Distributed warehouse real-time warehouse supplement early warning method based on convolutional miraculous network perception

Abstract

The invention discloses a distributed warehouse real-time warehouse supplement early warning method based on convolutional miraculous network perception, and the method comprises the following steps: obtaining the image information of warehouse goods, and carrying out the video intelligent recognition training of the warehouse goods through the image information of the warehouse goods; obtaining warehouse-in cargo data and warehouse-out cargo data according to the warehouse cargo identification features, and generating warehouse cargo data according to the warehouse-in cargo data and the warehouse-out cargo data; performing data cleaning on the warehouse cargo data through a KMP algorithm; warehouse filling feature extraction is carried out on the cleaned warehouse cargo data through a convolutional neural network; the cargo demand degree of the warehouse is obtained according to the warehouse supplementing features, and the warehouse supplementing early warning level of the warehouse is established through the warehouse supplementing features and the cargo demand degree; and according to the bin filling early warning grade, sending bin filling early warning, and obtaining a bin filling result. According to the invention, the cargo inventory state of the distributed warehouse can be monitored, and warehouse supplement early warning can be carried out according to the cargo demand quantity and the material consumption data.

Description

Technical Field

[0001] The present invention relates to the technical field of electric power material storage management, and in particular to a distributed warehouse real-time replenishment warning method based on convolutional neural network perception. Background Art

[0002] With the rapid rise of technologies such as the Internet of Things, big data, artificial intelligence, and machine learning, the warehousing management of power supplies is gradually moving towards automation and intelligence. For example, automated guided vehicles (AGVs), with their precise path planning capabilities, can automatically handle the handling of power supplies, effectively improving the efficiency and accuracy of material handling. The introduction of smart shelving systems has enabled automated storage and rapid retrieval of power supplies, significantly improving the utilization of warehouse space. Furthermore, intelligent warehouse management systems (WMS) have become the core support for power supply warehousing management. They enable real-time monitoring of material inbound and outbound operations, accurately monitoring inventory status, and fully informatizing and automating warehouse management processes, effectively reducing the errors and efficiency losses associated with manual operations.

[0003] In the actual management of power supplies, demand for them is influenced by a complex array of factors, including dynamic adjustments to grid construction plans, changes in maintenance schedules due to sudden power equipment failures, and shifts in the relationship between power supply and demand. However, existing warehouse early warning systems are mostly based on simple inventory threshold settings. These systems, relying solely on inventory thresholds, fail to comprehensively and timely account for these dynamic factors, leading to a disconnect between early warning information and actual power market demand, delaying project progress and impacting the stability of power supply. Summary of the Invention

[0004] The present invention provides a distributed warehouse real-time replenishment warning method based on convolutional neural network perception, which can monitor the inventory status of goods in distributed warehouses and provide replenishment warnings based on the demand for goods and material consumption data. The specific technical solution is as follows:

[0005] A distributed warehouse real-time replenishment warning method based on convolutional neural network perception includes the following steps:

[0006] Obtain image information of warehouse goods, conduct video intelligent recognition training on warehouse goods through the image information of warehouse goods and obtain recognition features;

[0007] Obtain incoming goods data and outgoing goods data based on warehouse goods identification features, and generate warehouse goods data based on the incoming goods data and outgoing goods data;

[0008] The warehouse cargo data is cleaned using the KMP algorithm; the cleaned warehouse cargo data is used to extract replenishment features through a convolutional neural network;

[0009] Obtain the warehouse's demand for goods based on replenishment characteristics, and establish a warehouse replenishment warning level based on the replenishment characteristics and demand for goods;

[0010] Issue a margin call warning based on the margin call warning level and obtain the margin call results.

[0011] Preferably, the obtaining of image information of warehouse goods, performing video intelligent recognition training on the warehouse goods through the image information of the warehouse goods and obtaining recognition features comprises the following steps:

[0012] Obtain image information of warehouse goods in six directions;

[0013] Build a warehouse cargo recognition model using convolutional neural networks;

[0014] The warehouse cargo recognition model obtains image information and trains it using the cross-entropy loss function.

[0015] The identification features are obtained based on the learning and training results of the warehouse cargo identification model.

[0016] Preferably, the cargo warehouse data includes cargo entry time, cargo exit time, cargo quantity, cargo type and consumption history information.

[0017] Preferably, the data cleaning of warehouse cargo data by using the KMP algorithm includes the following steps:

[0018] Get warehouse cargo data;

[0019] Set up a KMP matcher to perform anomaly detection on warehouse cargo data;

[0020] Complete and correct the data detected as abnormal.

[0021] Preferably, the method of extracting replenishment features from the warehouse cargo data after data cleaning through a convolutional neural network comprises the following steps:

[0022] Obtain warehouse cargo data that has completed data cleaning, perform convolution on the warehouse cargo data, and obtain warehouse data features;

[0023] Input warehouse data features into the pooling layer to simplify the warehouse data features;

[0024] The simplified warehouse data features are input into the fully connected layer for feature splicing, and the warehouse data features are corrected by consuming historical information to obtain the replenishment features.

[0025] Preferably, establishing a warehouse replenishment warning level based on replenishment characteristics and goods demand comprises the following steps:

[0026] Obtain the time feature and consumption feature of the replenishment feature;

[0027] Calculate the demand for goods based on time characteristics, consumption characteristics and inventory thresholds;

[0028] Set replenishment warning levels based on demand for goods.

[0029] Preferably, the method further includes inventory management and regional coordinated management of first aid kits within the region; the inventory management and regional coordinated management of first aid kits within the region includes the following steps:

[0030] Get the distribution of all first aid kits in the area;

[0031] Obtain historical demand information for first aid kits in the area and historical power emergency repair data in the area;

[0032] The inventory of first aid kits in the region is allocated and replenished based on the distribution of first aid kits, historical demand information and historical power emergency maintenance data.

[0033] Preferably, the replenishment result includes a first replenishment result and a second replenishment result; the first replenishment result is the inventory quantity of the corresponding goods within the time specified by the corresponding replenishment warning after the replenishment warning is issued; the second replenishment result is the inventory quantity of the corresponding goods within the time specified by twice the replenishment warning; the actual replenishment result is the replenishment operation finally executed and its effect.

[0034] A distributed warehouse real-time replenishment warning system based on convolutional neural network perception includes a first computing unit for acquiring image information of warehouse goods, performing video intelligent recognition training on the warehouse goods based on the image information of the warehouse goods, and obtaining recognition features;

[0035] A second calculation unit is used to obtain incoming goods data and outgoing goods data according to warehouse goods identification characteristics, and generate warehouse goods data according to the incoming goods data and outgoing goods data;

[0036] The third computing unit is used to clean the warehouse cargo data through an algorithm; the warehouse cargo data that has completed data cleaning is subjected to replenishment feature extraction through a convolutional neural network;

[0037] A fourth calculation unit is used to obtain the warehouse's goods demand based on the replenishment characteristics, and to establish the warehouse's replenishment warning level based on the replenishment characteristics and the goods demand;

[0038] The fifth calculation unit is used to issue a margin call warning according to the margin call warning level and obtain a margin call result.

[0039] A distributed warehouse real-time replenishment warning device based on convolutional neural network perception, the device includes a processor and a memory; the memory is used to store program code and transmit the program code to the processor;

[0040] The processor is used to execute the steps of the above-mentioned distributed warehouse real-time replenishment warning method based on convolutional network perception according to the instructions in the program code.

[0041] Compared with the prior art, the present invention has the following beneficial effects:

[0042] This invention captures image information of warehouse goods and uses video intelligent recognition technology to train and extract identification features. This enables accurate identification and data collection, providing reliable data support for subsequent analysis and decision-making. A KMP algorithm is used to clean data and a convolutional neural network is used to extract replenishment features, ensuring efficient data processing and accurate mining of key information. Based on these replenishment features, the system scientifically determines the demand for goods and establishes replenishment warning levels. This approach fully considers the complex influence of various factors on power supply demand, overcoming the limitations of traditional simple inventory threshold warning systems. It ensures that warning information closely aligns with actual power market demand, effectively avoiding project delays caused by inappropriate warnings and ensuring the stability of power supply supply. Furthermore, replenishment warnings are issued and replenishment results are obtained in a timely manner based on the warning level, enabling real-time replenishment warnings. This helps managers make quick decisions, rationally arrange procurement and replenishment, reduce inventory issues, optimize warehouse management processes, and improve overall operational efficiency. Furthermore, this solution integrates advanced technologies such as the Internet of Things, big data, and artificial intelligence to promote the automated and intelligent development of power supply warehouse management, enhancing the modernization and competitiveness of the power industry in this area. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following briefly describes the drawings required for the specific embodiments or the description of the prior art. Similar elements or parts are generally identified by similar reference numerals throughout the drawings. Elements or parts in the drawings are not necessarily drawn to scale.

[0044] Figure 1 Flow chart of the method of the present invention. DETAILED DESCRIPTION

[0045] 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 them. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0046] It will be understood that when used in this specification and the appended claims, the terms “comprises” and “comprising” indicate the presence of described features, integers, steps, operations, elements and / or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or groups thereof.

[0047] It should also be understood that the terms used in the present specification are only for the purpose of describing particular embodiments and are not intended to limit the present invention. As used in the present specification and the appended claims, the singular forms "a", "an", and "the" are intended to include the plural forms unless the context clearly indicates otherwise.

[0048] It should be further understood that the term "and / or" used in the present description and the appended claims refers to and includes any and all possible combinations of one or more of the associated listed items.

[0049] Example 1

[0050] As shown in the figure, a distributed warehouse real-time replenishment warning method based on convolutional neural network perception includes the following steps:

[0051] Obtain image information of warehouse goods, conduct video intelligent recognition training on warehouse goods through the image information of warehouse goods and obtain recognition features;

[0052] Obtain incoming goods data and outgoing goods data based on warehouse goods identification features, and generate warehouse goods data based on the incoming goods data and outgoing goods data;

[0053] The warehouse cargo data is cleaned using the KMP algorithm; the cleaned warehouse cargo data is used to extract replenishment features through a convolutional neural network;

[0054] Obtain the warehouse's demand for goods based on replenishment characteristics, and establish a warehouse replenishment warning level based on the replenishment characteristics and demand for goods;

[0055] Issue a margin call warning based on the margin call warning level and obtain the margin call results.

[0056] Cargo identification is the process of creating warehouse cargo identification features to identify incoming cargo. This is the foundation of the entire system. By uniquely identifying each batch of cargo, it provides a clear target for subsequent data collection and management. Warehouse cargo identification features are used to capture incoming and outgoing cargo data, which is then integrated to generate warehouse cargo data, comprehensively recording the flow of cargo within the warehouse and forming a comprehensive database.

[0057] The KMP algorithm is used to clean warehouse cargo data. The KMP (Knuth-Morris-Pratt) algorithm is an efficient string matching algorithm that can be used here to find and process abnormal patterns in data, such as incorrect formats and duplicate records, thereby improving data quality.

[0058] The cleaned data is fed into a convolutional neural network to extract replenishment features. Convolutional neural networks excel at processing spatial structures and characteristic patterns in data. Through a series of operations, such as convolution and pooling, they extract key features relevant to replenishment from complex warehouse inventory data. Based on these extracted replenishment features, the warehouse's inventory demand is assessed. Taking into account various factors, a specific computational model is used to determine inventory demand. Combining replenishment features with inventory demand, a replenishment warning level is established for the warehouse according to pre-defined rules. Based on the replenishment warning level, a corresponding replenishment warning is issued, prompting warehouse management to take action.

[0059] Example 2

[0060] This embodiment differs from the first embodiment in that the method of obtaining image information of warehouse goods, performing video intelligent recognition training on the warehouse goods based on the image information of the warehouse goods, and obtaining recognition features includes the following steps:

[0061] Obtain image information of warehouse goods in six directions;

[0062] Build a warehouse cargo recognition model using convolutional neural networks;

[0063] The warehouse cargo recognition model obtains image information and trains it using the cross-entropy loss function.

[0064] The identification features are obtained based on the learning and training results of the warehouse cargo identification model.

[0065] By combining multi-view image acquisition with convolutional neural networks, the system achieves intelligent identification and feature extraction of warehouse goods. First, the system collects image information from six angles of the warehouse goods to ensure comprehensive visual data. Then, a convolutional neural network is used to build a warehouse goods identification model, automatically extracting key features from the image through the convolutional layer. During the training phase, a cross-entropy loss function is used to evaluate the discrepancy between the model's predictions and the actual identification features, guiding the model's parameter adjustment to minimize error. Finally, the trained and optimized model outputs identification features, which are used in subsequent goods identification and classification tasks, enabling intelligent management of warehouse goods.

[0066] Example 3

[0067] The difference between this embodiment and embodiment 2 is that the cargo warehouse data includes cargo entry time, cargo exit time, cargo quantity, cargo type and consumption history information.

[0068] During the inbound and outbound processes, the system not only records basic information about the goods and their storage status, but also details the goods' inbound and outbound times, quantity, type, and consumption history. Inbound and outbound times reflect the goods' residence time and turnover frequency in the warehouse; quantity directly reflects inventory changes; type facilitates categorization and statistical analysis; and consumption history, by recording past power and material consumption, provides a crucial basis for forecasting future demand. This interconnected data forms a comprehensive collection of goods information, enabling in-depth analysis of goods flow patterns, consumption trends, and inventory fluctuations, providing robust data support for more precise replenishment strategies. The inclusion of consumption history information enables warehouse managers to better forecast market demand and plan inventory in advance based on the characteristics and consumption patterns of different goods, avoiding inventory overstocking or stockouts and improving warehouse management decision-making.

[0069] Example 4

[0070] This embodiment differs from the third embodiment in that the data cleaning of the warehouse cargo data by using the KMP algorithm includes the following steps:

[0071] Get warehouse cargo data;

[0072] Set up a KMP matcher to perform anomaly detection on warehouse cargo data;

[0073] Complete and correct the data detected as abnormal.

[0074] The KMP algorithm is used to set up a matcher, which scans the data according to the preset abnormal pattern rules to quickly locate data fragments that do not conform to the normal pattern. For example, for the cargo quantity field, the preset rule is that the quantity must be a non-negative integer. When the matcher traverses the data, if it finds a negative number or non-numeric character, it can be determined as abnormal data. For the time field, the correct time format is preset. If data that does not conform to the format appears, it is marked as abnormal. The detected abnormal data is processed according to the characteristics of the data and business rules. For missing values, if the storage time is missing and the time intervals between adjacent records have a certain pattern, it can be estimated and filled based on the pattern; for erroneous data, such as misspellings of cargo types, corrections are made after verification with the cargo database or manually. Through systematic anomaly detection and targeted data completion and correction, the accuracy and completeness of warehouse cargo data are effectively improved, ensuring the reliability of subsequent analysis and decision-making based on this data.

[0075] Example 5

[0076] This embodiment differs from the fourth embodiment in that the warehouse cargo data after data cleaning is subjected to replenishment feature extraction through a convolutional neural network, which includes the following steps:

[0077] Obtain warehouse cargo data that has completed data cleaning, perform convolution on the warehouse cargo data, and obtain warehouse data features;

[0078] Input warehouse data features into the pooling layer to simplify the warehouse data features;

[0079] The simplified warehouse data features are input into the fully connected layer for feature splicing, and the warehouse data features are corrected by consuming historical information to obtain the replenishment features.

[0080] After obtaining cleaned warehouse cargo data, a convolution operation is performed on it. The convolution layer slides a convolution kernel over the data, performing a weighted sum calculation. Different convolution kernels can extract different features. For example, a convolution kernel for time series data can extract the changing characteristics of cargo entry and exit times, while a convolution kernel for cargo quantity can capture quantity fluctuations, thereby converting the raw data into representative warehouse data features.

[0081] The extracted warehouse data features are fed into the pooling layer, which downsamples the data using methods such as max pooling or average pooling. Max pooling selects the maximum value in a local region, while average pooling calculates the average value in a local region. This approach reduces the data dimension and subsequent computation while preserving key features.

[0082] The simplified warehouse data features enter the fully connected layer, which concatenates all features to form a complete feature vector. Based on this, historical consumption information is introduced to modify the features. This historical consumption information reflects market demand for goods. Based on the correlation analysis between historical consumption data and current warehouse data features, the feature vector is adjusted to more accurately reflect actual demand for goods, ultimately generating replenishment features.

[0083] Example 6

[0084] This embodiment differs from Embodiment 5 in that establishing a warehouse replenishment warning level based on replenishment characteristics and goods demand includes the following steps:

[0085] Obtain the time feature and consumption feature of the replenishment feature;

[0086] Calculate the demand for goods based on time characteristics, consumption characteristics and inventory thresholds;

[0087] Set replenishment warning levels based on demand for goods.

[0088] The demand for goods is calculated based on time characteristics, consumption characteristics and inventory thresholds according to the following formula:

[0089] Demand = A*time characteristic + B*consumption characteristic + C*inventory threshold

[0090] The time feature analyzes time-related information in warehouse cargo data to extract the temporal distribution patterns of goods entering and leaving the warehouse at different time scales. For example, statistics are collected on daily, weekly, and monthly peak times for goods entering and leaving the warehouse, as well as the frequency of goods turnover in different time periods, to reflect the impact of time factors on goods demand. The consumption feature obtains goods consumption data and analyzes consumption quantity trends over different time periods, including the extent of growth or decline in material consumption and seasonal fluctuations in material consumption, thereby understanding the market demand for goods.

[0091] The demand for a product is calculated based on the separated time and consumption characteristics and a pre-set inventory threshold. A, B, and C are pre-determined weight coefficients, determined based on a variety of factors, including the warehouse's actual operations, the characteristics of the product, and market trends. The time, consumption, and inventory thresholds are weighted accordingly and then linearly combined to determine the demand for the product. For example, if the current period is high for electricity, the weight A corresponding to the time characteristic will be relatively large, indicating a more significant impact of time on demand. If the consumption of a particular product fluctuates significantly, the weight B of the consumption characteristic will be appropriately increased to emphasize the importance of material consumption on demand. The weight C of the inventory threshold reflects the influence of current inventory levels on demand calculations. This approach comprehensively considers multiple key factors to calculate a demand for the product that is more accurate and responsive to actual conditions.

[0092] Based on the calculated demand for goods, the replenishment warning level is set according to pre-established rules. When the calculated demand is low and the current inventory is relatively sufficient, with a large headroom to the inventory threshold, it indicates that the goods will not face significant demand pressure in the short term, and the warning level is set to a low level. If the demand is close to the inventory threshold, it indicates that inventory is relatively tight, and the demand for goods may exceed the inventory supply capacity at any time, and the warning level is set to a medium level. When the demand exceeds the inventory threshold, it means that the warehouse may soon be out of stock, and the warning level is set to a high level. The demand for goods is calculated based on time and consumption characteristics, and the inventory threshold and the weight of each factor are comprehensively considered through a formula. This fully considers the time pattern of power material consumption, the dynamic changes in market demand, and the current inventory status, making demand forecasts more accurate and facilitating the reasonable arrangement of inventory.

[0093] Example 7

[0094] This embodiment differs from Embodiment 6 in that the replenishment result includes a first replenishment result and a second replenishment result; the first replenishment result is the inventory quantity of the corresponding goods within the time specified by the replenishment warning after the replenishment warning is issued; the second replenishment result is the inventory quantity of the corresponding goods within the time specified by twice the replenishment warning; and the actual replenishment result is the replenishment operation finally executed and its effect.

[0095] After a replenishment warning is issued, the system begins recording the relevant replenishment results. The first replenishment result records the inventory quantity of the corresponding goods within the specified time period after the replenishment warning was issued. This record can reflect the direct impact of the replenishment warning on inventory in a relatively short period of time. The second replenishment result records the inventory quantity of the corresponding goods after twice the time period specified by the replenishment warning, observing inventory changes over a longer timeframe. The actual replenishment result records the final replenishment operation and its effectiveness. It includes detailed information such as the quantity of goods replenished, the replenishment time, and the inventory status after the replenishment, comprehensively reflecting whether the replenishment operation achieved the expected results. By recording the first and second replenishment results, the timeliness and effectiveness of the replenishment warning can be evaluated at different time scales, and the dynamic changes in inventory after the warning was issued can be understood. This helps warehouse managers analyze the rationality of replenishment operations, summarize lessons learned, adjust and optimize replenishment strategies based on different power material demand situations, and improve the level of refined warehouse management.

[0096] Example 8

[0097] This embodiment differs from embodiment 7 in that it further includes inventory management and regional coordinated management of first aid kits within the region; the inventory management and regional coordinated management of first aid kits within the region includes the following steps:

[0098] Get the distribution of all first aid kits in the area;

[0099] Obtain historical demand information for first aid kits in the area and historical power emergency repair data in the area;

[0100] The inventory of first aid kits in the region is allocated and replenished based on the distribution of first aid kits, historical demand information and historical power emergency maintenance data.

[0101] Regularly inventory all first aid kits in the area. Electronic scanning equipment installed at the kit storage location automatically reads the kit's identification information and compares the actual inventory quantity with the system records to ensure the accuracy of inventory data. Also, inspect the first aid kits for damage, expiration, etc., and mark and record any damaged or expired kits.

[0102] Collect historical power emergency repair data, including the time, location, type of repair incident, and the items in the first aid kit used. Using data analysis models, combined with factors such as the current grid operating conditions and weather forecasts, forecast the likelihood of power emergency repairs in different regions over the next period of time, and thus estimate the quantity and type of first aid kits required in each region. Evaluate the first aid kit inventory level in each region based on demand forecasts and pre-set safety stock standards. Determine which regions have a reasonable first aid kit inventory, which regions are at risk of stock shortages, and which regions may have inventory backlogs. Issue early warning signals for regions at higher risk of stock shortages, prompting timely restocking.

[0103] When emergency power repairs occur in a certain area and the existing inventory of first aid kits is insufficient to meet the demand, the system automatically searches for the inventory of first aid kits in surrounding areas. According to pre-set coordination rules, first aid kits are allocated from the nearest area with sufficient inventory first. During the allocation process, the first aid kit inventory information of each area is updated in real time, and detailed data such as the time, quantity, and destination of the allocation are recorded. For areas where inventory is reduced due to allocation, timely replenishment is arranged. Based on the actual needs and inventory consumption of each area, a scientific replenishment plan is formulated to ensure that first aid kits can be replenished to the corresponding locations in a timely and accurate manner and maintain a reasonable inventory level. At the same time, a close cooperative relationship is established with first aid kit suppliers to ensure that additional first aid kits can be quickly obtained in an emergency.

[0104] Example 9

[0105] A distributed warehouse real-time replenishment warning system based on convolutional neural network perception includes a first computing unit for acquiring image information of warehouse goods, performing video intelligent recognition training on the warehouse goods based on the image information of the warehouse goods, and obtaining recognition features;

[0106] A second calculation unit is used to obtain incoming goods data and outgoing goods data according to warehouse goods identification characteristics, and generate warehouse goods data according to the incoming goods data and outgoing goods data;

[0107] The third computing unit is used to clean the warehouse cargo data through an algorithm; the warehouse cargo data that has completed data cleaning is subjected to replenishment feature extraction through a convolutional neural network;

[0108] A fourth calculation unit is used to obtain the warehouse's goods demand based on the replenishment characteristics, and to establish the warehouse's replenishment warning level based on the replenishment characteristics and the goods demand;

[0109] The fifth calculation unit is used to issue a margin call warning according to the margin call warning level and obtain a margin call result.

[0110] Example 10

[0111] A distributed warehouse real-time replenishment warning device based on convolutional neural network perception, the device includes a processor and a memory; the memory is used to store program code and transmit the program code to the processor;

[0112] The processor is used to execute the steps of the above-mentioned distributed warehouse real-time replenishment warning method based on convolutional network perception according to the instructions in the program code.

[0113] In summary, this invention acquires image information of warehouse goods and uses video intelligent recognition technology to train and extract identification features. This allows for accurate identification and data collection of goods, providing reliable data support for subsequent analysis and decision-making. A KMP algorithm is used to clean data and a convolutional neural network is used to extract replenishment features, ensuring efficient data processing and accurate mining of key information. Furthermore, based on replenishment features, the demand for goods is scientifically determined and replenishment warning levels are established. This approach fully considers the reality that power material demand is influenced by multiple complex factors, overcoming the limitations of traditional simple inventory threshold warning systems. This ensures that warning information closely aligns with actual power market demand, effectively avoiding project delays caused by inappropriate warnings and ensuring the stability of power material supply. Furthermore, replenishment warnings are issued and replenishment results are obtained in a timely manner based on the warning level, enabling real-time replenishment warnings. This helps managers make quick decisions, rationally arrange procurement and replenishment, reduce inventory issues, optimize warehouse management processes, and improve overall operational efficiency. Furthermore, this solution integrates multiple advanced technologies, such as the Internet of Things, big data, and artificial intelligence, to promote the automated and intelligent development of power material warehouse management, enhancing the modernization and competitiveness of the power industry in material warehouse management.

[0114] Those skilled in the art will appreciate that the units of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition of each example has been generally described in terms of function in the above description. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0115] In the embodiments provided by the present invention, it should be understood that the division of units is merely a logical function division, and there may be other division methods in actual implementation, for example, multiple units can be combined into one unit, one unit can be split into multiple units, or some features can be ignored, etc.

[0116] In addition, the functional units in the various embodiments of the present invention may be integrated into a single processing unit, each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0117] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-0nly Memory), random access memory (RAM, Random Access Memory), mobile hard disk, magnetic disk or optical disk, etc., various media that can store program code.

[0118] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or make equivalent replacements for some or all of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the scope of the technical solutions of the embodiments of the present invention, and they should all be included in the scope of the claims and description of the present invention.

Claims

1. A distributed warehouse real-time replenishment warning method based on convolutional neural network perception, characterized by: The following steps are involved: Obtain image information of warehouse goods, conduct video intelligent recognition training on warehouse goods through the image information of warehouse goods and obtain recognition features; Obtain incoming goods data and outgoing goods data based on warehouse goods identification features, and generate warehouse goods data based on the incoming goods data and outgoing goods data; The warehouse cargo data is cleaned using the KMP algorithm; the cleaned warehouse cargo data is used to extract replenishment features through a convolutional neural network; Obtain the warehouse's demand for goods based on replenishment characteristics, and establish a warehouse replenishment warning level based on the replenishment characteristics and demand for goods; Issue a margin call warning based on the margin call warning level and obtain the margin call results.

2. A distributed warehouse real-time replenishment warning method based on convolutional neural network perception according to claim 1, characterized in that: The method of obtaining image information of warehouse goods, performing video intelligent recognition training on the warehouse goods based on the image information of the warehouse goods, and obtaining recognition features includes the following steps: Obtain image information of warehouse goods in six directions; Build a warehouse cargo recognition model using convolutional neural networks; The warehouse cargo recognition model obtains image information and trains it using the cross-entropy loss function. The identification features are obtained based on the learning and training results of the warehouse cargo identification model.

3. A distributed warehouse real-time replenishment warning method based on convolutional neural network perception according to claim 1, characterized in that: The cargo warehouse data includes cargo entry time, cargo exit time, cargo quantity, cargo type and consumption history information.

4. A distributed warehouse real-time replenishment warning method based on convolutional neural network perception according to claim 1, characterized in that: The data cleaning of warehouse cargo data by using the KMP algorithm includes the following steps: Get warehouse cargo data; Set up a KMP matcher to perform anomaly detection on warehouse cargo data; Complete and correct the data detected as abnormal.

5. A distributed warehouse real-time replenishment warning method based on convolutional neural network perception according to claim 1, characterized in that: The warehouse cargo data that has completed data cleaning is subjected to replenishment feature extraction through a convolutional neural network, including the following steps: Obtain warehouse cargo data that has completed data cleaning, perform convolution on the warehouse cargo data, and obtain warehouse data features; Input warehouse data features into the pooling layer to simplify the warehouse data features; The simplified warehouse data features are input into the fully connected layer for feature splicing, and the warehouse data features are corrected by consuming historical information to obtain the replenishment features.

6. A distributed warehouse real-time replenishment warning method based on convolutional neural network perception according to claim 1, characterized in that: The method of establishing a warehouse replenishment warning level based on replenishment characteristics and cargo demand includes the following steps: Obtain the time feature and consumption feature of the replenishment feature; Calculate the demand for goods based on time characteristics, consumption characteristics and inventory thresholds; Set replenishment warning levels based on demand for goods.

7. A distributed warehouse real-time replenishment warning method based on convolutional neural network perception according to claim 1, characterized in that: The replenishment results include the first replenishment result and the second replenishment result; the first replenishment result is the inventory quantity of the corresponding goods within the time specified by the corresponding replenishment warning after the replenishment warning is issued; the second replenishment result is the inventory quantity of the corresponding goods within the time specified by twice the replenishment warning; the actual replenishment result is the final executed replenishment operation and its effect.

8. A distributed warehouse real-time replenishment warning method based on convolutional neural network perception according to claim 1, characterized in that: The invention also includes inventory management and regional coordination management of first aid kits within the region; the inventory management and regional coordination management of first aid kits within the region includes the following steps: Get the distribution of all first aid kits in the area; Obtain historical demand information for first aid kits in the area and historical power emergency repair data in the area; The inventory of first aid kits in the region is allocated and replenished based on the distribution of first aid kits, historical demand information and historical power emergency maintenance data.

9. A distributed warehouse real-time replenishment warning system based on convolutional neural network perception, characterized by: The system comprises a first computing unit configured to obtain image information of warehouse goods, perform video intelligent recognition training on the warehouse goods through the image information of the warehouse goods, and obtain recognition features; A second calculation unit is used to obtain incoming goods data and outgoing goods data according to warehouse goods identification characteristics, and generate warehouse goods data according to the incoming goods data and outgoing goods data; The third computing unit is used to clean the warehouse cargo data through an algorithm; the warehouse cargo data that has completed data cleaning is subjected to replenishment feature extraction through a convolutional neural network; A fourth calculation unit is used to obtain the warehouse's goods demand based on the replenishment characteristics, and to establish the warehouse's replenishment warning level based on the replenishment characteristics and the goods demand; The fifth calculation unit is used to issue a margin call warning according to the margin call warning level and obtain a margin call result.

10. A distributed warehouse real-time replenishment warning device based on convolutional neural network perception, characterized in that: The device includes a processor and a memory; the memory is used to store program code and transmit the program code to the processor; The processor is used to execute the steps of the distributed warehouse real-time replenishment warning method based on convolutional network perception according to claims 1-7.

Images