Goods checking device for warehouse management and use method thereof

Abstract

The invention relates to the technical field of goods checking devices, in particular to a goods checking device for warehouse management and a using method thereof.The goods checking device for warehouse management comprises a warehouse, multiple sets of goods shelves, a fixing plate, a movable adjusting assembly, an intelligent checking unit and a warehouse management system; a bar code scanning gun and an RFID reader are fixedly connected to the outer portion of the fixing plate, and the movable adjusting assembly is arranged on the outer portion of the goods shelf and used for adjusting the position of the fixing plate. The intelligent inventory unit is used for dynamically optimizing the inventory efficiency and accuracy of the goods; the warehouse management system is used for storing data of goods, the bar code scanning gun and the RFID reader are driven to move through the moving adjusting assembly, checking recording of the goods is achieved, through control setting of the intelligent checking unit, the checking efficiency and accuracy of the goods are dynamically optimized, and the checking efficiency is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of goods inventory device, and in particular to a goods inventory device for warehouse management and a use method thereof. BACKGROUND

[0002] Under the rapid development of modern logistics industry, as the core link in the logistics industry chain, the management level of warehouse directly affects the operation efficiency and economic benefit of the whole logistics system. As a key component of warehouse management, goods inventory is an important means to ensure that accounts and goods are consistent, master inventory dynamics and optimize inventory structure.

[0003] At present, common goods inventory is mainly manual inventory, which relies on manual counting and recording by workers. This method not only consumes time and effort, but also is easily affected by human factors such as worker fatigue and negligence during the inventory process, resulting in a high inventory error rate. In particular, in large-scale warehouse environments with large quantities and various types of goods, the low efficiency of manual inventory is more prominent. SUMMARY

[0004] In view of the technical problem of high inventory error rate in the prior art, the present application provides a goods inventory device for warehouse management and a use method thereof.

[0005] The technical solution adopted by the present application is: a goods inventory device for warehouse management, characterized in that it comprises

[0006] a warehouse;

[0007] a plurality of shelves, each arranged in the warehouse;

[0008] a fixed plate, the outside of which is fixedly connected with a bar code scanner and an RFID reader;

[0009] a movement adjusting assembly arranged outside the shelf and used to adjust the position of the fixed plate;

[0010] an intelligent inventory unit used to dynamically optimize the inventory efficiency and accuracy of goods;

[0011] a warehouse management system used to store data of goods.

[0012] In one embodiment, the movement adjusting assembly comprises a first fixed frame and a second fixed frame, the first fixed frame is fixedly connected to the two sides of the shelf, a first threaded rod is rotatably connected in the first fixed frame, a first motor is fixedly connected to the top of the first fixed frame, and the output end of the first motor is fixedly connected with the first threaded rod.

[0013] The second fixed frame is arranged outside the first fixed frame, a second threaded rod is rotationally connected in the second fixed frame, a nut block is threadedly connected outside the first threaded rod and the second threaded rod, the nut block outside the first threaded rod is fixedly connected with a connecting block, the connecting block is fixedly connected with a connecting column outside, and the connecting column is fixedly connected with the second fixed frame; and the nut block outside the second threaded rod is fixedly connected with a fixed plate.

[0014] The intelligent inventory unit comprises a data acquisition and processing module, a goods classification and coding optimization module, an inventory area division and path planning module, an inventory device configuration and calibration module, a dynamic inventory cycle determination module, an inventory data real-time acquisition and transmission module, an inventory data verification and anomaly detection module, an anomaly data processing module, an inventory result statistical analysis module, and an inventory data updating module.

[0015] In one embodiment, the data acquisition and processing module acquires and pre-processes relevant data in the warehouse management system before goods inventory is conducted.

[0016] The data acquisition range includes the basic information of goods, inventory account information, and warehouse environment information.

[0017] The data acquisition method comprises: obtaining the inventory account information and the basic information of goods through the interface of the warehouse management system; and collecting the warehouse environment information by using a sensor network deployed in the warehouse.

[0018] The data preprocessing method is as follows:

[0019] Data cleaning: removing redundant data, error data, and data with excessive missing values; for data with fewer missing values, the mean filling method is used for processing, and the formula is: xfilled = 1n i=1 xi, wherein xfilled is the filled missing value, n is the number of valid data in the sample set of the data, and xi is the valid data in the sample set.

[0020] For inventory data with time series characteristics, the time series weighted filling formula is calculated as follows:

[0021] xfilled, t = t - kt + kwi x i i = t - kt + kwi;

[0022] Wherein xfilled, t is the filling value at time t, k is the time window size, wi = e-λi-t is the time weight, λ is the decay coefficient, and xi is the valid data at time i.

[0023] Data standardization: convert data of different magnitudes to a uniform magnitude, use Z-score standardization method, formula: xnorm=x-μσ, where xnorm is the standardized data, x is the original data, μ is the mean of the original data, σp is the standard deviation of the pth attribute, σ is the standard deviation of the original data;

[0024] Also includes data outlier correction, the specific calculation formula is as follows:

[0025] For the normalized xnorm>3 of abnormal value, using the truncated correction:

[0026] xcorr=μ+3σ,x>μ+3σμ-3σ,x<μ-3σx,other;

[0027] Where, xcorr is the corrected data.

[0028] In one embodiment, the goods classification and coding optimization module classifies the goods in the warehouse and optimizes the adjustment of the goods coding method;

[0029] Wherein, the goods classification and coding adjustment optimization method is different between the attributes of goods in different categories:

[0030] Goods classification: based on the attributes of goods, hierarchical clustering algorithm is used to classify goods, and the objective function of the algorithm is: J=k=1Kx∈Ckx-μk2, where J is the objective function value, K is the number of clustering categories, Ck is the kth clustering cluster, x is the sample data in the clustering cluster Ck, and μk is the mean vector of the clustering cluster Ck;

[0031] By minimizing the objective function J, the goods are divided into K different categories;

[0032] Also includes classification weight adjustment, the specific formula is as follows: based on the inventory fluctuation characteristics of goods, dynamic weight is given to the attribute vector, and the modified objective function is J'=k=1Kx∈Ckm=1Mwm×xm-μk,m2, where wm=σmp=1Mσp is the weight of the mth attribute, σm is the standard deviation of the attribute, σp is the standard deviation of the pth attribute, M is the total number of attributes, and μk,m is the mean of the kth attribute;

[0033] Coding adjustment optimization: based on the existing bar code or RFID coding, classification identification and location identification are added;

[0034] The new coding structure is: Code=P+T+S+N, where P is the warehouse partition identification, and is 2 bits, T is the goods classification identification, and is 3 bits, S is the shelf location identification, and is 4 bits, and N is the unique serial number of goods, and is 5 bits;

[0035] Through the coding, the storage location of the goods and the belonging category can be located;

[0036] The check bit is also coded, and the specific calculation formula is as follows: a check bit C is added at the end of the code, and the calculation formula is C=(i=114ai*2imod 2)mod 10, wherein ai is the numerical value of the 14 bits before coding, the check bit is used to detect the input error of the code, wherein imod 2 is the modulo operation of i to 2, the result is 0 or 1, which is used to determine the exponent of 2; mod 10 is used to take the modulo 10 of the result of the summation, and the result is the value of the check bit C.

[0037] In one embodiment, the inventory area division and path planning module divides the inventory area and plans the optimal inventory path, specifically including inventory area division and path planning.

[0038] Inventory area division: according to the layout of the warehouse, the distribution of the shelves and the classification result of the goods, the warehouse is divided into several independent inventory areas, and the division principle is that the number of goods in each area is balanced.

[0039] The greedy algorithm is used for area division, and the target is to minimize the total number of goods in each area, and the formula is: min sigma

[0040] 2=1M-1MQm-Q2, wherein sigma 2 is the variance of the number of goods in each area, M is the number of inventory areas, Qm is the number of goods in the mth area, and Q is the average value of the number of goods in all areas.

[0041] Set the area load balancing index: L=max Qm min Qm, and L<=1.2.

[0042] Path planning: for each inventory area, the Dijkstra algorithm is used to plan the optimal inventory path from the entrance to the exit, so that the walking distance of the inspector in the area is the shortest.

[0043] The Dijkstra algorithm calculates the shortest path from the starting point to each node, and the formula is: ds,v=min{ds,u+wu,v}, wherein ds,v is the shortest path length from the starting point s to node v, ds,u is the shortest path length from the starting point s to node u, and wu,v is the weight of the edge from node u to node v.

[0044] Set the fatigue coefficient correction path length formula: w′u,v=wu,v*1+alpha*t, wherein w′u,v is the corrected weight, alpha is the fatigue coefficient, and t is the cumulative time after the start of the inventory; by increasing the path weight with time, the path planning is adapted to the fatigue characteristics of the personnel, and the early inventory path is preferentially planned to cover nodes with longer distances.

[0045] In one embodiment, the inventory device configuration and calibration module calibrates the inventory device, in particular as follows:

[0046] Device configuration: according to the encoding type of the goods and the inventory requirement, configure the corresponding inventory device, including barcode scanner, RFID reader; for goods using optimized encoding, configure intelligent inventory terminal with wireless communication function, which can transmit inventory data to the warehouse management system in real time;

[0047] Device calibration:

[0048] Barcode scanner calibration: calibration by scanning standard barcode card, calculate the scanning error rate, formula: γ = NerrorNtotal x 100%, where γ is the scanning error rate, Nerror is the number of scanning errors, Ntotal is the total number of scanned barcodes; when γ exceeds 1%, the scanner needs to be adjusted or replaced;

[0049] Set dynamic scanning threshold calibration formula: according to the barcode printing quality score Q, correct the acceptable error rate threshold: γth = 1% + 100 - Q x 0.01%;

[0050] RFID reader calibration: by reading standard RFID tags, draw the reader's reading range curve, and calculate the reading sensitivity;

[0051] Reading sensitivity formula: S = 10lgPrPt, where S is the reading sensitivity, Pr is the signal power received by the reader, Pt is the signal power emitted by the tag;

[0052] Based on the environmental interference compensation formula: S' = S + 10lg1 + β x ρ, where S' is the compensated sensitivity, β is the interference coefficient, ρ is the density of the interference source, and the compensation compensates for the influence of environmental interference on sensitivity.

[0053] In one embodiment, the dynamic inventory cycle determination module determines the dynamic inventory cycle of different goods according to the turnover rate and importance of the goods;

[0054] Calculation of goods turnover rate: the turnover rate of goods refers to the ratio of the total quantity of goods out of the warehouse to the average inventory in a certain period, formula: T = QoutQstock, where T is the turnover rate of goods, Qout is the total quantity of goods out of the warehouse in a certain period, Qstock is the average inventory in that period, Qstock = Qstart + Qend2, Qstart is the initial inventory, Qend is the final inventory;

[0055] In addition, a weighted turnover rate formula is set: Tw=i=1nwi×Qout,iQstock, wherein wi=inn+1 / 2 is a time weight, and n is a number of statistical periods.

[0056] Dynamic inventory cycle calculation: the ABC classification method is combined with the turnover rate of goods to determine the inventory cycle.

[0057] For A-class goods, the inventory cycle is short; for C-class goods, the inventory cycle is long.

[0058] The calculation formula is: Ci=KTi×Vi, wherein Ci is the inventory cycle of the ith kind of goods, K is a constant, Ti is the turnover rate of the ith kind of goods, and Vi is the value coefficient of the ith kind of goods.

[0059] It also includes inventory fluctuation risk factor correction, and the inventory fluctuation risk factor correction formula is calculated as follows: Ci′=Ci×1-θ×Ri, wherein Ri=σstock,iQstock,i is an inventory fluctuation coefficient, and θ is a risk sensitivity coefficient.

[0060] In one of the embodiments, the inventory data real-time acquisition and transmission module acquires goods information in real time by using the configured inventory device, and transmits the data to the warehouse management system through wireless communication, and the specific method is as follows:

[0061] Data acquisition: the inventory device includes a barcode scanner and an RFID reader, and the code information is read by the barcode scanner, and for RFID goods, the tag information is automatically identified by the RFID reader;

[0062] The barcode scanner and the RFID reader automatically record the code, quantity, inventory time and other data of the goods;

[0063] It also includes multi-source data fusion, and the multi-source data fusion calculation formula is as follows: for goods using both barcodes and RFID, the quantity is a weighted fusion value Qfuse=α×Qbarcode+1-α×QRFID, wherein α is a credibility weight;

[0064] Data transmission: wireless sensor network is used for data transmission, and the data transmission rate formula is: R=Blog21+SN, wherein R is the data transmission rate, B is the channel bandwidth, S is the signal power of the receiving end, and N is the noise power;

[0065] It also includes dynamic transmission priority calculation, and the dynamic transmission priority calculation formula is as follows: Pi=Vi×Ti, which assigns transmission priorities to different categories of goods data, and uses a priority scheduling mechanism.

[0066] In one of the embodiments, the point data verification and anomaly detection module verifies the inventory data transmitted to the warehouse management system, detects and identifies abnormal data in the inventory process;

[0067] The consistency verification includes: comparing the inventory data with the account data in the warehouse management system, checking whether the information such as the goods code and the quantity is consistent or not;

[0068] The consistency verification formula is: ΔQi = Qinventory,i - Qbook,i, wherein ΔQi is the difference between the inventory quantity and the account quantity of the i-th goods, Qinventory,i is the inventory quantity, and Qbook,i is the account quantity;

[0069] When ΔQi = 0, the data is consistent; when ΔQi ≠ 0, the data is different;

[0070] The difference tolerance calculation also includes: the calculation formula is as follows: ΔQth,i = max(1, β × Qbook,i), wherein β is a tolerance coefficient (0.02 is taken), when ΔQi ≤ ΔQth,i, it is considered as an acceptable difference, and unnecessary abnormal processing is reduced;

[0071] The integrity verification includes: checking whether the inventory data is complete or not, whether there is missing goods or information or not;

[0072] The data integrity index is used for evaluation, and the formula is: I = NcompleteNtotal × 100%, wherein I is the data integrity index, Ncomplete is the number of complete inventory data, and Ntotal is the total number of data to be inventoried; when I ≥ 98%, it is considered that the data is complete;

[0073] The anomaly detection includes: the density-based local outlier factor algorithm is used for detecting abnormal data;

[0074] The LOF algorithm judges whether the sample point is an abnormal point by calculating the local outlier factor of the sample point, and the formula is: LOFp = o∈NkpLRDoLRDpNkp, wherein LOFp is the local outlier factor of the sample point p, Nkp is the k-neighborhood set of the sample point p, and LRDp is the local reachable density of the sample point p;

[0075] The anomaly weight correction also includes: the calculation formula is as follows: LOF'p = LOFp × 1 + ω × Vp, wherein Vp is a goods value coefficient, and ω is a weight coefficient, and the detection sensitivity of high-value goods anomaly is improved;

[0076] The abnormal data processing module analyzes and processes the detected abnormal data, and re-inventories the goods with problems;

[0077] Abnormal reason analysis: through the consultation of goods in and out of the warehouse records, warehouse environment data and inventory equipment log, analyze the cause of abnormal data;

[0078] Also includes abnormal reason probability prediction, the calculation formula is as follows: using naive Bayesian model to calculate each reason probability Pc|x=Px|cPcc'Px|c'Pc', wherein x is the abnormal feature vector, c is the abnormal reason category, auxiliary fast positioning reason;

[0079] Abnormal treatment measures: according to the abnormal reason to take corresponding treatment measures, such as replacing inventory equipment, registering and processing damaged goods, correcting in and out of the warehouse records, etc.;

[0080] Recheck operation: recheck the goods with abnormal data, calculate the recheck accuracy, the formula is: α=NcorrectNrecheck×100%, wherein α is the recheck accuracy, Ncorrect is the number of recheck correct goods, Nrecheck is the total number of recheck; When α≥99%, it is considered that the recheck result is reliable;

[0081] Also includes recheck sample size calculation, the calculation formula is as follows: for batch abnormal goods, the minimum recheck sample size n=z2×p×1-pe2, wherein z is the quantile corresponding to the confidence level, p is the estimated abnormal rate, e is the allowable error;

[0082] The inventory result statistical analysis module carries out statistics and analysis on the inventory data, and generates an inventory report;

[0083] Including account material coincidence rate calculation: the calculation formula is: β=Nmatch Ntotal×100%, wherein β is the account material coincidence rate, Nmatch is the number of account material coincidence, Ntotal is the total number of inventory goods;

[0084] Weight account material coincidence rate calculation, the calculation formula is: βw=i=1nVi×IΔQi=0i=1nVi×100%, wherein I· is the indicator function;

[0085] Inventory error rate calculation: the calculation formula is: δ=i=1nΔQii=1nQh ooki×100%, wherein δ is the inventory error rate, n is the number of inventory goods;

[0086] Data analysis: analyze the matching degree of inventory error rate, turnover rate and inventory cycle of different categories of goods, etc.;

[0087] Using correlation analysis method to study the relationship between goods turnover rate and inventory error rate, the correlation coefficient formula is:

[0088] r=i=1nTi-Tδi-δi=1nTi-T2i=1nδi-δ2;

[0089] Wherein r is a correlation coefficient, Ti is the turnover rate of the i-th goods, T is the average value of the turnover rate, δi is the inventory error rate of the i-th goods, and δ is the average value of the inventory error rate.

[0090] The inventory efficiency regression model is established: a multiple regression equation t=a+bρ+cT+∈ is established for the inventory time t and the goods density ρ and the turnover rate T, wherein a is a constant term, b and c are regression coefficients, and ∈ is an error term, and the contribution degree of the influencing factors to the efficiency is quantified.

[0091] The inventory data updating module updates the inventory data in the warehouse management system according to the inventory result, and continuously optimizes the inventory method and system;

[0092] Data updating: the inventory data after verification and processing is updated to the warehouse management system, so that the inventory data in the system is consistent with the actual inventory;

[0093] The updating formula is: Qnew,i=Qinventory,i, wherein Qnew,i is the updated inventory of the i-th goods;

[0094] Smooth updating calculation, the calculation formula is as follows: for the inventory data with violent fluctuations, exponential smoothing is used to update Qnew,i=α×Qinventory,i+1-α×Qold,i, wherein α is a smoothing coefficient;

[0095] System optimization: according to the inventory result analysis and feedback information, the inventory area division, path planning and inventory cycle are optimized and adjusted.

[0096] The beneficial effects of the present application are:

[0097] I. Compared with the prior art, in the present application, the barcode scanner and the RFID reader are moved by the mobile adjusting assembly to realize the inventory recording of the goods, and the dynamic optimization of the inventory efficiency and accuracy of the goods is realized through the control setting of the intelligent inventory unit, thereby improving the inventory efficiency.

[0098] Compared with the prior art, in the application, the classification weight adjustment is performed by dynamically assigning attribute weights, amplifying the influence of key attributes such as inventory fluctuation, improving the matching degree of the goods classification result and the actual inventory demand, greatly improving the inventory efficiency of goods in the same category, and reducing the inventory confusion caused by the mixed classification of goods with different attributes. The encoding check bit formula quickly detects encoding errors with simple mathematical operations, reduces the encoding error rate caused by manual input or scanning, and reduces subsequent data inconsistency problems. The newly added formula of inventory area division and path planning optimizes resource allocation. The formula of the device calibration link improves the adaptability of the device. The dynamic scanning threshold calibration formula adjusts the error threshold according to the quality of the barcode, improves the effective scanning rate of low-quality barcodes, and reduces unnecessary device adjustment. The environmental interference compensation is to compensate the RFID signal for interference sources such as metal and liquid, and to ensure the reading accuracy in complex environments. BRIEF DESCRIPTION OF DRAWINGS

[0099] Figure 1 is a structural schematic diagram of a warehouse in the application;

[0100] Figure 2 is a structural schematic diagram of a shelf in the application;

[0101] Figure 3 is a structural schematic diagram of a shelf in the application;

[0102] Figure 4 is a structural schematic diagram of a shelf in the application; Figure 3 is a structural schematic diagram of a shelf in the application;

[0103] In the figure, 1 is a warehouse, 2 is a shelf, 3 is a first fixed frame, 4 is a first motor, 5 is a first threaded rod, 6 is a connecting block, 7 is a second fixed frame, 8 is a connecting column, 9 is a fixed plate, 10 is a second motor, 11 is a second threaded rod, 12 is a nut block, 13 is a barcode scanner, and 14 is an RFID reader. DETAILED DESCRIPTION

[0104] In the description of the application, it should be noted that the terms "front", "upper", "lower", "left", "right", "vertical", "horizontal" and the like indicate the orientation or positional relationship shown in the drawings, and are only used to facilitate the description of the application and simplify the description, and do not indicate or imply that the devices or elements referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as a limitation on the application.

[0105] In the description of the present application, it should be noted that unless otherwise explicitly specified and limited, the terms "mounting", "connection", "connecting" should be understood broadly, for example, it can be fixed connection, or detachable connection, or integral connection; it can be direct connection, or indirect connection through intermediate medium, or internal communication of two elements. For those skilled in the art, the specific meaning of the above terms in the present application can be understood according to the specific circumstances.

[0106] The above and other aspects of the present application will become more apparent by describing in detail the preferred embodiments thereof with reference to the attached drawings in which: Figures 1-4 The present application is further described.

[0107] In order to solve the problems in the background art, the present application proposes the following technical solutions: A goods inventory device for warehouse management, comprising a warehouse 1, shelves 2, a fixed plate 9, a mobile adjusting assembly, an intelligent inventory unit and a warehouse management system, the shelves 2 are provided in groups and are arranged in the warehouse 1, the fixed plate 9 is externally fixedly connected with a bar code scanning gun 13 and an RFID reader 14, the mobile adjusting assembly is arranged outside the shelves 2 and is used for adjusting the position of the fixed plate 9; the intelligent inventory unit is used for dynamically optimizing the inventory efficiency and accuracy of goods; and the warehouse management system is used for storing the data of goods.

[0108] The mobile adjusting assembly comprises a first fixed frame 3 and a second fixed frame 7, the first fixed frame 3 is fixedly connected on both sides of the shelves 2, a first threaded rod 5 is rotatably connected in the first fixed frame 3, a first motor 4 is fixedly connected on the top of the first fixed frame 3, and the output end of the first motor 4 is fixedly connected with the first threaded rod 5;

[0109] The above technical solutions are explained as follows: In the mechanical adjusting aspect, the mobile adjusting assembly is the core of realizing flexible movement of the inventory device. The first fixed frame 3 fixed on both sides of the shelves 2 is internally provided with the first threaded rod 5, and the output end of the first motor 4 on the top is directly connected with the threaded rod, forming a driving structure. When the first motor 4 is started, the output torque drives the first threaded rod 5 to rotate, which is converted into linear motion by the screw transmission principle, driving the fixed plate 9 outside the second fixed frame 7 to move vertically along the shelves 2, and the second motor 10 drives the second threaded rod 11 to rotate, enabling the fixed plate 9 to move horizontally; so that the bar code scanning gun 13 and the RFID reader 14 can cover different shelf 2 areas, getting rid of the limitations of manual handheld devices and reducing the problem of missing scanning due to position deviation.

[0110] The intelligent inventory unit, as a core control module, realizes efficient inventorying relying on the dynamic optimization algorithm in the foregoing technical solution. Under the driving of the mobile adjustment assembly, the barcode scanner 13 and the RFID reader 14 perform double identification on the goods code. The collected data of the two devices are weighted and fused through a multi-source data fusion formula, and the code error is quickly detected in combination with a code check bit formula to ensure the accuracy of the basic data. At the same time, the intelligent inventory unit automatically plans the scanning priority according to the goods classification information and the dynamic inventorying period, preferentially completes the identification of high-value and high-turnover goods, matches the dynamic transmission priority mechanism, and uploads the data to the warehouse management system in real time.

[0111] The warehouse management system assumes the function of a data hub, compares the real-time data transmitted by the intelligent inventory unit with the pre-stored goods basic information and inventory account after receiving the real-time data. The inventorying difference is identified through a difference tolerance formula and an abnormal weight correction formula, and the abnormal data exceeding the threshold value triggers an early warning. The system also stores the inventorying process data, analyzes the device running state and the inventorying efficiency in combination with an efficiency regression model, and provides data support for the path optimization of the mobile adjustment assembly and the adjustment of the inventorying period.

[0112] In the overall operation process, the mobile adjustment assembly realizes full coverage scanning of the physical space, the intelligent inventory unit provides dynamic optimization supported by algorithms, the warehouse management system completes data storage and deep analysis, and the three work together to realize the automation, precision and intelligence of the inventorying process, effectively improving the inventorying efficiency and accuracy.

[0113] The intelligent inventory unit includes a data acquisition and processing module, a goods classification and code optimization module, an inventorying area division and path planning module, an inventorying device configuration and calibration module, a dynamic inventorying period determination module, an inventorying data real-time acquisition and transmission module, a point data verification and abnormality detection module, an abnormal data processing module, an inventorying result statistical analysis module, and an inventorying data updating module.

[0114] In this embodiment, the data acquisition and processing module acquires and pre-processes the relevant data in the warehouse management system before inventorying the goods.

[0115] The data acquisition range includes the basic information of the goods (such as the name, specification, model, supplier, and production date of the goods), the inventory account information (such as the on-hand inventory quantity, the warehousing time, and the warehousing time), and the warehouse environment information (such as the temperature, humidity, and illumination intensity of the warehouse 1).

[0116] The data acquisition method is to obtain the inventory account information and the basic information of the goods through the warehouse management system (WMS) interface, and to acquire the warehouse environment information by using the sensor network deployed in the warehouse 1, such as collecting the temperature data of the warehouse 1 by using the temperature sensor and collecting the humidity data by using the humidity sensor.

[0117] The data preprocessing method is as follows:

[0118] Data cleaning: removing redundant data, error data and data with too many missing values. For data with less missing values, the mean filling method is used for processing, the formula is: xfilled = 1 / n i=1 xi, wherein xfilled is the filled missing value, n is the number of valid data in the sample set of the data, xi is the valid data in the sample set.

[0119] For inventory data with time series characteristics (such as daily inventory), the time series weighted filling formula is calculated as follows:

[0120] xfilled,t = t-kt+kwi x i = t-kt+kwi;

[0121] Where xfilled,t is the filling value at time t, k is the time window size (usually 3-5), wi = e-λi-t is the time weight (λ is the decay coefficient, take 0.5-1), xi is the valid data at time i. The formula gives higher weight to the neighboring time points, improving the accuracy of time series data missing value filling.

[0122] The time series weighted filling formula gives higher weight to the neighboring time points, solving the problem that the traditional mean filling cannot reflect the time series characteristics of the data. Especially for inventory data with periodic fluctuations, the data outlier correction formula avoids the interference of abnormal data outside the standard deviation on the standardization result, improving the input data quality of subsequent clustering and analysis models.

[0123] Data standardization: converting data of different magnitudes to a unified magnitude, using Z-score standardization method, formula: xnorm = x-μ σ, wherein xnorm is the standardized data, x is the original data, μ is the mean of the original data, σ is the standard deviation of the original data, σp is the standard deviation representing the pth attribute, so as to eliminate the influence of data dimension, make different types of data comparable, and facilitate subsequent data analysis and model calculation.

[0124] It also includes data outlier correction, the specific formula is as follows:

[0125] For abnormal values after standardization xnorm>3, truncated correction is adopted:

[0126] xcorr = μ+3σ, x>μ+3σ μ-3σ, x<μ-3σ x, otherwise, wherein xcorr is the corrected data, avoiding the interference of extreme abnormal values on subsequent analysis.

[0127] The above technical solutions are explained as follows: comprehensive data collection and preprocessing, collecting multiple types of data including goods basis, inventory account and warehouse environment, etc., to ensure information integrity. Data cleaning uses mean filling and time series weighted filling, the latter gives higher weight to neighboring time points to improve the accuracy of filling missing values in time series data, data standardization eliminates the influence of dimension, the newly added abnormal value correction formula truncates extreme values to avoid their interference with subsequent analysis, improves the quality of input data, provides reliable data support for subsequent classification, planning, etc., and reduces the deviation caused by data problems.

[0128] In this embodiment, the goods classification and coding optimization module needs to scientifically classify the goods in the warehouse 1 and optimize the goods coding method to improve the efficiency of inventory.

[0129] The goods classification and coding optimization method is as follows:

[0130] Goods classification: based on the attributes of goods (such as weight, volume, value, turnover rate, etc.), the goods are classified using hierarchical clustering algorithm, and the objective function of the algorithm is: J = k = 1 K x e Ck x - muk 2, where J is the objective function value, K is the number of clusters, Ck is the kth cluster, x is the sample data in the cluster Ck, and muk is the mean vector of the cluster Ck.

[0131] By minimizing the objective function J, the goods are divided into K different categories, so that the similarity of the attributes of the goods in the same category is high, and the difference of the attributes of the goods between different categories is large;

[0132] It also includes classification weight adjustment, and the specific formula is as follows: considering the inventory fluctuation characteristics of goods, dynamic weights are given to the attribute vectors, and the modified objective function is J' = k = 1 K x e Ck m = 1 M wm x xm - muk, m 2, where wm = sm p = 1 M sp is the weight of the mth attribute (sm is the standard deviation of the attribute), M is the total number of attributes, sp is the standard deviation of the pth attribute, and muk, m is the mean of the kth attribute; by amplifying the weight of the attribute with large fluctuation, the sensitivity of classification to key attributes is improved.

[0133] Coding optimization: based on the existing bar code or RFID coding, classification identification and location identification are added;

[0134] The new coding structure is: Code = P + T + S + N, where P is the warehouse 1 partition identification (2 bits), T is the goods classification identification (3 bits), S is the shelf 2 location identification (4 bits), and N is the unique serial number of the goods (5 bits).

[0135] Through this coding, the storage location and the category to which the goods belong can be quickly located, and the recognition efficiency in the inventory process is improved.

[0136] The encoding check bit is also included, and the specific calculation formula is as follows: a check bit C is added at the end of the encoding, and the calculation formula is C=(i=114ai×2imod 2)mod10, wherein ai is the numerical value of the 14 bits before encoding, the check bit can quickly detect the encoding input error, and the error rate of manual input or scanning is reduced, wherein imod 2 is the modulo operation of i to 2, and the result is 0 or 1, which is used to determine the exponent of 2; mod10 is to take the modulo 10 of the result of the summation, and the result obtained is the value of the check bit C.

[0137] The classification weight adjustment formula amplifies the influence of key attributes such as inventory fluctuation by dynamically assigning attribute weights, so that the matching degree of the classification result of goods and the actual inventory demand is improved, and the inventory efficiency of goods in the same category is improved. The encoding check bit formula realizes fast detection of encoding errors through simple mathematical operations, reduces the encoding error rate caused by manual input or scanning, and reduces the subsequent data inconsistency problems caused by encoding errors.

[0138] The above technical solutions are explained as follows: The classification and encoding optimization of goods improves the inventory efficiency and accuracy. The classification based on the hierarchical clustering algorithm combined with the dynamic weight adjustment formula amplifies the influence of key attributes such as inventory fluctuation, so that the matching degree of classification and inventory demand is improved. The optimized encoding structure contains partition, classification, and location identification, and the new check bit formula quickly detects encoding errors through simple operations, reduces the encoding error rate, reduces the data inconsistency caused by encoding problems, and makes the positioning of goods more accurate.

[0139] In this embodiment, the inventory area division and path planning module reasonably divides the inventory area and plans the optimal inventory path, which can reduce invalid movement in the inventory process and improve the inventory efficiency, including inventory area division and path planning.

[0140] Inventory area division: according to the layout of the warehouse 1, the distribution of the shelves 2, and the classification result of the goods, the warehouse 1 is divided into several independent inventory areas, and the division principle is that the number of goods in each area is roughly balanced, and the distance between the areas is as far as possible to reduce the interference between the areas.

[0141] The greedy algorithm is used for area division, and the goal is to minimize the total number of goods in each area, and the formula is: minσ

[0142] 2=1M-1MQm-Q2, wherein σ2 is the variance of the number of goods in each area, M is the number of inventory areas,

[0143] Qm is the number of goods in the mth area, and Q is the average value of the number of goods in all areas.

[0144] Regional load balancing index: L = maxQmminQm, requires L≤1.2, to ensure that the inventory work load difference in each region is controlled within a reasonable range, avoiding excessive inventory pressure in some areas.

[0145] Path planning: for each inventory area, the Dijkstra algorithm is used to plan the optimal inventory path from the entrance to the exit, so that the walking distance of the inspector in the area is the shortest, because the inspector needs to walk in the warehouse 1 for inspection, and when the equipment is damaged, it can be found at the first time.

[0146] The core of Dijkstra algorithm is to calculate the shortest path from the starting point to each node, and the formula is: ds,v = min{ds,u +wu,v}, where ds,v is the shortest path length from the starting point s to node v, ds,u is the shortest path length from the starting point s to node u, and wu,v is the weight (i.e. distance) of the edge from node u to node v.

[0147] It also includes a fatigue coefficient correction path length formula: w′u,v = wu,v × 1 + α × t, where w′u,v is the corrected weight, α is the fatigue coefficient (0.01 / minute), and t is the cumulative time (minutes) since the start of inventory; by increasing the path weight over time, the path planning is more in line with the fatigue characteristics of personnel, and the early inventory path is preferentially planned to cover nodes with longer distances, because when scanning the barcode scanner, the user needs to walk in the warehouse 1 according to the specified path to observe.

[0148] The regional load balancing index avoids the problem of excessive local load that may occur in traditional variance minimization by controlling the workload difference in each region, and the work intensity of the inspector is more balanced. The fatigue coefficient correction path length formula takes into account the fatigue characteristics of personnel, so that the early inventory preferentially covers nodes with long distances, and the overall inventory path time is reduced.

[0149] The above technical solutions are explained as follows: reasonable division of inventory area and planning of path to reduce invalid movement. The regional division aims to minimize variance, and the new load balancing index controls the workload difference in each region to be low, avoiding excessive local pressure and making the work intensity of the inspector more balanced. The path planning uses the invented Dijkstra algorithm, combined with the fatigue coefficient correction formula, to increase the path weight over time, so that early inventory preferentially covers nodes with long distances, and the overall path time is reduced.

[0150] In this embodiment, the inventory device configuration and calibration module selects appropriate inventory devices and calibrates them to ensure that the devices can accurately obtain goods information during the inventory process.

[0151] Device configuration: according to the encoding type of the goods and the inventory requirements, configure the corresponding inventory devices, such as barcode scanner 13 and RFID reader 14;

[0152] For the goods with optimized coding, a smart inventory terminal with wireless communication function is configured, which can be installed outside the fixed plate 9. The terminal can transmit inventory data to the warehouse management system in real time. The smart inventory terminal configured with the coded goods is an intelligent device integrating data acquisition, processing and transmission. It takes hardware devices as the carrier and is usually connected with barcode scanning guns 13 or RFID readers and other acquisition components externally, which can accurately read the optimized coding information of the goods, including key data such as warehouse 1 partition, goods classification, shelf 2 position and unique serial number. The terminal has a built-in intelligent algorithm module, which can process the raw coding data in real time. The coding check bit formula is used to quickly detect coding input or scanning errors, reducing subsequent data inconsistency problems. At the same time, it is equipped with a wireless communication module, relying on a dynamic transmission priority formula, and classifying the inventory data according to the value and turnover rate of the goods, and transmitting the inventory data of high-value and high-turnover goods to the warehouse management system in priority, to ensure that the key data is uploaded in real time.

[0153] In addition, the smart inventory terminal can also receive the inventory tasks and path planning information issued by the warehouse management system, guide the inventory personnel to work according to the optimal path, and realize intelligent management and control of the inventory process. It is the core node connecting the front-end goods coding acquisition and the back-end system data management, effectively improving the efficiency and accuracy of coded goods inventory.

[0154] Device calibration:

[0155] Barcode scanning gun 13 calibration: calibration is performed by scanning a standard barcode card, and the scanning error rate is calculated, with the formula: γ = NerrorNtotal × 100%, where γ is the scanning error rate, Nerror is the number of scanned error barcodes, and Ntotal is the total number of scanned barcodes. When γ exceeds 1%, the scanning gun needs to be adjusted or replaced.

[0156] Dynamic scanning threshold calibration formula: according to the barcode printing quality score Q (0-100 points), the acceptable error rate threshold is modified: γth = 1% + 100-Q × 0.01%, which appropriately relaxes the error threshold for poor quality barcodes and reduces unnecessary device adjustments.

[0157] RFID reader 14 calibration: by reading standard RFID tags at different distances and angles, the reader's reading range curve is drawn, and the reading sensitivity is calculated. The reading sensitivity formula is: S = 10lgPrPt, where S is the reading sensitivity (dB), Pr is the signal power received by the reader, and Pt is the signal power emitted by the tag.

[0158] Also included is environmental interference compensation calculation, the calculation formula is as follows: S' = S + 10lg1 + β × ρ, wherein S' is the sensitivity after compensation, β is the interference coefficient (metal environment takes 0.8, liquid environment takes 0.5), ρ is the density of interference source (pieces / m 2 ), by compensating the influence of environmental interference on sensitivity, ensure the reading accuracy in different environments.

[0159] The dynamic scanning threshold calibration formula and the environmental interference compensation formula of the device calibration link improve the device adaptability. The dynamic scanning threshold calibration formula dynamically adjusts the error threshold according to the barcode quality, improves the effective scanning rate of low-quality barcodes, and reduces unnecessary device adjustment times. The environmental interference compensation compensates for the influence of metal, liquid and other interference sources on RFID signals.

[0160] The above technical solutions are explained as follows: device configuration and calibration ensure the accuracy and stability of data acquisition. According to the coding type, configure appropriate devices, and realize real-time data transmission of intelligent terminals. The dynamic threshold formula of the barcode scanner 13 calibration adjusts the error threshold according to the barcode quality, and the effective scanning rate of low-quality barcodes is improved. The RFID reader 14 calibration adds an environmental interference compensation formula, which compensates for metal, liquid and other interference sources, and improves the reading accuracy in complex environments.

[0161] In this embodiment, the dynamic inventory cycle determination module determines the dynamic inventory cycle of different goods according to the turnover rate and importance of the goods, avoids unnecessary frequent inventory, and improves the utilization rate of inventory resources.

[0162] Calculation of goods turnover rate: the goods turnover rate refers to the ratio of the total quantity of goods out of the warehouse to the average inventory in a certain period, and the formula is: T = Qout Qstock, wherein T is the goods turnover rate, Qout is the total quantity of goods out of the warehouse in a certain period, Qstock is the average inventory in the period, Qstock = Qstart + Qend 2, Qstart is the initial inventory, and Qend is the final inventory.

[0163] Also included is weighted turnover rate calculation, the calculation formula is as follows: considering the fluctuation of the quantity of goods out, Tw=i = 1 nwi × Qout,i Qstock, wherein wi = inn + 1 / 2 is the time weight (assign higher weight to recent goods out), and n is the number of statistical periods.

[0164] Dynamic inventory cycle calculation: ABC classification method combined with goods turnover rate is used to determine the inventory cycle. For A-class goods (high value, high turnover rate), the inventory cycle is shorter; for C-class goods (low value, low turnover rate), the inventory cycle is longer. The calculation formula is: Ci = KTix Vi, where Ci is the inventory cycle of the ith goods (days), K is a constant (set according to the actual situation of warehouse 1, generally take 365), Ti is the turnover rate of the ith goods, Vi is the value coefficient of the ith goods (A-class goods take 1.5, B-class goods take 1.0, C-class goods take 0.5).

[0165] It also includes inventory fluctuation risk factor correction, and the calculation formula is as follows: Ci' = Ci x 1-θ x Ri, where Ri = σstock,i Qstock,i is the inventory fluctuation coefficient σstock,i is the inventory standard deviation, θ is the risk sensitivity coefficient (take 0.3-0.5). For goods with large inventory fluctuation, the inventory out-of-control risk is reduced by shortening the inventory cycle.

[0166] The above technical solutions are explained as follows: dynamic inventory cycle determination realizes accurate allocation of inventory resources, calculates goods turnover through weighted turnover rate formula, gives higher weight to recent data, reduces seasonal goods turnover rate prediction error, combines dynamic inventory cycle formula of ABC classification method, introduces inventory fluctuation risk factor correction, shortens cycle for high fluctuation goods, reduces inventory out-of-control risk, and avoids excessive inventory of low fluctuation goods, so as to improve inventory resource utilization rate.

[0167] In this embodiment, the inventory data real-time acquisition and transmission module acquires goods information in real time during the inventory process by using the configured inventory equipment, and transmits the data to the warehouse management system through wireless communication technology.

[0168] Data acquisition: for bar code goods, the code information is read by a scanning gun; for RFID goods, the tag information is automatically identified by an RFID reader 14. The terminal device automatically records the code, quantity, inventory time and other data of the goods.

[0169] Multi-source data fusion formula: for goods using bar code and RFID at the same time, the quantity takes the weighted fusion value Qfuse = α x Qbarcode + 1-α x QRFID;

[0170] Where α is the credibility weight (dynamically adjusted according to the device error rate, α = γRFIDγbarcode + γRFID;

[0171] Data transmission: wireless sensor network (WSN) in the Internet of Things technology is used for data transmission, and the data transmission rate formula is: R = Blog21+ SN, where R is the data transmission rate (bps), B is the channel bandwidth (Hz), S is the signal power of the receiving end, and N is the noise power.

[0172] Dynamic transmission priority formula: different categories of goods data are assigned transmission priority Pi = Vi x Ti, and a priority scheduling mechanism is used to ensure that high-priority data (high-turnover goods of type A) is transmitted first, reducing the delay rate of critical data.

[0173] The above technical solutions are explained as follows: real-time collection and transmission of inventory data ensures timely and accurate data. The intelligent terminal barcode scanner 13 and RFID reader 14 scan and read the goods information according to the planned path, and the multi-source data fusion formula dynamically fuses barcode and RFID data, improving the accuracy of quantity collection and solving the problem of single device data deviation. WSN is used to transmit data, and the dynamic transmission priority formula ensures that high-value and high-turnover goods data are transmitted first, and the average delay of critical data is reduced. Encryption ensures data security, and real-time transmission allows the system to obtain inventory data in a timely manner.

[0174] In this embodiment, the point data verification and anomaly detection module verifies the inventory data transmitted to the warehouse management system, detects and identifies abnormal data in the inventory process.

[0175] Consistency verification: compare the inventory data with the book data in the warehouse management system to check whether the information such as goods code and quantity is consistent.

[0176] The consistency verification formula is: ΔQi = Qinventory, i - Qbook, i, where ΔQi is the difference between the inventory quantity and the book quantity of the i-th goods, Qinventory, i is the inventory quantity, and Qbook, i is the book quantity.

[0177] When ΔQi = 0, the data is consistent; when ΔQi ≠ 0, there is a difference in the data.

[0178] Difference tolerance formula: ΔQth, i = max1, β x Qbook, i, where β is the tolerance coefficient (0.02), and when ΔQi ≤ ΔQth, i, it is considered as acceptable difference, reducing unnecessary abnormal processing;

[0179] Integrity check: Check if the inventory data is complete, whether there is missing goods or information. Adopt data integrity index for evaluation, formula is: I=NcompleteNtotalx100%, wherein I is data integrity index, Ncomplete is the number of complete inventory data, Ntotal is the total number of data to be checked. When I≥98%, it is considered that the data is complete.

[0180] Abnormality detection: adopt density-based local outlier factor (LOF) algorithm to detect abnormal data.

[0181] LOF algorithm judges whether the sample point is an abnormal point by calculating the local outlier factor of the sample point, the formula is:

[0182] LOFp=o∈NkpLRDoLRDpNkp;

[0183] Wherein LOFp is the local outlier factor of sample point p, Nkp is the k-neighbor set of sample point p, LRDp is the local reachable density of sample point p.

[0184] Abnormal weight correction formula: LOF'p=LOFp x 1+ωxVp, wherein Vp is the value coefficient of goods, ω is the weight coefficient (0.5), which improves the detection sensitivity of high-value goods anomaly.

[0185] The above technical solutions are explained as follows: data verification and abnormality detection effectively identify and process problem data, consistency check compares inventory and account data, difference tolerance formula reduces small difference misjudgment and avoids invalid processing. Integrity check ensures data integrity, abnormality detection based on LOF algorithm combined with weight correction formula improves the detection sensitivity of high-value goods anomaly, and the abnormality detection rate of A-class goods is improved. Accurate identification of abnormal data provides clear direction for subsequent processing, reduces missed detection and false detection, reduces the risk of account and material inconsistency caused by abnormal data not processed, and ensures the reliability of inventory results.

[0186] In the embodiment, the abnormal data processing module analyzes and processes the detected abnormal data, and re-inventories the goods with problems.

[0187] Abnormal reason analysis: through checking the warehouse in and out records, warehouse environment data and inventory equipment log, etc., the causes of abnormal data are analyzed, such as inventory equipment failure, goods damage, in and out record error, etc.

[0188] Abnormal reason probability prediction formula: adopt naive Bayes model to calculate each reason probability Pc|x=Px|cPcc'Px|c'Pc', wherein x is abnormal feature vector, c is abnormal reason category, which helps to quickly locate the reason.

[0189] Abnormality treatment measures: take appropriate measures according to the cause of the abnormality, such as replacing the inventory equipment, registering and processing damaged goods, and correcting the warehouse entry and exit records.

[0190] Re-inventory operation: re-inventory the goods with abnormal data (re-inventory), calculate the re-inventory accuracy rate, and the formula is: α = NcorrectNrecheck×100%, where α is the re-inventory accuracy rate, Ncorrect is the number of re-inventory correct goods, and Nrecheck is the total number of re-inventory goods. When α ≥ 99%, it is considered that the re-inventory result is reliable.

[0191] Re-inventory sample size calculation formula: for batch abnormal goods, the minimum re-inventory sample size n = z2×p×1-pe2, where z is the quantile corresponding to the confidence level (1.96 for 95% confidence), p is the estimated abnormal rate, and e is the allowable error (0.05), to ensure the statistical reliability of the re-inventory result.

[0192] The above technical solutions are explained as follows: abnormal data processing and re-inventory to improve the reliability of inventory results. Through multi-dimensional analysis of abnormal reasons, the newly added probability prediction formula helps to quickly locate the problem. For re-inventory of abnormal goods, a sample size calculation formula is used to determine a reasonable sample size to avoid sample size bias caused by insufficient sample size.

[0193] In this embodiment, the inventory result statistical analysis module performs statistical analysis on the inventory data to generate an inventory report, providing a basis for warehouse management decisions.

[0194] Inventory result statistics:

[0195] Account-goods matching rate: the calculation formula is: β = Nmatch Ntotal×100%, where β is the account-goods matching rate, Nmatch is the number of account-goods matching goods, and Ntotal is the total number of inventory goods.

[0196] Weighted account-goods matching rate:

[0197] βw=i=1nVi×IΔQi=0i=1nVi×100%;

[0198] Where I· is an indicator function that highlights the weight of high-value goods matching.

[0199] Inventory error rate: the calculation formula is: δ = i = 1nΔQii = 1nQh ooki × 100%, where δ is the inventory error rate, and n is the number of inventory goods.

[0200] Data analysis: analyze the matching degree of inventory error rate, turnover rate, and inventory cycle of different categories of goods. Use correlation analysis method to study the relationship between goods turnover rate and inventory error rate, and the correlation coefficient formula is:

[0201] r = i = 1 nTi - Tdi - di = 1 nTi - T2i = 1 n di - d2;

[0202] wherein r is a correlation coefficient, Ti is the turnover rate of the i-th kind of goods, T is the average value of the turnover rate, di is the inventory error rate of the i-th kind of goods, and d is the average value of the inventory error rate.

[0203] A regression model of inventory efficiency is established: a multiple regression equation t = a + b p + c T + e is established for the inventory time t, the density p of the goods, and the turnover rate T, wherein a is a constant term, b and c are regression coefficients, and e is an error term, so as to quantify the contribution degree of the influencing factors to the efficiency.

[0204] The above technical solutions are explained as follows: the inventory result statistics and analysis provide a basis for warehouse decision-making. The account-goods coincidence rate and the error rate are counted, and the weighted account-goods coincidence rate highlights the high-value goods, which is more in line with the actual management needs. The correlation analysis studies the relationship between the turnover rate and the error rate, and the newly added efficiency regression model quantifies the contribution degree of the influencing factors. The weak links in the inventory are found out through data analysis, and the optimization direction is clear.

[0205] In this embodiment, the inventory data updating module updates the inventory data in the warehouse management system according to the inventory results, and continuously optimizes the inventory method and system.

[0206] Data updating: the inventory data after verification and processing is updated to the warehouse management system, so as to ensure that the inventory data in the system is consistent with the actual inventory;

[0207] The update formula is: Qnew,i = Qinventory,i, wherein Qnew,i is the inventory quantity of the i-th kind of goods after updating.

[0208] Smooth updating formula: for inventory data with severe fluctuations, the exponential smoothing is used to update Qnew,i = a x Qinventory,i + 1 - a x Qold,i, wherein a is a smoothing coefficient (0.7-0.9), so as to avoid severe fluctuations of the system data.

[0209] System optimization: according to the inventory result analysis and feedback information, the inventory area division, path planning, inventory cycle, etc. are optimized and adjusted. The feedback control algorithm is used to realize the dynamic optimization of the system, and the control deviation formula is: et = Qdesiredt - Qactualt, wherein et is the control deviation at time t, Qdesiredt is the expected inventory quantity at time t (obtained according to historical data and market demand prediction), and Qactualt is the actual inventory quantity at time t (i.e. the inventory quantity after updating).

[0210] Adaptive learning rate optimization formula: control parameter update step size \eta t = \eta 0 \times e ^ (-\kappa \times e t), wherein \eta 0 is an initial learning rate, \kappa is a decay coefficient, the larger the deviation, the smaller the step size, avoiding optimization process shock, improving system stability.

[0211] The above technical solutions are explained as follows: data updating and system optimization realize continuous improvement of inventory. The inventory data after verification is updated to the system, and the smoothing update formula reduces the inventory data fluctuation, avoiding the interference of violent fluctuation on management. The feedback control algorithm is used to dynamically optimize the inventory elements, and the adaptive learning rate formula makes the control parameter update more stable, avoiding optimization shock. The system continuously adjusts the regional division, path planning, etc. according to the feedback, combined with equipment maintenance and upgrading, to ensure that the inventory method adapts to the change of the warehouse environment, realize continuous optimization of the inventory process, and improve the stability and efficiency of the warehouse management system.

[0212] In the application, the classification weight adjustment amplifies the influence of key attributes such as inventory fluctuation by dynamically assigning attribute weights, so that the matching degree of the goods classification result and the actual inventory demand is improved, the inventory efficiency of goods in the same category is greatly improved, and the inventory confusion caused by mixed classification of goods with different attributes is reduced. The encoding check bit formula quickly detects encoding errors with simple mathematical operations, reduces the encoding error rate caused by manual input or scanning, and reduces subsequent data inconsistency problems. The newly added formula of inventory area division and path planning optimizes resource allocation. The formula of the device calibration link improves the adaptability of the device. The dynamic scanning threshold calibration formula adjusts the error threshold according to the quality of the barcode, improves the effective scanning rate of low-quality barcodes, and reduces unnecessary device adjustment. The environmental interference compensation formula compensates for the RFID signal for interference sources such as metals and liquids, ensuring reading accuracy in complex environments.

[0213] The standard parts used in the application can be purchased from the market, and the special-shaped parts can be ordered according to the description and the drawings, and the specific connection mode of each part adopts the conventional means such as bolts, rivets and welding in the prior art, the mechanical parts and equipment adopt conventional models in the prior art, and the circuit connection adopts conventional connection mode in the prior art, which will not be described in detail here, and the contents not described in detail in the specification belong to the prior art known to those skilled in the art.

[0214] Although embodiments of the application have been shown and described, the scope of the application is defined by the appended claims and their equivalents.

Claims

1. A goods inventory counting device for warehouse management, characterized in that, include Warehouse (1); Shelves (2), which are provided in several groups and are all located in the warehouse (1); A fixing plate (9) is fixedly connected to the outside of the fixing plate (9) with a barcode scanner (13) and an RFID reader (14); A movable adjustment component is disposed outside the shelf (2) and is used to adjust the position of the fixed plate (9); Intelligent inventory counting unit is used to dynamically optimize the efficiency and accuracy of inventory counting; A warehouse management system is used to store data about goods.

2. The inventory counting device for warehouse management according to claim 1, characterized in that, in, The movable adjustment component includes a first fixed frame (3) and a second fixed frame (7). The first fixed frame (3) is fixedly connected to both sides of the shelf (2). A first threaded rod (5) is rotatably connected inside the first fixed frame (3). A first motor (4) is fixedly connected to the top of the first fixed frame (3). The output end of the first motor (4) is fixedly connected to the first threaded rod (5). The second fixing frame (7) is located outside the first fixing frame (3). A second threaded rod (11) is rotatably connected inside the second fixing frame (7). Nut blocks (12) are threadedly connected to the outside of both the first threaded rod (5) and the second threaded rod (11). A connecting block (6) is fixedly connected to the nut block (12) outside the first threaded rod (5). A connecting column (8) is fixedly connected to the outside of the connecting block (6). The connecting column (8) is fixedly connected to the second fixing frame (7). A fixing plate (9) is fixedly connected to the nut block (12) outside the second threaded rod (11). The intelligent inventory unit includes a data acquisition and processing module, a goods classification and coding optimization module, an inventory area division and route planning module, an inventory equipment configuration and calibration module, a dynamic inventory cycle determination module, a real-time inventory data acquisition and transmission module, a point data verification and anomaly detection module, an anomaly data processing module, an inventory result statistical analysis module, and an inventory data update module.

3. The inventory counting device for warehouse management according to claim 2, characterized in that, Before conducting inventory checks, the data acquisition and processing module collects and preprocesses relevant data from the warehouse management system. The data collection scope includes: basic information about goods, inventory records, and warehousing environment information; Data acquisition methods: Obtain inventory account information and basic cargo information through the warehouse management system interface; collect warehouse environment information using the sensor network deployed in the warehouse (1); The data preprocessing methods are as follows: Data cleaning: Remove redundant data, erroneous data, and data with too many missing values; for data with few missing values, the mean imputation method is used, with the formula: xfilled = 1 / n * ni = 1 / xi, where xfilled is the missing value after imputation, n is the number of valid data in the sample set, and xi is the number of valid data in the sample set. For inventory data with time-series characteristics, the time-series weighted filling formula is used for calculation, as follows: xfilled,t=i=t-kt+kwi×xii=t-kt+kwi; Where xfilled,t is the filled value at time t, k is the time window size, wi=e-λi-t is the time weight, λ is the decay coefficient, and xi is the effective data at time i; Data standardization: Convert data of different magnitudes into a uniform magnitude using the Z-score standardization method. The formula is: xnorm = x - μσ, where xnorm is the standardized data, x is the original data, μ is the mean of the original data, and σ is the standard deviation of the original data. This also includes data outlier correction, with the specific calculation formula as follows: For outliers with a normalized xnorm > 3, a truncation correction is used: xcorr=μ+3σ,x>μ+3σμ-3σ,x<μ-3σx,others; Where xcorr is the corrected data.

4. The inventory counting device for warehouse management according to claim 3, characterized in that, The goods classification and coding optimization module classifies the goods in the warehouse (1) and adjusts the goods coding method; The method for adjusting the attribute differences of goods between different categories in goods classification and coding is as follows: Goods classification: Goods are classified based on their attributes. A hierarchical clustering algorithm is used to classify goods. The objective function of the algorithm is: J = k = 1 Kx∈Ckx-μk2, where J is the objective function value, K is the number of clusters, Ck is the k-th cluster, x is the sample data in cluster Ck, and μk is the mean vector of cluster Ck. By minimizing the objective function J, the goods are divided into K different categories; It also includes classification weight adjustment, the specific formula of which is as follows: Based on the fluctuation characteristics of goods inventory, dynamic weights are assigned to the attribute vectors, and the objective function is modified to J′; J′=k=1Kx∈Ckm=1Mwm×xm-μk,m2; Where wm = σmp = 1 / Mσp is the weight of the m-th attribute, σm is the standard deviation of the attribute, σp is the standard deviation of the p-th attribute, M is the total number of attributes, and μk,m is the mean of the m-th attribute in the k-th class. Coding adjustment: Add classification and location identifiers to the existing barcode or RFID coding; The new coding structure is: Code = P + T + S + N, where P is the warehouse (1) partition identifier, which is 2 bits, T is the goods category identifier, which is 3 bits, S is the shelf (2) location identifier, which is 4 bits, and N is the goods unique serial number, which is 5 bits; This code allows us to pinpoint the storage location and category of goods. It also includes encoding check bits, and the specific calculation formula is as follows: Add a check bit C at the end of the encoding. The calculation formula is C = (i = 114ai × 2i mod 2) mod 10, where ai is the numerical value of the first 14 bits of the encoding. The check bit is used to detect encoding input errors. Here, i mod 2 is the modulo operation of i with 2, and the result is 0 or 1, which is used to determine the exponent of 2; mod 10 means taking the result of the sum modulo 10 again, and the result is the value of the check bit C.

5. The inventory counting device for warehouse management according to claim 4, characterized in that, The inventory area division and route planning module divides the inventory area and plans the inventory route, specifically including inventory area division and route planning. Inventory area division: Based on the layout of warehouse (1), the distribution of shelves (2) and the classification results of goods, warehouse (1) is divided into several independent inventory areas. The principle of division is that the quantity of goods in each area is balanced. A greedy algorithm is used to divide the region, with the goal of minimizing the variance of the total number of goods in each region. The formula is: minσ2=1M-1MQm-Q2, where σ2 is the variance of the number of goods in each region, M is the number of inventory regions, Qm is the number of goods in the m-th region, and Q is the average number of goods in all regions. Set the regional load balancing index as follows: L = maxQmminQm, and L ≤ 1.2; Path planning: For each inventory area, Dijkstra's algorithm is used to plan the inventory path from the entrance to the exit.

6. The inventory counting device for warehouse management according to claim 5, characterized in that, The inventory equipment configuration and calibration module calibrates the inventory equipment, and the specific method is as follows: Equipment configuration: Based on the coding type of the goods and the inventory requirements, configure the inventory equipment, including barcode scanner (13) and RFID reader (14); For goods that use coding, a smart inventory terminal with wireless communication function is configured, which can transmit inventory data to the warehouse management system in real time. Equipment calibration: Barcode scanner (13) calibration: Calibrate by scanning a standard barcode card and calculate the scanning error rate. The formula is: γ = Nerror / Ntotal × 100%, where γ is the scanning error rate, Nerror is the number of barcodes that are scanned incorrectly, and Ntotal is the total number of barcodes scanned. When γ exceeds 1%, the scanner needs to be adjusted or replaced. Set the dynamic scanning threshold calibration formula: Based on the barcode printing quality score Q, correct the acceptable error rate threshold: γth = 1% + 100 - Q × 0.01%; RFID reader (14) calibration: By reading standard RFID tags, plot the reader's reading range curve and calculate the reading sensitivity; It also includes read sensitivity, which is calculated using the formula: S = 10lgPrPt, where S is the read sensitivity, Pr is the signal power received by the reader, and Pt is the signal power transmitted by the tag. It also includes environmental interference compensation. The calculation formula for environmental interference compensation is as follows: S′=S+10lg1+β×ρ, where S′ is the sensitivity after compensation, β is the interference coefficient, and ρ is the interference source density. It compensates for the impact of environmental interference on sensitivity. Environmental interference compensation is for compensating for the impact of metal and liquid interference sources on RFID signals.

7. The inventory counting device for warehouse management according to claim 6, characterized in that, The dynamic inventory cycle determination module determines the dynamic inventory cycle for different goods based on their turnover rate and importance. Goods turnover rate calculation: Goods turnover rate is the ratio of the total amount of goods shipped out to the average inventory level within a certain period. The formula is: T = QoutQstock, where T is the goods turnover rate, Qout is the total amount of goods shipped out within a certain period, Qstock is the average inventory level within that period, and Qstock = Qstart + Qend2, where Qstart is the beginning inventory level and Qend is the ending inventory level. In addition, the weighted turnover ratio formula is set as follows: Tw=i=1nwi×Qout,iQstock, where wi=inn+1 / 2 is the time weight and n is the number of statistical periods; Dynamic inventory cycle calculation: The inventory cycle is determined by combining the ABC classification method with the inventory turnover rate; For Category A goods, the inventory counting period is shorter; for Category C goods, the inventory counting period is longer. The calculation formula is: Ci = KTi × Vi, where Ci is the inventory cycle of the i-th type of goods, K is a constant, Ti is the turnover rate of the i-th type of goods, and Vi is the value coefficient of the i-th type of goods. It also includes the adjustment of the inventory volatility risk factor, which is calculated as follows: Ci′=Ci×1-θ×Ri, where Ri=σstock,iQstock,i is the inventory volatility coefficient and θ is the risk sensitivity coefficient.

8. The inventory counting device for warehouse management according to claim 7, characterized in that, The real-time inventory data acquisition and transmission module uses configured inventory equipment to collect goods information in real time and transmits the data to the warehouse management system via wireless communication, as follows: Data collection: The inventory equipment includes a barcode scanner (13) and an RFID reader (14). The barcode scanner (13) reads the coded information, and for RFID goods, the RFID reader (14) automatically identifies the label information. The barcode scanner (13) and RFID reader (14) automatically record the goods' code, quantity, and inventory time data; It also includes multi-source data fusion. The multi-source data fusion calculation formula is as follows: For goods that use both barcode and RFID, the quantity is taken as the weighted fusion value Qfuse=α×Qbarcode+1-α×QRFID, where α is the confidence weight; Data transmission: Wireless sensor network is used for data transmission. The data transmission rate formula is: R = Blog21 + SN, where R is the data transmission rate, B is the channel bandwidth, S is the signal power of the receiver, and N is the noise power. It also includes dynamic transmission priority calculation, the formula for which is as follows: Pi = Vi × Ti, which assigns transmission priorities to different categories of cargo data and adopts a priority scheduling mechanism.

9. A warehouse management inventory counting device according to claim 7, characterized in that, The point data verification and anomaly detection module verifies the inventory data transmitted to the warehouse management system and detects and identifies abnormal data during the inventory process. This includes consistency verification: comparing the inventory data with the accounting data in the warehouse management system to check whether the goods codes and quantity information are consistent; The consistency verification formula is: ΔQi=Qinventory,i-Qbook,i, where ΔQi is the difference between the inventory quantity and the book quantity of the i-th type of goods, Qinventory,i is the inventory quantity, and Qbook,i is the book quantity; When ΔQi = 0, the data is consistent; When ΔQi≠0, the data differs; It also includes the calculation of the difference tolerance, the calculation formula is as follows: ΔQth,i=max1,β×Qbook,i, where β is the tolerance coefficient (taken as 0.02). When ΔQi≤ΔQth,i, it is considered an acceptable difference, reducing unnecessary abnormal handling; Integrity verification: Check whether the inventory data is complete and whether there are any missing goods or information; The data integrity index is used for evaluation, and the formula is: I = Ncomplete / Ntotal × 100%, where I is the data integrity index, Ncomplete is the number of complete inventory data, and Ntotal is the total number of data that should be inventoried; when I ≥ 98%, the data is considered complete. Anomaly detection: A density-based local outlier factor algorithm is used to detect anomalous data; The LOF algorithm determines whether a sample is an outlier by calculating the local outlier factor of the sample point. The formula is: LOFp = o∈NkpLRDoLRDpNkp, where LOFp is the local outlier factor of the sample point p, Nkp is the k-nearest neighbor set of the sample point p, and LRDp is the local reachability density of the sample point p. It also includes anomaly weight correction, calculated as follows: LOF′p=LOFp×1+ω×Vp, where Vp is the cargo value coefficient and ω is the weight coefficient, which improves the detection sensitivity of anomalies in high-value cargoes; The abnormal data processing module analyzes and processes the detected abnormal data and reviews the problematic goods. Anomaly Cause Analysis: By reviewing goods entry and exit records, warehousing environment data, and inventory equipment logs, the causes of the abnormal data were analyzed. It also includes anomaly cause probability prediction, calculated using the following formula: The probability of each cause is calculated using a Naive Bayes model: Pc|x=Px|cPcc′Px|c′Pc′, where x is the anomaly feature vector and c is the anomaly cause category, which helps to quickly locate the cause. Abnormal Handling Measures: Take corresponding measures according to the cause of the abnormality, such as replacing inventory equipment, registering and processing damaged goods, and correcting inbound and outbound records; Recounting operation: Recount goods with abnormal data and calculate the recount accuracy rate. The formula is: α = Ncorrect / Nrecheck × 100%, where α is the recount accuracy rate, Ncorrect is the number of goods correctly recounted, and Nrecheck is the total number of goods recounted. When α ≥ 99%, the recount result is considered reliable. It also includes the calculation of the reconnaissance sample size, and the calculation formula is as follows: For batch abnormal goods, the minimum reconnaissance sample size n = z2 × p × 1 - pe2, where z is the quantile corresponding to the confidence level, p is the estimated abnormality rate, and e is the allowable error; The inventory result statistical analysis module statistically analyzes the inventory data and generates an inventory report; Including the calculation of the consistency rate between accounts and physical inventory: The calculation formula is: β = Nmatch / Ntotal × 100%, where β is the consistency rate between accounts and physical inventory, Nmatch is the quantity of goods that match the accounts, and Ntotal is the total quantity of goods counted. The formula for calculating the consistency rate of assets and liabilities is: βw=i=1nVi×IΔQi=0i=1nVi×100%, where I· is the indicator function; Inventory error rate calculation: The calculation formula is: δ=i=1nΔQii=1nQh ooki×100%, where δ is the inventory error rate and n is the number of types of goods to be inventoried; Data analysis: Analyze the matching degree between inventory error rate, turnover rate and inventory cycle of different categories of goods; Correlation analysis was used to study the relationship between cargo turnover rate and inventory error rate. The correlation coefficient formula is as follows: r=i=1nTi-Tδi-δi=1nTi-T2i=1nδi-δ2; Where r is the correlation coefficient, Ti is the turnover rate of the i-th type of goods, T is the average turnover rate, δi is the inventory error rate of the i-th type of goods, and δ is the average inventory error rate. Inventory efficiency regression model establishment: Establish a multiple regression equation for inventory time t and cargo density ρ and turnover rate T: t = a + bρ + cT + ∈, where a is a constant term, b and c are regression coefficients, and ∈ is an error term, quantifying the contribution of influencing factors to efficiency; The inventory data update module updates the inventory data in the warehouse management system based on the inventory results, and continuously optimizes the inventory method and system. Data update: Update the verified and processed inventory data to the warehouse management system to ensure that the inventory data in the system is consistent with the actual inventory; The update formula is: Qnew,i = Qinventory,i, where Qnew,i is the inventory of the i-th type of goods after the update; The smoothing update calculation is as follows: For volatile inventory data, exponential smoothing is used to update Qnew,i = α × Qinventory,i+1 - α × Qold,i, where α is the smoothing coefficient; System optimization: Based on the analysis of inventory results and feedback information, optimize and adjust the division of inventory areas, route planning, and inventory cycle.

10. A method of using a cargo inventory counting device for warehouse management, characterized in that: Specifically, the following steps are included: Step 1: First, define the scope of data collection, covering basic cargo information and warehousing environment information. Then, perform data cleaning to remove redundant, erroneous, and excessively missing data, and fill in data with fewer missing values. Next, standardize the data to eliminate the impact of different data volumes and correct outliers. Step 2: Classify the goods in the warehouse (1), and then optimize the goods coding. On the basis of the existing barcode or RFID coding, add warehouse (1) partition identifier, goods classification identifier, shelf (2) location identifier and goods unique serial number to form a new coding structure. At the same time, add a check bit at the end of the code. Step 3: Combine the warehouse (1) layout, shelf (2) distribution and goods classification results to divide the inventory area. When dividing, ensure that the quantity of goods in each area is balanced, and at the same time ensure that the difference in inventory workload in each area is within the set range. For each inventory area, plan the optimal inventory path from the entrance to the exit, and adopt the path planning idea that minimizes the walking distance. Step 4: Configure appropriate inventory equipment according to the type of goods code and inventory requirements, calibrate the barcode scanner (13), calculate the scanning error rate by scanning the standard barcode card, calibrate the RFID reader (14), determine its reading range and sensitivity by reading the standard RFID tag at different distances and angles, and compensate for the sensitivity according to the interference in the warehouse (1) environment. Step 5: Calculate the inventory turnover rate. Calculate the total outbound volume and average inventory level of goods within a certain period. Consider the time fluctuation of outbound volume and assign a higher weight to recent outbound volume to calculate the weighted turnover rate. Step 6: By controlling the movement adjustment component, the barcode scanner and RFID reader (14) scan or read the goods according to the planned inventory path; Step 7: Perform consistency verification on the inventory data transmitted to the system, compare the inventory quantity with the book quantity, calculate the difference, and set a reasonable tolerance for the difference based on the condition of the goods. The difference is considered acceptable if it is within the tolerance range. Step 8: For the detected abnormal data, review the goods entry and exit records, warehousing environment data, and inventory equipment logs, and use a probabilistic prediction model to analyze the cause of the abnormality; Step 9: Compile the inventory results, calculate the consistency rate between the physical inventory and the written records, and simultaneously calculate the weighted average consistency rate based on the value of the goods, highlighting the consistency of high-value goods; calculate the inventory error rate to reflect the overall error situation of the inventory. Step 10: Update the verified and processed inventory data to the warehouse management system. For inventory data with drastic fluctuations, use a smooth update method.

Images