the humidity data are filled according to the interpolation results to obtain complete humidity data.

2. A feed production data optimization collection method according to claim 1, characterized in that: The adjacent data point set includes: Taking the missing data point as the center, a preset number of adjacent data points are evenly selected on both sides to construct an adjacent data point set, wherein the adjacent data point set cannot include the missing point. In response to the insufficient number of data points on one side of the missing point, the missing point is supplemented on the other side.

3. A feed production data optimization collection method according to claim 1, characterized in that: The obtaining of the prominence of each adjacent data point includes: Take any missing data point as a marked missing point, obtain a set of adjacent data points of the marked missing point, take any adjacent point in the set of adjacent data points as a marked adjacent point, traverse and calculate the absolute value of the numerical difference between the marked adjacent data point and other adjacent data points, sum up the absolute values ​​of all numerical differences, and obtain the sum of the numerical differences of the marked adjacent data points; The absolute difference between the maximum and minimum values ​​in the adjacent point set and the maximum and minimum values ​​in the adjacent data point set excluding the marked adjacent data point are calculated respectively, and the absolute difference value plus the product of the sum of the hyperparameters and the sum of the numerical differences is taken as the numerical prominence of the marked adjacent point.

4. A feed production data optimization collection method according to claim 1, characterized in that: The obtaining of the prominence of each adjacent data point further includes: Take any missing data point as a marked missing point, obtain a set of adjacent data points of the marked missing point, take any adjacent point in the set of adjacent data points as a marked adjacent point, traverse and calculate the absolute value of the numerical difference between the marked adjacent data point and other adjacent data points, sum up the absolute values ​​of all numerical differences, and obtain the sum of the numerical differences of the marked adjacent data points; The variance of the values ​​in the adjacent data set of the marked missing point is taken as the overall discrete degree, the variance of the values ​​of the remaining data points after removing the marked adjacent points is taken as the local discrete degree, the difference between the overall discrete degree and the local discrete degree is normalized, and the product of the normalized difference and the sum of the numerical differences is taken as the numerical prominence of the marked adjacent points.

5. A feed production data optimization collection method according to claim 1, characterized in that: The noise level includes: Calculate the local mean of the variation between each adjacent data point of each missing data point and the data points corresponding to the sampling moments on both sides of the adjacent data point, and calculate the overall mean of the variation between all two adjacent data points of each missing data point; The absolute difference between the local mean change and the overall mean change is calculated, and the product of the absolute difference and the sum of the hyperparameters and the numerical prominence is taken as the noise level of each adjacent data point.

6. A feed production data optimization collection method according to claim 1, characterized in that: The noise level also includes: Count the number of times that the positive and negative attributes of the numerical difference between each adjacent data point of each missing point and the previous sampling moment are opposite to the positive and negative attributes of the numerical difference between all other adjacent adjacent data points; Calculate the local mean of the variation between each adjacent data point of each missing data point and the corresponding data points at the sampling moments on both sides of the adjacent data point, and calculate the overall mean of the variation between all two adjacent data points of each missing data point; calculate the absolute difference between the local mean of variation and the overall mean of variation, and take the product of the sum of the absolute difference and the hyperparameter, the number of opposites, and the degree of numerical prominence as the noise level of each adjacent data point.

7. A feed production data optimization collection method according to claim 1, characterized in that: The interpolation result satisfies the following relationship: ； In the formula, Indicates The interpolation result of missing data points is Indicates different preset quantities Group a set of adjacent data points, Indicates of the missing data points The mean humidity value of a set of adjacent data points. Indicates of the missing data points The noise level of the group of adjacent data points, Indicates An exponential function with base .

8. A feed production data optimization collection method according to claim 1, characterized in that: The complete humidity data obtained also includes: The integrity of the filled humidity data is verified to check whether there are new missing values ​​or outliers, and the numerical differences between the filled data and adjacent data points are compared to ensure that the filled data points maintain reasonable consistency in numerical values ​​with the surrounding data points.

9. A feed production data optimization collection method according to claim 8, characterized in that: The comparing of the value difference between the filled data and the adjacent data points also includes: The error value after filling is calculated by using mean square error, and the accuracy of the filled data points is determined based on the error value and the preset error threshold; In response to the error value being greater than a preset threshold, it is considered that the accuracy of the filled data point does not meet the requirement, and the difference calculation is performed again. Otherwise, it is considered that the accuracy requirement of the data point is met.

10. A feed production data optimization collection system, characterized in that: include: A processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the feed production data optimization collection method according to any one of claims 1 to 9 is implemented.

Images