CN111260141B - Storage distribution method based on user information - Google Patents

Storage distribution method based on user information Download PDF

Info

Publication number
CN111260141B
CN111260141B CN202010061305.9A CN202010061305A CN111260141B CN 111260141 B CN111260141 B CN 111260141B CN 202010061305 A CN202010061305 A CN 202010061305A CN 111260141 B CN111260141 B CN 111260141B
Authority
CN
China
Prior art keywords
height
age
user
weight
distance
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
Application number
CN202010061305.9A
Other languages
Chinese (zh)
Other versions
CN111260141A (en
Inventor
马立玲
郭建
王军政
赵江波
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Beijing Institute of Technology BIT
Original Assignee
Beijing Institute of Technology BIT
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Beijing Institute of Technology BIT filed Critical Beijing Institute of Technology BIT
Priority to CN202010061305.9A priority Critical patent/CN111260141B/en
Publication of CN111260141A publication Critical patent/CN111260141A/en
Application granted granted Critical
Publication of CN111260141B publication Critical patent/CN111260141B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

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
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
    • 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
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • 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
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/047—Probabilistic or stochastic networks
    • 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
    • G06N3/084—Backpropagation, e.g. using gradient descent
    • 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
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0631—Resource planning, allocation, distributing or scheduling for enterprises or organisations
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/60—Analysis of geometric attributes
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
    • G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
    • G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
    • G06V40/16—Human faces, e.g. facial parts, sketches or expressions
    • G06V40/172—Classification, e.g. identification
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10024—Color image
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20076—Probabilistic image processing
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20081—Training; Learning
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20084—Artificial neural networks [ANN]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30196—Human being; Person
    • G06T2207/30201—Face
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
    • G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
    • G06V40/16—Human faces, e.g. facial parts, sketches or expressions
    • G06V40/178—Human faces, e.g. facial parts, sketches or expressions estimating age from face image; using age information for improving recognition

Abstract

The invention discloses a storage allocation method based on user information. The invention can help a specific user to obtain the most suitable space, and enhances the intelligence of the warehousing system and the reasonability of allocation.

Description

Storage distribution method based on user information
Technical Field
The invention belongs to the field of storage and deposit, and particularly relates to a method for distributing storage positions according to user information.
Background
At present, a large number of storage cabinets and storage boxes are installed and used in a plurality of scenes such as supermarkets, stations, schools, communities and the like, and certain convenience is brought to people. However, the using method is single, generally, the user manually applies for storage, the storage system gives a paper storage certificate, and then the user opens or closes the storage space of the user according to the certificate. The storage space of users is often randomly distributed, and the old, children and disabled people cannot obtain the simplest storage space matched with the height of the users in the presence of a large number of storage cabinets and storage boxes. In addition, the existing storage systems are generally arranged and used according to groups, and when the number of users is large, new users often need to check each group of storage cabinets or storage boxes one by one to find out free storage space, which also brings unnecessary troubles to users.
Disclosure of Invention
In view of this, the present invention provides a dynamic warehousing allocation method, which can help a specific user to obtain the most suitable space, and enhance the intelligence of the warehousing system and the reasonability of allocation.
In order to solve the technical problem, the invention is realized as follows:
a warehousing distribution method based on user information comprises the following steps:
step 1, acquiring the age and height of a user;
step 2, determining the distance, the size and the height of the storage box suitable for the user according to the age and the height;
and 3, distributing storage boxes matched with the distance, the size and the height for the user.
Optionally, in step 1, the age and height of the user are obtained as follows: the age and the height of the user are obtained by shooting the image of the user for image recognition.
Optionally, when the distance, the size and the height of the storage box suitable for the user are determined in the step 2, further processing is carried out by combining the age weight and the height weight; the age weight and the height weight are determined according to the data acquisition accuracy and the specificity; the specificity refers to the difference degree between the data acquired in the step 1 and a default standard value.
Optionally, the step 2 adopts a neural network to determine the distance, size and height suitable for the user; the input of the neural network is age, age weight, height weight, and the output is distance, size and height.
Optionally, the step 3 includes:
step 31: judging whether the number of the residual storage boxes is higher than a set threshold value or not; if so, go to step 32; otherwise, executing step 33;
step 32: adopting an optimal search scheme: searching storage boxes meeting requirements according to the distance, size and height of a proper user and distributing the storage boxes to the user; if not, go to step 33;
step 33: adopting a suboptimal search scheme: and scoring the degree of each residual storage box meeting the requirement, and selecting the storage box with the highest score to distribute to users.
Optionally, the optimal search scheme is:
constructing a storage tree of a storage system: determining the sub-node of the 2 nd layer according to the number of the groups of the storage boxes, setting the sub-node of the 3 rd layer according to the number of the layers where the storage boxes are located, setting the sub-node of the 4 th layer according to the distance, and setting the sub-node of the 5 th layer according to the size;
starting from a father node of a storage tree, accessing child nodes which belong to the same group as a storage box group generating an allocation request, and continuing to search child nodes which meet other requirements of the current child nodes if the current node meets one of the requirements according to a depth-first search principle; and if the current child node does not meet the requirement, searching the brother node of the current child node, and repeating the steps until the nodes meeting all the requirements are found.
Optionally, the suboptimal search scheme is:
step a: determining an age weight w based on data acquisition accuracy and specificityageAnd height weight wheight(ii) a The specificity refers to the difference between the measured value and the default standard value;
step b: calculating distance weight when the age weight is greater than the height weight
Figure BDA0002374587260000031
High and low weights of
Figure BDA0002374587260000032
The magnitude weight is
Figure BDA0002374587260000033
When the age weight is less than or equal to the height weight, the height weight is calculated as
Figure BDA0002374587260000034
The magnitude weight is
Figure BDA0002374587260000035
Distance weighted as
Figure BDA0002374587260000036
Step c: traversing all the rest empty boxes, and calculating the score of each empty box; the calculation method comprises the following steps: the size only meets the requirements and does not meet the requirements, if the size meets the requirements, the scores of the size weights are accumulated, and if the size does not meet the requirements, no score is accumulated; the distance has two conditions of requiring distance inside and distance outside, the storage box accumulates the score of the distance weight in the distance, and accumulates the score according to the proportion of exceeding the distance outside the distance, and the larger the exceeding distance is, the less the score is accumulated, and the minimum score is 0; the height of the storage box is within the required height range and outside the required height range, the scores of the weights of the height of the storage box are accumulated within the height range, the scores of the storage box are accumulated outside the height range according to the proportion exceeding the height range, and the larger the exceeding height range is, the fewer the scores are accumulated, and the minimum score is 0.
Optionally, the age weight is determined by:
wage＝k1·accuracyage+k2·|agemeasure-agestandard|
wherein, wageAs age weight, accuracyageAge data acquisition accuracy, | agemeasure-agestandard| represents specificity of age; k is a radical of1And k2Weighted weights, age, for data acquisition accuracy and specificity, respectivelymeasureIs the age value, age, obtained in step 1standardSetting a default standard value for the set age;
the height weight is determined in the following way:
wheight＝k3·accuracyheight+k4·|hmeasure-heightstandard|
wherein, wheightAs a height weight, accuracyheightFor height data acquisition accuracy, | hmeasure-heightstandardL represents the specificity of height; k is a radical of3And k4Weighted weights, h, for data acquisition accuracy and specificity, respectivelymeasureHeight value, height, obtained in step 1standardDefaults standard value for the set height.
Optionally, the data acquisition accuracy of the age and the age is obtained by: calculating the user image by using the neural network for estimating the age, outputting the probability that the user belongs to various ages by using the output node, and obtaining the age of the user according to the output resultmeasure(ii) a The probability output by the output node of the neural network is the data acquisition accuracy of the age, accuracycacyage；
The height data acquisition accuracy rate acquisition mode is as follows: obtaining the user image when the user is at the ground mark point, estimating the distance s between the face and the camera according to the proportion of the face image in the whole image, and obtaining the image of the user when the user is at the ground mark point
Figure BDA0002374587260000041
The horizontal distance l between the camera and the ground mark point can be obtained1(ii) a By means of1The accuracy rate of calculating the height estimation with the real value l is as follows:
Figure BDA0002374587260000042
optionally, the principle of determining the distance, size and height of the storage box suitable for the user in step 2 is as follows: the height of the storage box is matched with the height of a user; if the user ages too much or too little, the closer the storage box is, the larger the storage box is, and the height of the storage box is matched with the capacity of the user for taking and placing articles.
Has the advantages that:
(1) the invention fully utilizes the user information to distribute different storage schemes for different user identities, so that each user obtains the most suitable storage space. The storage cabinet completely abandons the allocation strategy of randomly allocating storage space in the traditional method, and gives intelligence to the storage cabinet. The invention is especially significant for special people such as the old, the disabled and the children, and can greatly improve the use experience of the users.
(2) According to the invention, the age and the height are adjusted by using the weight, and the accuracy (including the data acquisition accuracy and specificity) of the physical quantities of the age information and the height information is added into the consideration of storage distribution, so that the distribution rationality is improved.
(3) The invention adopts the neural network to construct the mapping relation between the age, the age weight, the height and the height weight and the size, the height and the distance of the storage box, and reduces the realization difficulty of the distribution strategy.
(4) According to the invention, the threshold value of the storage box is set, the optimal distribution scheme is used only when more storage boxes are available, and the suboptimal scheme is directly used when the storage boxes are insufficient, so that the problem of time waste caused by the fact that the optimal scheme is executed first and the suboptimal scheme is executed after the optimal scheme is searched for in each time is solved.
(5) In the suboptimal scheme, the calculation modes of distance weight, high weight, low weight and size weight are skillfully designed, so that more reasonable scores can be obtained when the weights are used for scoring.
(6) The invention also provides a preferable determination scheme of the age weight and the height weight, the calculation is simple, and the accuracy of the user data is effectively represented.
(7) When the optimal search scheme is implemented, the depth-first search algorithm is used, the occupied memory is small, and the method is suitable for solving the search problem of the storage cabinet with a large scale.
Drawings
Fig. 1 is a flowchart of a warehouse allocation method based on user information according to the present invention.
FIG. 2 is a schematic of the calculated height of the present invention.
Fig. 3 is a block diagram of a neural network used in the present invention.
Fig. 4 is a flowchart of step four in fig. 1.
Detailed Description
The invention provides a storage distribution method based on user information, which has great practical reference significance for improving the use and management mode of a large storage cabinet. The traditional locker has simple structure and single use mode, all users are regarded as the same type, only one allocation scheme is randomly allocated, and special use requirements are difficult to meet. In addition, even an ordinary user may need to perform a plurality of operations when using the device, which wastes user time. According to the invention, after the user agrees, the age and height information of the user is obtained, and the distance, size and height of the storage box suitable for the user are found according to the age and height, so that the storage box with the suitable distance, size and height is distributed for the user. And finally, the position of the storage box can be sent to a terminal execution mechanism according to a distribution result, and one-time distribution is completed.
According to the invention, not only the height of the user but also the age of the user are considered, and the storage box is distributed according to the two information and the distribution rule, so that each user can obtain the most suitable storage space. The storage cabinet completely abandons the allocation strategy of randomly allocating storage space in the traditional method, and gives intelligence to the storage cabinet. The invention is especially significant for special people such as the old, the disabled and the children, and can greatly improve the use experience of the users.
When the storage box is distributed to the user, the storage box is further distributed according to the age weight and the height weight, and the age weight and the height weight mainly reflect the availability degree of the data to decision making. The age weight and the height weight are added into the consideration factor of the storage distribution, so that the distribution rationality can be improved. When the combined consideration is carried out, the combined calculation such as multiplication, addition and the like can be directly carried out on the age/height data by utilizing the corresponding weight, and then a proper storage box is searched according to the adjusted age/height according to the distribution strategy. However, because the relationship between the age and the height and the size, the height and the distance of the allocated storage box is complex, the neural network is selected as the mapping network in the following embodiment, and the method has the advantages of high accuracy and simple parameter setting. Besides the age and the height, the input information of the neural network is added with the age weight and the height weight, so that the accuracy of the data is added into the network operation.
To make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings, and it is apparent that the described embodiments are some, but not all embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Fig. 1 is a flowchart of a warehouse allocation method based on user information according to the present invention, as shown in fig. 1, including the following steps:
step one, acquiring age of usermeasureAnd height hmeasure。
The embodiment acquires the age and the height of the user by shooting the image of the user for image recognition. Particularly, the information acquisition equipment is composed of the camera with the holder, and the information acquisition equipment can directly shoot images after users put forward using requirements.
AgemeasureThe acquisition mode is as follows: the method of extracting a face image from a shot image and obtaining the age of a user from the face image generally adopts the identification of specific features, such as wrinkles, hair color and the like, has a mature and open technology, and is not described in detail herein.
Height hmeasureThe acquisition mode is as follows: referring to fig. 2, when a user stands at a ground mark point position when using the device, the cradle head automatically adjusts the angle to enable the mark point in the face of the user, such as a binocular center, to coincide with the center of an image, records the inclination angle alpha of the central axis and the horizontal direction of the cradle head at the moment, generally arranges the cradle head at a higher position, thus ensuring that the inclination angle alpha is between 0 degree and 90 degrees all the time, and can know the height measurement value h according to the geometric relationshipmeasureL · tan α; h can be obtained when the horizontal distance l between the camera and the ground mark point is knownmeasureI.e. the difference between the user and the standard height, may represent the height of the user.
Step two, calculating age weight wageAnd height weight wheight。
As previously mentioned, the age weight wageAnd height weight wheightReflecting the availability of the corresponding data to the decision. The present invention utilizes data acquisition accuracy and specificity to normalize this weight. Wherein the content of the first and second substances,the higher the accuracy rate is, the more decision optimization can be performed according to the physical quantity. Specificity represents the degree of distinction between this physical quantity and a common, default parameter that does not require optimization, and can generally be expressed as the difference between the measured value and the default standard value, with the higher the specificity, the more emphasis is placed on this physical quantity in the optimization.
Based on the above analysis, the age weight w in this embodimentageThe determination method comprises the following steps:
wage＝k1·accuracyage+k2·|agemeasure-agestandard|
wherein, accuracyageAge data acquisition accuracy, | agemeasure-agestandard| is specificity of age; k is a radical of1And k2The weighting (positive values), which are the data acquisition accuracy and specificity, respectively, can be adjusted as needed. age (age)measureIs the age measurement obtained in step onestandardDefault standard value for set age.
Wherein, accuracyageThe method can be directly obtained according to the neural network of the estimated age, namely, the neural network of the estimated age is used for calculating the user image, each output node of the neural network outputs the probability that the user belongs to various ages, and the age corresponding to the person with the maximum probability is the estimated age of the user. The probability value output by the output node of the neural network can be used as the age accuracy rate accuracyage。
Also, the height weight w in this embodimentheightThe determination method comprises the following steps:
wheight＝k3·accuracyheight+k4·|hmeasure-heightstandard|
wherein, accuracyheightFor height data acquisition accuracy, | hmeasure-heightstandardL represents the specificity of height; k is a radical of3And k4Weighted weights (positive values), which are data acquisition accuracy and specificity, respectively, can be adjusted as needed。hmeasureHeight measurement obtained for step onestandardDefaults standard value for the set height.
Wherein, accuracyheightThe calculation method of (2) may be: obtaining the user image when the user is at the ground mark point, estimating the distance s between the face and the camera according to the proportion of the face image in the whole image, and calculating the distance between the face and the camera according to the distance s
Figure BDA0002374587260000081
The horizontal distance l between the camera and the ground mark point can be obtained1. By means of1The accuracy rate of calculating the height estimation with the real value l is as follows:
Figure BDA0002374587260000082
when l is1When equal to l, accuracyheight＝100％。
And step three, determining the distance, the size and the height of the storage box suitable for the user according to the age, the height and the combination of the age weight and the height weight.
The determination principle may be: the height of the storage box is matched with the height of a user; if the storage box is too old or too small, the closer the storage box is, the larger the storage box is, and the height of the storage box is matched with the capacity of a user for taking and placing articles.
The corresponding relations among the age, the height, the age weight, the height weight, the distance, the size and the height are complex, so that the embodiment of the invention adopts the neural network to determine the distance, the size and the height of the storage box.
As shown in fig. 3, the BP neural network with 4 input nodes and 3 output nodes is adopted in the present embodiment, wherein the 4 input nodes are respectively age, age weight, height and height weight, and the 3 output nodes are respectively distance, size and height of the required storage box. Setting the number of hidden layers to be 3-4 layers, and carrying out Gaussian truncation distribution initialization on the initial weight of the network. Setting an exponential decay learning rate, generally setting an initial learning rate to be 0.1, a decay coefficient to be 0.9 and a decay step length to be 50 rounds; the momentum term is set, typically to a momentum coefficient of 0.5.
Training data is acquired and divided into a training set and a test set. The training data of the neural network in the invention is structured, and can be constructed according to the desire of the warehousing system manager, such as users with the height of 90% being possibly 160 cm and the age of 99% being possibly 89 years, and the requirement of the users is a large box at the 4 th layer and the 5 th layer within 1m from the camera. The principle of data structure is that firstly all users preferentially obtain the storage box which is matched with the height of the users and is close to the height of the users, secondly, the old people preferentially obtain the large storage box which is medium in height and is close to the height of the users, and the young users preferentially obtain the storage box with lower height and the like. The neural network is trained by utilizing the training set, the neural network can learn the requirement calculation idea after a large amount of training, and then the training is carried out by utilizing the test set so as to further determine the modification of the parameters in the algorithm. Also considering that the age is "old" and the height is not clearly defined in the divisions "super high", etc., the advantages of the neural network can be exploited to train the network using only extreme data as described in the example, i.e. extra old and extra high user data, for the network to handle ambiguous classification problems.
During training, a training set is fed into a neural network, forward propagation is carried out according to a network structure, and a loss function is calculated according to calculation records and sample types, and is generally taken as a cross entropy loss function. And performing back propagation according to the loss function value, and updating the weight parameters and the bias parameters of the weight network. And repeating the execution until the loss function value meets the requirement.
And step four, distributing storage boxes matched with the distance, the size and the height for users.
When the storage box is selected according to the determined distance, size and height, storage boxes meeting the conditions can be searched from the rest of the storage boxes to serve as an optimal scheme. If the search is not achieved, the box meeting partial conditions is selected, namely a suboptimal scheme is considered. When the warehouse system is large in scale, it is very time-consuming to carry out the operation according to the sequence of searching the optimal scheme without solution and then calculating the suboptimal scheme, and most of the calculation amount in the worst case is invalid, so that it is practical to use simple criteria to avoid the worst case.
According to the invention, through setting the threshold value, the quick and simple switching between the two schemes can be realized, and the operation efficiency of the system is improved. As shown in fig. 4, the process includes the steps of:
step 41: judging whether the number of the residual storage boxes is higher than a set threshold value or not; if so, the optimal solution can be found with a higher probability, and then step 42 is executed; otherwise, step 43 is executed.
Step 42: adopting an optimal search scheme: searching a storage box meeting the requirements according to the distance, the size and the height of a proper user; if not, step 43 is performed.
In this step, firstly, according to the hardware configuration and the operation state of the warehousing system, a storage tree is designed and constructed: generally, the number d of child nodes with the depth of 2 is determined according to the number of storage cabinets2(child node record group) in which the number of child nodes d having a depth of 3 is set according to the number of layers3(number of child node recording layers) and the number d of child nodes having a depth of 4 is set according to the distance4(child node recording distance) and the number d of child nodes having a depth of 5 is set according to the size5(child node record size).
During searching, the child nodes which belong to the same group with an information acquisition module (positioned in a storage box group generating an allocation request) are accessed from a father node of a storage tree, and then according to the principle of depth-first searching, if the current node meets one of the requirements, the child nodes meeting other requirements of the current child node are continuously searched; and if the current child node does not meet the requirement, searching the brother node of the current child node, and repeating the steps to find out the nodes meeting all the requirements.
If the nodes meeting all the requirements are found, the positions of the nodes are recorded, the tree structure of the warehousing system is updated, and other requests are prevented from continuously using the nodes. And sending the node position to a terminal execution mechanism to finish one-time distribution. If no satisfactory node is found, step 43 is executed.
Step 43: adopting a suboptimal search scheme: and scoring the degree of each residual storage box meeting the requirement, and selecting the storage box with the highest score to distribute to users.
The method specifically comprises the following substeps:
step a: acquiring the age weight w calculated in the step 2ageAnd height weight wheight。
Step b: calculating distance weight when the age weight is greater than the height weight
Figure BDA0002374587260000101
High and low weights of
Figure BDA0002374587260000102
The magnitude weight is
Figure BDA0002374587260000103
Therefore, the distance weight is ensured to be constantly greater than the high and low weights, and the high and low weights are constantly greater than the large and small weights, so that the core intention of helping the old user walk less is embodied;
when the age weight is less than or equal to the height weight, the height weight is
Figure BDA0002374587260000111
The magnitude weight is
Figure BDA0002374587260000112
Distance weighted as
Figure BDA0002374587260000113
Therefore, the height weight is constantly larger than the size weight, and the size weight is constantly larger than the distance weight, so that the core intention of helping a user with special height to find a proper box is reflected;
it is noted that the weight of each user, the size weight, and the distance weight are dynamically adjusted, and the values are related to the age weight and the height weight, which ensures that different users can compare the weights with each other. And if one box is a sub-optimal scheme of a plurality of users at the same time, the box is distributed to the user with the highest score according to the self weight rule.
Step c: and traversing all the rest empty boxes according to a snake shape, and calculating the score of each empty box. The calculation method comprises the following steps: the size only meets the requirements and does not meet the requirements, if the size meets the requirements, the scores of the size weights are accumulated, and if the size does not meet the requirements, no score is accumulated; the distance has two conditions of requiring distance inside and distance outside, the storage box accumulates the score of the distance weight in the distance, and accumulates the score according to the proportion of exceeding the distance outside the distance, and the larger the exceeding distance is, the less the score is accumulated, and the minimum score is 0; the height of the storage box is within the required height range and outside the required height range, the scores of the weights of the height of the storage box are accumulated within the height range, the scores of the storage box are accumulated outside the height range according to the proportion exceeding the height range, and the larger the exceeding height range is, the fewer the scores are accumulated, and the minimum score is 0.
Step d: and (4) obtaining the position of the box with the highest score by using a quick sorting algorithm for the calculated score list. The dynamic allocation in the invention is also embodied in that for different backgrounds, such as different time periods of each day and different entrances of the storage system, the allocation schemes given by the system are not in a constant one-to-one correspondence relationship, but the warehousing schemes meeting the user requirements are given according to the storage state at the time when the use requirements are provided.
In the prior art, when the conventional warehousing scheme is used, a plurality of storage cabinets are often required to be accessed to find the free storage cabinets in a time period with a large user quantity, and the problem can be effectively solved by uniformly deciding through a plurality of storage systems. In practical implementation, each storage cabinet runs one storage allocation algorithm of the invention, and the running state of the storage system consisting of a plurality of storage cabinets at the current time is shared. When a certain locker searches the optimal scheme, the box is directly occupied, and the running state of the warehousing system is changed. However, when calculating a suboptimal solution, finding a suboptimal box of a current user cannot be directly occupied, and whether the box is also the suboptimal solution of other users in a fixed time interval needs to be judged, if so, scores need to be compared; if not, it may be occupied.
Step five: and sending the distributed positions of the storage boxes to a terminal execution mechanism, and responding by the action of the terminal execution mechanism.
In summary, the above description is only a preferred embodiment of the present invention, and is not intended to limit the scope of the present invention. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims (8)

1. A warehousing distribution method based on user information is characterized by comprising the following steps:
step 1, acquiring the age and height of a user;
step 2, determining the distance, the size and the height of the storage box suitable for the user according to the age and the height;
step 3, allocating storage boxes matched with the distance, the size and the height for users:
step 31: judging whether the number of the residual storage boxes is higher than a set threshold value or not; if so, go to step 32; otherwise, executing step 33;
step 32: adopting an optimal search scheme: searching storage boxes meeting requirements according to the distance, size and height of a proper user and distributing the storage boxes to the user; if not, go to step 33;
the optimal search scheme is as follows: constructing a storage tree of a storage system: determining the sub-node of the 2 nd layer according to the number of the groups of the storage boxes, setting the sub-node of the 3 rd layer according to the number of the layers where the storage boxes are located, setting the sub-node of the 4 th layer according to the distance, and setting the sub-node of the 5 th layer according to the size;
starting from a father node of a storage tree, accessing child nodes which belong to the same group as a storage box group generating an allocation request, and continuing to search child nodes which meet other requirements of the current child nodes if the current node meets one of the requirements according to a depth-first search principle; if the current child node does not meet the requirement, searching brother nodes of the current child node, and repeating the steps until all the nodes meeting the requirement are found;
step 33: adopting a suboptimal search scheme: and scoring the degree of each residual storage box meeting the requirement, and selecting the storage box with the highest score to distribute to users.
2. The method of claim 1, wherein the step 1 of obtaining the age and height of the user is: the age and the height of the user are obtained by shooting the image of the user for image recognition.
3. The method of claim 1, wherein the step 2 of determining the distance, size and height of the storage box suitable for the user is further processed in combination with an age weight and a height weight; the age weight and the height weight are determined according to the data acquisition accuracy and the specificity; the specificity refers to the difference degree between the data acquired in the step 1 and a default standard value.
4. The method of claim 3, wherein the step 2 employs a neural network to determine a distance, size and height suitable for the user; the input of the neural network is age, age weight, height weight, and the output is distance, size and height.
5. The method of claim 1, wherein the suboptimal search scheme is:
step a: determining an age weight w based on data acquisition accuracy and specificityageAnd height weight wheight(ii) a The specificity refers to the difference between the measured value and the default standard value;
step b: calculating distance weight when the age weight is greater than the height weight
Figure FDA0003396950830000021
High and low weights of
Figure FDA0003396950830000022
The magnitude weight is
Figure FDA0003396950830000023
When the age weight is less than or equal to the height weight, the height weight is calculated as
Figure FDA0003396950830000024
The magnitude weight is
Figure FDA0003396950830000025
Distance weighted as
Figure FDA0003396950830000026
Step c: traversing all the rest empty boxes, and calculating the score of each empty box; the calculation method comprises the following steps: the size only meets the requirements and does not meet the requirements, if the size meets the requirements, the scores of the size weights are accumulated, and if the size does not meet the requirements, no score is accumulated; the distance has two conditions of requiring distance inside and distance outside, the storage box accumulates the score of the distance weight in the distance, and accumulates the score according to the proportion of exceeding the distance outside the distance, and the larger the exceeding distance is, the less the score is accumulated, and the minimum score is 0; the height of the storage box is within the required height range and outside the required height range, the scores of the weights of the height of the storage box are accumulated within the height range, the scores of the storage box are accumulated outside the height range according to the proportion exceeding the height range, and the larger the exceeding height range is, the fewer the scores are accumulated, and the minimum score is 0.
6. The method according to claim 3 or 5, wherein the age weight is determined by:
wage＝k1·accuracyage+k2·|agemeasure-agestandard|
wherein, wageAs age weight, accuracyageAge data acquisition accuracy, | agemeasure-agestandard| represents specificity of age; k is a radical of1And k2Weighted weights, age, for data acquisition accuracy and specificity, respectivelymeasureIs the age value, age, obtained in step 1standardSetting a default standard value for the set age;
the height weight is determined in the following way:
wheight＝k3·accuracyheight+k4·|hmeasure-heightstandard|
wherein, wheightAs a height weight, accuracyheightFor height data acquisition accuracy, | hmeasure-heightstandardL represents the specificity of height; k is a radical of3And k4Weighted weights, h, for data acquisition accuracy and specificity, respectivelymeasureHeight value, height, obtained in step 1standardDefaults standard value for the set height.
7. The method of claim 6, wherein the age and age data acquisition accuracy is obtained by: calculating the user image by using the neural network for estimating the age, outputting the probability that the user belongs to various ages by using the output node, and obtaining the age of the user according to the output resultmeasure(ii) a The probability output by the output node of the neural network is the data acquisition accuracy of the age, accuracycacyage；
The height data acquisition accuracy rate acquisition mode is as follows: obtaining the user image when the user is at the ground mark point, estimating the distance s between the face and the camera according to the proportion of the face image in the whole image, and obtaining the image of the user when the user is at the ground mark point
Figure FDA0003396950830000031
The horizontal distance l between the camera and the ground mark point can be obtained1(ii) a By means of1The accuracy rate of calculating the height estimation with the real value l is as follows:
Figure FDA0003396950830000032
8. the method of claim 1, wherein the distance, size and height of the storage compartment suitable for the user are determined in step 2 by: the height of the storage box is matched with the height of a user; if the user ages too much or too little, the closer the storage box is, the larger the storage box is, and the height of the storage box is matched with the capacity of the user for taking and placing articles.
CN202010061305.9A 2020-01-19 2020-01-19 Storage distribution method based on user information Active CN111260141B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010061305.9A CN111260141B (en) 2020-01-19 2020-01-19 Storage distribution method based on user information

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010061305.9A CN111260141B (en) 2020-01-19 2020-01-19 Storage distribution method based on user information

Publications (2)

Publication Number Publication Date
CN111260141A CN111260141A (en) 2020-06-09
CN111260141B true CN111260141B (en) 2022-03-18

Family

ID=70952344

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010061305.9A Active CN111260141B (en) 2020-01-19 2020-01-19 Storage distribution method based on user information

Country Status (1)

Country Link
CN (1) CN111260141B (en)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113538802A (en) * 2021-06-25 2021-10-22 华录智达科技股份有限公司 Optimal addressing method for intelligent key cabinet of bus station
CN115331357A (en) * 2022-05-19 2022-11-11 拉扎斯网络科技（上海）有限公司 Request processing method and device, electronic equipment and computer readable storage medium

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106204948A (en) * 2016-07-11 2016-12-07 商汤集团有限公司 Locker management method and locker managing device
CN106708047A (en) * 2016-12-21 2017-05-24 精效新软新技术（北京）有限公司 Intelligent article delivery robot device and control method
CN107016800A (en) * 2016-07-15 2017-08-04 苏州市凯发金属制品有限公司 Intelligence letter lodge and its application method
CN108416940A (en) * 2018-03-07 2018-08-17 深圳万发创新进出口贸易有限公司 A kind of locker managing device
CN209202426U (en) * 2018-08-14 2019-08-06 殷肇良 A kind of self-service access material cabinet of practical activity

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104978346A (en) * 2014-04-09 2015-10-14 阿里巴巴集团控股有限公司 User evaluation information providing method and user evaluation information providing system
US10585703B2 (en) * 2017-06-03 2020-03-10 Apple Inc. Dynamic operation allocation for neural networks

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106204948A (en) * 2016-07-11 2016-12-07 商汤集团有限公司 Locker management method and locker managing device
CN107016800A (en) * 2016-07-15 2017-08-04 苏州市凯发金属制品有限公司 Intelligence letter lodge and its application method
CN106708047A (en) * 2016-12-21 2017-05-24 精效新软新技术（北京）有限公司 Intelligent article delivery robot device and control method
CN108416940A (en) * 2018-03-07 2018-08-17 深圳万发创新进出口贸易有限公司 A kind of locker managing device
CN209202426U (en) * 2018-08-14 2019-08-06 殷肇良 A kind of self-service access material cabinet of practical activity

Also Published As

Publication number Publication date
CN111260141A (en) 2020-06-09

Similar Documents

Publication Publication Date Title
CN108509463B (en) Question response method and device
CN111104595B (en) Deep reinforcement learning interactive recommendation method and system based on text information
CN110647904B (en) Cross-modal retrieval method and system based on unmarked data migration
CN107885778B (en) Personalized recommendation method based on dynamic near point spectral clustering
CN111260141B (en) Storage distribution method based on user information
CN109960737B (en) Remote sensing image content retrieval method for semi-supervised depth confrontation self-coding Hash learning
CN109104688A (en) Wireless network access point model is generated using aggregation technique
CN103309869B (en) Method and system for recommending display keyword of data object
CN104731882B (en) A kind of adaptive querying method that weighting sequence is encoded based on Hash
CN109635083A (en) It is a kind of for search for TED speech in topic formula inquiry document retrieval method
CN111914162A (en) Method for guiding personalized learning scheme based on knowledge graph
CN103945238A (en) Community detection method based on user behaviors
CN115861738A (en) Category semantic information guided remote sensing target detection active sampling method
CN114637881B (en) Image retrieval method based on multi-agent metric learning
CN108984711A (en) A kind of personalized APP recommended method based on layering insertion
CN108647739B (en) Social network community discovery method based on improved density peak clustering
CN106529490A (en) System and method for realizing handwriting identification based on sparse auto-encoding codebook
CN109635004A (en) A kind of object factory providing method, device and the equipment of database
CN111291625B (en) Friend recommendation method and system based on face retrieval
CN115759679A (en) Grouping method and device
CN115730248A (en) Machine account detection method, system, equipment and storage medium
CN110008975A (en) Social networks navy detection method based on Danger Immune theory
CN109472712A (en) A kind of efficient Markov random field Combo discovering method strengthened based on structure feature
CN115329210A (en) False news detection method based on interactive graph layered pooling
CN112733035A (en) Knowledge point recommendation method and device based on knowledge graph, storage medium and electronic device

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