Pig industry prediction method and device

Abstract

The invention discloses a live pig industry prediction method and device, and relates to the technical field of live pig breeding monitoring, and the method comprises the steps: collecting the influence data of the live pig industry in a set region range; the influence data comprises multiple pieces of index data, each piece of index data is index data of a date before a preceding period corresponding to a prediction date, and each piece of index data corresponds to one preceding period; and inputting the influence data into a live pig slaughtering number prediction model to obtain a live pig slaughtering number prediction value of the prediction date in the set area range. According to the invention, the effectiveness of marketing quantity prediction is improved.

Description

Technical Field

[0001] The present application relates to the field of pig farming monitoring technology, and in particular to a pig industry prediction method and device. Background Art

[0002] The pig industry in a certain country is mainly based on statistical survey data, supplemented by management, supervision and service process data. The overall development is still at a relatively low level, which can easily lead to cyclical fluctuations such as insufficient or oversupply of pigs. At present, there is an urgent need to effectively monitor and warn of pig industry data. Summary of the invention

[0003] The purpose of this application is to provide a pig industry forecasting method and device to effectively predict the number of pigs slaughtered.

[0004] To achieve the above objectives, this application provides the following solutions:

[0005] In a first aspect, the present application provides a pig industry forecasting method, comprising:

[0006] Collect the impact data of the pig industry within the set area; the impact data includes a plurality of indicator data, each indicator data is the indicator data of the date before the corresponding lead period of the forecast date, and each indicator data corresponds to a lead period;

[0007] The influencing data is input into a pig slaughter number prediction model to obtain a predicted value of the pig slaughter number on the predicted date within the set area.

[0008] Optionally, the multiple indicator data include the proportion of sows in breeding, the price of live pigs, the average cost of piglets per head, the average feed cost per head and the average epidemic prevention cost per head.

[0009] Optionally, the lead period for the proportion of sows is 10 months, the lead period for the live pig price is 12 months, the lead period for the average piglet cost per head is 6 months, the lead period for the average feed cost per head is 12 months, and the lead period for the average epidemic prevention cost per head is 1 month or 12 months.

[0010] Optionally, the pig output prediction model is a multivariate regression model.

[0011] Optionally, collect impact data on the pig industry within the set area, including:

[0012] The farms within the set area are sampled by random sampling, and the proportion of sows capable of reproduction, the average piglet cost per pig, the average feed cost per pig, and the average epidemic prevention cost per pig are collected;

[0013] The sampling method is adopted to sample the collective markets within the set area, and the live pig prices of the sampled collective markets are collected;

[0014] The impact data of the pig industry in the set area is calculated based on the proportion of sows in the sampled farms, the average cost of piglets per head, the average feed cost per head and the average epidemic prevention cost per head, as well as the live pig prices in the sampled markets.

[0015] Optionally, after inputting the impact data into a pig output prediction model to obtain a predicted value of the pig output on the predicted date within the set area, the pig industry prediction method further includes:

[0016] According to the formula QX CMEAT，T =∑ CLIVE (IOLVMT CLIVE，CMEAT，T ×QDP CLIVE，T ) Forecast pork production;

[0017] Among them, QX CMEAT，T For pork production, IOLVMT CLIVE，CMEAT，T is carcass weight, QDP CLIVE，T is the predicted value of the number of live pigs slaughtered, and CLIVE represents the set of live pigs.

[0018] Optionally, after inputting the impact data into a pig output prediction model to obtain a predicted value of the pig output on the predicted date within the set area, the pig industry prediction method further includes:

[0019] According to the formula QDT C，T =QDF C，T +QDL C，T +QDS C，T +QDP C，T +QDO C，T +QDW C，T +STV C，T forecasting pork demand;

[0020] Among them, QDT C，T is the total pork demand, QDF C，T The amount of pork required for a household, QDL C，T QDS is the market demand for pork. C，T Pork quantity required for schools, QDP C，T To process the pork demand, QDO C，T To replace the pork demand, STV C，T is the pork import demand, QDW C，T For QDF C，T 、QDL C，T , QDS C，T , QDPC，T 、QDO C，T and STV C，T Other pork demand.

[0021] Optionally, after inputting the influencing data into a pig output prediction model to obtain a predicted value of the pig output on the predicted date within the set area, the pig industry prediction method further includes: issuing a warning message when the predicted value of the pig output exceeds a preset range.

[0022] In a second aspect, the present application provides a pig industry prediction device, comprising:

[0023] A data collection module is used to collect the impact data of the pig industry within a set area; the impact data includes a plurality of indicator data, each indicator data is the indicator data of the date before the corresponding lead period of the forecast date, and each indicator data corresponds to a lead period;

[0024] The pig output number prediction module is used to input the influencing data into the pig output number prediction model to obtain the predicted value of the pig output number on the predicted date within the set area.

[0025] Optionally, the multiple indicator data are respectively the proportion of sows, live pig price, average piglet cost per head, average feed cost per head and average epidemic prevention cost per head; the lead period for the proportion of sows is 10 months, the lead period for the live pig price is 12 months, the lead period for the average piglet cost per head is 6 months, the lead period for the average feed cost per head is 12 months, and the lead period for the average epidemic prevention cost per head is 1 month or 12 months.

[0026] According to the specific embodiments provided in this application, this application discloses the following technical effects:

[0027] The present application provides a pig industry prediction method and device, which collects impact data of the pig industry in a set area; the impact data includes multiple indicator data, each indicator data is the indicator data of the date before the corresponding lead period of the prediction date, and each indicator data corresponds to a lead period. The time limit of the impact of each indicator data on the pig output is taken into account, that is, the lead period, thereby improving the effectiveness of the output prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0029] Figure 1A schematic diagram of a process flow of a pig industry prediction method provided in one embodiment of the present application;

[0030] Figure 2 A schematic diagram of the fluctuation cycle of the pig industry provided in one embodiment of the present application;

[0031] Figure 3 A schematic diagram of the correlation between the fluctuation range of sow inventory and the fluctuation range of live pig prices provided in one embodiment of the present application. DETAILED DESCRIPTION

[0032] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0033] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below with reference to the accompanying drawings and specific implementation methods.

[0034] This application provides a pig industry prediction method, such as Figure 1 As shown, the pig industry prediction method includes steps 101 and 102.

[0035] Step 101: Collect the impact data of the pig industry in the set area; the impact data includes a plurality of indicator data, each indicator data is the indicator data of the date before the corresponding lead period of the forecast date, and each indicator data corresponds to a lead period.

[0036] Step 102: Input the influencing data into a pig slaughter number prediction model to obtain a predicted value of the pig slaughter number on the predicted date within the set area.

[0037] The impact data of the pig industry in the set area are collected; the impact data include multiple indicator data, each indicator data is the indicator data of the date before the corresponding lead period of the prediction date, each indicator data corresponds to a lead period, and the time period of the impact of each indicator data on the pig output is taken into account, that is, the lead period, thereby improving the effectiveness and accuracy of the output forecast.

[0038] In an exemplary embodiment, in order to comprehensively monitor and timely warn the development of the pig industry, the pig industry monitoring and early warning system includes not only the monitoring of the breeding link, but also the monitoring of the circulation link. In the breeding link, it includes not only the monitoring of breeding villages and households, but also the monitoring of large-scale breeding farms. In the circulation link, it includes not only the monitoring of prices, but also the monitoring of transaction volume. The following describes the sample selection techniques for the pig breeding link and the circulation link respectively.

[0039] 1. Sample selection technology in the breeding process

[0040] The sampling of monitoring villages, monitoring households and scale farms in the pig breeding process adopts the method of "stratification, three-stage, relevant sign queuing, symmetrical and equidistant sampling". "Relevant sign queuing, symmetrical and equidistant sampling" is a sampling technique with very good sampling effect. The specific sampling method is as follows:

[0041] Phase 1: Sample counties are selected from the whole province. The whole province is the target area to be monitored, that is, the regional scope is set. First, according to the annual output of live pigs, all counties in the province are divided into two categories (i.e., two layers) of main pig-producing counties and non-main pig-producing counties, with an annual output of 200,000 pigs as the dividing point. Then, the main and non-main pig-producing counties are ranked according to the output of live pigs in 2006, and 15 counties are selected as sample counties by symmetrical and equidistant sampling, for a total of 30 sample counties.

[0042] The second stage: select administrative villages from sample counties. In each sample county, all administrative villages were ranked according to the number of pigs slaughtered in 2006, and 10 administrative villages were selected as fixed-point monitoring villages using the symmetrical equidistant sampling method, for a total of 300 fixed-point monitoring villages.

[0043] The third stage: The monitoring counties select monitoring households. In each monitoring county, the number of pigs sold each year is divided into two scales, large and small. One household is selected from each scale, and two households in total are selected as cost-benefit fixed-point monitoring households. The average scale of these two households is required to be equal to the average scale of the administrative village. There are 90 fixed-point monitoring households in total.

[0044] Selection of large-scale farm samples. In all pig-producing counties in the province, farms with an annual output of more than 500 pigs are included in the monitoring scope, which currently amounts to more than 12,000 large-scale farms.

[0045] (II) Sample selection technology in the circulation link

[0046] In the circulation link, the main monitoring is the price of live pigs and pork, as well as the amount of pork transactions. There are 26 counties in the province that monitor prices and 17 counties that monitor transaction volumes. The sample selection method is the "stratified, two-stage sampling" method, as follows:

[0047] Phase 1: Select monitoring counties from the whole province. Considering the province as one layer, select monitoring counties according to the stratified sampling method. The allocation of 26 monitoring counties mainly considers their pig output and pork sales, and the selection is carried out according to the principle of "selecting large counties" (i.e. counties with the largest pig output). Among the selected price monitoring counties, half of the counties are selected as transaction volume monitoring counties at the same time, and the principle is also "selecting large counties".

[0048] The second stage: The monitoring counties select the collective markets. In each monitoring county, 3-5 collective markets are selected as price monitoring points. In the monitoring counties with small price changes, 3 collective markets are selected; in the monitoring counties with large price changes, 5 collective markets are selected. The principle of selecting collective markets is to select collective markets with large transaction volumes, while considering geographical location and degree of work cooperation.

[0049] In an exemplary embodiment, step 101 specifically includes steps 1011 to 1013 .

[0050] Step 1011: Sampling farms within the set area is carried out by random sampling, and the proportion of sows capable of reproduction, the average piglet cost per head, the average feed cost per head and the average epidemic prevention cost per head of the sampled farms are collected.

[0051] Step 1012: sampling the collective markets within the set area by a sampling method, and collecting the live pig prices of the sampled collective markets.

[0052] Step 1013: Calculate the impact data of the pig industry in the set area based on the proportion of sows in the sampled farms, the average piglet cost per pig, the average feed cost per pig and the average epidemic prevention cost per pig, as well as the live pig price of the sampled collective markets.

[0053] The sampling error estimation techniques are as follows:

[0054] 1.1 Selection of sampling error estimation methods

[0055] The sampling of samples for the pig monitoring system adopts the method of "stratified, three-stage, symmetrical and equidistant sampling with relevant signs in queue", which is a combination of different sampling methods and belongs to complex sample sampling design. For complex sample sampling design, the estimation of sampling error cannot adopt the sampling error calculation formula of a certain sampling method (such as stratified sampling, equidistant sampling), but adopts the "jackknife method", "self-help method", "random group" and other methods. Among them, the "jackknife method" is an estimation method for complex sample sampling error, so this application adopts this method to estimate the sampling error.

[0056] The specific process of estimating sampling error using the "jackknife method" is: divide all samples of live pigs into k random groups (k is generally about 15, which can be adjusted according to the number of all samples), usually using the random starting point equidistant sampling method, and select k sets of small samples from all samples as k random groups. The sample size of each random group is m = n / k, where n is the total sample size. represents the estimate of θ obtained from the full sample, Indicates discarding the αth group of observations and using The estimator is constructed in the same way. The so-called "dummy value" is defined as The jackknife estimate of θ is defined as all The average right The jackknife variance estimate of is:

[0057] and So the above formula can be used as The variance estimate of . After the jackknife variance is estimated, the jackknife standard deviation SE (the absolute value of the sampling error) and the sampling error coefficient CV (the relative number of the sampling error) can be obtained.

[0058] Jackknife standard deviation Sampling Error Coefficient

[0059] In actual work, the confidence level for evaluating sampling error is generally set at 95%, that is, the sampling error coefficient is considered at a 95% level of confidence. Since the probability corresponding to the 95% confidence level is t=1.96, the sampling error coefficient at the 95% confidence level is the above-mentioned sampling error coefficient CV multiplied by 1.96.

[0060] In an exemplary embodiment, the sampling error of the pig monitoring system is evaluated, that is, the sampling error of the pig farms and the market is evaluated.

[0061] The sampling errors of pig inventory, pig slaughter, and breeding sow inventory are acceptable, and the sampling errors of pig breeding income and breeding costs per pig meet the usual requirements. Tables 1 and 2 list the sampling errors of the main indicators of pig breeding. It can be seen that at a confidence level of 95%, the sampling error coefficients of pig inventory, pig slaughter, and breeding sow inventory are 9.26%, 8.72%, and 13.07%, respectively. This level is slightly higher, but acceptable. The sampling error coefficients of income and cost per pig are 4.83% and 4.93%, respectively, which are less than 5%, meeting the usual requirements for sampling errors. In this monitoring system, the net income per pig is obtained by subtracting the average breeding cost per pig from the average breeding income per pig. Since the sampling errors of income and cost are small, it can be judged that the sampling error of the net income per pig is also small.

[0062] Table 1 Sampling errors of main indicators of pig breeding quantity

[0063]

[0064] Table 2 Sampling errors of main indicators of pig farming cost and benefit

[0065] Income per fattened pig sold Cost per fattened pig sold Sample mean 1108 1138 Jackknife Variance 746.47 818.34 Jackknife standard deviation SE 27.32 28.61 Sampling error coefficient CV 2.47％ 2.51％ Sampling error coefficient at 95% confidence level 4.83％ 4.93％ Lower confidence limit 1054.45 1081.93 Upper confidence limit 1161.55 1194.07

[0066] In large-scale continuous surveys, sample rotation is the most important measure to prevent sample aging and ensure sample representativeness. The pig monitoring and early warning system adopted sample rotation technology at the beginning of its establishment, and carried out regular sample rotation. Since large-scale farms are fully counted, sample rotation is not required; price and transaction volume monitoring adheres to the principle of "selecting large" when rotating. The sample rotation technology for breeding villages and breeders (breeding farms) is detailed as follows.

[0067] 1. Basic principles

[0068] 1. Only sample villages and households are rotated, and sample counties are generally not rotated (to reduce costs).

[0069] 2. Sample villages and households are only passively adjusted. That is, the sample village is adjusted only when it must be adjusted for some reason. The sample village needs to be adjusted in the following cases: The sample village used to raise pigs, but now it does not.

[0070] (II) Sample rotation method

[0071] 1. Determine the position of the sample villages to be rotated each time based on the queue order of the original sample villages.

[0072] 2. Use the pig breeding data from the previous year to compile a sampling frame, and adopt "relevant sign queuing, symmetrical and equidistant sampling" to select a new set of sample villages.

[0073] 3. In the newly selected sample villages, the original sample villages will be replaced according to the “same number replacement principle”. This method can ensure the representativeness of the sample and prevent sample aging.

[0074] In order to fully understand the dynamic development factors of the pig industry and deeply grasp the changes in supply and demand, we started from the aspects of production, consumption, price, income, performance and biological characteristics of pigs, and selected the key indicators that best represent all aspects of the pig industry and predict future development trends. Finally, 33 sub-items of indicators in four categories were selected. Some indicators include "pig inventory, pig output, sow inventory, piglet inventory, piglet price, feed costs, epidemic prevention costs, piglet costs, pork prices, piglet prices, live pig prices, corn and other feed raw materials and finished product prices."

[0075] Before data collection for this application, the concept of collection indicators, the time of data collection and the process of data collection are standardized. By standardizing the concept of collection indicators, the connotation and extension of data collection of relevant indicators such as pig inventory, breeding sows, piglets and pig output are unified and standardized.

[0076] The time for collecting pig data is standardized to facilitate unified operation and effective data collection in various places. It is stipulated that the monthly monitoring report of standardized scale farms and the monthly monitoring report of the slaughter volume of culled sows shall be collected and reported on the 8th of each month; the monthly monitoring report of fixed points shall be collected and summarized on the 25th of each month and reported on the same day at the end of the month; the weekly report on pig product price information and transaction volume shall be collected on Wednesday every week and reported on Thursday.

[0077] The standardized data collection process includes: village-level information collectors go to breeding farms to collect data within the specified collection period every month (or collect data by telephone inquiries, and the inquiries require random checks of data), and then report the paper (or electronic) reports directly to the county-level statisticians. After review, verification or rejection, the county-level statisticians directly enter the data into the computer terminal system and report it directly to the ministry-level central database (before the data is stored, it must be reviewed, verified, or rejected by the provincial-level statisticians for final confirmation).

[0078] Price and transaction volume collection process. Every week within the fixed data collection period, market collectors go to fixed stalls (or farmers) in the market to collect relevant price and transaction volume data (or collect data by telephone inquiries). The collected data will then be submitted to county-level statisticians in the form of paper reports or electronic reports. The county-level statisticians will then enter the data into a computer terminal (or mobile phone APP terminal) after review, verification or rejection and report it directly to the ministerial central database (before data storage, it must be reviewed, verified, or rejected by provincial-level statisticians for final confirmation).

[0079] In an exemplary embodiment, the present application also includes reporting the impact data of the pig industry through a data collection and reporting platform. The data collection and reporting platform integrates advanced Java 2 Platform Enterprise Edition (Java 2 Platform Enterprise Edition, J2EE) technology architecture, microservice technology framework, multi-layer development technology, cloud virtualization technology, large database storage, distributed file storage technology, message queue, load balancing technology, business intelligence (Business Intelligence, BI) multi-dimensional data analysis and other 6 major main technologies to build a data collection and reporting platform. Through the use of the network platform, it promotes the functions of flexible data collection methods, automatic report generation, intelligent constraint detection, etc. The monitoring data transmission realizes the comprehensive management from collection content, collection process, collection style to collection information release, review, reporting, summary, analysis, urging, re-reporting, assessment and scoring, which is simple and fast, free to convert, automatic reminder, efficient and fast.

[0080] 1. J2EE technology architecture, microservice technology framework. The system platform adopts a multi-layer technology architecture based on J2EE technology, and combines advanced design and development such as service-oriented architecture (SOA) technology, cloud technology, and big data technology, making full use of the characteristics of various advanced technologies to achieve full integration of advanced technologies and livestock statistics monitoring business. At the same time, the deployment is split according to the business functions in the application to improve the system's high fault tolerance, rapid evolution and iteration. At the same time, the application is effectively split to achieve agile development and deployment to meet the changing needs of livestock statistics business. In particular, the microservice framework can integrate different systems (homomorphic and heterogeneous) such as data collection, analysis, aggregation and reporting systems through unified standards, improve the reuse of business in different systems, and quickly adapt to changes in livestock statistics business.

[0081] 2. Multi-layer development technology. It adopts a five-layer structure of front-end user interface (UI) layer, control layer, business layer, persistence layer and data service layer. The front-end UI layer uses the EasyUI framework to display data lists, and uses Highcharts to provide rich chart components to enrich the client's interaction and display capabilities. It mainly handles user operations on the system and feeds back the user's operation results to the user. The control layer is divided into the control logic layer and the service layer. The control logic layer mainly performs protocol identification, encoding processing, task distribution and other processing on the client's request; the service layer provides log, exception, security, general business rollback, data permissions, cache and other services for the overall application. The business layer is divided into the request processing layer, view service layer, and business component layer. It mainly handles page jumps, data conversion, log processing, exception handling, business logic, operations and calls between classes. The persistence (O / RMapping) layer maps the object-oriented design structure with the relational data structure in the database, and is responsible for data access between the application and the database. The data service layer is responsible for the storage of persistent business data.

[0082] 3. Cloud virtualization technology. Cloud virtualization technology can expand the capacity of hardware, simplify the software reconfiguration process, reduce software virtual machine-related overhead and support a wider range of operating systems. Virtualization technology can be used to isolate software applications from the underlying hardware. It includes a split mode that divides a single resource into multiple virtual resources, and an aggregation mode that integrates multiple resources into one virtual resource. This system organically integrates the hardware and software resources in the cloud through virtualization, standardization and automation, and delivers services in the cloud to users through the network. The system is deployed using the cloud host of the Information Center of the Ministry of Agriculture and Rural Affairs to ensure unlimited expansion of hardware resources and increase response resources on demand. Monitor the use and efficiency of resources to reduce the investment in increasing resources.

[0083] 4. Message queue and load balancing technology. Achieve high availability, high load and high concurrency.

[0084] By expanding the capacity of application service nodes for cluster deployment and separating the application server from the database server and file server, the system IO efficiency is greatly improved while data security is greatly enhanced. The Nginx load balancing software service is used to distribute requests. The load balancing service not only provides load balancing, but also provides failure detection function. When a node in the background cluster goes down, it will automatically jump to the node running normally to ensure the user experience. At the same time, in order to improve availability, disaster recovery backup is added in the cloud to prevent unavailability problems caused by node failure. Online and offline simultaneous backup can select different backup strategies according to different failure requirements.

[0085] Redis queues are used to solve the problem of high system concurrency. Data is temporarily stored in the Redis queue, and the background service process obtains the information that needs to be stored. Redis can read at a speed of 110,000 times / s and write at a speed of 81,000 times / s. It also supports a variety of storage data types, including strings, lists, sets, ordered sets, and hash types. At the same time, these data types support push / pop, add / remove, intersection and union, difference, and more operations. It has a significant effect on improving data concurrency, solving the CPU and memory pressure of application servers, reducing IO operations, and alleviating IO pressure.

[0086] 5. Large database storage and distributed file storage technology. Oracle11g is used for data storage, supporting relational object-based, distributed database products. It supports multi-user, large transaction volume transaction processing, data security and integrity control, supports distributed data processing, and is portable. The database provides access technology based on client / server multiple industrial standards. FastDFS distributed file system is used to manage files, solving the problems of large-capacity storage and load balancing. It fully realizes mechanisms such as redundant backup, load balancing, and linear expansion, and focuses on indicators such as high availability and high performance.

[0087] 6. BI intelligent multi-dimensional data analysis, custom report templates. The platform uses Cognos to establish a unified metadata model for reports. The unified metadata model provides a unified and consistent view for applications. Customized reports can be made in the browser with flexible formats and rich elements, and timely open queries can be made through Query Studio. Using Cognos' unique drill through, slice and dice, and pivot functions, animal husbandry statistical data information can be quickly, consistently, and interactively accessed from multiple angles, thereby gaining a deeper understanding of the data, effectively linking various related information, and being able to delve into the detailed data of interest while analyzing the summary data, so as to have a more comprehensive understanding of the situation and make correct decisions.

[0088] In an exemplary embodiment, the present application also adopts data analysis early warning technology.

[0089] 1. Data Analysis Technology

[0090] 1. Comparative analysis techniques

[0091] The comparative analysis techniques used in this application mainly include comparative analysis, month-on-month analysis, year-on-year analysis and fixed base ratio analysis. Comparative analysis is to compare objective things in order to understand the essence and laws of things and make correct evaluations. For example, when analyzing the fluctuations in pig prices, the pig prices in each province can be compared with the national pig prices to determine whether the supply and demand relationship in each province is coordinated with the supply and demand relationship in the national market. Month-on-month analysis is the ratio of the level in the reporting period to the level in the previous period, indicating the trend of changes in the phenomenon from period to period. For example, in the process of monitoring the pig industry, the month-on-month changes in the number of pigs and the number of sows in each month are usually calculated and compared. Year-on-year analysis is expanded according to time, i.e., year, quarter, month, date, etc., and the actual number of occurrences in this period is compared with the historical figures of the same caliber to produce dynamic relative indicators to reveal the level of development and growth rate, expressed as percentages or multiples. In the process of monitoring the pig industry, the year-on-year changes in the number of pigs and the number of sows in each month are usually calculated. Fixed base ratio analysis is the ratio of the level in the reporting period to the level in a fixed period, indicating the overall level of change of this phenomenon over a longer period of time. Fixed base ratio growth rate = (current period - base period) / base period × 100%. In the process of pig industry monitoring, fixed base indexes of pig stocks and breeding sow stocks are usually calculated.

[0092] (II) Cycle Analysis Technology

[0093] Cycle analysis refers to finding the periodicity of data changes and analyzing the cycle patterns in the process of long-term time series data changes. Cycle analysis can be used to study the cycles of an industry in the process of long-term historical changes. The "pig cycle" has long plagued the pig industry and has attracted widespread attention in the industry. After years of data accumulation, the cyclical fluctuation patterns and fluctuation ranges of the pig industry can be clearly observed. Studies have found that since 2003, the cyclical fluctuation patterns of live pig, piglet and pork prices have been relatively obvious, basically a four-year cycle, and the fluctuation range has become larger and larger. Based on this, it can be judged that 2018 is the end of the previous cycle and the beginning of a new cycle. Therefore, it can be judged that the overall pig price in 2019 will show an upward trend. Figure 2 shown.

[0094] (III) Correlation analysis technology

[0095] Correlation analysis is a basic tool for causal analysis. Through correlation analysis, the substitution relationship and correlation between analysis indicators can be determined. The correlation between two different variables is judged to determine whether there is a correlation between the variables. Correlation analysis is used to study the relationship between two variables (X, Y) and determine the degree of their connection. The determination method can be observed intuitively from a scatter plot, or a more accurate judgment can be obtained by calculating the correlation coefficient. The calculation process can be directly obtained through corresponding software, such as SPSS, Eviews software, etc. The value range of the correlation coefficient r is -1≦r≦1. When r=-1, the two variables are completely negatively correlated, when r=1, the two variables are completely positively correlated, and when r=0, the two variables are completely unrelated. Generally speaking, the correlation coefficient of two variables will not have such an extreme value, but is located in a certain interval in the middle. For example, according to the law of pig herd turnover, the current number of sows in stock is closely related to the number of commercial pigs on the market ten months later. The correlation coefficient between the two reaches 0.6831. The change in the number of sows in stock in the current period has a direct impact on the changes in the market price of live pigs in the later period. Figure 3 shown.

[0096] Through the test results of the factors affecting pig production risks, this application shows that the proportion of sows x1, live pig price x2, piglet cost x3, per head feed cost x4 and per head epidemic prevention cost x5 all have an impact on the number of pigs slaughtered, but the leading length of their impact needs to be analyzed using the vector autoregression model (VAR model) of the econometric software package (Eviews software). Live pig price refers to the average price of live pigs.

[0097] In an exemplary embodiment, the multiple indicator data are the proportion of sows capable of reproduction, the price of live pigs, the average cost of piglets per head, the average feed cost per head and the average epidemic prevention cost per head.

[0098] The leading period for the proportion of sows is 10 months, that is, changes in the proportion of sows in this period will affect the number of pigs slaughtered 10 months later. This conclusion is in line with general production experience.

[0099] The lead period of the live pig price is 12 months, that is, the change of the live pig price in this period will affect the number of live pigs sold 12 months later. The high or low live pig price will inevitably affect the enthusiasm of farmers for pig farming. Due to the long production cycle of live pigs, it usually takes one year for the market price factor to affect the number of live pigs sold.

[0100] The lead time for the average piglet cost per head is 6 months. It takes about 6 months from the purchase of piglets to the slaughter of live pigs by farmers. Changes in piglet costs will affect the number of piglets purchased by farmers, and in turn affect the number of live pigs slaughtered a few months later.

[0101] The lead time of the above-mentioned per-pig feed cost is 12 months. Since farmers have a certain amount of pig stocks, the sudden increase in feed prices will not have much impact on the number of pigs slaughtered in the near future. The current increase in feed prices will affect farmers to reduce the number of piglets purchased, which will have a certain impact on the number of pigs slaughtered in the future.

[0102] The lead period for the per-head epidemic prevention costs is 1 month or 12 months. The lead period for the per-head epidemic prevention costs for fattening pigs is 1 month, and the lead period for the per-head epidemic prevention costs for breeding sows is 12 months. An epidemic outbreak will directly lead to an increase in epidemic prevention costs, and infection with the disease will cause the death of pigs in a short period of time, affecting the pig output; an epidemic outbreak will also affect the number of sows on hand, resulting in the impact on the number of pigs slaughtered in the next period; in addition, the outbreak of an epidemic requires a certain amount of reaction time. After an epidemic breaks out, farmers take a wait-and-see attitude towards restocking, and wait until the epidemic has a cure or stabilizes before restocking again, which usually takes 1 to 2 years.

[0103] The pig industry monitoring and early warning analysis shows that the number of pigs slaughtered is intrinsically linked to the proportion of sows, the average price of live pigs, piglet costs, feed costs per pig, and epidemic prevention costs per pig. However, it is still uncertain whether these factors are correlated with the number of pigs slaughtered and the size of the correlation, so a measurement model is established to further analyze them.

[0104] The pig output prediction model is a multiple regression model.

[0105] The multiple regression model is expressed as: y = a + b1x1 + b2x2 + b3x3 + b4x4 + b5x5 + μ, where y is the predicted value of the number of pigs slaughtered, a is a constant term, u is a disturbance term, and b1, b2, b3, b4 and b5 are coefficients.

[0106] The multivariate regression model is obtained by fitting data based on historical data sets. Each sample data in the historical data set includes input data and output data. The input data is the indicator data of each corresponding leading period, and the output data is the predicted value of the number of live pigs slaughtered.

[0107] In the statistical analysis software (Stata software), the number of live pigs slaughtered is taken as the explained variable, and the proportion of sows, live pig price, piglet cost, per head feed cost and per head epidemic prevention cost are taken as explanatory variables to obtain the regression equation and exponential regression equation. The regression equation is the multiple regression model.

[0108] The regression equation is:

[0109] y=1558.057+73.52599x1+93.27704x2-2.529015x3-2.266197x4-21.35572x5;

[0110] Exponential regression equation:

[0111] lny=13.1298+0.6232lnx1+0.6796lnx2-0.4603lnx3-1.0210lnx4-0.1116lnx5.

[0112] The coefficient before the explanatory variable in the multiple regression model is called the partial regression coefficient, which means that when other explanatory variables remain unchanged, a certain explanatory variable changes by one unit and the explained variable y changes on average, that is, the degree of influence of a certain explanatory variable on the explained variable y. x1, x2, x3, x4 and x5 are all coefficients. The coefficient of x1 in the exponential regression equation is 0.6232, indicating that for every one percentage point increase in the proportion of sows, the number of pigs slaughtered after 12 months will increase by 0.6232 percentage points.

[0113] In order to predict the possible impact of pig industry policies and animal diseases on the future pig industry, the project team constructed the China Pig Sector Model (CPSM). The pig industry equilibrium model quantitatively describes the production, consumption, trade and price of the pig industry and its influencing factors through equations, and establishes the relationship between them. The current pig price is jointly determined by pig supply, market demand and international market supply. When the market is cleared, the market equilibrium is achieved. This model is a rule simulation and external shock prediction and early warning tool. It can be used to quantitatively simulate and evaluate the impact of various rules or external factors on the development of the pig industry, and provide technical support for the research on the development of the pig industry.

[0114] The pig industry equilibrium model consists of a series of equations such as production, consumption, trade, price and supply and demand balance. The equations include relevant factors that affect endogenous variables, and achieve supply and demand balance and market clearing through price interaction. At present, the model consists of 36 sets of equations, a total of 566 single equations, 23 sets of variables, 566 endogenous variables and several exogenous variables. The main equations are:

[0115] 1. Pork production equation: QX CMEAT,T =∑ CLIVE (IOLVMT CLIVE,CMEAT,T ×QDP CLIVE,T ).

[0116] Among them, QX CMEAT,T For pork production, IOLVMT CLIVE,CMEAT,T is carcass weight, QDP CLIVE,T It is the predicted value of live pig output. The subscript T represents time, CLIVE represents live pigs, and CMEAT represents meat.

[0117] 2. Pork Demand Equation: QDT C，T =QDFC，T +QDL C,T +QDS C，T +QDP C，T +QDO C，T +QDW C，T +STV C，T .

[0118] Among them, QDT C,T is the total pork demand, QDF C，T The amount of pork required for a household, QDL C,T QDS is the market demand for pork. C，T Pork quantity required for schools, QDP C，T To process the pork demand, QDO C,T To replace the pork demand, QDW C，T For other pork demand, STV C，T The demand for pork imports.

[0119] 3. Consumer Price Equation: PD C，T =PX C，T ×(1+margD C ).

[0120] Among them, PD C，T is the pork consumption price, PX C,T is the pork production cost, margD C It is the average profit ratio of live pigs.

[0121] 4. Market Clearing Equation: QX C，T +QM C，T -QE C，T =QDT C，T .

[0122] Among them, QX C,T is the pork import volume, QM C,T is the pork production during the cycle, QE C，T The market pork inventory.

[0123] After step 102, the pig industry forecasting method further includes: according to the formula QX CMEAT,T =∑ CLIVE (IOLVMT CLIVE,CMEAT,T ×QDP CLIVE,T ) Forecast pork production.

[0124] After step 102, the pig industry forecasting method further includes: according to the formula QDT C，T =QDF C，T +QDL C，T +QDS C，T +QDP C,T+QDO C，T +QDW C，T +STV C，T Forecasting pork demand.

[0125] After step 102, the pig industry prediction method further includes: when the predicted value of the number of pigs slaughtered exceeds a preset range, issuing a warning message.

[0126] For example, in order to evaluate and simulate the impact that a certain swine fever epidemic may have on the later pig market, the model designed relevant plans and calculated its impact results. The model results show that if there is no swine fever epidemic, the national pork production of a certain country in 2020 will be 54.38 million tons, the net import will be 1.17 million tons, and the direct food consumption of residents will be 39.6 million tons. Affected by a certain swine fever epidemic, the initial sow inventory in 2020 decreased by 8%. In this case, the model predicts that the national pork production will decrease by 5.3% in 2020. If imports do not increase, pork prices will rise by 19.5%. Due to the price increase, the direct food consumption of residents will decrease by 5.4%. If the domestic supply and demand gap is adjusted by increasing imports, the domestic market demand and prices of a certain country will remain unchanged, and the decline in pork production will be greater than the assumption that imports remain unchanged. Under a medium impact, the domestic pork production of a certain country will decrease by 10.3%, and imports will increase by 2.19 times, that is, to 4.96 million tons, an increase of 3.41 million tons. According to customs data monitoring, a certain country imported a total of 5 million tons of pork and pork offal in 2020, of which 4.39 million tons was pork, which is basically consistent with the expected data.

[0127] This application can strengthen the monitoring of the pig industry by effectively predicting the number of live pigs slaughtered, and improve the development level of the pig industry through digital transformation.

[0128] Based on the same inventive concept, the embodiment of the present application also provides a pig industry prediction device for implementing the pig industry prediction method involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in one or more pig industry prediction device embodiments provided below can refer to the limitations of the pig industry prediction method above, and will not be repeated here.

[0129] In an exemplary embodiment, the present application provides a pig industry prediction device comprising:

[0130] The data collection module is used to collect the impact data of the pig industry within the set area; the impact data includes multiple indicator data, each indicator data is the indicator data of the date before the corresponding lead period of the forecast date, and each indicator data corresponds to a lead period.

[0131] The pig output number prediction module is used to input the influencing data into the pig output number prediction model to obtain the predicted value of the pig output number on the predicted date within the set area.

[0132] The multiple indicator data are respectively the proportion of sows, live pig prices, average piglet cost per head, average feed cost per head and average epidemic prevention cost per head; the lead period for the proportion of sows is 10 months, the lead period for the live pig prices is 12 months, the lead period for the average piglet cost per head is 6 months, the lead period for the average feed cost per head is 12 months, and the lead period for the average epidemic prevention cost per head is 1 month or 12 months.

[0133] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0134] This article uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method and core ideas of this application. At the same time, for those skilled in the art, according to the ideas of this application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.

Claims

1. A pig industry forecasting method, characterized in that: The pig industry forecasting method comprises: Collect the impact data of the pig industry within the set area; the impact data includes a plurality of indicator data, each indicator data is the indicator data of the date before the corresponding lead period of the forecast date, and each indicator data corresponds to a lead period; The influencing data is input into a pig slaughter number prediction model to obtain a predicted value of the pig slaughter number on the predicted date within the set area.

2. The pig industry forecasting method according to claim 1, characterized in that: The multiple indicator data are the proportion of sows in breeding, the price of live pigs, the average cost of piglets per head, the average feed cost per head and the average epidemic prevention cost per head.

3. The pig industry forecasting method according to claim 2, characterized in that: The lead period for the proportion of sows is 10 months, the lead period for the live pig price is 12 months, the lead period for the average piglet cost per head is 6 months, the lead period for the average feed cost per head is 12 months, and the lead period for the average epidemic prevention cost per head is 1 month or 12 months.

4. The pig industry forecasting method according to claim 1, characterized in that: The pig output prediction model is a multiple regression model.

5. The pig industry forecasting method according to claim 2, characterized in that: Collect the impact data of the pig industry within the set area, including: The farms within the set area are sampled by random sampling, and the proportion of sows capable of reproduction, the average piglet cost per pig, the average feed cost per pig, and the average epidemic prevention cost per pig are collected; The sampling method is adopted to sample the collective markets within the set area, and the live pig prices of the sampled collective markets are collected; The impact data of the pig industry in the set area is calculated based on the proportion of sows in the sampled farms, the average cost of piglets per head, the average feed cost per head and the average epidemic prevention cost per head, as well as the live pig prices in the sampled markets.

6. The pig industry forecasting method according to claim 1, characterized in that: After inputting the impact data into the pig output prediction model to obtain the predicted value of the pig output on the predicted date within the set area, the pig industry prediction method further includes: According to the formula QX CMEAT，T =∑ CLIVE (IOLVMT CLIVE，CMEAT，T ×QDP CLIVE，T ) Forecast pork production; Among them, QX CMEAT，T For pork production, IOLVMT CLIVE，CMEAT，T is carcass weight, QDP CLIVE，T is the predicted value of the number of live pigs slaughtered, and CLIVE represents the set of live pigs.

7. The pig industry forecasting method according to claim 6, characterized in that: After inputting the impact data into the pig output prediction model to obtain the predicted value of the pig output on the predicted date within the set area, the pig industry prediction method further includes: According to the formula QDT C，T =QDF C，T +QDL C，T +QDS C，T +QDP C，T +QDO C，T +QDW C，T +STV C，T forecasting pork demand; Among them, QDT C，T is the total pork demand, QDF C，T The amount of pork required for a household, QDL C，T QDS is the market demand for pork. C，T Pork quantity required for schools, QDP C，T To process the pork demand, QDO C，T To replace the pork demand, STV C，T is the pork import demand, QDW C，T For QDF C，T 、QDL C，T , QDS C，T , QDP C，T 、QDO C，T and STV C，T Other pork demand.

8. The pig industry forecasting method according to claim 1, characterized in that: After inputting the influencing data into the pig output prediction model to obtain the predicted value of the pig output on the predicted date within the set area, the pig industry prediction method also includes: issuing a warning message when the predicted value of the pig output exceeds a preset range.

9. A pig industry prediction device, characterized in that: The pig industry prediction device comprises: A data collection module is used to collect the impact data of the pig industry within a set area; the impact data includes a plurality of indicator data, each indicator data is the indicator data of the date before the corresponding lead period of the forecast date, and each indicator data corresponds to a lead period; The pig output number prediction module is used to input the influencing data into the pig output number prediction model to obtain the predicted value of the pig output number on the predicted date within the set area.

10. The pig industry prediction device according to claim 9, characterized in that: The multiple indicator data are respectively the proportion of sows, live pig prices, average piglet cost per head, average feed cost per head and average epidemic prevention cost per head; the lead period for the proportion of sows is 10 months, the lead period for the live pig prices is 12 months, the lead period for the average piglet cost per head is 6 months, the lead period for the average feed cost per head is 12 months, and the lead period for the average epidemic prevention cost per head is 1 month or 12 months.

Images