CN114266395A - Cigarette logistics distribution center information system based on combined prediction method - Google Patents

Abstract

The invention discloses a cigarette logistics distribution center information system based on a combined prediction method, which comprises a cigarette demand combined prediction model: constructing a cigarette combination demand model of a time sequence according to the influence factors, and predicting the logistics demand of the cigarette area; the cigarette industry and commerce finished product preposed warehouse based on the logistics alliance: establishing a logistics union of industrial and commercial enterprises, establishing a site selection mathematical model to determine the optimal site selection longitude and latitude according to the lowest cost and the optimal service principle, establishing a cigarette finished product prepositive library of the industrial enterprise in the area of the optimal site selection to store partial cigarettes with high market demands, and providing the industrial warehousing and delivery service for commercial enterprises; intelligent warehouse management system: the electronic tags are associated with cigarette information based on the RFID technology, precise intelligent management is carried out on cigarettes in different industries in a warehouse, integration of warehousing and delivery is achieved, and seamless connection with an engineering system I is achieved. The invention improves the industrial transportation efficiency, reduces the commercial shortage of goods and enhances the sorting and distribution timeliness.

Description

Cigarette logistics distribution center information system based on combined prediction method

Technical Field

The invention relates to the field of tobacco logistics distribution, in particular to a cigarette logistics distribution center information system based on a combined prediction method.

Background

Due to the practical requirements of building a modern tobacco economic system and promoting the high-quality development of the tobacco industry, the construction of a logistics distribution center and a distribution center in a propulsion area needs to be explored. Therefore, promoting regional logistics construction and integrating industrial and commercial logistics business are inevitable trends of high-quality logistics development in the tobacco industry. At present, due to the fact that cigarette market demands are distributed unevenly, scientific guidance is lacked in logistics work, effective connection is not formed between industrial enterprises and commercial enterprises, cigarette transport vehicles of industrial enterprises are low in full load rate, long in-transit time and high in transport cost; the inventory turnover of the commercial enterprises is low, and the capital occupation cost is high.

Disclosure of Invention

The invention aims to overcome the defects of the prior art and provides a cigarette logistics distribution center information system based on a combined prediction method, so as to solve the problems that the distribution of cigarette market demands is uneven, the logistics work lacks scientific guidance, the industrial and commercial enterprises do not form effective connection, the cigarette transport vehicles of the industrial enterprises have low full-load rate, long in-transit time and high transport cost; the inventory turnover of the commercial enterprises is low, and the capital occupation cost is high.

In order to achieve the purpose, the technical scheme adopted by the invention is as follows: a cigarette logistics distribution center information system based on a combined prediction method comprises

The cigarette demand combination prediction model comprises the following steps: constructing a cigarette combination demand model of a time sequence according to the influence factors, and predicting the logistics demand of the cigarette area;

the cigarette industry and commerce finished product preposed warehouse based on the logistics alliance: establishing a logistics union of industrial and commercial enterprises, establishing a site selection mathematical model to determine the optimal site selection longitude and latitude according to the lowest cost and the optimal service principle, establishing a cigarette finished product prepositive library of the industrial enterprise in the area of the optimal site selection to store partial cigarettes with high market demands, and providing the industrial warehousing and delivery service for commercial enterprises;

intelligent warehouse management system: the electronic tags are associated with cigarette information based on the RFID technology, the cigarettes in different industries in the warehouse are precisely and intelligently managed, the integration of warehouse entry and exit is realized, and the electronic tags are in seamless joint with an engineering system I; the first engineering system is specifically a national bureau first engineering system.

Preferably, the influencing factors include historical subscription data, seasonal variations, holidays, prices.

Preferably, the construction process of the cigarette demand combination prediction model comprises the following steps,

s1, analyzing the cigarette market capacity and demand: comprehensively analyzing the whole market capacity of the cigarettes and other influencing factors around two levels of a macroscopic environment and a cigarette market; the whole cigarette market capacity comprises the market capacity of a standing population, the gift sending capacity of the standing population and the purchasing capacity of a tourist population, and the other influence factors comprise seasons, festivals and holidays;

s2, establishing a seasonal ARIMA model: establishing an autoregressive moving average model aiming at the cigarette historical ordering data, then introducing the quarterly and holidays in the influence factors, and establishing a seasonal ARIMA model;

s3, establishing a cigarette demand combination prediction model based on machine learning: on the basis of cigarette historical ordering data, seasonal characteristic values are constructed, the characteristics of a machine learning algorithm are fused, a semi-supervised machine learning cigarette demand prediction model is constructed, and the monthly cigarette ordering amount is predicted by using a rolling type single-step cyclic prediction method.

Preferably, the overall market capacity includes

Standing population market capacity: the smoking angle is the regular population, the smoking rate, the daily average smoking amount and the number of days of residence in province;

capacity of daily life: the gift consumption is the number of smokers with suitable age in the population of the ordinary living, the gift sending rate and the gift sending amount in the whole year;

consumption capacity of an aged smoker: the consumption of smokers with suitable age is divided into the number of the population with suitable age, the ratio of the smokers and the average smoking amount of the smokers;

market capacity of tourism population: the purchase angle is the number of tourist population, the purchase rate and the purchase amount.

Preferably, the population of the regular living is set to live for 6 months or more, the smoker of the suitable age is set to 18 to 64 years old, and the population of the tourist is set to live for less than 1 month.

Preferably, the construction process of the seasonal ARIMA model comprises the following steps,

s1, sequence difference: selecting monthly cigarette ordering data of a certain time period to perform model fitting, establishing a time-ordering quantity sequence diagram, and performing stability inspection and seasonal difference on the data respectively;

s2, smoothness checking: performing stationarity test on the original data, and when the p value of the test result is less than 0.005, proving that the original data is stable and purely random; when the stability test is not passed, carrying out difference calculation, and when the p value of the original data after 1-order difference is less than 0.005, passing the stability test;

s3, seasonal decomposition: carrying out periodic decomposition on the original data, wherein the periodicity is 12, extracting seasonal indexes to form a periodic time sequence, carrying out 2-order difference on the periodic time sequence to obtain a 1-order difference and 2-order seasonal difference sequence diagram, and determining the orders of d to be 1 and 2 respectively;

s4, p, q value determination: selecting an autoregressive order p and a moving average term number q of a seasonal ARIMA model, and determining through a 1-order difference, a 2-order seasonal autocorrelation coefficient ACF graph and a partial autocorrelation coefficient PACF graph; the p and q values are determined as (1,1) through analysis, and the p, d and q values are determined, so that the cigarette demand prediction model is ARIMA (1,1,1) (1,2,1), namely the seasonal ARIMA model.

Preferably, the construction process of the cigarette demand combination prediction model based on machine learning comprises the following steps,

s1, time matrix conversion: training and fitting the model, and assuming that the time sequence is at different time points t₁,t₂,…,t_nObservation of (2)The value may be A (t)₁)，A(t₂)，...，A(t_n) Is shown if t_n+1The observed value of the time can be expressed as

A(t_n+1)＝f(A(t_n)，A(t_n-1)，...，A(t_n-k+1))

Where k < n, then t is indicated_n+1The observation value of the moment is a function expression of the observation values of the k previous time points, and is predicted in a rolling mode according to a time sequence;

s2, cutting the data set: performing machine learning to perform test set and training set division on the data, wherein the training set is used for training the model, and the test set judges the final training effect according to the training result of the training set;

s2, predicting the result: after 14 iterations, the prediction model test results are obtained, the results in the graph are converted into a table, and the error is calculated.

Preferably, the process for constructing the pre-library of finished products of cigarette industry and commerce based on the logistics alliance comprises the following steps,

s1, establishing a preposed database addressing evaluation model: determining alternative points of the pre-database by a qualitative and quantitative combined method, converting the alternative points into a mathematical formula by aiming at the minimum value of the sum of storage cost, change cost, transportation cost and goods loss cost and the maximum service reliability, and establishing a pre-database site selection mathematical model;

s2, determining the optimal point of site selection: according to the established preposed base addressing mathematical model, programming solution is carried out by utilizing Python language to obtain an optimal solution of an objective function, the longitude and latitude of the area where the optimal solution is located are obtained, and the map is utilized to carry out visual conversion to obtain the location corresponding to the longitude and latitude.

Preferably, the construction process of the pre-database addressing evaluation model comprises the following steps,

s1, decision variables and parameter definition:

l-a collection of industrial enterprises, L ═ (1,2, … …, p);

i-set of pre-library candidate points, I ═ 1,2, … …, m;

j-stream center demand point set, J ═ (1,2, … …, n);

h_il-unit price of transportation from factory i to warehouse i;

w_il-traffic from factory i to warehouse i;

q_il-mileage from factory i to warehouse i;

d_j-demand d for each logistics centre demand point j;

C_i-setting a fixed cost for the warehouse candidate node i;

S_i-maximum supply of warehouse candidate nodes i;

c_ij-a shipping mileage from the warehouse candidate node i to the logistics center point j;

r_ij-unit price from warehouse candidate node i to logistics center j;

x_ijthe quantity of materials for warehousing the warehouse alternative node i to the logistics center j;

z_i-representing a cost z of selecting i alternative nodes to build a bin;

y_i-selecting warehouse alternate point i as warehouse, y _i1 is ═ 1; if not, then y_i＝0；

The transportation cost of each industrial company for delivering goods to the front-end warehouse is the product of the transportation unit price, the transportation amount and the mileage, and the total cost T is expressed as:

similarly, the expression of the total transportation cost Q from the front-end warehouse to each logistics center is as follows:

s2, mathematical modeling: based on the above conditions, a mathematical expression is established

The restriction condition group (4) is explained

1) The demand point j demands the satisfaction degree;

2) the number of products provided by the front-end warehouse is not more than the maximum capacity of the facility;

3) the total transportation volume from each industrial enterprise to the preposed warehouse i is less than the maximum capacity of the facility;

4) the delivery amount of the warehouse i to the demand point j is not negative;

5) the number of warehouse alternate points is at least 1.

Preferably, the intelligent warehouse management system comprises

A basic management module: as a basic layer, the system is mainly responsible for realizing basic information maintenance, wherein the basic information maintenance comprises information maintenance of a goods main dimension, a warehouse, a storage area, a goods shelf, a goods location and a strategy;

an operation management module: the method comprises the steps of warehousing management, warehousing management and ex-warehouse management, wherein the warehousing management is used for warehousing entry management, warehousing goods position distribution and warehousing confirmation, the ex-warehouse management is used for inventory management, damage and overflow management and goods position adjustment, and the ex-warehouse management is used for ex-warehouse entry management, ex-warehouse goods position distribution and ex-warehouse confirmation;

a comprehensive analysis module: the method comprises inventory analysis, warehouse-in and warehouse-out analysis and warehouse age analysis, wherein the inventory analysis is used for inventory data query and handheld PDA inventory display, the warehouse-in and warehouse-out analysis is used for warehouse-in data query and warehouse-out data query, and the warehouse age analysis is used for warehouse age screening and warehouse age data query.

The invention has the beneficial effects that:

1. the invention is based on the cigarette time sequence demand prediction research of a combined prediction method, scientifically predicts and analyzes the cigarette market change situation by utilizing advanced prediction methods such as machine learning, ARIMA and the like, realizes the autonomous controllability of an industry prediction method, improves the industry intelligent prediction level and dynamically masters the market change.

2. According to the invention, through the modes of logistics union, resource sharing, common warehouse building, and common factory and merchant sharing and common management of the front warehouse, the dynamic change of the attributes of the front warehouse is realized, the industrial transportation efficiency is improved, the commercial shortage of goods is reduced, and the timeliness of sorting and distribution is enhanced.

3. The intelligent warehousing management system realizes the whole-process intelligent management of work such as warehousing management, goods allocation to ex-warehouse confirmation and the like of cigarettes, ensures high integration and information sharing inside and outside a management information system, and improves the resource reuse rate; and the open interface design ensures strong adaptability of future system expansion and secondary development.

4. The intelligent warehousing management system mode of the invention adopts a front-end warehouse and a set of client-side systems to dock a 1-to-N mode of a plurality of industrial enterprise managers, thus greatly simplifying the docking work with each industrial enterprise and not needing to deploy independent client-side software and hardware respectively.

Drawings

FIG. 1 is a sequence chart of a date order quantity of the present invention;

FIG. 2 is a sequence diagram of 1 st order difference &2 nd order seasonal difference of the present invention;

FIG. 3 is a diagram of a 1-stage seasonal differential auto-correlation ACF according to the present invention;

FIG. 4 is a graph of 1-stage seasonal partial autocorrelation PACF of the present invention;

FIG. 5 is a graph of a 2-stage seasonal differential auto-correlation ACF according to the present invention;

FIG. 6 is a graph of a 2-stage seasonal partial auto-correlation PACF of the present invention;

FIG. 7 is a diagram of ARIMA cigarette demand prediction results in view of seasonal influences in accordance with the present invention;

FIG. 8 is a graph of non-seasonal ARIMA cigarette demand prediction results of the present invention;

FIG. 9 is a graph of the output of the test results of the present invention;

FIG. 10 is a diagram of a pre-warehouse distribution model of the business association of the present invention;

FIG. 11 is a diagram of an optimal solution for Python objective function according to the present invention;

FIG. 12 is a visual depiction of the prepositioned library addressing optimal point of the present invention;

FIG. 13 is a functional block diagram of the intelligent warehouse management system according to the present invention;

fig. 14 is a functional structure diagram of the system of the present invention.

Detailed Description

In order to make the objects, technical solutions and advantages of the present invention more apparent, the technical solutions of the present invention will be clearly and completely described below with reference to the specific embodiments of the present invention and the accompanying drawings.

Examples

In this embodiment, taking the Sichuan tobacco market as an example, the functional structure of the system of the present invention is shown in FIG. 14.

1. Cigarette time series demand prediction research based on combined prediction method

1.1 cigarette market Capacity and demand analysis in Sichuan province

The premise of forecasting the logistics demand is to accurately master the cigarette market development condition. The method disclosed by the invention is used for comprehensively analyzing the cigarette market and the influence factors around two levels of a macroscopic environment and the cigarette market.

1) Overall market volume of cigarette in Sichuan province

The cigarette market capacity in Sichuan province consists of three parts, namely:

one permanent population market capacity:

smoking angle is defined as population, smoking rate, average smoking amount per day, residence days in province

Definition of standing population: living for more than 6 months;

second, the capacity of the regular living population for delivering gifts:

gift consumption is 18-64 years old population, gift rate and whole year gift amount

Definition of standing population: living for more than 6 months;

wherein 18-64 years old are smokers of suitable age, and the consumption calculation method for smokers of suitable age is as follows:

the consumption of smokers with suitable age is divided into the number of the population with suitable age, the ratio of the smokers and the average smoking amount of the smokers

(iii) buying angle ═ number of tourist population x buying tobacco rate x buying tobacco quantity

Definition of tourist population: the population with residence time less than 1 month is called tourist population, including the purposes of travel, play, investigation, short-term service worker, etc.

Through calculation and market total reduction, the cigarette capacity of the whole province in 2020 is determined to be 257.8 ten thousand boxes, the error range is 256.5-259.1 ten thousand boxes, wherein the cigarette capacity of the 18-64-year-old resident population is 212.7 ten thousand boxes, the cigarette capacity of the 64-year-old resident population is 32.9 ten thousand boxes, and the cigarette consumption of the tourist population is 9.5 ten thousand boxes.

2) Other factors

Because the cigarettes are obviously fluctuated due to the fact that holidays of spring festival, national day and the like are close to during sale, when market demands are predicted, besides market capacity, influence factors such as seasons, holidays and the like need to be considered in a key mode.

1.2 cigarette demand prediction analysis based on time series prediction-ARIMA model

According to the method, the data are analyzed and found to be time series with linear characteristics aiming at the cigarette historical ordering data in Sichuan province, and an autoregressive moving average model (ARIMA) is established for the data, so that the method has better fitting performance and prediction accuracy. Therefore, the method selectively establishes an ARIMA model to predict the cigarette demand market. However, since the historical ordering data of the cigarettes in Sichuan province has strong seasonal characteristics, seasonal influence factors need to be introduced during modeling, and a seasonal ARIMA model needs to be established, so that the fitting degree of the model is enhanced, and the prediction accuracy of the model is improved.

1) Sequence difference

And (3) selecting cigarette monthly ordering data from 10 months to 2021 months in 2006 in Sichuan province to perform model fitting to predict the future cigarette ordering amount. Firstly, a time-order quantity sequence is established as shown in fig. 1, and as can be seen from the sequence chart, the sequence chart contains obvious seasonal components and has obvious changes, and for the characteristics, the data are respectively subjected to stationarity check and seasonal differentiation.

2) Stationarity test

The ARIMA model requires that the sequence is a stationary sequence, so that the stationary test (white noise test) is carried out on the original data, and when the p value of the test result is less than 0.005, the data is proved to be stationary and purely random. The p value of the cigarette historical ordering raw data is 0.01573855, and the cigarette does not pass the stability testIf the difference is needed, the p value of the order data after 1-order difference is calculated to be 6.53619543 multiplied by 10^-14And the stability can be checked.

3) Seasonal decomposition

The original data is periodically decomposed, the periodicity is 12, seasonal indexes are extracted, a periodic time sequence is formed, 2-order difference is carried out on the periodic time sequence, and a 1-order difference & 2-order seasonal difference sequence diagram is obtained and is shown in figure 2.

Therefore, the order of d can be determined to be 1 and 2, respectively.

4) p, q value determination

The autoregressive order p and the moving average number q of terms of the seasonal ARIMA model are selected and determined through an ACF (seasonal autocorrelation coefficient) graph and a PACF (partial autocorrelation coefficient) graph of 1 order difference and 2 order seasonality, as shown in figures 3-6.

As can be seen, the graphs all showed streaks, and the p and q values were both (1,1) by analysis. And determining the p, d and q values, and obtaining an ARIMA (1,1,1) (1,2,1) as a cigarette demand prediction model.

5) Comparison of predicted results

Comparing ARIMA (1,1,1) (1,2,1) model considering seasonal influence factors with ARIMA (1,1,1) model prediction results:

TABLE 1 comparison of predicted results

Index (I)	Consider seasonal ARIMA (1,1,1) (1,2,1)	Non-seasonal ARIMA (1,1,1)
			Stationarity R2	0.871	0.708
Significance test	0.002	0.000
			Mean Absolute Error (MAE)	7.08％	23.56％

Referring to fig. 7-8, it is apparent from the prediction results that the average absolute error of the ARIMA prediction model considering the seasonal influence factors can be reduced to 7%, the prediction results are more accurate, and the fitting degree is higher.

1.3 cigarette demand combination prediction model analysis based on machine learning

The ARIMA prediction model can only capture the linear relation of data essentially and can not capture the nonlinear relation, so in order to further improve the prediction precision, the invention adopts a machine learning method, constructs 72 characteristic values of spring festival, year, month and the like on the basis of the cigarette historical ordering data, integrates the characteristics of various machine learning algorithms, constructs a semi-supervised machine learning cigarette demand prediction model, and predicts the monthly cigarette ordering amount by using a rolling type single-step cyclic prediction method.

1) Time matrix conversion

In a machine learning-based time series prediction model, it is usually necessary to train and fit the model, assuming that at different time points t for the time series₁，t₂，...，t_nThe observed value of (a) can be represented by A (t)₁)，A(t₂)，...，A(t_n) Is shown if t_n+1The observed value at a time can be expressed as:

A(t_n+1)＝f(A(t_n)，A(t_n-1)，...，A(t_n-k+1))

where k < n, then t is indicated_n+1The observations at a time are a functional representation of the observations at its first k time points. Time series rolling prediction forms such asAs shown in Table 2, the present invention sets the value of the cycle period k to 12.

TABLE 2 time series rolling prediction table

The prediction of the cigarette order quantity can be understood as that the target value of the 13 th month, namely the cigarette order quantity of the 13 th month is solved by utilizing the first 12 input characteristic sequences.

2) Data set slicing

And performing machine learning to perform test set and training set division on the data, wherein the training set is used for training the model, and the test set judges the final training effect according to the training result of the training set. In the existing four-chuan province, 177 cigarette historical ordering data are counted in the period from 2006 to 10-2021 and the model is considered to be fully trained as far as possible to meet the prediction requirement, the first 171 data are selected as a training set, and the last 6 data are selected as a test set

Training set: TrainX belongs to R^12×171(ii) a Training set labels: TrainY ∈ R^1×6

And (3) test set: TestX ∈ R^12×171(ii) a Test set labeling: TestY ∈ R^1×171

The dimension of a single sample of the training data is 12, which means that 13 th time point data is predicted by using the previous 12 adjacent time point data.

3) Predicted results

After 14 iterations, the test result of the prediction model is shown in fig. 9, the result in the graph is converted into a table, and the calculation error is shown in table 3:

TABLE 3 comparison of true and predicted results

	1 month in 2021	2 months in 2021	3 months in 2021	2021 year 4 month	2021 year 5 month	6 months in 2021
							True value	27079042	9402202	12883436	12569556	12186087	11721043
Prediction value	26662123	10479039	12172066	12413363	12959105	12356890
							Absolute error	0.015	-0.11	0.055	0.012	-0.063	-0.054

Through adjustment, the Average error (Average MAE) of the model is 5.2%, and compared with a seasonal ARIMA model, the cigarette prediction model based on the combination of various machine learning methods has the advantages that the prediction precision is further improved, so that the cigarette market analysis in the tobacco industry is more scientific and accurate.

2. Construction of preposed warehouse of Chuan finished products of cigarette industry and commerce

2.1 preposition database site selection evaluation model

2.1.1 Prebase addressing influencing factors and targets

1 influencing factor

Influence factors of the site selection of the pre-database can be divided into cost factors and non-cost factors according to a key index of cost, and the influence factors are specifically shown in table 4:

TABLE 4 various influencing factors for preposed base addressing

The whole process of logistics transportation on a tobacco supply chain strictly complies with the national tobacco transportation standard, and finished cigarettes are dispatched by an industrial company and transported to a logistics center of a commercial enterprise for warehousing, sorting, ex-warehouse and other operations; on the other hand, the existing storage center in Sichuan province, which is partially idle due to aging of equipment and other factors, still needs to pay a large amount of management cost every year. Therefore, in addition to the traditional influence factors such as geographical distribution and transportation cost of tobacco logistics centers in various cities (states), special factors such as inventory resources and upgrade and reuse of old plants are comprehensively considered.

2 site selection target

Aiming at the factors, the invention mainly considers the following targets during address selection:

the cost is minimized.

The primary goal in site selection is to minimize the overall cost of transportation and facilities. The total cost of facilities is often constant, and the transportation cost is determined by the transportation quantity, the transportation distance and the transportation unit price. In order to realize the scale effect, the transportation quantity must reach the transportation batch, which is beneficial to reducing the transportation cost; the transport distance is determined by the logistics front-end warehouse, and the total transport distance can be reduced by reasonable logistics warehouse site selection, so that the transport cost is reduced; and the unit price of transportation depends on the transportation mode and the transportation batch.

Maximization of the object flow

The maximization of the material flow rate is realized on the basis of the minimization of the material flow cost. The service area of the cigarette pre-warehouse in Sichuan province is distributed in cities (states) in Sichuan province, the service area is wide, the southeast is more economic, the customer demand is larger, and the material flow is larger. The larger object flow rate reflects the higher processing capacity and operation level of the prepositive warehouse.

Service reliability maximization

The address selection of the preposed library meets the optimization of delivery time, delivery distance, delivery speed and timeliness, and embodies higher service level.

High development potential

The preposed library has long service time and high construction cost, and cannot be easily moved and changed after being built, so that the invention fully considers the development potential and the economic development level of the surrounding environment and is beneficial to the expansion situation of self service in the site selection process.

2.1.2 Pre-selection of addresses and mathematical modeling

1 preselection address

Aiming at the problems of low loading rate of industrial transport vehicles, small, scattered and urgent urban orders and the like, a front-end library is established as a middle-end node of a logistics distribution network of a cigarette area in Sichuan province, each industrial enterprise selects a part of cigarettes with high sales volume in Sichuan and transports the cigarettes to the front-end library in a full load mode, and when urgent orders are generated in each city (state), the cigarettes in the front-end library are transported to a corresponding logistics center of commerce by the front-end library, and reference is made to fig. 10.

The method comprises the following steps of collecting geographic positions of 38 tobacco companies logistics centers and part of branch companies in each city (state) in Sichuan province, calling a Baidu map interface through Python, and converting the geographic positions and longitude and latitude, wherein the display part in the text is shown in a table 5:

TABLE 5 longitude and latitude of tobacco logistics center in Sichuan province, city, or State

2 mathematical modeling

The alternative points of the pre-database are determined by a qualitative and quantitative combined method, and converted into a mathematical formula to establish a pre-database site selection mathematical model by aiming at the minimum sum of storage cost, change cost, transportation cost and goods loss cost and the maximum service reliability.

Decision variables and parameter definitions

L-a collection of industrial enterprises, L ═ (1,2, … …, p);

i-set of pre-library alternate points, I ═ 1,2, … …, m)

J-stream center demand point set, J ═ (1,2, … …, n);

h_il-unit price of transportation from factory i to warehouse i;

w_il-traffic from factory i to warehouse i;

q_il-mileage from factory i to warehouse i;

d_j-demand d for each logistics centre demand point j;

C_i-setting a fixed cost for the warehouse candidate node i;

S_i-maximum supply of warehouse candidate nodes i;

c_ij-a shipping mileage from the warehouse candidate node i to the logistics center point j;

r_ij-unit price from warehouse candidate node i to logistics center j;

x_ijthe quantity of materials for warehousing the warehouse alternative node i to the logistics center j;

z_i-representing selection of i candidate nodes for binning intoZ is a radical;

y_i-selecting warehouse alternate point i as warehouse, y _i1 is ═ 1; if not, then y_i＝0。

The transportation cost of each industrial company for delivering goods to the front-end warehouse is the product of the transportation unit price, the transportation amount and the mileage, and the total cost T is expressed as:

similarly, the expression of the total transportation cost Q from the front-end warehouse to each logistics center is as follows:

second, mathematical modeling

Based on the above conditions, a mathematical expression is established:

the restriction condition set (4) is explained:

1) the demand point j demands the satisfaction degree;

2) the number of products provided by the front-end warehouse is not more than the maximum capacity of the facility;

3) the total transportation volume from each industrial enterprise to the preposed warehouse i is less than the maximum capacity of the facility;

4) the delivery amount of the warehouse i to the demand point j is not negative;

5) the number of warehouse alternate points is at least 1.

3 site selection optimal point determination

According to the established preposed base addressing mathematical model, the Python language is used for programming and solving to obtain the optimal solution of the objective function, as shown in FIG. 11, the longitude and latitude of the region where the optimal solution is located are (103.78, 30.84).

The map is used for visual conversion, as shown in fig. 12, and the place corresponding to the longitude and latitude is known as the city of the city river weir.

The invention has the characteristics of preposed warehouse construction: the invention combines industrial enterprises with commercial companies, the cigarettes planned for sale are intensively transported to the preposed warehouse by the industry according to the sale plan made by the industry and the commerce in a warehouse moving mode, and meanwhile, the logistics distribution center can communicate with the industrial enterprises in time according to the early warning condition of the warehouse inventory so as to initiate activities such as warehouse separation, replenishment, warehouse moving and the like;

the preposed warehouse can realize accurate management of cigarettes in different industries while sharing the stock information of the industry and the commerce, complete the integrated work of warehousing and warehousing, and is in seamless butt joint with the national bureau I engineering system, so that the aims of improving the operation efficiency, reducing the operation cost, realizing the intellectualization of management, standardizing the process, visualizing the stock and realizing the service trace are achieved.

3. Intelligent warehouse management system

The intelligent warehousing management system mode of the invention adopts a front-end warehouse and a set of client systems to dock a 1-to-N mode of a plurality of industrial enterprise managers, thus greatly simplifying the docking work with each industrial enterprise without respectively deploying independent client software and hardware; the system is based on the RFID technology, the background database is utilized to associate the electronic tag with the cigarette carton information on the tray to which the electronic tag is attached, so that the electronic tag corresponds to the cigarette carton information on the tray, and the barcode information of the cigarette carton on the whole tray can be acquired by collecting the data of the electronic tag.

Aiming at the cigarettes arriving, the mobile handheld terminal PDA is utilized to scan the bar codes of the cigarettes arriving one project one by one, and then the bar code data is written into the electronic tags of the response trays through the UHF ultrahigh frequency RFID fixed read-write equipment. Scanning the electronic tags of the cigarettes in the whole pallet by using RFID equipment to realize the code scanning and warehousing of the whole pallet; in a warehousing link, stack information is adjusted by using RFID mobile equipment; and in the sorting and ex-warehouse link, reading cigarette bar code data from the electronic tag through RFID equipment, and finishing sorting and ex-warehouse scanning.

The intelligent warehouse management system comprises basic management, operation management and comprehensive analysis, and reference is made to fig. 13.

3.1 basic management Module

The basic management module is a basic layer of the intelligent warehousing management system and is mainly responsible for maintaining basic information such as a goods owner, a warehouse, a storage area, a goods shelf, a goods location and the like.

Owner information maintenance, which is mainly used for maintaining owner types, owner codes and owner names;

the warehouse information maintenance mainly maintains key information such as warehouse names, warehouse contents, types and the like;

maintaining information of the storage area, mainly maintaining information such as the type of the storage area, whether to set up the storage and the like;

maintaining shelf information, namely maintaining the number of rows and columns of goods on a shelf according to actual conditions;

maintaining goods position information, mainly maintaining the storage type and the maximum storage capacity of cigarettes in the goods position;

and strategy information maintenance, which mainly maintains an warehouse entry and exit strategy and a cargo space adjusting strategy and establishes an allocation idea for automatic allocation of cargo spaces.

3.2 Job management Module

The operation management module is divided into warehousing management, warehousing management and ex-warehouse management, and by means of RFID technology, PDA handheld mobile terminals, telescopic cars and other tools, the whole intelligent management process from cigarette warehousing note management, goods allocation, inventory management to ex-warehouse confirmation and other work is achieved, the cigarette operation management operation efficiency is greatly improved, and the high-quality development process of the intelligent management equipment is accelerated.

Storage management

The warehousing management system is in butt joint with an industrial and commercial on-the-way system and an industrial logistics system, and obtains on-the-way information of industrial vehicles in real time so as to make harvesting preparation in advance.

When industrial cigarettes are put in storage, the storage management system is in butt joint with a 'project' system of the national Bureau, industrial cigarette storage documents are obtained in real time, and storage operation tasks are automatically generated according to the storage documents. The warehousing management system performs automatic goods allocation according to a preset warehousing strategy, can perform manual adjustment, transmits goods allocation data to the handheld PDA after allocation is completed, the handheld PDA performs cigarette shelving according to goods allocation results, clicks a warehousing completion button after shelving completion, returns shelving completion information to the warehousing system, increases corresponding goods main inventory in the warehouse, and completes warehousing operation.

Management in warehouse

When inventory checking is carried out, an inventory checking process can be initiated through the warehousing system, the platform computer can check inventory list information, inventory personnel can carry out mobile inventory checking through the handheld PDA, actual data are returned to the warehousing system after inventory checking is finished, and the warehousing system generates a difference list.

The warehousing system can generate profit and loss documents according to the checking difference sheets to initiate profit and loss processes, and can also directly initiate profit and loss processes to complete profit and loss services.

And (3) carrying out goods location adjustment business through a goods location adjustment function according to the storage requirement of the cigarettes in the warehouse, sending an adjustment document to the handheld PDA after the adjustment is finished, clicking a 'adjustment finishing' button of the handheld PDA after the adjustment is finished, feeding back the adjustment finishing information to the warehousing system, and finishing the adjustment single flow.

Management of ex-warehouse

When industrial cigarettes are delivered from the warehouse, the warehouse management system is in real-time butt joint with the national bureau I engineering system, industrial cigarette delivery documents (including delivery and sale delivery) are obtained in real time, and delivery operation tasks are automatically generated according to the delivery documents.

The warehousing management system performs automatic goods allocation according to a preset ex-warehouse strategy, can perform manual adjustment, transmits goods allocation data to the handheld PDA after allocation is completed, the handheld PDA performs cigarette shelving according to goods allocation results, clicks an ex-warehouse completion button after shelving is completed, returns the off-shelf completion information to the warehousing system, reduces corresponding goods main inventory in the warehouse, and completes ex-warehouse operation.

3.3 Integrated analysis

Based on daily operation and warehousing operation requirements, the intelligent warehouse management system of the front-end warehouse needs to integrate relevant data for data storage, so that the industrial and commercial enterprises can master the goods flow condition in real time, query the data condition, perform specific analysis on specific data, further realize the datamation and visualization of inventory management, and strengthen the dynamic interconnection among the industrial and commercial enterprises in the tobacco industry.

The accuracy of delivery data acquisition, inventory data query and handheld PDA inventory display are improved, and the inventory information of cigarettes in the warehouse by each owner can be queried; the storage condition of the cigarettes in the warehouse is displayed in real time through the handheld PDA;

the standardization of in-warehouse cigarette management and the query of in-warehouse and out-warehouse data are ensured, the information of each warehouse owner in the warehouse can be queried within a certain period of time, the warehousing order is transmitted to a warehousing system, and relevant information is automatically and dynamically fed back to the system;

the method realizes the operation trace management, the library age screening and the library age data query of the cigarettes flowing from the industry to the commercial enterprises in the whole process, can query the age information of the cigarettes stored in the library, performs the library age screening, analyzes the turnover rate of the cigarettes, and prepares for the selection, purchase and transportation of the next-stage cigarette specification.

The above-described embodiments are intended to illustrate rather than to limit the invention, and any modifications and variations of the present invention are within the spirit of the invention and the scope of the claims.

Claims (10)

1. A cigarette logistics distribution center information system based on a combined prediction method comprises

The cigarette demand combination prediction model comprises the following steps: constructing a cigarette combination demand model of a time sequence according to the influence factors, and predicting the logistics demand of the cigarette area;

the cigarette industry and commerce finished product preposed warehouse based on the logistics alliance: establishing a logistics union of industrial and commercial enterprises, establishing a site selection mathematical model to determine the optimal site selection longitude and latitude according to the lowest cost and the optimal service principle, establishing a cigarette finished product prepositive library of the industrial enterprise in the area of the optimal site selection to store partial cigarettes with high market demands, and providing the industrial warehousing and delivery service for commercial enterprises;

intelligent warehouse management system: the electronic tags are associated with cigarette information based on the RFID technology, the cigarettes in different industries in the warehouse are precisely and intelligently managed, the integration of warehouse entry and exit is realized, and the electronic tags are in seamless joint with an engineering system I; the first engineering system is specifically a national bureau first engineering system.

2. The cigarette logistics distribution center information system based on combination prediction method according to claim 1, wherein the influence factors comprise historical ordering data, seasonal variation, holidays, prices.

3. The cigarette logistics distribution center information system based on the combined prediction method according to claim 2, wherein the construction process of the cigarette demand combined prediction model comprises the following steps,

s1, analyzing the cigarette market capacity and demand: comprehensively analyzing the whole market capacity of the cigarettes and other influencing factors around two levels of a macroscopic environment and a cigarette market; the whole cigarette market capacity comprises the market capacity of a standing population, the gift sending capacity of the standing population and the purchasing capacity of a tourist population, and the other influence factors comprise seasons, festivals and holidays;

s2, establishing a seasonal ARIMA model: establishing an autoregressive moving average model aiming at the cigarette historical ordering data, then introducing the quarterly and holidays in the influence factors, and establishing a seasonal ARIMA model;

s3, establishing a cigarette demand combination prediction model based on machine learning: on the basis of cigarette historical ordering data, seasonal characteristic values are constructed, the characteristics of a machine learning algorithm are fused, a semi-supervised machine learning cigarette demand prediction model is constructed, and the monthly cigarette ordering amount is predicted by using a rolling type single-step cyclic prediction method.

4. The cigarette logistics distribution center information system based on combined prediction method according to claim 3, wherein the overall market capacity comprises

Standing population market capacity: the smoking angle is the regular population, the smoking rate, the daily average smoking amount and the number of days of residence in province;

capacity of daily life: the gift consumption is the number of smokers with suitable age in the population of the ordinary living, the gift sending rate and the gift sending amount in the whole year;

consumption capacity of an aged smoker: the consumption of smokers with suitable age is divided into the number of the population with suitable age, the ratio of the smokers and the average smoking amount of the smokers;

market capacity of tourism population: the purchase angle is the number of tourist population, the purchase rate and the purchase amount.

5. The cigarette logistics distribution center information system based on the combined prediction method according to claim 4, wherein the population of the regular living is set to live for more than 6 months, the smoker of the suitable age is set to age 18-64 years, and the population of the tourist is set to live for less than 1 month.

6. The cigarette logistics distribution center information system based on the combined prediction method according to claim 3, wherein the construction process of the seasonal ARIMA model comprises the following steps,

s1, sequence difference: selecting monthly cigarette ordering data of a certain time period to perform model fitting, establishing a time-ordering quantity sequence diagram, and performing stability inspection and seasonal difference on the data respectively;

s2, smoothness checking: performing stationarity test on the original data, and when the p value of the test result is less than 0.005, proving that the original data is stable and purely random; when the stability test is not passed, carrying out difference calculation, and when the p value of the original data after 1-order difference is less than 0.005, passing the stability test;

s3, seasonal decomposition: carrying out periodic decomposition on the original data, wherein the periodicity is 12, extracting seasonal indexes to form a periodic time sequence, carrying out 2-order difference on the periodic time sequence to obtain a 1-order difference and 2-order seasonal difference sequence diagram, and determining the orders of d to be 1 and 2 respectively;

s4, p, q value determination: selecting an autoregressive order p and a moving average term number q of a seasonal ARIMA model, and determining through a 1-order difference, a 2-order seasonal autocorrelation coefficient ACF graph and a partial autocorrelation coefficient PACF graph; the p and q values are determined as (1,1) through analysis, and the p, d and q values are determined, so that the cigarette demand prediction model is ARIMA (1,1,1) (1,2,1), namely the seasonal ARIMA model.

7. The cigarette logistics distribution center information system based on the combination prediction method according to claim 3, wherein the construction process of the cigarette demand combination prediction model based on the machine learning comprises the following steps,

s1, time matrix conversion: training and fitting the model, and assuming that the time sequence is at different time points t₁,t₂,…,t_nThe observed value of (a) can be represented by A (t)₁),A(t₂),…,A(t_n) Is shown if t_n+1The observed value of the time can be expressed as

A(t_n+1)＝f(A(t_n),A(t_n-1),…,A(t_n-k+1))

Where k < n, then t is indicated_n+1The observation value of the moment is a function expression of the observation values of the k previous time points, and is predicted in a rolling mode according to a time sequence;

s2, cutting the data set: performing machine learning to perform test set and training set division on the data, wherein the training set is used for training the model, and the test set judges the final training effect according to the training result of the training set;

s2, predicting the result: after 14 iterations, the prediction model test results are obtained, the results in the graph are converted into a table, and the error is calculated.

8. The cigarette logistics distributed allocation center information system based on the combined prediction method according to claim 1, wherein the construction process of the cigarette industry and commerce finished product pre-library based on the logistics alliance comprises the following steps,

s1, establishing a preposed database addressing evaluation model: determining alternative points of the pre-database by a qualitative and quantitative combined method, converting the alternative points into a mathematical formula by aiming at the minimum value of the sum of storage cost, change cost, transportation cost and goods loss cost and the maximum service reliability, and establishing a pre-database site selection mathematical model;

s2, determining the optimal point of site selection: according to the established preposed base addressing mathematical model, programming solution is carried out by utilizing Python language to obtain an optimal solution of an objective function, the longitude and latitude of the area where the optimal solution is located are obtained, and the map is utilized to carry out visual conversion to obtain the location corresponding to the longitude and latitude.

9. The cigarette logistics distributed allocation center information system based on the combined prediction method according to claim 8, wherein the construction process of the pre-base site selection evaluation model comprises the following steps,

s1, decision variables and parameter definition:

l — the collection of industrial enterprises, L ═ (1,2, … …, p);

i — set of pre-library candidate points, I ═ 1,2, … …, m;

j — the set of stream center demand points, J ═ (1,2, … …, n);

h_il-unit price of transportation from factory i to warehouse i;

w_il-traffic from factory i to warehouse i;

q_il-mileage from factory i to warehouse i;

d_j-the demand d of each logistics centre demand point j;

C_i-setting a fixed cost for the warehouse candidate node i;

S_i-maximum supply of warehouse candidate nodes i;

c_ijthe transportation mileage of the warehouse from the warehouse alternative node i to the logistics center point j;

r_ij-unit price from warehouse candidate node i to logistics center j;

x_ijthe quantity of materials for warehousing the warehouse alternative nodes i to the logistics center j;

z_i-representing the cost z of selecting i alternative nodes to bin;

y_i-selecting warehouse alternate point i as warehouse, y_i1 is ═ 1; if not, then y_i＝0；

The transportation cost of each industrial company for delivering goods to the front-end warehouse is the product of the transportation unit price, the transportation amount and the mileage, and the total cost T is expressed as:

similarly, the expression of the total transportation cost Q from the front-end warehouse to each logistics center is as follows:

s2, mathematical modeling: based on the above conditions, a mathematical expression is established

The restriction condition group (4) is explained

1) The demand point j demands the satisfaction degree;

2) the number of products provided by the front-end warehouse is not more than the maximum capacity of the facility;

3) the total transportation volume from each industrial enterprise to the preposed warehouse i is less than the maximum capacity of the facility;

4) the delivery amount of the warehouse i to the demand point j is not negative;

5) the number of warehouse alternate points is at least 1.

10. The cigarette logistics distribution center information system based on combination prediction method according to claim 1, wherein the intelligent warehouse management system comprises

A basic management module: as a basic layer, the system is mainly responsible for realizing basic information maintenance, wherein the basic information maintenance comprises information maintenance of a goods main dimension, a warehouse, a storage area, a goods shelf, a goods location and a strategy;

an operation management module: the method comprises the steps of warehousing management, warehousing management and ex-warehouse management, wherein the warehousing management is used for warehousing entry management, warehousing goods position distribution and warehousing confirmation, the ex-warehouse management is used for inventory management, damage and overflow management and goods position adjustment, and the ex-warehouse management is used for ex-warehouse entry management, ex-warehouse goods position distribution and ex-warehouse confirmation;

a comprehensive analysis module: the method comprises inventory analysis, warehouse-in and warehouse-out analysis and warehouse age analysis, wherein the inventory analysis is used for inventory data query and handheld PDA inventory display, the warehouse-in and warehouse-out analysis is used for warehouse-in data query and warehouse-out data query, and the warehouse age analysis is used for warehouse age screening and warehouse age data query.

Images