Library inventory management method and system based on big data analysis
By dynamically adjusting the maintenance cycle and priority of library books through big data analysis and edge vision sensors, the problems of low efficiency and unreasonable resource allocation in library book maintenance and management have been solved, achieving efficient and economical library inventory management.
Patent Information
- Application Number
- CN202511093702.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-08-06
- Publication Date
- 2025-11-07
AI Technical Summary
In existing technologies, library book maintenance and management rely on fixed-cycle manual inspections, which leads to low efficiency and slow response. It is unable to cope with the accelerated wear and tear of frequently borrowed books in a timely manner, lacks intelligent damage assessment and prioritization, resulting in unreasonable allocation of maintenance resources. Valuable documents are not repaired in time, while ordinary books are over-maintained, leading to high maintenance costs.
This paper adopts a library inventory management method based on big data analysis. By acquiring information from the library inventory database and the borrowing database, it uses PAC principal component analysis, K-means clustering, ARMA time-moving autoregressive model and edge vision sensor to establish book borrowing popularity trend indicators and damage assessment models. It dynamically adjusts the maintenance cycle and priority of books, and combines machine vision inspection to replace manual inspection, so as to realize intelligent resource allocation.
It improved the efficiency of library book maintenance, enabled adaptive response to high-demand books, reduced ineffective maintenance expenditures for low-value books, ensured that precious documents received key protection, and significantly reduced management costs and improved inventory management efficiency.
Smart Images
Figure CN120911894A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the technical field of big data analysis, in particular to a library inventory management method and system based on big data analysis. BACKGROUND
[0002] The current library book maintenance management mainly relies on fixed period manual inspection method, which leads to low maintenance efficiency and lag response, and cannot timely respond to the accelerated wear of high-frequency borrowed books; at the same time, there is a lack of intelligent damage assessment and priority determination system, which causes unreasonable allocation of maintenance resources, leading to that precious literature cannot be repaired in time while ordinary books are over-maintained, so that the actual demand leads to high maintenance cost. SUMMARY
[0003] To solve the above technical problems, a library inventory management method and system based on big data analysis are provided, which solves the above problems that the current library book maintenance management mainly relies on fixed period manual inspection method, which leads to low maintenance efficiency and lag response, and cannot timely respond to the accelerated wear of high-frequency borrowed books; at the same time, there is a lack of intelligent damage assessment and priority determination system, which causes unreasonable allocation of maintenance resources, leading to that precious literature cannot be repaired in time while ordinary books are over-maintained, so that the actual demand leads to high maintenance cost.
[0004] To achieve the above purpose, the technical scheme adopted by the present application is:
[0005] A library inventory management method based on big data analysis, comprising:
[0006] S1, based on the library inventory database, obtaining the basic information of each book in the library, initializing the ideal maintenance period limit of each book in the library according to the book type corresponding to each book basic information;
[0007] S2, based on the library borrowing database, obtaining the historical borrowing information and external influencing factors of each book, analyzing the borrowing heat trend of each book in the historical borrowing information, and generating a library book borrowing heat trend index;
[0008] S3, compensating and fitting the maintenance period limit of each book in the library according to the library book borrowing heat trend index, generating the actual maintenance period limit of each book in the library;
[0009] S4, based on the edge visual sensor, obtaining the library borrowed book return image data, establishing a book return image loss analysis model, evaluating the damage state of the target book, correcting the actual maintenance period limit of each book in the library, and generating a library inventory book maintenance priority.
[0010] Preferably, based on the Internet, the audience big data corresponding to the known book categories is obtained;
[0011] The PAC principal component analysis is utilized to perform data dimension reduction on the audience big data corresponding to the known book categories, so as to obtain audience big data corresponding to the known book categories after dimension reduction;
[0012] According to the audience big data corresponding to the known book categories after dimension reduction, K-means clustering is utilized, a plurality of initial clustering clusters are preset, and the audience big data corresponding to the book is iteratively divided, so as to generate a known book category corresponding to the audience preference graph; the audience big data after dimension reduction includes: borrowing age, borrowing gender, borrowing book category
[0013] Based on the audience preference graph corresponding to the known book category, the audience of the known book category is divided according to the age, and the audience age distribution preference array of the known book category is obtained, and the audience age distribution preference matrix A of the known book category is formed;
[0014]
[0015] Wherein, a ij is the jth audience age distribution preference array of the known i book categories, m is the total number of book categories, and n is the total number of audience age distribution preference arrays.
[0016] Preferably, based on the audience age distribution preference matrix of the known book category, the borrowing times of the audience age distribution of the book are counted, and the audience borrowing frequency time sequence characteristic data of the known book category is determined;
[0017] Based on the known book category of the audience, the basic maintenance cycle time limit of the library inventory book is preset;
[0018] Based on the ARMA time moving autoregressive model, the audience borrowing frequency time sequence characteristic data of the known book category is taken as the input, and the audience borrowing heat index of the known book category is taken as the output;
[0019] The known book category of the audience borrowing demand index is utilized to compensate the basic maintenance cycle time limit of the library inventory book, and the ideal maintenance cycle time limit of each book of the library is initialized.
[0020] Preferably, based on the library borrowing database, the historical borrowing information of each book is obtained, the borrowing unit time is taken as the observation window, and the historical borrowing information of each book is taken as the observation attribute, so as to obtain the historical borrowing time sequence data of each book of the library;
[0021] The historical borrowing time sequence data of each book in the library is substituted into an ARMA time moving autoregressive model to generate a reader group borrowing heat index of each historical book in the library, and a time sequence of the reader group borrowing heat index of each historical book in the library is determined
[0022] The time sequence of the reader group borrowing heat index of each historical book in the library is decomposed by using SLT decomposition to obtain a reader group borrowing heat multi-dimensional index of each historical book in the library; the reader group borrowing heat multi-dimensional index of each historical book in the library includes a reader group borrowing heat trend index, a reader group borrowing heat seasonal index and a reader group borrowing heat residual index of each historical book in the library;
[0023] Based on the reader group borrowing heat multi-dimensional index of each historical book in the library, an external factor of the change of the reader group borrowing multi-dimensional index of the historical book is marked, and the external factor is recorded as a reader group borrowing heat attention external factor object of each historical book in the library;
[0024] The reader group borrowing heat attention external factor object of each historical book in the library is normalized;
[0025] According to the entropy weight method, the proportion of the reader group borrowing heat attention external factor object of each historical book in the library in the total is verified, and the reader group borrowing heat attention external factor object of each historical book in the library is given a weight;
[0026] Based on the weight of the reader group borrowing heat attention external factor object of each historical book in the library, the reader group borrowing heat attention external factor object of each historical book in the library is recorded as a reader group borrowing heat attention external factor object of each historical book in the library, and an environment influence time-varying vector of the reader group borrowing heat of each historical book in the library is calculated.
[0027] Preferably, based on the library borrowing database, the historical borrowing information of each real-time book in the library is obtained;
[0028] Based on the Internet, real-time network book discussion hot topics are obtained, and real-time network book hot feature data is extracted by using an NLP natural language analysis model;
[0029] Based on the reader group borrowing heat attention external factor object of each historical book in the library, the historical borrowing information of each real-time book in the library is screened to obtain a real-time borrowing heat attention external factor of each real-time book in the library;
[0030] The difference coefficient between the real-time borrowing heat attention external factor of each real-time book in the library and the reader group borrowing heat attention external factor object of each historical book in the library is calculated, and the weight of the real-time borrowing heat attention external factor of each book in the library is determined.
[0031] Based on the real-time borrowing hotness of each book in the library, the external factor weight is calculated based on the real-time borrowing hotness of each book in the library, and the real-time borrowing hotness of each book in the library is calculated based on the real-time borrowing hotness of each book in the library.
[0032] Based on the real-time borrowing hotness of each book in the library, the external factor weight is calculated based on the real-time borrowing hotness of each book in the library, and the real-time borrowing hotness of each book in the library is calculated based on the real-time borrowing hotness of each book in the library.
[0033]
[0034] Wherein, H i is the borrowing hotness trend index of the i-th book in the library, and a is the borrowing trust coefficient of the book, is the j-th audience group borrowing hotness environment influence time-varying vector of the i-th book in the library at the t-th unit time, is the j-th audience group borrowing hotness environment influence time-varying vector of the i-th book in the library at the t-th unit time.
[0035] Preferably, based on Random Forest, according to the audience group preference graph corresponding to the known book category, the maintenance period limit of each book in the library is taken as the leaf node, and the audience group borrowing frequency time sequence feature data set of the known book category is taken as the root node, the maintenance period limit decision tree of the known book category is established, and the library book maintenance period limit compensation update random forest model is established.
[0036] Based on the library book maintenance period limit compensation update random forest model, the library book borrowing hotness trend index is taken as the input, and the actual maintenance period limit of each book in the library is taken as the output.
[0037] Preferably, based on the edge visual sensor, the multi-angle image data of the library book return is obtained;
[0038] The multi-angle image data of the library book return is preprocessed, the multi-angle image contour area of the library book return is determined according to edge detection, and the multi-angle image contour area is cropped by using U-Net++, so as to obtain the damage image feature data of the library book return;
[0039] The damage image feature pixels in the damage image feature data of the library book return are normalized;
[0040] Based on Multi-Task Learning multi-task target learning, the book return image loss analysis model is established, the damage score and damage type classification of the library book return are taken as the objective function, the internal and external tearing and page loss of the book are taken as the physical consistency constraint, the damage image feature data of the library book return is taken as the output, and the evaluation value corresponding to the damage image of the library book return is substituted into the Sigmoid function to generate the damage score and damage type classification of the library book return;
[0041] Using NSGA-II multi-objective optimization genetic algorithm, based on the damage type classification of the library book return, the maintenance cost required for the maintenance action of the target book under the damage type classification is estimated, it is judged whether the maintenance cost exceeds the book value, if yes, it is judged to abandon the maintenance, if not, it is judged to correct the actual maintenance cycle time limit of each book in the library according to the damage score of the library book return, and the maintenance priority of the library inventory book is determined.
[0042] Further, a library inventory management system based on big data analysis is used to realize a library inventory management method based on big data analysis as described above, comprising:
[0043] The initial maintenance cycle module, the book borrowing heat module, the maintenance cycle correction module and the maintenance priority generation module;
[0044] The initial maintenance cycle module is used to obtain the basic information of each book in the library based on the library inventory database, and initialize the ideal maintenance cycle time limit of each book in the library according to the book type corresponding to each book basic information;
[0045] The book borrowing heat module is used to obtain the historical borrowing information and external influencing factors of each book based on the library borrowing database, analyze the borrowing heat trend of each book in the historical borrowing information, and generate the library book borrowing heat trend index;
[0046] The maintenance cycle correction module is electrically connected with the initial maintenance cycle module and the initial maintenance cycle module, and the maintenance cycle correction module is used to compensate and fit the maintenance cycle time limit of each book in the library according to the library book borrowing heat trend index, and generate the actual maintenance cycle time limit of each book in the library;
[0047] The maintenance priority generation module is electrically connected with the maintenance cycle correction module, and the maintenance priority generation module is used to obtain the library borrowing book return image data based on the edge visual sensor, establish the book return image loss analysis model, evaluate the damage state of the target book, correct the actual maintenance cycle time limit of each book in the library, and generate the maintenance priority of the library inventory book.
[0048] Compared with the prior art, the present application has the beneficial effects that:
[0049] The present application provides a library inventory management scheme based on big data analysis, which improves the efficiency of library book maintenance according to a three-level optimization system of basic cycle-borrowing popularity compensation-visual damage correction and edge computing technology, realizes a dynamic adjustment mechanism to make the maintenance response of high-demand books self-adaptive, uses machine vision detection to replace manual inspection, and improves the detection efficiency of books; at the same time, with the help of a double screening mechanism and a cost-value analysis model, the invalid maintenance expenditure of low-value books is greatly reduced, the precious literature is protected, and the efficiency of library inventory management is improved and the management cost is reduced. BRIEF DESCRIPTION OF DRAWINGS
[0050] Figure 1 It is a library inventory management method based on big data analysis;
[0051] Figure 2 It is a library inventory management system framework based on big data analysis. DETAILED DESCRIPTION
[0052] The following description is used to disclose the present application so that those skilled in the art can implement the present application. The preferred embodiments in the following description are only as examples, and other obvious modifications can be thought of by those skilled in the art.
[0053] Referring to Figure 1 The library inventory management method based on big data analysis comprises the following steps:
[0054] Step one, based on the library inventory database, obtaining the basic information of each book in the library, initializing the ideal maintenance cycle time limit of each book in the library according to the book type corresponding to each book basic information;
[0055] The step one comprises the following contents:
[0056] Based on the Internet, obtaining the audience group big data corresponding to the known book category;
[0057] Using PAC principal component analysis, the audience group big data corresponding to the known book category is subjected to data dimension reduction, and the audience group dimension reduction big data corresponding to the known book category is obtained;
[0058] According to the audience group dimension reduction big data corresponding to the known book category, using K-means clustering, presetting a plurality of clustering initial clusters, dividing and iterating the audience group dimension reduction big data corresponding to the book to generate the audience group preference graph corresponding to the known book category; the audience group dimension reduction big data comprises: borrowing age, borrowing gender, borrowing book category
[0059] Based on the known book category corresponding to the audience group preference graph, the audience group age distribution preference matrix A of the known book category is obtained by dividing the audience group of the known book category according to the age, and the audience group age distribution preference matrix A of the known book category is established.
[0060]
[0061] Wherein, a ij is the jth audience group age distribution preference matrix of the known i book category, m is the total number of book categories, and n is the total number of audience group age distribution preference matrices;
[0062] -----
[0063] Based on the known book category corresponding to the audience group preference graph, the audience group age distribution preference matrix A of the known book category is obtained by dividing the audience group of the known book category according to the age, and the audience group age distribution preference matrix A of the known book category is established.
[0064] As further content, when the audience group age distribution borrowing times of the book is less than the average value, the audience group borrowing low frequency book of the book category is given as the average frequency, so as to avoid the maintenance period of individual book to deviate from the actual maintenance time limit;
[0065] Based on the known book category corresponding to the audience group preference graph, the audience group age distribution preference matrix A of the known book category is obtained by dividing the audience group of the known book category according to the age, and the audience group age distribution preference matrix A of the known book category is established.
[0066] As further content, since the library borrowing book cannot be monitored in real time, but based on the known book category corresponding to the audience group big data, the age distribution is given to the preference book category basic maintenance period limit of different age groups according to the experience method, so as to ensure that the book is matched with the use intensity and the maintenance period, and is maintained in time.
[0067] Based on ARMA time moving autoregressive model, the audience group borrowing frequency time sequence characteristic data of the known book category is used as input, and the audience group borrowing heat index of the known book category is used as output;
[0068] The known book category audience group borrowing demand index is used to compensate the basic maintenance period limit of the library inventory book, and the ideal maintenance period limit of each book of the library is initialized;
[0069] In use, the contents in the above steps are combined,
[0070] As a further content, the current library inventory management method adopts a static fixed maintenance cycle and ignores the actual use difference of books, sets a strategy only based on physical properties, lacks correlation analysis of audience group characteristics (such as age and borrowing times), accumulates borrowing data but stays in shallow statistics without mining time sequence characteristics and group preferences, and relies on manual rules to cause subjective and inefficient decision-making, which finally leads to resource mismatch of insufficient maintenance of popular books and excessive maintenance of unpopular books.
[0071] The scheme constructs an audience group preference graph through PAC dimension reduction and K-means clustering, and realizes dynamic maintenance cycle optimization based on reader age characteristics and borrowing behavior patterns by combining an ARMA time sequence prediction model, which not only avoids resource waste of the traditional fixed maintenance cycle strategy, but also predicts borrowing peaks in advance and adjusts maintenance plans, quantifies the influence difference of different reader groups on book wear and tear, ensures timely maintenance of high-frequency borrowed books (such as children's picture books), reduces the invalid maintenance cost of low-frequency books (such as academic monographs), and significantly improves management efficiency and resource utilization compared with the traditional method.
[0072] Step two, based on the library borrowing database, obtain the historical borrowing information and external influencing factors of each book, analyze the borrowing heat trend of each book in the historical borrowing information, and generate a library book borrowing heat trend index;
[0073] The step two includes the following contents:
[0074] Based on the library borrowing database, obtain the historical borrowing information of each book, take the borrowing unit time as the observation window, and take the historical borrowing information of each book as the observation attribute to obtain the historical borrowing time sequence data of each book in the library;
[0075] Based on the historical borrowing time sequence data of each book in the library, substitute the ARMA time moving autoregressive model to generate the audience group borrowing heat index of each historical book in the library, and determine the audience group borrowing heat index time sequence of each historical book in the library
[0076] Using SLT decomposition, the audience group borrowing heat index time sequence of each historical book in the library is decomposed to obtain the audience group borrowing heat multi-dimensional index of each historical book in the library; the audience group borrowing heat multi-dimensional index of each historical book in the library includes: the audience group borrowing heat trend index, the audience group borrowing heat seasonal index and the audience group borrowing heat residual index of each historical book in the library;
[0077] Mark the external factors of the multi-dimensional audience borrowing heat index of the historical books of the library according to the external factor object of the multi-dimensional audience borrowing heat index of the historical books of the library;
[0078] The external factor object of the multi-dimensional audience borrowing heat index of the historical books of the library is normalized;
[0079] According to the entropy weight method, verify the proportion of the external factor object of the multi-dimensional audience borrowing heat index of the historical books of the library in the total under the unit time, and give the external factor object of the multi-dimensional audience borrowing heat index of the historical books of the library a weight;
[0080] Based on the weight of the external factor object of the multi-dimensional audience borrowing heat index of the historical books of the library, the external factor object of the multi-dimensional audience borrowing heat index of the historical books of the library is recorded, and the environmental influence time-varying vector of the historical books of the library is calculated.
[0081] ----
[0082] Based on the library borrowing database, the real-time historical borrowing information of each book of the library is obtained;
[0083] Based on the Internet, the real-time network book discussion hot topic is obtained, and the real-time network book hot feature data is extracted by using the NLP natural language analysis model;
[0084] As a further content, since the daily borrowing quantity of the book under the influence of linear factors is usually in a predictable normal distribution range under high latitude observation, however, when the book is filmed or faces the recommendation of network celebrities, the exposure of the book will increase exponentially, resulting in a sharp increase in the borrowing quantity of the offline book, therefore, by focusing on the real-time network book discussion hot topic, the environmental influence time-varying vector of the audience borrowing heat of the historical books is updated in real time, so as to avoid the local loss of the borrowing heat trend of each book in the historical borrowing information, resulting in inaccurate book maintenance period;
[0085] Based on the use of the external factor object of the multi-dimensional audience borrowing heat index of the historical books of the library, the historical borrowing information of each book of the library is screened, and the real-time borrowing heat external factor of each book of the library is obtained.
[0086] The difference coefficient between the real-time borrowing heat external factor of each book of the library and the external factor object of the multi-dimensional audience borrowing heat index of the historical books of the library is calculated, and the real-time borrowing heat external factor weight of each book of the library is determined;
[0087] Based on the real-time borrowing heat of each book in the library, the external factor weight is calculated according to the real-time borrowing heat of each book in the library, and the external factor weight is calculated according to the real-time borrowing heat of each book in the library.
[0088] Based on the real-time borrowing heat of each book in the library, the external factor weight is calculated according to the real-time borrowing heat of each book in the library, and the external factor weight is calculated according to the real-time borrowing heat of each book in the library.
[0089]
[0090] Wherein, H i is the borrowing heat trend index of the i-th book in the library, and a is the borrowing trust coefficient of the book, is the borrowing heat environment influence time-varying vector of the j-th audience group of the i-th book in the library at the t-th unit time, is the borrowing heat environment influence time-varying vector of the j-th audience group of the i-th book in the library at the t-th unit time;
[0091] In use, the contents in the above steps are combined,
[0092] As further content, the current library in borrowing heat analysis only relies on the simple statistics of static historical borrowing data and cannot capture the sudden hot events (such as the borrowing surge caused by the film adaptation), completely ignores the external influence factors such as network public opinion, uses the rough moving average method and cannot effectively separate the seasonal, trend and noise components, and the weight distribution of external factors depends on subjective experience and lacks data support, resulting in a serious disconnection between maintenance resource allocation and real borrowing demand.
[0093] The present scheme realizes multi-dimensional accurate prediction of borrowing heat through SLT time sequence decomposition, NLP public opinion analysis and dynamic time-varying modeling, can identify long-term trend and periodic fluctuation, and can capture sudden hot events such as film adaptation in real time; through the entropy weight method, the weight of external factors is adjusted adaptively, the hot books are quickly identified and the maintenance priority is automatically adjusted, and through the ARMA model, the noise interference is effectively filtered, the response speed of traditional method is improved, and the timeliness and accuracy of library resource allocation are significantly improved.
[0094] Step three, according to the library book borrowing heat trend index, the maintenance cycle time limit of the initialized library each book is compensated and fitted, and the actual maintenance cycle time limit of the library each book is generated;
[0095] The step three includes the following contents:
[0096] Based on the Random Forest, according to the audience preference graph corresponding to the known book category, the maintenance period limit of each book in the library is taken as the leaf node, and the borrowing frequency time series feature data set of the audience of the known book category is taken as the root node, a maintenance period limit change decision tree of the known book category is established, and a library book maintenance period limit compensation update random forest model is formed;
[0097] Based on the library book maintenance period limit compensation update random forest model, the library book borrowing heat trend index is taken as the input, and the actual maintenance period limit of each book in the library is taken as the output;
[0098] In use, the contents in the above steps are combined,
[0099] As further content, an intelligent decision model is constructed based on the Random Forest algorithm, and precise dynamic adjustment of the maintenance period is realized by integrating multi-dimensional data features: the complex nonlinear relationship between borrowing heat and maintenance demand can be captured, and various special scenarios (such as short-term surge of examination books and sudden demand for bestsellers) can be automatically identified; at the same time, interpretable decision quantization support is provided to realize adaptive optimization of the maintenance management needs of the book category, and the timeliness and adaptability of the maintenance strategy are significantly improved.
[0100] Step four, based on the edge visual sensor, the library borrowing book return image data is obtained, a book return image damage analysis model is established, the damage state of the target book is evaluated, the actual maintenance period limit of each book in the library is corrected, and the library inventory book maintenance priority is generated;
[0101] The step four includes the following contents:
[0102] Based on the edge visual sensor, multi-angle image data of the library book return is obtained;
[0103] The multi-angle image data of the library book return is preprocessed, the multi-angle image contour area of the library book return is determined according to edge detection, and the multi-angle image contour area is cropped using U-Net++ to obtain damage image feature data of the library book return;
[0104] The damage image feature pixels in the damage image feature data of the library book return are normalized;
[0105] Based on Multi-Task Learning multi-task target learning, a book return image damage analysis model is established, taking the damage score and damage type classification of the library book when it is returned as the objective function, taking the physical consistency constraint of the book tearing inside and outside and the page loss, taking the damage image feature data of the library book when it is returned as the output, and taking the evaluation value corresponding to the damage image of the library book when it is returned into the Sigmoid function to generate the damage score and damage type classification of the library book when it is returned;
[0106] Using NSGA-II multi-objective optimization genetic algorithm, based on the damage type classification of the library book when it is returned, the maintenance cost required for the maintenance action of the target book under the damage type classification is estimated, it is judged whether the maintenance cost exceeds the book value, if yes, it is determined to abandon the maintenance, if not, it is determined to correct the actual maintenance cycle time limit of each book in the library according to the damage score of the library book when it is returned, and the maintenance priority of the library inventory book is determined.
[0107] In use, in combination with the contents in the above steps,
[0108] As further content, the present scheme realizes the library book maintenance management through the deep integration of intelligent visual detection and multi-objective decision system: advanced image analysis technology is used to quickly identify various types of explicit and implicit damage, a three-dimensional evaluation system is established to comprehensively consider the book damage condition, repair cost and literature value, a dynamic priority algorithm is used to realize the immediate response to the emergency situation, and the predictive maintenance capability can identify the high-wear book category in advance, so as to generate a precise and economic and efficient book maintenance management system, which significantly improves the book protection effect and reduces the maintenance cost.
[0109] Referring to Figure 2 As shown in the figure, a library inventory management system based on big data analysis includes:
[0110] An initial maintenance cycle module, a book borrowing popularity module, a maintenance cycle correction module and a maintenance priority generation module;
[0111] The initial maintenance cycle module is used to obtain the basic information of each book in the library based on the library inventory database, and initialize the ideal maintenance cycle time limit of each book in the library according to the book type corresponding to each book basic information;
[0112] The book borrowing popularity module is used to obtain the historical borrowing information and external influencing factors of each book based on the library borrowing database, analyze the borrowing popularity trend of each book in the historical borrowing information, and generate a library book borrowing popularity trend index;
[0113] The maintenance cycle correction module is electrically connected with the initial maintenance cycle module and the initial maintenance cycle module, and is used for compensating and fitting the maintenance cycle time limit of each book of the library according to the library book borrowing heat trend index, and generating the actual maintenance cycle time limit of each book of the library.
[0114] The maintenance priority generation module is electrically connected with the maintenance cycle correction module, and is used for acquiring the library book return image data based on the edge visual sensor, establishing a book return image loss analysis model, evaluating the damage state of the target book, correcting the actual maintenance cycle time limit of each book of the library, and generating the maintenance priority of the library inventory book.
[0115] The basic principle, main features and advantages of the present application are shown and described above. It should be understood by those skilled in the art that the present application is not limited by the above examples, and the above examples and descriptions in the specification are only the principles of the present application. Without departing from the spirit and scope of the present application, various changes and improvements can be made to the present application, and these changes and improvements all fall within the scope of the claimed present application. The scope of protection required by the present application is defined by the appended claims and their equivalents.
Claims
1. A library inventory management method based on big data analysis, characterized by, Comprise: S1, based on the library inventory database, obtain the basic information of each book in the library, initialize the ideal maintenance cycle time limit of each book in the library according to the book type corresponding to each book basic information; S2, based on the library lending database, obtain the historical lending information and external influencing factors of each book, analyze the lending trend of each book in the historical lending information, and generate the lending trend index of the library books; S3, according to the library book lending trend index, the maintenance cycle time limit of the initialized library is compensated and fitted, and the actual maintenance cycle time limit of each book in the library is generated; S4, based on the edge visual sensor, the library lending book return image data is obtained, the book return image loss analysis model is established, the target book damage state is evaluated, the actual maintenance cycle time limit of each book in the library is corrected, and the library inventory book maintenance priority is generated.
2. The library inventory management method based on big data analysis according to claim 1, wherein, The S1 comprises: Based on the Internet, obtain the audience group big data corresponding to the known book category; Using PAC principal component analysis, the audience group big data corresponding to the known book category is subjected to data dimension reduction to obtain the audience group dimension reduction big data corresponding to the known book category; According to the audience group dimension reduction big data corresponding to the known book category, K-means clustering is used, a plurality of clustering initial clusters are preset, the audience group dimension reduction big data corresponding to the book is divided and iterated, and the audience group preference graph corresponding to the known book category is generated; the audience group dimension reduction big data comprises: lending age, lending gender, lending book category Based on the audience group preference graph corresponding to the known book category, the audience group age distribution preference matrix A of the known book category is obtained by dividing the audience group age of the known book category, and the audience group age distribution preference matrix A of the known book category is established; wherein a ij is the age distribution bias array of the jth audience group for the ith book category, m is the total number of book categories, and n is the total number of audience age distribution bias arrays.
3. The library inventory management method based on big data analysis according to claim 2, wherein, The S1 further comprises: Based on the audience group age distribution preference matrix of the known book category, the lending times of the audience group age distribution of the book are counted, and the audience group lending frequency time sequence characteristic data of the known book category is determined; Based on the audience group of the known book category, the basic maintenance cycle time limit of the preset library inventory book is determined; Based on the ARMA time moving autoregressive model, the audience group lending frequency time sequence characteristic data of the known book category is taken as the input, and the audience group lending heat index of the known book category is taken as the output; Using the audience group lending demand index of the known book category, the basic maintenance cycle time limit of the library inventory book is compensated, and the ideal maintenance cycle time limit of each book in the library is initialized.
4. The library inventory management method based on big data analysis according to claim 3, wherein, The S2 comprises: Based on the library lending database, the historical lending information of each book is obtained, the lending unit time is taken as the observation window, and the historical lending information of each book is taken as the observation attribute to obtain the historical lending time sequence data of each book in the library; Based on the historical lending time sequence data of each book in the library, the ARMA time moving autoregressive model is substituted to generate the audience group lending heat index of each historical book in the library, and the audience group lending heat index time sequence of each historical book in the library is determined The SLT decomposition is used to decompose the audience group borrowing heat index time series of each historical book of the library, and a multi-dimensional audience group borrowing heat index of each historical book of the library is obtained; the multi-dimensional audience group borrowing heat index of each historical book of the library includes: audience group borrowing heat trend index, audience group borrowing heat seasonal index and audience group borrowing heat residual index of each historical book of the library; Based on the multi-dimensional audience group borrowing heat index of each historical book of the library, the external factors of the change of the multi-dimensional audience group borrowing index of the historical book are marked, and the audience group borrowing heat external factor object of each historical book of the library is recorded; The audience group borrowing heat external factor object of each historical book of the library is normalized; According to the entropy weight method, the proportion of the audience group borrowing heat external factor object of each historical book of the library in the total is verified, and the weight of the audience group borrowing heat external factor object of each historical book of the library is given; Based on the audience group borrowing heat external factor object weight of each historical book of the library, the audience group borrowing heat external factor object of each historical book of the library is recorded, and the environmental influence time-varying vector of the audience group borrowing heat of each historical book of the library is calculated.
5. The library inventory management method based on big data analysis according to claim 4, wherein, The S2 further includes: Based on the library borrowing database, the historical borrowing information of each book of the library is obtained; Based on the Internet, the real-time network book discussion hot topic is obtained, and the real-time network book hot feature data is extracted by using the NLP natural language analysis model; Based on the audience group borrowing heat external factor object of each historical book of the library, the historical borrowing information of each book of the library is screened, and the real-time borrowing heat external factor of each book of the library is obtained; The difference coefficient between the real-time borrowing heat external factor of each book of the library and the audience group borrowing heat external factor object of each historical book of the library is calculated, and the real-time borrowing heat external factor weight of each book of the library is determined; Based on the real-time borrowing heat external factor weight of each book of the library and the real-time borrowing heat external factor of each book of the library, the audience group borrowing heat environmental influence time-varying vector of each book of the library is calculated; Based on the audience group borrowing heat environmental influence time-varying vector of each historical book of the library and the audience group borrowing heat environmental influence time-varying vector of each book of the library, the library book borrowing heat trend index is calculated, and the method is as follows: wherein H i is the library's i-th book borrowing heat trend index, and a is the borrowing trust coefficient of the book, is the historical i-th book's j-th audience borrowing heat environment influence time-varying vector of the library at the t-th unit time, is the real-time i-th book's j-th audience borrowing heat environment influence time-varying vector of the library at the t-th unit time.
6. The library inventory management method based on big data analysis according to claim 5, wherein, The S3 includes: Based on the Random Forest, the audience group preference graph corresponding to the known book category is used as the root node, the maintenance cycle time limit of each book of the library is used as the leaf node, and the maintenance cycle time limit change decision tree of the known book category is established, and the library book maintenance cycle time limit compensation update random forest model is established. The library book maintenance cycle time limit compensation update random forest model is based on the library book borrowing heat trend index as input, and the actual maintenance cycle time limit of each book in the library as output.
7. The library inventory management method based on big data analysis according to claim 6, wherein, The S4 includes: Based on the edge visual sensor, the multi-angle image data of the library book return is obtained; The multi-angle image data of the library book return is preprocessed, the multi-angle image contour area of the library book return is determined according to edge detection, and the multi-angle image contour area is cropped by using U-Net++, to obtain the damage image feature data of the library book return; The damage image feature pixels in the damage image feature data of the library book return are normalized; Based on Multi-Task Learning multi-task target learning, a book return image loss analysis model is established, the damage score and damage type classification of the library book return are taken as the objective function, the internal and external tearing and page loss of the book are taken as the physical consistency constraint, the damage image feature data of the library book return is taken as the output, and the evaluation value corresponding to the damage image of the library book return is substituted into the Sigmoid function to generate the damage score and damage type classification of the library book return; Using NSGA-II multi-objective optimization genetic algorithm, based on the damage type classification of the library book return, the maintenance cost required for the maintenance action of the target book under the damage type classification is estimated, it is judged whether the maintenance cost exceeds the book value, if yes, it is judged to abandon the maintenance, if not, it is judged to correct the actual maintenance cycle time limit of each book in the library according to the damage score of the library book return, and the maintenance priority of the library inventory book is determined.
8. A library inventory management system based on big data analysis, characterized by, For realizing the library inventory management method based on big data analysis as claimed in any one of claims 1-7, comprising: initial maintenance cycle module, book borrowing heat module, maintenance cycle correction module and maintenance priority generation module; The initial maintenance cycle module is used for obtaining the basic information of each book in the library based on the library inventory database, initializing the ideal maintenance cycle time limit of each book in the library according to the book type corresponding to each book basic information; The book borrowing heat module is used for obtaining the historical borrowing information and external influencing factors of each book based on the library borrowing database, analyzing the borrowing heat trend of each book in the historical borrowing information, and generating the library book borrowing heat trend index; The maintenance cycle correction module is electrically connected with the initial maintenance cycle module and the initial maintenance cycle module, and is used for compensating and fitting the maintenance cycle time limit of each book in the library according to the library book borrowing heat trend index, to generate the actual maintenance cycle time limit of each book in the library. The maintenance priority generation module is electrically connected with the maintenance cycle correction module. The maintenance priority generation module is used for obtaining the book return image data of the library lending books based on the edge visual sensor, establishing a book return image loss analysis model, evaluating the damage state of the target books, correcting the actual maintenance cycle time limit of each book of the library, and generating the maintenance priority of the library inventory books.