CN108564326B - Order prediction method and device, computer readable medium and logistics system - Google Patents
Order prediction method and device, computer readable medium and logistics system Download PDFInfo
- Publication number
- CN108564326B CN108564326B CN201810357890.XA CN201810357890A CN108564326B CN 108564326 B CN108564326 B CN 108564326B CN 201810357890 A CN201810357890 A CN 201810357890A CN 108564326 B CN108564326 B CN 108564326B
- Authority
- CN
- China
- Prior art keywords
- sequence
- neural network
- order
- network model
- dimensional signal
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/08—Logistics, e.g. warehousing, loading or distribution; Inventory or stock management
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Economics (AREA)
- General Physics & Mathematics (AREA)
- Human Resources & Organizations (AREA)
- Strategic Management (AREA)
- Entrepreneurship & Innovation (AREA)
- Development Economics (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- Marketing (AREA)
- General Business, Economics & Management (AREA)
- Operations Research (AREA)
- Biomedical Technology (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Game Theory and Decision Science (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
A prediction method and device, computer readable medium and logistics system of orders are provided, the prediction method of orders comprises: acquiring a time sequence corresponding to a historical order; extracting a characteristic sequence based on the time sequence; generating a two-dimensional signal feature map based on the feature sequence; and constructing a neural network model based on the two-dimensional signal characteristic diagram, and performing order prediction according to the constructed neural network model. By applying the scheme, the accuracy of order prediction can be improved.
Description
Technical Field
The embodiment of the invention relates to the field of logistics, in particular to a method and a device for predicting orders, a computer readable medium and a logistics system.
Background
The order prediction of the whole vehicle logistics can enable scheduling personnel to prepare in advance and be not rainy and mushy, so that the scheduling of transportation resources is more reasonable, and the accurate order prediction of the whole vehicle logistics plays an important role in the whole vehicle logistics.
The time series may describe discrete values of variables over time, such as variables of temperature, humidity, stock price, car sales trends, and the like. The time-series prediction means that a value of one unknown time point (t +1) or one unknown time period (t +1, t + n) is determined based on values of the time series at the current time point t and past time points (t-1), (t-2), … …, and (t-n-1). In general, a time series predictor may depend on one or more other time series in addition to past time. Thus, a time series prediction problem refers to a regression problem in a high dimensional space. In a high dimensional space, a predictor is a highly non-linear function of its past values and other related time series. The order prediction problem of the whole vehicle logistics can be essentially regarded as a time series prediction problem. The whole vehicle logistics order prediction problem can be regarded as a time series prediction problem in nature, so that the problem can be solved through a time series prediction model, such as a traditional Auto Regression Moving Average (ARMA) model, a feedforward Neural Network model, a Recurrent Neural Network (RNN), a Long-Short Memory-variable Neural Network (LSTM), and the like. When the models solve the whole vehicle logistics order prediction problem, logistics order data are combined into one-dimensional data to serve as input, and then time correlation on the data is extracted.
The existing forecasting method of the whole vehicle logistics order takes a synthesized one-dimensional sequence as model input for learning and forecasting by a simple method of combining a plurality of time sequences or a plurality of characteristic sequences as input.
The method for predicting the whole vehicle logistics order is poor in accuracy.
Disclosure of Invention
The technical problem solved by the embodiment of the invention is how to improve the accuracy of order prediction.
In order to solve the above technical problem, an embodiment of the present invention provides an order prediction method, where the method includes: acquiring a time sequence corresponding to a historical order; extracting a characteristic sequence based on the time sequence; generating a two-dimensional signal feature map based on the feature sequence; and constructing a neural network model based on the two-dimensional signal characteristic diagram, and performing order prediction according to the constructed neural network model.
Optionally, the obtaining a time sequence corresponding to the order includes: acquiring original data corresponding to a historical order; and preprocessing the original data to obtain a time sequence corresponding to the order.
Optionally, the pre-treatment comprises at least one of: abnormal value processing and missing value processing.
Optionally, the extracting the feature sequence includes: and extracting a characteristic sequence based on a wavelet transform algorithm.
Optionally, the generating the two-dimensional signal feature map includes: dividing each characteristic sequence into a plurality of sequence segments with the length of n, wherein n is a positive integer; copying sequence segments corresponding to different characteristic sequences line by line to generate m-line sequence segments, wherein the m-line sequence segments meet the condition that different characteristic sequence lines are adjacent in pairs, and m is a positive integer; and generating a two-dimensional signal feature map of m x n based on the m-row sequence segments.
Optionally, the segmenting each feature sequence into a plurality of sequence segments with a length of n includes: based on the shift operation, each feature sequence is segmented into a plurality of sequence segments with the length of n.
Optionally, the building the neural network model includes: constructing a neural network model; and training the neural network model based on the two-dimensional signal characteristic diagram to obtain parameters of the neural network model.
Optionally, after the neural network model is built, the order prediction method further includes: acquiring online data of an order; training and updating the neural network model based on the online data.
Optionally, the neural network is: a convolutional neural network.
An embodiment of the present invention provides an order prediction apparatus, including: the first acquisition unit is suitable for acquiring a time sequence corresponding to a historical order; an extraction unit adapted to extract a feature sequence based on the time sequence; a generating unit adapted to generate a two-dimensional signal feature map based on the feature sequence; and the construction unit is suitable for constructing a neural network model based on the two-dimensional signal characteristic diagram and carrying out order prediction according to the constructed neural network model.
Optionally, the first obtaining unit includes: the first acquisition subunit is suitable for acquiring original data corresponding to the historical order; and the second acquisition subunit is suitable for preprocessing the original data to acquire the time sequence corresponding to the order.
Optionally, the pre-treatment comprises at least one of: abnormal value processing and missing value processing.
Optionally, the extracting unit is adapted to extract the feature sequence based on a wavelet transform algorithm.
Optionally, the generating unit includes: a segmentation subunit adapted to segment each feature sequence into a plurality of sequence segments of length n, where n is a positive integer; the replicon unit is suitable for replicating the sequence segments corresponding to different characteristic sequences line by line to generate m-line sequence segments, wherein the m-line sequence segments meet the condition that every two of different characteristic sequence lines are adjacent, and m is a positive integer; a generating subunit adapted to generate a two-dimensional signal profile of m x n based on the m rows of sequence segments.
Optionally, the dividing subunit is adapted to divide each feature sequence into a plurality of sequence segments with a length of n based on a shift operation.
Optionally, the construction unit comprises: a construction subunit adapted to construct a neural network model based on the neural network model parameters; the training subunit is suitable for training the neural network model based on the two-dimensional signal characteristic diagram to obtain parameters of the neural network model; and the prediction subunit is suitable for performing order prediction according to the trained neural network model.
Optionally, the order prediction apparatus further includes: the second acquisition unit is suitable for acquiring online data of the order; and the updating unit is suitable for training and updating the neural network model based on the online data.
Optionally, the neural network is: a convolutional neural network.
Embodiments of the present invention provide a computer-readable storage medium, on which computer instructions are stored, and when the computer instructions are executed, the method steps are executed.
The embodiment of the present invention provides a logistics system, which includes a memory and a processor, where the memory stores computer instructions executable on the processor, and the processor executes the computer instructions to perform the steps of any one of the above methods.
Compared with the prior art, the technical scheme of the embodiment of the invention has the following beneficial effects:
according to the embodiment of the invention, a two-dimensional signal characteristic diagram is generated based on the characteristic sequence, then a neural network model is constructed based on the two-dimensional signal characteristic diagram, and order prediction is carried out according to the constructed neural network model. Because the one-dimensional correlation sequences are combined into the two-dimensional characteristic diagram, and the neural network is utilized to learn the relevance between the time adjacent points of each sequence and the adjacent sequence points at the same time, the inclusion characteristics of the signal characteristic diagram are learned to the maximum extent, and the accuracy of order prediction can be effectively improved.
Furthermore, abnormal values can be eliminated by processing the abnormal values of the original data corresponding to the historical orders, so that the influence of the abnormal values on the accuracy of subsequent prediction is avoided.
Furthermore, the missing value processing is carried out on the original data corresponding to the historical order, so that the accuracy of order prediction is improved.
Drawings
FIG. 1 is a flow chart of a method for predicting an order according to an embodiment of the present invention;
FIG. 2 is a flow chart of another method for forecasting orders provided by embodiments of the present invention;
FIG. 3 is a flow chart of a method for forecasting an order according to another embodiment of the present invention;
fig. 4 is a schematic structural diagram of an order prediction apparatus according to an embodiment of the present invention.
Detailed Description
The existing method can not well learn the relevance among a plurality of sequences or a plurality of characteristics, so that the characteristics contained in the input data can not be well learned by a machine learning model. Therefore, the accuracy of order prediction is poor.
According to the embodiment of the invention, a two-dimensional signal characteristic diagram is generated based on the characteristic sequence, then a neural network model is constructed based on the two-dimensional signal characteristic diagram, and order prediction is carried out according to the constructed neural network model. Because the one-dimensional correlation sequences are combined into the two-dimensional characteristic diagram, and the neural network is utilized to learn the relevance between the time adjacent points of each sequence and the adjacent sequence points at the same time, the inclusion characteristics of the signal characteristic diagram are learned to the maximum extent, and the accuracy of order prediction can be effectively improved.
In order to make the aforementioned objects, features and advantages of the present invention comprehensible, embodiments accompanied with figures are described in detail below.
Referring to fig. 1, an embodiment of the present invention provides an order prediction method, which may include the following steps:
step S101, acquiring a time sequence corresponding to the historical order.
In a specific implementation, in order to predict an order, a time series corresponding to a historical order, that is, related data of the historical order, needs to be obtained first, and then future data of the order, that is, a regression problem of a high-dimensional space, needs to be predicted based on the related data of the historical order. In a high dimensional space, a predictor is a highly non-linear function of its past values and other related time series.
In specific implementation, a desired time sequence may not be directly obtained, and only raw data such as whole vehicle logistics order time sequence data and stock sequence data corresponding to a historical order may be obtained. Therefore, the original data corresponding to the historical order can be obtained, and then the original data is preprocessed through a certain algorithm and logic, for example, data cleaning is carried out, and the time sequence corresponding to the order is obtained.
In an embodiment of the present invention, the preprocessing includes: and processing abnormal values. For example, the original data corresponding to the historical order is checked, and the abnormal value is removed, so that the influence of the abnormal value on the accuracy of the subsequent prediction is avoided.
In another embodiment of the present invention, the pre-processing comprises: and (5) processing missing values. For example, the original data corresponding to the historical order is checked and the corresponding missing value is filled up, so that the accuracy of order prediction is improved.
In yet another embodiment of the present invention, missing value processing and outlier processing exist simultaneously.
It is understood that other pre-processing may be performed, and will not be described in detail herein.
And S102, extracting a characteristic sequence based on the time sequence.
In a specific implementation, the feature sequence corresponding to the time sequence may be extracted based on a wavelet transform algorithm.
In a specific implementation, for the preprocessed time series, all the features may be extracted first, for example, M different features are extracted, and then correlation analysis and judgment are performed to select N most relevant feature sequences from the M feature sequences, where N, M are positive integers, and N is less than or equal to M.
And step S103, generating a two-dimensional signal feature map based on the feature sequence.
In a specific implementation, each signature sequence may be segmented into a plurality of sequence segments of length n, where n is a positive integer; then, the sequence segments corresponding to different characteristic sequences are copied row by row to generate m rows of sequence segments, wherein the m rows of sequence segments meet the condition that every two of different characteristic sequence rows are adjacent, m is a positive integer, namely, the sequence segments corresponding to the same index of each characteristic sequence are copied to generate m rows of sequence segments, any two of the m rows of sequence segments are positioned in two adjacent rows of sequence segments corresponding to the same index of each characteristic sequence, m is a positive integer, and each row is the sequence segment corresponding to the same index of the characteristic sequence; and finally, generating a two-dimensional signal feature map of m x n based on the m-row sequence segments.
In an embodiment of the present invention, each signature sequence is divided into a plurality of sequence segments with length n based on a shift operation.
In a specific implementation, the shift operation may be to shift one bit at a time, or may be to shift more than two bits at a time, and may be flexibly selected according to an actual situation, which is not limited in the embodiment of the present invention.
For example, 5 feature sequences, a, b, c, d, and e respectively, are extracted based on step S102, and each feature sequence has a length of L, that is, each feature sequence contains L bits of information. Firstly, by shift slice segmentation, 5 characteristic sequences with length L are respectively segmented into L-n sequence segments with length n, wherein the sequence segments respectively comprise: the sequence segment 1 corresponds to the 1 st to the nth bit of the characteristic sequence, the sequence segment 2 corresponds to the 2 nd to the n +1 th bit of the characteristic sequence, … …, and the sequence segment L-n corresponds to the L-n th to the L-th bit of the characteristic sequence; then m row sequence segments meeting pairwise adjacency between different characteristic sequence rows are generated, and for 5 characteristic sequences a, b, c, d and e, the m row sequence segments meeting pairwise adjacency between the rows are generated as follows: abcdeacebda, m is 11, and the adjacent between any two different characteristic sequence lines is satisfied. Finally, based on the m-row sequence segments, a two-dimensional signal feature map of m × n can be generated.
And step S104, constructing a neural network model based on the two-dimensional signal characteristic diagram, and performing order prediction according to the constructed neural network model.
In a specific implementation, the neural network model parameters may be calculated based on the two-dimensional signal feature map; and then constructing a neural network model based on the neural network model parameters.
In a specific implementation, the Neural Network model can be constructed by using a Convolutional Neural Network (CNN), which is a feed-forward Neural Network whose artificial neurons can respond to a part of surrounding units in a coverage range and has excellent performance for large-scale image processing. It includes a convolutional layer (alternating volumetric layer) and a pooling layer (pooling layer). The method has good characteristics of displacement invariance, weight sharing and the like, and can well extract abstract features.
In a specific implementation, the characteristic diagram can be divided into a training part, a verifying part and a testing part, and the training part is input into the neural network model to complete the process of training and adjusting parameters. The trained neural network model can be used as an order prediction model.
In the implementation, as time goes by, new logistics order data can be continuously input into the neural network system, so that the neural network model can be trained and updated based on real-time online data, and the convolutional neural network model which is trained and completed offline before is updated through relearning of the online data, so that the logistics order data can be advanced with time.
By applying the scheme, the one-dimensional correlation sequences are combined into the two-dimensional characteristic diagram, and the neural network is utilized to learn the relevance between the time adjacent points of each sequence and the adjacent sequence points at the same time, so that the inclusion characteristics of the signal characteristic diagram are learned to the greatest extent, and the accuracy of order prediction can be effectively improved.
In order to make the present invention more understandable and practical for those skilled in the art, the embodiment of the present invention shows the comparison result of the accuracy of the order prediction between the solution provided by the embodiment of the present invention and the existing solution, as shown in the graph.
TABLE 1
Referring to table 1, the accuracy of the order prediction of the scheme provided by the embodiment of the present invention and the existing scheme is verified based on the entire vehicle logistics order data in 2013-2016. 2013-2015 three-year data is used as training data to generate a neural network model, then 2016 order data is predicted and compared with 2016 real order data to give an error rate, and a specific calculation formula is as follows: error rate is (actual value-predicted value)/actual value.
As can be seen from table 1, by using the scheme provided by the embodiment of the present invention, the error rate of the order prediction can be effectively reduced, so as to provide higher accuracy of the order prediction.
To enable those skilled in the art to better understand and implement the present invention, another order forecasting method is provided in the embodiment of the present invention, as shown in fig. 2.
Referring to fig. 2, the order prediction method may include the steps of:
step S201, obtain offline order data.
Step S202, preprocessing the order data.
And step S203, extracting a characteristic sequence based on the order data.
And step S204, constructing a two-dimensional signal characteristic diagram, and dividing the two-dimensional signal characteristic diagram into a training set, a verification set and a test set.
And step S205, constructing a convolutional neural network.
In a specific implementation, the convolutional neural network is a convolutional neural network model for order prediction.
And S206, training the convolutional neural network based on the two-dimensional signal characteristic diagram.
In a specific implementation, the convolutional neural network may be trained based on the two-dimensional signal feature map to modify parameters of the convolutional neural network.
Step S207, determining whether the convolutional neural network meets a stop condition, executing step S208 when the convolutional neural network meets the stop condition, otherwise executing step S206.
In a specific implementation, the stop condition may be: based on the validation set, the computed error of the convolutional neural network does not decrease anymore, i.e. the convolutional neural network converges.
Step S208, determining whether the error of the convolutional neural network meets a preset threshold, and if the error of the convolutional neural network meets the preset threshold, performing step S209, otherwise performing step S210.
And step S209, storing the convolutional neural network model for order prediction.
And step S210, adjusting the structure and parameters of the convolutional neural network, and repeatedly executing the step S205.
To enable those skilled in the art to better understand and implement the present invention, another order forecasting method is provided in the embodiment of the present invention, as shown in fig. 3.
Step S301, obtaining online order data.
Step S302, preprocessing order data.
Step S303, extracting a characteristic sequence based on the order data.
And step S304, constructing a two-dimensional signal characteristic diagram.
And S305, updating the convolutional neural network model based on online learning.
And step S306, predicting the order based on the convolutional neural network model.
In step S307, the prediction result is output.
In order to make those skilled in the art better understand and implement the present invention, the embodiment of the present invention further provides a forecasting apparatus capable of implementing the above forecasting method for orders, as shown in fig. 4.
Referring to fig. 4, the order prediction apparatus includes: a first acquisition unit 41, an extraction unit 42, a generation unit 43, and a construction unit 44, wherein:
the first obtaining unit 41 is adapted to obtain a time sequence corresponding to a historical order.
The extraction unit 42 is adapted to extract a feature sequence based on the time sequence.
The generating unit 43 is adapted to generate a two-dimensional signal feature map based on the feature sequence.
The constructing unit 44 is adapted to construct a neural network model based on the two-dimensional signal characteristic diagram, and perform order prediction according to the constructed neural network model.
In a specific implementation, the first obtaining unit 41 may include: a first acquisition subunit 411 and a second acquisition subunit 412, wherein:
the first obtaining subunit 411 is adapted to obtain original data corresponding to the historical order.
The second obtaining subunit 412 is adapted to pre-process the original data to obtain a time sequence corresponding to the order.
In an embodiment of the invention, the pre-treatment comprises at least one of: abnormal value processing and missing value processing.
In an embodiment of the present invention, the extracting unit 42 is adapted to extract the feature sequence based on a wavelet transform algorithm.
In a specific implementation, the generating unit 43 includes: a splitting subunit 431, a replication subunit 432, and a generating subunit 433, wherein:
the segmentation subunit 431 is adapted to segment each feature sequence into a plurality of sequence segments with a length of n, where n is a positive integer.
The replicon unit 432 is adapted to replicate the sequence segments corresponding to different characteristic sequences row by row to generate m rows of sequence segments, where the m rows of sequence segments satisfy that two of the different characteristic sequence rows are adjacent to each other.
The generating subunit 433 is adapted to generate a two-dimensional signal feature map of m × n based on the m-line sequence segments.
In an embodiment of the present invention, the dividing subunit 431 is adapted to divide each feature sequence into a plurality of sequence segments with a length of n based on a shift operation.
In a specific implementation, the building unit 44 includes: a construction subunit 441, a training subunit 442, and a prediction subunit 443, wherein:
the constructing subunit 441 is adapted to construct a neural network model based on the neural network model parameters.
The training subunit 442 is adapted to train the neural network model based on the two-dimensional signal feature map, and obtain parameters of the neural network model.
The forecasting subunit 443 is adapted to perform order forecasting according to the trained neural network model.
In a specific implementation, the order prediction apparatus 40 may further include: a second acquisition unit (not shown) and an update unit (not shown), wherein:
the second obtaining unit is suitable for obtaining the online data of the order.
The updating unit is suitable for training and updating the neural network model based on the online data.
In an embodiment of the present invention, the neural network is: a convolutional neural network.
In a specific implementation, the workflow and the principle of the order prediction apparatus 40 may refer to the description of the method provided in the above embodiment, and are not described herein again.
An embodiment of the present invention provides a computer-readable storage medium, which is a non-volatile storage medium or a non-transitory storage medium, and on which a computer instruction is stored, where the computer instruction executes, when running, any of the steps corresponding to the foregoing methods, and details are not described here again.
The embodiment of the present invention provides a logistics system, which includes a memory and a processor, where the memory stores a computer instruction capable of running on the processor, and the processor executes, when running the computer instruction, any of the steps corresponding to the above methods, which is not described herein again.
Those skilled in the art will appreciate that all or part of the steps in the methods of the above embodiments may be implemented by associated hardware instructed by a program, which may be stored in a computer-readable storage medium, and the storage medium may include: ROM, RAM, magnetic or optical disks, and the like.
Although the present invention is disclosed above, the present invention is not limited thereto. Various changes and modifications may be effected therein by one skilled in the art without departing from the spirit and scope of the invention as defined in the appended claims.
Claims (18)
1. A method for predicting an order, comprising:
acquiring a time sequence corresponding to a historical order;
extracting a characteristic sequence based on the time sequence;
generating a two-dimensional signal feature map based on the feature sequence;
constructing a neural network model based on the two-dimensional signal characteristic diagram, and predicting orders according to the constructed neural network model;
wherein the generating a two-dimensional signal feature map comprises:
dividing each characteristic sequence into a plurality of sequence segments with the length of n, wherein n is a positive integer;
copying sequence segments corresponding to different characteristic sequences line by line to generate m-line sequence segments, wherein the m-line sequence segments meet the condition that different characteristic sequence lines are adjacent in pairs, and m is a positive integer;
and generating a two-dimensional signal feature map of m x n based on the m-row sequence segments.
2. The order prediction method of claim 1, wherein the obtaining of the time sequence corresponding to the historical order comprises:
acquiring original data corresponding to a historical order;
and preprocessing the original data to obtain a time sequence corresponding to the order.
3. The method of predicting an order according to claim 2, wherein said preprocessing comprises at least one of: abnormal value processing and missing value processing.
4. The method of predicting an order according to claim 1, wherein said extracting a sequence of features comprises: and extracting a characteristic sequence based on a wavelet transform algorithm.
5. The method of predicting an order according to claim 1, wherein said dividing each signature sequence into a plurality of sequence segments of length n comprises:
based on the shift operation, each feature sequence is segmented into a plurality of sequence segments with the length of n.
6. The method of predicting an order according to claim 1, wherein said constructing a neural network model comprises:
constructing a neural network model;
and training the neural network model based on the two-dimensional signal characteristic diagram to obtain parameters of the neural network model.
7. The method of predicting an order according to claim 6, further comprising, after constructing the neural network model:
acquiring online data of an order;
training and updating the neural network model based on the online data.
8. The method of predicting an order according to claim 6 or 7, wherein said neural network is: a convolutional neural network.
9. An order prediction apparatus, comprising:
the first acquisition unit is suitable for acquiring a time sequence corresponding to a historical order;
an extraction unit adapted to extract a feature sequence based on the time sequence;
a generating unit adapted to generate a two-dimensional signal feature map based on the feature sequence;
the building unit is suitable for building a neural network model based on the two-dimensional signal characteristic diagram and predicting orders according to the built neural network model;
wherein the generating unit includes:
a segmentation subunit adapted to segment each feature sequence into a plurality of sequence segments of length n, where n is a positive integer;
the replicon unit is suitable for replicating the sequence segments corresponding to different characteristic sequences line by line to generate m-line sequence segments, wherein the m-line sequence segments meet the condition that every two of different characteristic sequence lines are adjacent, and m is a positive integer;
a generating subunit adapted to generate a two-dimensional signal profile of m x n based on the m rows of sequence segments.
10. The order prediction device according to claim 9, wherein the first acquisition unit includes:
the first acquisition subunit is suitable for acquiring original data corresponding to the historical order;
and the second acquisition subunit is suitable for preprocessing the original data to acquire the time sequence corresponding to the order.
11. The apparatus for predicting an order according to claim 10, wherein said preprocessing comprises at least one of: abnormal value processing and missing value processing.
12. The apparatus according to claim 9, wherein the extraction unit is adapted to extract the feature sequence based on a wavelet transform algorithm.
13. The order prediction device according to claim 9, wherein the segmentation subunit is adapted to segment each feature sequence into a plurality of sequence segments of length n based on a shift operation.
14. The order prediction device of claim 9, wherein the building unit comprises: a construction subunit adapted to construct a neural network model based on the neural network model parameters;
the training subunit is suitable for training the neural network model based on the two-dimensional signal characteristic diagram to obtain parameters of the neural network model;
and the prediction subunit is suitable for performing order prediction according to the trained neural network model.
15. The order forecasting device of claim 14, further comprising:
the second acquisition unit is suitable for acquiring online data of the order;
and the updating unit is suitable for training and updating the neural network model based on the online data.
16. The apparatus for predicting an order according to claim 14 or 15, wherein said neural network is: a convolutional neural network.
17. A computer readable storage medium having computer instructions stored thereon, wherein the computer instructions when executed perform the steps of the method of any one of claims 1 to 8.
18. A logistics system comprising a memory and a processor, the memory having stored thereon computer instructions executable on the processor, wherein the processor, when executing the computer instructions, performs the steps of the method of any one of claims 1 to 8.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810357890.XA CN108564326B (en) | 2018-04-19 | 2018-04-19 | Order prediction method and device, computer readable medium and logistics system |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810357890.XA CN108564326B (en) | 2018-04-19 | 2018-04-19 | Order prediction method and device, computer readable medium and logistics system |
Publications (2)
Publication Number | Publication Date |
---|---|
CN108564326A CN108564326A (en) | 2018-09-21 |
CN108564326B true CN108564326B (en) | 2021-12-21 |
Family
ID=63535754
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810357890.XA Active CN108564326B (en) | 2018-04-19 | 2018-04-19 | Order prediction method and device, computer readable medium and logistics system |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108564326B (en) |
Families Citing this family (14)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109543924A (en) * | 2018-12-20 | 2019-03-29 | 上海德启信息科技有限公司 | Goods amount prediction technique, device and computer equipment |
CN109685276B (en) * | 2018-12-27 | 2021-01-01 | 拉扎斯网络科技(上海)有限公司 | Order processing method and device, electronic equipment and computer readable storage medium |
CN110084438A (en) * | 2019-05-09 | 2019-08-02 | 上汽安吉物流股份有限公司 | Prediction technique and device, the logistics system and computer-readable medium of order |
CN110309948A (en) * | 2019-05-09 | 2019-10-08 | 上汽安吉物流股份有限公司 | Complete vehicle logistics order forecast method and device, logistics system and computer-readable medium |
CN110110932A (en) * | 2019-05-09 | 2019-08-09 | 上汽安吉物流股份有限公司 | Order forecast method and device, logistics system and computer-readable medium |
CN110097320A (en) * | 2019-05-09 | 2019-08-06 | 上汽安吉物流股份有限公司 | Order forecast method and device, logistics system and computer-readable medium |
CN110309947A (en) * | 2019-05-09 | 2019-10-08 | 上汽安吉物流股份有限公司 | Complete vehicle logistics order forecast method and device, logistics system and computer-readable medium |
CN110110931A (en) * | 2019-05-09 | 2019-08-09 | 上汽安吉物流股份有限公司 | Order forecast method and device, logistics system and computer-readable medium |
CN110084437A (en) * | 2019-05-09 | 2019-08-02 | 上汽安吉物流股份有限公司 | Prediction technique and device, the logistics system and computer-readable medium of order |
CN110458664B (en) * | 2019-08-06 | 2021-02-02 | 上海新共赢信息科技有限公司 | User travel information prediction method, device, equipment and storage medium |
CN110689278A (en) * | 2019-10-11 | 2020-01-14 | 珠海格力电器股份有限公司 | Sheet metal material management method and system, storage medium and electronic equipment |
CN111461815B (en) * | 2020-03-17 | 2023-04-28 | 上海携程国际旅行社有限公司 | Order recognition model generation method, recognition method, system, equipment and medium |
CN113435968B (en) * | 2021-06-23 | 2024-02-02 | 南京领行科技股份有限公司 | Network appointment vehicle dispatching method and device, electronic equipment and storage medium |
CN114677087A (en) * | 2022-03-31 | 2022-06-28 | 上海圆徕科技有限公司 | Vehicle combination unmanned aerial vehicle cooperative distribution method |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107239532A (en) * | 2017-05-31 | 2017-10-10 | 北京京东尚科信息技术有限公司 | Data digging method and device |
CN107679462A (en) * | 2017-09-13 | 2018-02-09 | 哈尔滨工业大学深圳研究生院 | A kind of depth multiple features fusion sorting technique based on small echo |
CN107909206A (en) * | 2017-11-15 | 2018-04-13 | 电子科技大学 | A kind of PM2.5 Forecasting Methodologies based on deep structure Recognition with Recurrent Neural Network |
Family Cites Families (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105224984B (en) * | 2014-05-31 | 2018-03-13 | 华为技术有限公司 | A kind of data category recognition methods and device based on deep neural network |
CN105373840B (en) * | 2015-10-14 | 2018-12-11 | 深圳市天行家科技有限公司 | In generation, drives order forecast method and generation drives Transport capacity dispatching method |
-
2018
- 2018-04-19 CN CN201810357890.XA patent/CN108564326B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107239532A (en) * | 2017-05-31 | 2017-10-10 | 北京京东尚科信息技术有限公司 | Data digging method and device |
CN107679462A (en) * | 2017-09-13 | 2018-02-09 | 哈尔滨工业大学深圳研究生院 | A kind of depth multiple features fusion sorting technique based on small echo |
CN107909206A (en) * | 2017-11-15 | 2018-04-13 | 电子科技大学 | A kind of PM2.5 Forecasting Methodologies based on deep structure Recognition with Recurrent Neural Network |
Also Published As
Publication number | Publication date |
---|---|
CN108564326A (en) | 2018-09-21 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108564326B (en) | Order prediction method and device, computer readable medium and logistics system | |
CN111210024B (en) | Model training method, device, computer equipment and storage medium | |
Kim et al. | A capsule network for traffic speed prediction in complex road networks | |
CN109255505B (en) | Short-term load prediction method of multi-model fusion neural network | |
CN115018021B (en) | Machine room abnormity detection method and device based on graph structure and abnormity attention mechanism | |
CN110909803A (en) | Image recognition model training method and device and computer readable storage medium | |
US20200372342A1 (en) | Systems and methods for predictive early stopping in neural network training | |
CN108960537B (en) | Logistics order prediction method and device and readable medium | |
CN107609667B (en) | Heat supply load prediction method and system based on Box _ cox transformation and UFCNN | |
CN111369299A (en) | Method, device and equipment for identification and computer readable storage medium | |
CN109214559B (en) | Logistics service prediction method and device and readable storage medium | |
CN110390425A (en) | Prediction technique and device | |
CN109741268B (en) | Damaged image complement method for wall painting | |
CN111950810B (en) | Multi-variable time sequence prediction method and equipment based on self-evolution pre-training | |
CN111144548A (en) | Method and device for identifying working condition of pumping well | |
CN103324954A (en) | Image classification method based on tree structure and system using same | |
CN113449919B (en) | Power consumption prediction method and system based on feature and trend perception | |
Ibragimovich et al. | Effective recognition of pollen grains based on parametric adaptation of the image identification model | |
CN114548591A (en) | Time sequence data prediction method and system based on hybrid deep learning model and Stacking | |
Usman et al. | Data analytics for short term price and load forecasting in smart grids using enhanced recurrent neural network | |
Prabowo et al. | Continually learning out-of-distribution spatiotemporal data for robust energy forecasting | |
CN116627773B (en) | Abnormality analysis method and system of production and marketing difference statistics platform system | |
CN116543153A (en) | Semi-supervised point cloud semantic segmentation method based on selective active learning | |
CN113822441B (en) | Decision model training method, device, terminal equipment and storage medium | |
CN115081609A (en) | Acceleration method in intelligent decision, terminal equipment and storage medium |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |