An industrial AI logistics warehouse system based on the Internet of Things

Abstract

The application relates to the technical field of logistics warehouses, and particularly discloses an industrial AI logistics warehouse system based on the Internet of Things, which comprises a cargo identification module, a risk discrimination module, a warehouse entry discrimination module, a warehouse entry prediction module and an AI analysis module. The cargo identification module acquires the warehouse idle state, the vehicle cargo capacity and the cargo category, and compares the same to determine the cargo that cannot be stored in the warehouse. The risk discrimination module analyzes the warehouse entry risk value to determine whether the warehouse entry risk exists. If the risk exists, the warehouse entry discrimination module analyzes the expected warehouse capacity required by the cargo that cannot be stored in the warehouse, generates a warehouse entry analysis signal, the warehouse entry prediction module acquires the actual warehouse entry amount, draws an analysis curve, acquires the predicted total warehouse entry amount under a linear rule, the AI analysis module acquires the predicted total warehouse entry amount under a nonlinear rule, comprehensively judges whether the warehouse can meet the warehouse entry demand and calculates the warehouse entry waiting time, and effectively supports the warehouse entry management of the logistics warehouse.

Description

Technical Field

[0001] The present invention relates to the field of logistics warehouse technology, and in particular to an industrial AI logistics warehouse system based on the Internet of Things. Background Art

[0002] In the context of the booming logistics industry today, traditional logistics warehouse management technology has gradually exposed many difficult-to-overcome defects in the face of growing cargo traffic and complex and changing storage needs.

[0003] Existing technologies generally lack cargo identification and capacity analysis capabilities, making it difficult to effectively determine whether cargo can be successfully put into storage before the arrival of the cargo transport vehicle. Existing technologies are unable to achieve efficient integration and comparison of warehouse idle status and vehicle cargo loading information, resulting in the inability to timely detect potential warehousing risks.

[0004] In existing technologies, when there is a risk of goods entering the warehouse, there is a lack of effective monitoring and analysis of the dynamic changes in the goods waiting to be shipped out and waiting to be stored in the warehouse. It is impossible to accurately estimate the actual available capacity of the warehouse before the arrival of the transportation, and it is even more difficult to determine whether it can meet the needs of goods that cannot be stored in the warehouse. Existing technologies are unable to conduct in-depth analysis and prediction of the changing trends of the actual warehousing volume, and cannot provide data support for warehouse management decisions, which can easily lead to backlogs or delays of goods.

[0005] To this end, the present invention provides an industrial AI logistics warehouse system based on the Internet of Things. Summary of the Invention

[0006] The purpose of the present invention is to provide an industrial AI logistics warehouse system based on the Internet of Things to solve the problems in the above background.

[0007] The purpose of the present invention can be achieved through the following technical solutions:

[0008] An industrial AI logistics warehouse system based on the Internet of Things, including:

[0009] The cargo identification module obtains the warehouse availability status and stores it in the warehouse database. It also obtains the cargo capacity and cargo category, compares the cargo capacity and cargo category with the warehouse availability status in the warehouse database, and marks cargo that cannot be placed in the warehouse as unavailable for storage.

[0010] Risk Identification Module: This module analyzes the location information of the cargo transport vehicles corresponding to the goods that cannot enter the warehouse to obtain the transport arrival time. It then calculates the transport arrival time and the proximity warning time to obtain the time proximity ratio. Based on the time proximity ratio and the number of goods that cannot enter the warehouse, it performs a numerical calculation to obtain the warehouse entry risk value Rc. This value is then compared with the warehouse entry risk threshold to determine whether there is a risk in the cargo transport vehicle entering the warehouse.

[0011] Warehouse entry identification module: If there is a risk in the entry of goods into the warehouse by the freight transport vehicle, the expected available capacity of the warehouse and the warehouse capacity required for goods that cannot be entered into the warehouse are analyzed. If the expected available capacity is higher than the warehouse capacity required for goods that cannot be entered into the warehouse, a warehouse entry analysis signal is generated;

[0012] Warehouse entry prediction module: Based on the warehouse entry analysis signal, the module obtains the actual number of goods that can be entered into the warehouse in real time, obtains the actual warehouse entry volume, plots the change curve of the actual warehouse entry volume, and performs linear law analysis. If the change curve conforms to the linear law, the linear predicted total warehouse entry volume is obtained. If the change curve does not conform to the linear law, an AI analysis signal is generated.

[0013] AI analysis module: Based on AI analysis signals, a cargo prediction model is constructed to calculate the predicted total warehousing volume under nonlinear rules. Combined with the predicted total warehousing volume under linear rules, it is determined whether the warehouse can meet the warehousing demand of goods that cannot be stored before the arrival time of transportation. If it cannot be met, a numerical analysis is performed on the time to meet the demand for goods that cannot be stored to obtain the warehousing waiting time.

[0014] As a further technical solution of the present invention: the method for obtaining the warehousing risk value Rc is:

[0015] Perform data analysis on the location information of the cargo transport vehicles corresponding to the cargo that cannot be put into the warehouse to obtain the time proximity ratio Tb;

[0016] Analyze the quantity of goods that cannot be put into storage and obtain the goods not put into storage ratio Wr;

[0017] Calculate the warehousing risk value Rc based on the time proximity ratio Tb and the goods not warehousing ratio Wr;

[0018] By formula: Get the entry risk value Rc, where a and b are preset proportional coefficients;

[0019] The method to determine whether there is a risk in the warehousing of goods on cargo transport vehicles is:

[0020] If the warehousing risk value Rc ≥ the warehousing risk threshold, it indicates that there is a risk in the goods entering the warehouse of the freight transport vehicle.

[0021] As a further technical solution of the present invention: the time closeness ratio Tb is obtained as follows:

[0022] Based on the location information of the cargo transport vehicle corresponding to the cargo that cannot be put into the warehouse, the transportation arrival time of the cargo transport vehicle is obtained;

[0023] The transport arrival time is subtracted from the approach warning time to obtain the time approach difference;

[0024] The time approach difference is processed by ratio with the approach warning time to obtain the time approach ratio, which is marked as Tb;

[0025] The method for obtaining the goods un-warehousing ratio Wr is as follows:

[0026] Obtain the number of goods that cannot be stored in the transport vehicle and the number of items transported by the vehicle, and perform a ratio calculation on the number of goods that cannot be stored and the number of items transported by the vehicle to obtain the goods not stored ratio, which is marked as Wr.

[0027] As a further technical solution of the present invention: generating the warehousing analysis signal includes:

[0028] The method for analyzing the expected available capacity and the warehouse capacity required for goods that cannot be stored is as follows:

[0029] If there is a risk in the cargo transport vehicle entering the warehouse, obtain the warehouse's cargo waiting to be shipped out or received before the transport arrives from the warehouse database;

[0030] The warehouse database updates the status of goods waiting to be shipped out and in, and obtains the warehouse's expected available capacity before the transport arrives from the warehouse database;

[0031] Analyze the expected available capacity of the warehouse and the warehouse capacity required for goods that cannot be stored to obtain the expected available capacity;

[0032] If the expected available capacity is higher than the warehouse capacity required for the goods that cannot be stored, a storage analysis signal is generated.

[0033] As a further technical solution of the present invention: the variation curve of the actual warehouse volume and the method of performing linear law analysis are as follows:

[0034] Based on the warehouse entry analysis signal, during the monitoring period, the system obtains data from the warehouse database in real time, analyzes the actual quantity of goods that the warehouse can partially meet the needs of goods that cannot be entered into the warehouse, and obtains the actual warehouse entry quantity;

[0035] Obtain the actual warehouse entry volume at different monitoring times within the monitoring period, and draw a time-actual warehouse entry volume change curve in a two-dimensional rectangular coordinate system;

[0036] Connect the two endpoints of the time-actual warehouse quantity change curve with a straight line to obtain the cargo change reference line;

[0037] Based on the time-actual warehouse volume change curve, analyze the linear regularity of the change curve;

[0038] The process of analyzing the linear regularity of the change curve is as follows:

[0039] Calculation time - whether there is an inflection point in the curve of the actual inventory volume change. If there is no inflection point, generate an AI analysis signal;

[0040] If there is an inflection point, obtain the X-axis coordinate value of the time-actual warehouse entry change curve corresponding to the inflection point, that is, the time value. Divide the analysis segments into multiple segments according to the time value corresponding to the inflection point to obtain the time analysis segments.

[0041] Obtain the starting endpoint of the cargo change reference line within the time analysis segment, and draw a tangent line between the inflection point of the time-actual warehouse quantity change curve and the starting endpoint of the cargo change reference line;

[0042] Get the angle between the tangent line and the cargo change reference line to get the curve analysis angle, and mark the curve analysis angle as θ i , i=1,2,...,n, n is the total number of curve analysis angles;

[0043] Compare the curve analysis angle with the preset curve analysis angle threshold, and take the curve analysis angle whose curve analysis angle is lower than the curve analysis angle threshold as the low deviation analysis angle, and mark the low deviation analysis angle as Dp j , j is the number of the low-bias analysis angle, j = 1, 2, ..., m, m is the total number of low-bias analysis angles.

[0044] As a further technical solution of the present invention: the method for judging whether the change curve conforms to the linear law is:

[0045] Based on low bias analysis angle Dp j Calculate the linear law value Xg through the formula;

[0046] By formula: Get the linear law value Xg, where c and d are preset proportional coefficients;

[0047] Compare the linear law value Xg with the preset linear law threshold to determine whether the time-actual warehouse volume change curve is a linear law change;

[0048] If the linear law value Xg is lower than the linear law threshold, it indicates that the time-actual inventory change curve is nonlinear, and an AI analysis signal is generated;

[0049] If the linear law value Xg is higher than the linear law threshold, it indicates that the time-actual warehouse volume change curve is a linear change.

[0050] As a further technical solution of the present invention: the linear prediction of the total warehouse volume is obtained as follows:

[0051] If the time-actual warehouse quantity change curve is linear, calculate the slope of the goods change reference line;

[0052] If the slope of the goods change reference line is positive, it indicates that the time-actual warehouse quantity change curve shows a linear growth trend, indicating that the actual warehouse quantity is increasing. The least squares method is used to fit the time-actual warehouse quantity change curve to obtain the warehouse fitting line, which is then drawn in a two-dimensional rectangular coordinate system.

[0053] Extend the warehousing fitting line to the transportation arrival time, obtain the Y-axis coordinate value corresponding to the transportation arrival time, that is, the predicted warehousing quantity under the transportation arrival time under the linear law, and obtain the linear predicted total warehousing quantity.

[0054] As a further technical solution of the present invention: the method for obtaining the predicted total warehousing amount under the nonlinear law is:

[0055] Based on AI analysis signals, we use the decision tree algorithm in AI to build a cargo forecasting model to analyze and predict the actual warehouse volume;

[0056] Based on the cargo forecasting model, the transport arrival time is input to obtain the predicted total warehousing volume under nonlinear rules.

[0057] As a further technical solution of the present invention: the cargo prediction model is constructed as follows:

[0058] S1. Obtain the actual warehouse entry quantity at each monitoring moment during the monitoring period, build a buffer data sequence, and obtain the warehouse storage capacity and transportation arrival time to build a basic data set;

[0059] S2. Preprocess the basic data set to fill the missing values ​​in the basic data set and correct the outliers using the principle of 3 times the standard deviation;

[0060] S3. Divide the preprocessed basic data set into a training set and a test set in a ratio of 7:3. The training set is used to train the cargo prediction model, and the test set is used to verify the results of the cargo prediction model.

[0061] S4. Select the DecisionTreeRegressor class of the sklearn library in Python to initialize the decision tree model, set the maximum depth of the decision tree max_depth = 5, the minimum number of sample splits min_samples_split = 10, and the minimum number of sample leaf nodes min_samples_leaf = 5, use the divided training set to import the decision tree model, and use the fit method for training;

[0062] S5. Use the test set to verify the model training results. If the mean absolute error (MSE) is lower than 5, it indicates that the model training meets expectations. Otherwise, adjust the maximum depth (max_depth) of the decision tree to 3-10 and the minimum number of sample splits (min_samples_split) to 5-20, and retrain the model until the mean absolute error (MSE) requirement is met.

[0063] As a further technical solution of the present invention: the method for obtaining the warehousing waiting time is:

[0064] Compare the total volume of goods entering the warehouse based on the linear law forecast and the cargo forecast with the expected available capacity to determine whether the warehouse can meet the demand for goods that cannot be entered within the transport arrival time.

[0065] If the total predicted warehousing volume is lower than the expected available capacity, it means that the warehouse cannot meet the warehousing demand of the goods that cannot be stored at the time of transport arrival. The waiting time for the goods to be stored is calculated;

[0066] Based on the cargo linear law prediction and cargo forecasting model, the warehouse can calculate the time when it can meet the warehousing demand of the goods that cannot be put into the warehouse, and obtain the predicted warehousing time;

[0067] The predicted warehousing time is subtracted from the transportation arrival time to obtain the warehousing waiting time.

[0068] Beneficial effects of the present invention:

[0069] (1) Through the cargo identification module and risk assessment module, the cargo capacity of the cargo transport vehicle and the idle status of the warehouse can be obtained, and the location of the transport vehicle and the quantity of cargo can be integrated to accurately calculate the risk value of entering the warehouse. This enables warehouse managers to understand the risks of cargo entering the warehouse in advance and make response plans during the transportation of cargo, thereby alleviating the backlog of cargo at the warehouse entrance.

[0070] (2) The cargo prediction model constructed with the help of the AI ​​analysis module can predict the warehousing situation under nonlinear laws. Combined with the analysis results of linear laws, it can comprehensively estimate the cargo warehousing volume in different periods, provide a decision-making basis for warehouse management, and help ensure the efficient and orderly development of cargo warehousing operations. BRIEF DESCRIPTION OF THE DRAWINGS

[0071] The present invention will be further described below with reference to the accompanying drawings.

[0072] Figure 1 It is a flow chart of cargo identification and risk determination in the present invention;

[0073] Figure 2 It is a flow chart of warehousing analysis and prediction in the present invention;

[0074] Figure 3This is a system module diagram of an industrial AI logistics warehouse system based on the Internet of Things of the present invention. DETAILED DESCRIPTION

[0075] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative efforts shall fall within the scope of protection of the present invention.

[0076] Example 1

[0077] See also Figure 1 As shown, the present invention is an industrial AI logistics warehouse system based on the Internet of Things, including:

[0078] The cargo identification module obtains the warehouse availability status and stores it in the warehouse database. It also obtains the cargo capacity and cargo category, compares the cargo capacity and cargo category with the warehouse availability data in the warehouse database, and marks cargo that cannot be placed in the warehouse as unavailable for storage.

[0079] In some embodiments, the warehouse idle state data is obtained, wherein the warehouse idle state data includes: the category of the idle shelves and the storage capacity of the shelves, and the warehouse idle state data is stored in a warehouse database;

[0080] According to the cargo transport list, obtain the cargo capacity and cargo category of the cargo transport vehicle when the cargo is loaded, and send the cargo capacity and cargo category to the system through the IoT device;

[0081] The system compares the cargo capacity and cargo category with the warehouse's cargo database and marks cargo that the warehouse cannot currently store as unavailable for storage;

[0082] It should be noted that the system compares the cargo capacity and cargo category with the warehouse database in the following ways: the cargo capacity is compared with the storage capacity of the current shelf one by one, and the cargo category is compared with the category of the free shelves one by one. The cargo that cannot be stored in the warehouse is marked as unavailable for storage.

[0083] Risk Identification Module: This module analyzes the location information of the cargo transport vehicles corresponding to the goods that cannot enter the warehouse to obtain the transport arrival time. It then calculates the transport arrival time and the proximity warning time to obtain the time proximity ratio. Based on the time proximity ratio and the number of goods that cannot enter the warehouse, it performs a numerical calculation to obtain the warehouse entry risk value Rc. This value is then compared with the warehouse entry risk threshold to determine whether there is a risk in the cargo transport vehicle entering the warehouse.

[0084] Based on the location information of the cargo transport vehicle corresponding to the cargo that cannot be put into the warehouse, the transportation arrival time of the cargo transport vehicle is obtained;

[0085] In some embodiments, the transport location information of the transport vehicle is analyzed using an API of a map software, wherein the location information includes: the GPS location of the vehicle and the location of the warehouse, and the remaining transport distance is obtained. The remaining transport distance and the average speed of the transport vehicle are combined to perform a numerical calculation to obtain the transport arrival time of the cargo transport vehicle;

[0086] The transport arrival time is subtracted from the approach warning time to obtain the time approach difference;

[0087] The time approach difference is processed by ratio with the approach warning time to obtain the time approach ratio, which is marked as Tb;

[0088] It should be noted that the proximity warning time is set by professionals in this field based on their experience;

[0089] Obtain the number of goods that cannot be stored in the transport vehicle and the number of items transported by the vehicle. Ratio the number of goods that cannot be stored to the number of items transported by the vehicle to obtain the goods not stored ratio, which is marked as Wr.

[0090] Calculate the warehousing risk value Rc based on the time proximity ratio Tb and the goods not warehousing ratio Wr;

[0091] By formula: Get the entry risk value Rc, where a = 0.628, b = 0.354;

[0092] Compare the warehousing risk value Rc with the warehousing risk threshold to determine whether there is a risk in the cargo transport vehicle entering the warehouse;

[0093] It should be noted that the risk of goods entering the warehouse on freight transport vehicles means that not all the transported goods can be placed in the warehouse;

[0094] If the entry risk value Rc ≥ the entry risk threshold, it indicates that there is a risk in the entry of goods into the warehouse by the freight transport vehicle;

[0095] If the entry risk value Rc is less than the entry risk threshold, it indicates that the entry risk value of the cargo transport vehicle is within a controllable range, and it is still necessary to continuously monitor the changes in the entry risk value Rc;

[0096] The technical solution of this embodiment is: obtain the idle status of the warehouse and store it in the warehouse database, obtain the cargo capacity and cargo category, compare and judge the cargo capacity and cargo category with the data in the warehouse database, mark the cargo that cannot be placed in the warehouse as cargo that cannot be put into the warehouse, perform data analysis on the position information of the cargo transport vehicles corresponding to the cargo that cannot be put into the warehouse, obtain the transportation arrival time, perform numerical calculation on the transportation arrival time and the approach warning time, obtain the time approach ratio, perform numerical calculation based on the time approach ratio and in combination with the number of cargo that cannot be put into the warehouse, obtain the warehouse entry risk value Rc, and compare it with the warehouse entry risk threshold to determine whether there is a risk of cargo entry of the cargo transport vehicle, and provide data support for subsequent cargo warning analysis of the warehouse by making an initial judgment on the risk value of the transport vehicle.

[0097] Example 2

[0098] like Figure 2 As shown, the present invention is an industrial AI logistics warehouse system based on the Internet of Things, which also includes:

[0099] Warehouse entry identification module: If there is a risk of cargo entry into the warehouse of a freight transport vehicle, the module analyzes the expected available capacity and the warehouse capacity required for goods that cannot be entered into the warehouse. If the expected available capacity is higher than the warehouse capacity required for goods that cannot be entered into the warehouse, a warehouse entry analysis signal is generated;

[0100] If there is a risk in the cargo transport vehicle entering the warehouse, obtain the warehouse's cargo waiting to be shipped out or received before the transport arrives from the warehouse database;

[0101] It should be noted that the goods waiting to be shipped out of the warehouse refer to the goods that the warehouse is preparing to ship out before the transportation arrives, and the goods waiting to be received in the warehouse refer to the goods that the warehouse is preparing to receive in before the transportation arrives.

[0102] The warehouse database updates the status of goods waiting to be shipped out and in, and obtains the warehouse's expected available capacity before the transport arrives from the warehouse database;

[0103] Analyze the expected available capacity of the warehouse and the warehouse capacity required for goods that cannot be stored to obtain the expected available capacity;

[0104] If the expected available capacity is lower than the warehouse capacity required for the goods that cannot be stored, an entry warning signal is sent to the system;

[0105] If the expected available capacity is higher than the warehouse capacity required for the goods that cannot be received, it means that the expected capacity of the shelves to be shipped out before the transport arrives can meet the required capacity of the goods that cannot be received. However, it is still necessary to analyze the actual shelves that have been shipped out to generate a warehouse analysis signal.

[0106] Warehousing prediction module: Based on the warehousing analysis signal, the module obtains the actual quantity of goods that the warehouse can partially meet to meet the demand of goods that cannot be stored in the warehouse in real time, obtains the actual warehousing quantity, draws the change curve of the actual warehousing quantity, and performs linear law analysis on the change curve. If the change curve conforms to the linear law, the linear predicted total warehousing quantity is obtained. If the change curve does not conform to the linear law, an AI analysis signal is generated.

[0107] Based on the warehouse entry analysis signal, during the monitoring period, the system obtains data from the warehouse database in real time, analyzes the actual quantity of goods that the warehouse can partially meet the needs of goods that cannot be entered into the warehouse, and obtains the actual warehouse entry quantity;

[0108] Obtain the actual warehouse entry volume at different monitoring times within the monitoring period, plot the time-actual warehouse entry volume curve in a two-dimensional rectangular coordinate system, with time as the X-axis and the actual warehouse entry volume as the Y-axis;

[0109] Connect the two endpoints of the time-actual warehouse quantity change curve with a straight line to obtain the cargo change reference line;

[0110] It should be noted that the actual warehousing quantity changes in real time as the goods to be stored and shipped change;

[0111] Based on the time-actual warehouse volume change curve, analyze the linear regularity of the change curve;

[0112] Specifically, the analysis process of the linear regularity of the change curve is as follows:

[0113] Calculation time - whether there is an inflection point in the curve of the actual inventory volume change. If there is no inflection point, generate an AI analysis signal;

[0114] If there is an inflection point, obtain the X-axis coordinate value of the time-actual warehouse entry change curve corresponding to the inflection point, that is, the time value. Divide the analysis segments into multiple segments according to the time value corresponding to the inflection point to obtain the time analysis segments.

[0115] Obtain the starting endpoint of the cargo change reference line within the time analysis segment, and draw a tangent line between the inflection point of the time-actual warehouse quantity change curve and the starting endpoint of the cargo change reference line;

[0116] Get the angle between the tangent line and the cargo change reference line to get the curve analysis angle, and mark the curve analysis angle as θ i , i=1,2,...,n, n is the total number of curve analysis angles;

[0117] For example, the inflection point coordinates of the time-actual warehouse quantity change curve are marked as (x1, y1), and the coordinates of the starting end point of the cargo change reference line are marked as (x0, y0). According to the formula: Get the curve analysis angle θ1;

[0118] Compare the curve analysis angle with the curve analysis angle threshold, and take the curve analysis angle whose curve analysis angle is lower than the curve analysis angle threshold as the low deviation analysis angle, and mark the low deviation analysis angle as Dp j , j is the number of the low-bias analysis angle, j = 1, 2, ..., m, m is the total number of low-bias analysis angles;

[0119] The curve analysis angles whose curve analysis angles are higher than the curve analysis angle threshold are marked as high-bias analysis angles;

[0120] It should be noted that the curve analysis angle threshold is set by professionals in this field based on experience;

[0121] Based on low bias analysis angle Dp j , calculate the linear law value Xg through the formula;

[0122] By formula: Obtain the linear law value Xg, where c = 0.65, d = 0.35;

[0123] It should be noted that m and n are calculated as a ratio, and the result is the proportion of low-bias analysis angles in the curve analysis angles. The greater the proportion of low-bias analysis angles in the curve analysis angles, the greater the number of curve analysis angles with smaller values ​​between all tangent lines and the cargo change reference line, and the closer the time-actual warehouse quantity change curve is to the linear trend of the cargo change reference line.

[0124] Compare the linear law value Xg with the linear law threshold to determine whether the time-actual warehouse volume change curve is a linear change;

[0125] If the linear law value Xg is lower than the linear law threshold, it indicates that the time-actual inventory change curve is nonlinear, and an AI analysis signal is generated;

[0126] If the linear law value Xg is higher than the linear law threshold, it means that the time-actual warehouse volume change curve is a linear change;

[0127] If the time-actual warehouse quantity change curve is linear, calculate the slope of the goods change reference line;

[0128] If the slope of the goods change reference line is positive, it indicates that the time-actual warehouse quantity change curve shows a linear growth trend, indicating that the actual warehouse quantity is increasing. The least squares method is used to fit the time-actual warehouse quantity change curve to obtain the warehouse fitting line, which is then drawn in a two-dimensional rectangular coordinate system.

[0129] Extend the warehousing fitting line to the transport arrival time, obtain the Y-axis coordinate value corresponding to the transport arrival time, that is, the predicted warehousing quantity under the transport arrival time under the linear law, and obtain the linear predicted total warehousing quantity;

[0130] It should be noted that the actual warehousing volume is obtained by analyzing and predicting the existing goods that have not been shipped out or received in the warehouse. The expected available capacity is under ideal conditions.

[0131] If the slope of the cargo change reference line is negative, it indicates that the time-actual warehouse quantity change curve shows a linear downward trend, and a warehouse entry warning signal is generated;

[0132] AI Analysis Module: Based on AI analysis signals, a cargo prediction model is constructed to calculate the total predicted warehousing volume under nonlinear rules. Combined with the predicted warehousing volume under linear rules and the expected available capacity of the warehouse, it is determined whether the warehouse can meet the warehousing demand of the goods that cannot be stored before the transport arrives. If it cannot be met, a numerical analysis is performed on the time it takes to meet the warehousing demand of the goods that cannot be stored to determine the warehousing waiting time.

[0133] Based on AI analysis signals, we use the decision tree algorithm in AI to build a cargo forecasting model to analyze and predict the actual warehouse volume;

[0134] Specifically, the cargo forecasting model is constructed as follows:

[0135] S1. Obtain the actual warehouse entry quantity at each monitoring moment during the monitoring period, build a buffer data sequence, and obtain the warehouse storage capacity and transportation arrival time to build a basic data set;

[0136] S2. Data preprocessing of the basic data set, including: filling missing values ​​in the basic data set and correcting outliers using the principle of 3 times the standard deviation;

[0137] It should be noted that missing values ​​are filled based on the mean of the data series. If the arrival time of a transport at a certain moment is far beyond the normal range, it is calculated based on three times the standard deviation of the historical transport time, and the calculation result is used to correct the outlier.

[0138] S3. Divide the preprocessed basic data set into a training set and a test set in a ratio of 7:3. The training set is used to train the cargo prediction model, and the test set is used to verify the results of the cargo prediction model.

[0139] S4. Select the DecisionTreeRegressor class of the sklearn library in Python to initialize the decision tree model, set the maximum depth of the decision tree max_depth = 5, the minimum number of sample splits min_samples_split = 10, and the minimum number of sample leaf nodes min_samples_leaf = 5, use the divided training set to import the decision tree model, and use the fit method for training;

[0140] S5. Use the test set to verify the model training results. If the mean absolute error (MSE) is lower than 5, it indicates that the model training meets expectations. Otherwise, adjust the maximum depth (max_depth) of the decision tree to 3-10 and the minimum number of sample splits (min_samples_split) to 5-20, and retrain the model until the mean absolute error (MSE) requirement is met.

[0141] Based on the cargo forecasting model, the transport arrival time is input to obtain the predicted total warehousing volume under nonlinear rules;

[0142] Compare the total volume of goods entering the warehouse based on the linear law forecast and the cargo forecast with the expected available capacity to determine whether the warehouse can meet the demand for goods that cannot be entered within the transport arrival time.

[0143] If the total predicted warehousing volume is higher than the expected available capacity, it means that the warehouse can meet the warehousing demand for goods that could not be stored at the time of transport arrival;

[0144] If the total predicted warehousing volume is lower than the expected available capacity, it means that the warehouse cannot meet the warehousing demand of the goods that cannot be stored at the time of transportation arrival, and the waiting time for the goods to be stored is calculated;

[0145] Based on the cargo linear law prediction and cargo forecasting model, the warehouse can calculate the time when it can meet the warehousing demand of the goods that cannot be put into the warehouse, and obtain the predicted warehousing time;

[0146] The difference between the predicted warehousing time and the transportation arrival time is processed to obtain the warehousing waiting time, and the warehousing waiting time is sent to the system;

[0147] The technical solution of this embodiment is as follows: if there is a risk in the entry of goods into the warehouse by the cargo transport vehicle, the expected available capacity and the warehouse capacity required for the goods that cannot be entered into the warehouse are analyzed. If the expected available capacity is higher than the warehouse capacity required for the goods that cannot be entered into the warehouse, a warehouse entry analysis signal is generated. Based on the warehouse entry analysis signal, the actual quantity of goods that the warehouse can partially meet to meet the demand for the goods that cannot be entered into the warehouse is obtained to obtain the actual warehouse entry quantity. A change curve of the actual warehouse entry quantity is drawn, and a linear law analysis is performed on the change curve. If the change curve conforms to the linear law, the predicted total warehouse entry quantity is obtained. If the change curve does not conform to the linear law, an AI analysis signal is generated. Based on the AI ​​analysis signal, a cargo prediction model is constructed to calculate the predicted total warehouse entry quantity under the nonlinear law. Combined with the predicted total warehouse entry quantity under the linear law and the expected available capacity of the warehouse, it is determined whether the warehouse can meet the warehouse entry demand of the goods that cannot be entered into the warehouse within the transportation arrival time. If it cannot be met, the time to meet the warehouse entry demand of the goods that cannot be entered into the warehouse is numerically analyzed to obtain the warehouse entry waiting time. By analyzing the warehouse entry waiting time, it is beneficial to ensure the efficient and orderly implementation of the goods entry operation.

[0148] The above is a detailed description of an embodiment of the present invention. However, the content is only a preferred embodiment of the present invention and should not be considered to limit the scope of the present invention. All equivalent changes and improvements made within the scope of the present invention should still fall within the scope of the present invention.

Claims

1. An industrial AI logistics warehouse system based on the Internet of Things, characterized by: include: The cargo identification module obtains the warehouse availability status and stores it in the warehouse database. It also obtains the cargo capacity and cargo category, compares the cargo capacity and cargo category with the warehouse availability status in the warehouse database, and marks cargo that cannot be placed in the warehouse as unavailable for storage. Risk Identification Module: This module analyzes the location information of the cargo transport vehicles corresponding to the goods that cannot enter the warehouse to obtain the transport arrival time. It then calculates the transport arrival time and the proximity warning time to obtain the time proximity ratio. Based on the time proximity ratio and the number of goods that cannot enter the warehouse, it performs a numerical calculation to obtain the warehouse entry risk value Rc. This value is then compared with the warehouse entry risk threshold to determine whether there is a risk in the cargo transport vehicle entering the warehouse. Warehouse entry identification module: If there is a risk in the entry of goods into the warehouse by the freight transport vehicle, the expected available capacity of the warehouse and the warehouse capacity required for goods that cannot be entered into the warehouse are analyzed. If the expected available capacity is higher than the warehouse capacity required for goods that cannot be entered into the warehouse, a warehouse entry analysis signal is generated; Warehouse entry prediction module: Based on the warehouse entry analysis signal, the module obtains the actual number of goods that can be entered into the warehouse in real time, obtains the actual warehouse entry volume, plots the change curve of the actual warehouse entry volume, and performs linear law analysis. If the change curve conforms to the linear law, the linear predicted total warehouse entry volume is obtained. If the change curve does not conform to the linear law, an AI analysis signal is generated. AI analysis module: Based on AI analysis signals, a cargo prediction model is constructed to calculate the predicted total warehousing volume under nonlinear rules. Combined with the predicted total warehousing volume under linear rules, it is determined whether the warehouse can meet the warehousing demand of goods that cannot be stored before the arrival time of transportation. If it cannot be met, a numerical analysis is performed on the time to meet the demand for goods that cannot be stored to obtain the warehousing waiting time.

2. The industrial AI logistics warehouse system based on the Internet of Things according to claim 1 is characterized by: The method for obtaining the warehousing risk value Rc is as follows: Perform data analysis on the location information of the cargo transport vehicles corresponding to the cargo that cannot be stored to obtain the time proximity ratio Tb; Analyze the quantity of goods that cannot be put into storage and obtain the goods not put into storage ratio Wr; Calculate the warehousing risk value Rc based on the time proximity ratio Tb and the goods not warehousing ratio Wr; By formula: Get the entry risk value Rc, where a and b are preset proportional coefficients; The method to determine whether there is a risk in the warehousing of goods on cargo transport vehicles is: If the warehousing risk value Rc ≥ the warehousing risk threshold, it indicates that there is a risk in the cargo transport vehicle entering the warehouse.

3. The industrial AI logistics warehouse system based on the Internet of Things according to claim 2 is characterized by: The time close ratio Tb is obtained as follows: Based on the location information of the cargo transport vehicle corresponding to the cargo that cannot be put into the warehouse, the transportation arrival time of the cargo transport vehicle is obtained; The transport arrival time is subtracted from the approach warning time to obtain the time approach difference; The time approach difference is processed by ratio with the approach warning time to obtain the time approach ratio, which is marked as Tb; The method for obtaining the goods un-warehousing ratio Wr is as follows: Obtain the number of goods that cannot be stored in the transport vehicle and the number of items transported by the vehicle, and perform a ratio calculation on the number of goods that cannot be stored and the number of items transported by the vehicle to obtain the goods not stored ratio, which is marked as Wr.

4. The industrial AI logistics warehouse system based on the Internet of Things according to claim 1 is characterized in that: Generating the warehouse entry analysis signal includes: The method for analyzing the expected available capacity and the warehouse capacity required for goods that cannot be stored is as follows: If there is a risk in the cargo transport vehicle entering the warehouse, obtain the warehouse's cargo waiting to be shipped out or received before the transport arrives from the warehouse database; The warehouse database updates the status of goods waiting to be shipped out and in, and obtains the warehouse's expected available capacity before the transport arrives from the warehouse database; Analyze the expected available capacity of the warehouse and the warehouse capacity required for goods that cannot be stored to obtain the expected available capacity; If the expected available capacity is higher than the warehouse capacity required for the goods that cannot be stored, a storage analysis signal is generated.

5. The industrial AI logistics warehouse system based on the Internet of Things according to claim 1 is characterized in that: The variation curve of the actual warehouse volume and the method of performing linear law analysis are as follows: Based on the warehouse entry analysis signal, during the monitoring period, the system obtains data from the warehouse database in real time, analyzes the actual quantity of goods that the warehouse can partially meet the needs of goods that cannot be entered into the warehouse, and obtains the actual warehouse entry quantity; Obtain the actual warehouse entry volume at different monitoring times within the monitoring period, and draw a time-actual warehouse entry volume change curve in a two-dimensional rectangular coordinate system; Connect the two endpoints of the time-actual warehouse quantity change curve with a straight line to obtain the cargo change reference line; Based on the time-actual warehouse volume change curve, analyze the linear regularity of the change curve; The process of analyzing the linear regularity of the change curve is as follows: Calculation time - whether there is an inflection point in the curve of the actual inventory volume change. If there is no inflection point, generate an AI analysis signal; If there is an inflection point, obtain the X-axis coordinate value of the time-actual warehouse entry change curve corresponding to the inflection point, that is, the time value. Divide the analysis segments into multiple segments according to the time value corresponding to the inflection point to obtain the time analysis segments. Obtain the starting endpoint of the cargo change reference line within the time analysis segment, and draw a tangent line between the inflection point of the time-actual warehouse quantity change curve and the starting endpoint of the cargo change reference line; Get the angle between the tangent line and the cargo change reference line to get the curve analysis angle, and mark the curve analysis angle as θ i , i=1,2,...,n, n is the total number of curve analysis angles; Compare the curve analysis angle with the preset curve analysis angle threshold, and take the curve analysis angle whose curve analysis angle is lower than the curve analysis angle threshold as the low deviation analysis angle, and mark the low deviation analysis angle as Dp j , j is the number of the low-bias analysis angle, j = 1, 2, ..., m, m is the total number of low-bias analysis angles; Based on low bias analysis angle Dp j Determine whether the change curve conforms to the linear law.

6. The industrial AI logistics warehouse system based on the Internet of Things according to claim 5 is characterized by: The judgment method of whether the change curve conforms to the linear law is: Based on low bias analysis angle Dp j , calculate the linear law value Xg through the formula; By formula: Get the linear law value Xg, where c and d are preset proportional coefficients; Compare the linear law value Xg with the preset linear law threshold to determine whether the time-actual warehouse volume change curve is a linear law change; If the linear law value Xg is lower than the linear law threshold, it indicates that the time-actual inventory change curve is nonlinear, and an AI analysis signal is generated; If the linear law value Xg is higher than the linear law threshold, it indicates that the time-actual warehouse volume change curve is a linear change.

7. The industrial AI logistics warehouse system based on the Internet of Things according to claim 6, characterized in that: The linear prediction of the total amount of warehousing is obtained as follows: If the time-actual warehouse quantity change curve is linear, calculate the slope of the goods change reference line; If the slope of the goods change reference line is positive, it indicates that the time-actual warehouse quantity change curve shows a linear growth trend, indicating that the actual warehouse quantity is increasing. The least squares method is used to fit the time-actual warehouse quantity change curve to obtain the warehouse fitting line, which is then drawn in a two-dimensional rectangular coordinate system. Extend the warehousing fitting line to the transportation arrival time, obtain the Y-axis coordinate value corresponding to the transportation arrival time, that is, the predicted warehousing quantity under the transportation arrival time under the linear law, and obtain the linear predicted total warehousing quantity.

8. The industrial AI logistics warehouse system based on the Internet of Things according to claim 1, characterized in that: The method for obtaining the predicted total warehousing volume under the nonlinear law is as follows: Based on AI analysis signals, we use the decision tree algorithm in AI to build a cargo forecasting model to analyze and predict the actual warehouse volume; Based on the cargo forecasting model, the transport arrival time is input to obtain the predicted total warehousing volume under nonlinear rules.

9. The industrial AI logistics warehouse system based on the Internet of Things according to claim 8, characterized in that: The cargo prediction model is constructed as follows: The cargo forecasting model is constructed as follows: S1. Obtain the actual warehouse entry quantity at each monitoring moment during the monitoring period, build a buffer data sequence, and obtain the warehouse storage capacity and transportation arrival time to build a basic data set; S2. Preprocess the basic data set to fill the missing values ​​in the basic data set and correct the outliers using the principle of 3 times the standard deviation; S3. Divide the preprocessed basic data set into a training set and a test set in a ratio of 7:

3. The training set is used to train the cargo prediction model, and the test set is used to verify the results of the cargo prediction model. S4. Select the DecisionTreeRegressor class of the sklearn library in Python to initialize the decision tree model, set the maximum depth of the decision tree max_depth = 5, the minimum number of sample splits min_samples_split = 10, and the minimum number of sample leaf nodes min_samples_leaf = 5, use the divided training set to import the decision tree model, and use the fit method for training; S5. Use the test set to verify the model training results. If the mean absolute error (MSE) is lower than 5, it indicates that the model training meets expectations. Otherwise, adjust the maximum depth (max_depth) of the decision tree to 3-10 and the minimum number of sample splits (min_samples_split) to 5-20, and retrain the model until the mean absolute error (MSE) requirement is met.

10. The industrial AI logistics warehouse system based on the Internet of Things according to claim 1, characterized in that: The method for obtaining the warehousing waiting time is as follows: Compare the total volume of goods entering the warehouse based on the linear law forecast and the cargo forecast with the expected available capacity to determine whether the warehouse can meet the demand for goods that cannot be entered within the transport arrival time. If the total predicted warehousing volume is lower than the expected available capacity, it means that the warehouse cannot meet the warehousing demand of the goods that cannot be stored at the time of transport arrival. The waiting time for the goods to be stored is calculated; Based on the cargo linear law prediction and cargo forecasting model, the warehouse can calculate the time when it can meet the warehousing demand of the goods that cannot be put into the warehouse, and obtain the predicted warehousing time; The predicted warehousing time is subtracted from the transportation arrival time to obtain the warehousing waiting time.

Images