Intelligent logistics and storage management system based on internet of things

Abstract

The application discloses an intelligent logistics warehouse management system based on Internet of Things and relates to the technical field of logistics warehouse management.The system comprises an initial target identification area setting module, a cargo order quantity data acquisition and analysis module, a data analysis and feature extraction module, a long short-term memory network prediction module and an intelligent target identification area adjustment module; the initial target identification area setting module sets an initial target identification area range for a machine vision system in an initial stage of cargo sorting, identifies an identification code on the cargo in the initial target identification area range and determines cargo information; the system intelligently adjusts the target identification area of the machine vision system by combining real-time cargo order quantity data analysis and long short-term memory network model prediction, solves the identification difficulty caused by label concentration when the order quantity surges, effectively improves identification precision, sorting speed and system adaptability, reduces false identification and sorting and ensures the efficiency and accuracy of the logistics process.

Description

Technical Field

[0001] This invention relates to the field of logistics and warehousing management technology, and more specifically to an intelligent logistics and warehousing management system based on the Internet of Things. Background Technology

[0002] IoT-based smart logistics and warehousing management refers to utilizing IoT technology to deploy smart devices such as sensors, RFID tags, and cameras within the warehouse to achieve real-time monitoring, tracking, and management of goods, equipment, and the environment. Data collected by these devices enables precise inventory management, temperature and humidity control, goods location tracking, and automated processing, such as automated sorting and handling. Furthermore, combined with big data analytics and machine learning technologies, the system can predict inventory demand, optimize warehouse layout, and improve operational efficiency, achieving intelligent and automated management throughout the entire process. This reduces manual intervention and improves warehouse operational efficiency, accuracy, and flexibility. Ultimately, IoT-based smart logistics and warehousing management can reduce costs, increase customer satisfaction, and adapt to the rapid changes and complex needs of modern supply chains.

[0003] In IoT-based smart logistics warehousing management, automated sorting refers to the process of quickly and accurately classifying and handling goods through automated systems. This process typically relies on conveyor belt systems to move goods from one area to another, while the sorting task is completed by automated sorting equipment. Specifically, the conveyor belt delivers goods to the sorting area or in front of the sorting machine based on their type, size, weight, and label information. The intelligent sorting system uses sensors, visual recognition technology, RFID readers, and other means to identify each item in real time and make corresponding sorting decisions. These sorting decisions can be executed by robotic arms, automated sorting gates, or robots to accurately deliver goods to designated storage locations or transport channels. IoT technology plays a crucial role in this process, ensuring the accuracy of information during sorting through real-time monitoring and data exchange, thus ensuring efficient and low-error warehousing management.

[0004] In existing technologies, goods sorting typically relies on machine vision systems to identify codes (such as barcodes, QR codes, or RFID tags) on goods to determine their type, specifications, and other information. Machine vision systems use high-resolution cameras and image processing algorithms to scan tags and transmit the extracted data to a central control system for analysis, thereby determining the sorting method. However, when order volumes surge and tags become densely packed, the target recognition area of ​​the vision system cannot adaptively adjust, leading to multiple tags clustering in the same area and becoming difficult to distinguish effectively. Because the camera's focal length and resolution are fixed, the system cannot flexibly respond to changes in goods density, potentially resulting in blurred or overlapping tag details, especially in high-density areas. This situation severely impacts recognition accuracy, potentially leading to information extraction failures, classification errors, and even incorrect sorting and order delivery, resulting in serious consequences.

[0005] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0006] The purpose of this invention is to provide an IoT-based intelligent logistics and warehousing management system. This system combines real-time order volume data analysis and long short-term memory network model prediction to intelligently adjust the target recognition area of ​​the machine vision system. This addresses the recognition difficulties caused by dense labeling during order surges. When order volume increases, the system shrinks the recognition area, focusing on denser areas to improve processing efficiency and reduce misidentification. When order volume decreases, the system expands the recognition area to ensure efficient processing of low-density areas. This effectively improves recognition accuracy and sorting speed, reduces classification errors caused by overlapping or blurred labels, significantly enhances sorting accuracy and efficiency, avoids missorting and delivery problems, and ensures a smooth and efficient logistics process, thus solving the problems mentioned in the background technology.

[0007] To achieve the above objectives, the present invention provides the following technical solution: an intelligent logistics and warehousing management system based on the Internet of Things, including an initial target identification area setting module, a goods order volume data collection and analysis module, a data analysis and feature extraction module, a long short-term memory network prediction module, and an intelligent target identification area adjustment module;

[0008] The initial target recognition area setting module sets an initial target recognition area range for the machine vision system in the initial stage of goods sorting. The identification code on the goods is identified within the initial target recognition area range to determine the goods information.

[0009] The goods order volume data collection and analysis module acquires and analyzes the current goods order volume data in real time during goods sorting.

[0010] The data analysis and feature extraction module establishes an analysis set based on real-time order volume data according to different time periods and cargo flow, and extracts feature data related to surges in order volume from it under the monitoring window;

[0011] The Long Short-Term Memory Network (LSTM) prediction module inputs the extracted features into a pre-trained LTM network model. Based on the prediction results of the LTM network model, it accurately predicts the changing trend of goods order volume.

[0012] The intelligent target recognition area adjustment module intelligently adjusts the initial target recognition area based on the prediction results of the long short-term memory network model. Specifically, when the order volume surges, the recognition area is automatically reduced to more efficiently focus on label recognition in dense areas. Conversely, when the order volume decreases, the target recognition area is expanded to efficiently handle goods in low-density areas.

[0013] Preferably, the target recognition area range refers to the initial scanning area set by the machine vision system during the goods sorting process, which is the size and location of the area used to identify the identification code on the goods.

[0014] Preferably, feature data related to order volume surges are extracted from the analysis set. The extracted features include the time interval between order volume peaks and the density of each order occupying the logistics channel per unit time. Within the monitoring window, after analyzing the extracted features, order peak interval time factor and order density factor are generated respectively. The order peak interval time factor quantifies the suddenness and density of order traffic, reflecting whether the time interval between order peaks shortens sharply when order volume surges. The order density factor quantifies the concentration of orders in the logistics channel per unit time, reflecting the congestion of the goods storage and processing area when orders surge.

[0015] Preferably, the order peak interval time factor and order density factor extracted from the analysis set under the monitoring window are input into a pre-trained long short-term memory network model. The long short-term memory network model generates an order increment reference value, and the change trend of goods order volume is predicted based on the order increment reference value.

[0016] Preferably, within the monitoring window, the specific steps for generating the order peak interval time factor after analyzing the time interval characteristics between extracted order volume peaks are as follows:

[0017] Based on real-time order data, peak order volume moments are identified. The peak order volume is the point in time when order volume reaches its highest point. By calculating the time interval between adjacent peak order volumes, a series of peak interval time series is obtained. The expression for the peak interval time series is as follows: ,in, This represents the peak interval time for the i-th order. This represents the time point of the i-th order peak, and n represents the total number of all order peaks;

[0018] By weighting the impact of different time intervals, the urgency of order surges is reflected. The weighted expression is: ,in, Indicates the weighted peak interval time. These are weighting coefficients, which depend on the short-term volatility of the time interval. This is the sum of all weighting coefficients;

[0019] The order peak interval time factor is generated by weighting the peak interval time to quantify the degree of shortening of the order peak interval time during a surge in order traffic, thus measuring the suddenness and density of order traffic. The calculation expression for the order peak interval time factor is as follows: ,in, This is the peak order interval time factor. This is the average of the weighted peak interval times. The maximum interval time, The minimum interval time, The standard deviation of the weighted peak interval time. and This is an adjustment coefficient used to control sensitivity to suddenness and density.

[0020] Preferably, the specific steps for generating an order density factor after analyzing the density characteristics of each order occupying the logistics channel within a unit of time within the monitoring window are as follows:

[0021] Within the monitoring window, first calculate the space occupied by each order within the logistics channel, assuming the space occupied by each order is... , Let be the space occupied by the u-th order within the monitoring window, where u is the order number, representing different orders. Then, calculate the total space occupied by all orders within a unit of time. The expression for this calculation is: ,in, This represents the total space occupied by all orders within a unit of time, where N is the total number of orders processed within a unit of time.

[0022] Based on the layout and volume design of the logistics channel, the maximum capacity is determined. Then, the ratio of the total space occupied by all orders per unit time to the maximum capacity is used to generate a load factor, which quantifies the load level of the logistics channel area. The load factor represents the degree to which goods are currently occupying the logistics channel. The calculation expression is as follows: Where C is the maximum capacity of the logistics channel area. It is the load factor;

[0023] The order density factor is generated by combining the total space occupancy of all orders per unit time with the load factor. This factor reflects the density of goods in the logistics channel. The calculation expression for the order density factor is as follows: ,in, Indicates the order density factor. This indicates the space occupied by all orders within a unit of time under the monitoring window, reflecting the total demand for orders.

[0024] Preferably, the initial target recognition region is intelligently adjusted based on the prediction results of the Long Short-Term Memory network model. The specific steps are as follows:

[0025] First, based on the output of the Long Short-Term Memory network model, we obtain the order increment reference value, which reflects the future trend of the order volume.

[0026] Based on the generated order increment reference value, the adjustment ratio of the initial target recognition area is calculated. The adjustment ratio will affect the size of the initial target recognition area. The formula for calculating the adjustment ratio is: Where k is an adjustment factor used to control the adjustment range of the target recognition region, and its value ranges from (0, 1). This serves as a reference value for order increments predicted by the Long Short-Term Memory (LSTM) network model. It is a constant representing the sensitivity of changes in the order increment reference value, controlling the degree of exponential decay.

[0027] Preferably, the initial target recognition area is adjusted according to the calculated adjustment ratio. Specifically, based on changes in the order increment reference value, the system will intelligently adjust the size of the recognition area. The expression for the target recognition area adjustment is: ,in, The area of ​​the initial target recognition region. The adjusted target recognition area area. Further increase or decrease the adjustment range.

[0028] The technical effects and advantages provided by the present invention in the above technical solution are as follows:

[0029] This invention combines the analysis of real-time order volume data with predictions from a long short-term memory network model, enabling the system to intelligently adjust the target recognition area of ​​the machine vision system. This solves the recognition difficulty caused by dense labels during order surges. When order volume surges, the system automatically shrinks the target recognition area, focusing on label recognition in dense areas, improving system processing efficiency and reducing the risk of misidentification. Conversely, when order volume decreases, the system expands the recognition area, ensuring that goods in low-density areas can be processed efficiently. In this way, the system not only improves recognition accuracy and sorting speed but also enhances adaptability to dynamic changes, effectively reducing classification errors caused by overlapping or blurred labels, significantly improving sorting accuracy and efficiency, avoiding missorting and order delivery problems, and ensuring a smooth and efficient logistics process. Attached Figure Description

[0030] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0031] Figure 1 This is a schematic diagram of the modules of the Internet of Things-based intelligent logistics and warehousing management system of the present invention. Detailed Implementation

[0032] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, they are provided so that the description of this disclosure will be more complete and fully convey the concept of the exemplary embodiments to those skilled in the art.

[0033] This invention provides, for example Figure 1 The IoT-based intelligent logistics and warehousing management system shown includes an initial target recognition area setting module, a goods order volume data collection and analysis module, a data analysis and feature extraction module, a long short-term memory network prediction module, and an intelligent target recognition area adjustment module.

[0034] The initial target recognition area setting module sets an initial target recognition area range for the machine vision system in the initial stage of goods sorting. The system then uses this initial target recognition area range to identify the identification codes (such as barcodes, QR codes, or RFID tags) on the goods and determine the goods information.

[0035] The target recognition area range refers to the initial scanning area set by the machine vision system during the goods sorting process, which is the size and location of the area used to identify identification codes (such as barcodes, QR codes, or RFID tags) on the goods.

[0036] The initial scanning area should cover most of the labels or identification codes on goods under normal circumstances to ensure efficient identification of standard goods. The initial scanning area should be set based on the actual type and specifications of the goods, as well as the size and distribution of the identification codes (barcodes, QR codes, or RFID tags) used. The selection of the initial identification area should take into account the most common goods stacking and label distribution patterns to ensure that the machine vision system can operate normally and effectively identify goods information in most situations.

[0037] The goods order volume data collection and analysis module acquires and analyzes the current goods order volume data in real time during goods sorting.

[0038] The key to this step is synchronizing data with the warehouse management system (WMS) through integrated Internet of Things (IoT) technology to ensure real-time updates of order data. This data reflects not only the quantity of goods at any given time but also information such as the type and specifications of the goods. This real-time data forms the basis for subsequent analysis and adjustments. By monitoring and recording changes in order volume, a better understanding of goods flow patterns and demand fluctuations can be achieved, supporting the subsequent training and regulation of long short-term memory network models.

[0039] The data analysis and feature extraction module establishes an analysis set based on real-time order volume data according to different time periods and cargo flow, and extracts feature data related to surges in order volume from it under the monitoring window;

[0040] Feature data related to order surges are extracted from the analysis set. These features include the time interval between order peaks and the density of each order occupying a logistics channel per unit time. Within the monitoring window, the extracted features are analyzed to generate an order peak interval time factor and an order density factor. The order peak interval time factor quantifies the suddenness and density of order traffic, reflecting whether the time interval between order peaks shortens sharply during an order surge, thereby assessing the urgency of centralized order processing. The order density factor quantifies the concentration of orders in the logistics channel per unit time, reflecting the congestion level of the goods storage and processing areas during an order surge.

[0041] Short intervals between order volume peaks typically indicate a surge in orders. When order volume increases rapidly, the order processing rate may not keep up, leading to a significant shortening of the time interval between peaks. This phenomenon usually occurs during periods of high demand or promotional activities, when the system needs to process a large number of orders in a short period. In such situations, the concentration of orders and increased processing pressure can overload the logistics system and affect processing efficiency. Therefore, monitoring the time interval between order peaks can effectively predict trends of order volume surges.

[0042] Within the monitoring window, the specific steps for generating the order peak interval time factor after analyzing the time interval characteristics between extracted order volume peaks are as follows:

[0043] Based on real-time order data, peak order volume moments are identified. The peak order volume is the point in time when order volume reaches its highest point. By calculating the time interval between adjacent peak order volumes, a series of peak interval time series is obtained. The expression for the peak interval time series is as follows: ,in, This represents the peak interval time for the i-th order. This represents the time point of the i-th order peak, and n represents the total number of all order peaks;

[0044] The purpose of this step is to identify the time distribution of order peaks and capture key time points of order traffic changes.

[0045] By weighting the impact of different time intervals, the urgency of order surges is reflected. The weighted expression is: ,in, Indicates the weighted peak interval time. The weighting factor depends on the short-term volatility of the time interval (e.g., if a time interval is particularly short, it is given a higher weight, i.e., a weighting factor). This is the sum of all weighting coefficients;

[0046] The purpose of this step is to identify abrupt changes in the peak order intervals, providing a dynamic evaluation basis for subsequent calculations.

[0047] The order peak interval time factor is generated by weighting the peak interval time to quantify the degree of shortening of the order peak interval time during a surge in order traffic, thus measuring the suddenness and density of order traffic. The calculation expression for the order peak interval time factor is as follows: ,in, This is the peak order interval time factor. This is the average of the weighted peak interval times. The maximum interval time, The minimum interval time, The standard deviation of the weighted peak interval time. and This is an adjustment coefficient used to control sensitivity to suddenness and intensity.

[0048] This step combines the average, minimum, maximum, and standard deviation of the weighted peak interval time to reflect the urgency of changes in the peak order interval, thereby generating an order peak interval time factor that can accurately assess the urgency of concentrated order processing.

[0049] As can be seen from the order peak interval time factor, within the monitoring window, the smaller the value of the order peak interval time factor generated after analyzing the time interval characteristics between the extracted order volume peaks, the more significant the order volume surge. This is because the order peak interval time factor reflects the suddenness and density of order traffic, especially during order volume surges, when the time interval between peaks usually shortens dramatically. When order volume surges, the processing time of multiple orders becomes closer, leading to a shorter time interval between adjacent order peaks. Therefore, a smaller order peak interval time factor means a stronger suddenness in order traffic changes, requiring the system to process more orders in a shorter time. Conversely, when order volume surges are not significant, the peak interval time is longer, and the order peak interval time factor value is larger, indicating a more gradual change in order traffic and lower system pressure. Therefore, a smaller order peak interval time factor value reflects a higher urgency and surge in order processing.

[0050] A high density of orders occupying a logistics channel per unit of time usually indicates a surge in order volume. In a logistics system, when order volume increases rapidly, storage and processing space becomes increasingly congested. Specifically, order density refers to the number of orders processed in a logistics channel per unit of time. Higher density indicates that more goods need to be processed in the same time period, leading to increased space occupancy and channel congestion. This phenomenon is often a sign of an order surge.

[0051] Within the monitoring window, the specific steps for generating an order density factor after analyzing the density characteristics of each order occupying the logistics channel within a unit of time are as follows:

[0052] Within the monitoring window, first calculate the space occupied by each order within the logistics channel, assuming the space occupied by each order is... , Let be the space occupied by the u-th order within the monitoring window, where u is the order number, representing different orders. Then, calculate the total space occupied by all orders within a unit of time. The expression for this calculation is: ,in, This represents the total space occupied by all orders within a unit of time, where N is the total number of orders processed within a unit of time.

[0053] The space occupied by each order can be determined by its physical dimensions (such as length, width, and height) and its location within the storage area or logistics channel. For non-standard shaped goods, the space occupied also needs to be comprehensively assessed by considering the three-dimensional arrangement of the goods. The purpose of this step is to quantify the space occupied by each order and provide a basis for subsequent density calculations.

[0054] Based on the layout and volume design of the logistics channel, the maximum capacity is determined (the maximum capacity is usually determined comprehensively based on factors such as the channel size and storage capacity of the storage area). Then, the ratio of the total space occupied by all orders per unit time to the maximum capacity is used to generate a load factor, which quantifies the load level of the logistics channel area. The load factor represents the degree to which goods are currently occupying the logistics channel, and the calculation expression is as follows: Where C is the maximum capacity of the logistics channel area. It is the load factor;

[0055] If the load factor is close to 1, it means that the logistics channel area is close to full capacity, which means that the order volume has surged and the logistics channel is highly congested.

[0056] The order density factor is generated by combining the total space occupancy of all orders per unit time with the load factor. This factor reflects the density of goods in the logistics channel. The calculation expression for the order density factor is as follows: ,in, Indicates the order density factor. This indicates the space occupied by all orders within a unit of time under the monitoring window, reflecting the total demand for orders;

[0057] By order density factor It can quantify the concentration of orders in the logistics channel per unit of time, reflecting the congestion of goods storage and processing areas when orders surge.

[0058] As can be seen from the order density factor, within the monitoring window, the higher the value of the order density factor generated after analyzing the density characteristics of each order occupying the logistics channel per unit time, the more significant the surge in order volume. A higher order density factor indicates a higher density of logistics channels occupied by each order per unit time within the monitoring window, meaning more orders per unit space or channel, reflecting a surge in order volume and increased congestion. This increased density means a heavier load on the goods storage and processing areas, requiring the system to respond more sensitively to these changes to avoid processing bottlenecks or congestion. Conversely, a lower order density factor indicates relatively even use of logistics channel space, with less pronounced order volume surges. Therefore, the order density factor, as an indicator reflecting surges, has significant practical value in goods sorting systems, helping to dynamically adjust the system's operating status and sorting strategies.

[0059] The Long Short-Term Memory Network (LSTM) prediction module inputs the extracted features into a pre-trained LTM network model. Based on the prediction results of the LTM network model, it accurately predicts the changing trend of goods order volume.

[0060] The peak order interval time factor and order density factor extracted from the analysis set under the monitoring window are input into a pre-trained long short-term memory network model. The long short-term memory network model generates an order increment reference value, and the change trend of goods order volume is predicted based on the order increment reference value.

[0061] A pre-trained Long Short-Term Memory (LSTM) network model refers to a neural network model that has been thoroughly trained using historical data and known patterns of order volume changes, enabling it to make effective predictions. During training, the LSTM network continuously adjusts its internal weights through backpropagation, allowing it to capture the complex nonlinear relationship between input features (such as peak order interval time factors and order density factors) and the target variable (order increment reference value). By progressively "memorizing" the temporal characteristics of order volume changes during training, the LSTM can accurately predict future order volume fluctuations when faced with new inputs.

[0062] Specifically, the LSTM model is a deep learning method specifically designed for processing time-series data, effectively addressing the long-term dependencies inherent in such data. In the task of predicting order volume, the LSTM model first receives time-series data features such as the peak order interval time factor and order density factor through the input layer. Then, it performs feature extraction and temporal pattern learning through multiple LSTM units. Each LSTM unit contains a "memory unit" that stores and updates long-term information, enabling the model to remember long-term trends affecting order volume. Through the training process, the LSTM learns to identify and utilize regular patterns at different stages of order volume changes, thereby generating predictive order increment reference values. These order increment reference values ​​reflect how order volume will fluctuate or increase in the future, providing data support for the system to dynamically adjust its order processing strategy.

[0063] The advantage of a pre-trained LSTM model lies in its ability to learn effective predictive patterns from a large amount of historical data during the training phase, avoiding the time and computational cost of retraining the model in practical applications. During training, the LSTM model can accurately capture the periodicity and trends of order volume fluctuations through its Long Short-Term Memory (LSTM) mechanism, especially when facing complex situations such as seasonal changes and sudden surges in demand. Because the LSTM network has strong memory capabilities, it can retain key temporal information from historical data, thus exhibiting high accuracy and stability in order volume prediction.

[0064] This pre-trained LSTM model is crucial for dynamically adjusting the response strategy of the goods sorting system. Specifically, by inputting the extracted order peak interval time factor and order density factor, the LSTM model can not only accurately predict the future trend of order volume changes, but also provide the system with real-time predictions based on order increment reference values. This helps the system react promptly when order volume surges, avoiding congestion and resource waste. Through this intelligent prediction and response, the goods sorting system can make more efficient dynamic adjustments, thereby improving overall logistics efficiency and customer satisfaction.

[0065] The Long Short-Term Memory (LSTM) network model is not specifically limited here, but it can achieve the function of factoring the peak order interval time. and order density factor As a feature vector, it is used for comprehensive analysis to generate a reference value for order increment. Any machine learning model is acceptable. To achieve the technical solution of this invention, this invention provides a specific implementation method; order increment reference value. The generated calculation formula is: In the formula, , These are the peak order interval time factors. and order density factor The preset proportional coefficient, and , All are greater than 0.

[0066] As can be seen from the order increment reference value, within the monitoring window, the smaller the performance value of the order peak interval time factor generated after analyzing the time interval characteristics between the extracted order volume peaks, the larger the performance value of the order density factor generated after analyzing the density characteristics of the logistics channels occupied by each order within a unit time. In other words, the larger the performance value of the order increment reference value generated when predicting the changing trend of goods order volume through the pre-trained long short-term memory network model, the more obvious the surge in order volume, and vice versa.

[0067] The intelligent target recognition area adjustment module intelligently adjusts the initial target recognition area based on the prediction results of the long short-term memory network model. Specifically, when the order volume surges, the recognition area is automatically reduced to more efficiently focus on the label recognition of dense areas, improve the system's processing capacity, and reduce the risk of misidentification. Conversely, when the order volume decreases, the target recognition area is expanded to efficiently handle goods in low-density areas.

[0068] Based on the prediction results of the Long Short-Term Memory (LSTM) network model, the goal of intelligently adjusting the initial target recognition region is to dynamically adjust the target recognition region range of the machine vision system by evaluating the order increment reference value (the rate and trend of order volume change) in real time.

[0069] Based on the prediction results of the Long Short-Term Memory network model, the initial target recognition region is intelligently adjusted. The specific steps are as follows:

[0070] First, based on the output of the Long Short-Term Memory network model, we obtain the order increment reference value, which reflects the future trend of the order volume.

[0071] A higher order increment reference value indicates a surge in order volume, while a lower order increment reference value indicates a decrease or stabilization in order volume.

[0072] Based on the generated order increment reference value, the adjustment ratio of the initial target recognition area is calculated. This adjustment ratio affects the size of the initial target recognition area, thus determining whether to shrink or expand the recognition area. If the order increment reference value indicates a surge in order volume, the target recognition area will shrink; conversely, it will expand. The expression for calculating the adjustment ratio is: Where k is an adjustment factor used to control the adjustment range of the target recognition region, and its value is generally between (0, 1). This serves as a reference value for order increments predicted by the Long Short-Term Memory (LSTM) network model. It is a constant representing the sensitivity to changes in the order increment reference value, controlling the degree of exponential decay. This parameter can adjust the system's sensitivity to surges in order volume.

[0073] If the order increment reference value Larger (indicating a surge in order volume). The value will approach 0, indicating that the recognition area will be reduced; if the order increment reference value Smaller (reduced order volume) A value close to 1 indicates that the recognition area will expand.

[0074] The initial target recognition area is adjusted according to the calculated adjustment ratio. The initial target recognition area refers to the area used by the machine vision system to scan product labels under normal circumstances. Based on changes in the order increment reference value, the system will intelligently adjust the size of the recognition area to more efficiently sort goods and identify labels. The expression for the target recognition area adjustment is: ,in, The area of ​​the initial target recognition region. The adjusted target recognition area area. Further increase or decrease the adjustment range, depending on the trend of order volume changes, allowing for stronger area reduction during surges or more moderate area expansion during troughs;

[0075] The adjusted identification area will automatically shrink or expand based on the prediction results of the order increment reference value, so as to better adapt to the current changes in order volume and optimize the sorting process.

[0076] Based on the predictions of the Long Short-Term Memory (LSTM) network model, the intelligent adjustment of the initial target recognition area dynamically optimizes the operational efficiency and accuracy of the machine vision system to cope with fluctuations in order volume. When order volume surges, the system automatically narrows the recognition area, focusing on densely packed areas of goods, reducing unnecessary scanning interference, improving label recognition accuracy and processing speed, and avoiding errors or delays caused by information overload. Conversely, when order volume decreases, the system expands the recognition area, increasing coverage of low-density goods areas, thus avoiding missed detections due to an insufficiently small area. This adjustment strategy ensures that the sorting system can optimize its operating range in real time according to changes in order volume, maximizing processing efficiency and resource utilization while reducing error rates and operational delays, providing strong support for efficient management in dynamic logistics environments.

[0077] This invention combines the analysis of real-time order volume data with predictions from a long short-term memory network model, enabling the system to intelligently adjust the target recognition area of ​​the machine vision system. This solves the recognition difficulty caused by dense labels during order surges. When order volume surges, the system automatically shrinks the target recognition area, focusing on label recognition in dense areas, improving system processing efficiency and reducing the risk of misidentification. Conversely, when order volume decreases, the system expands the recognition area, ensuring that goods in low-density areas can be processed efficiently. In this way, the system not only improves recognition accuracy and sorting speed but also enhances adaptability to dynamic changes, effectively reducing classification errors caused by overlapping or blurred labels, significantly improving sorting accuracy and efficiency, avoiding missorting and order delivery problems, and ensuring a smooth and efficient logistics process.

[0078] The foregoing has only described certain exemplary embodiments of the present invention by way of illustration. Undoubtedly, those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the foregoing drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. An intelligent logistics and warehousing management system based on the Internet of Things, characterized in that, It includes an initial target recognition region setting module, a goods order volume data collection and analysis module, a data analysis and feature extraction module, a long short-term memory network prediction module, and an intelligent target recognition region adjustment module; The initial target recognition area setting module sets an initial target recognition area range for the machine vision system in the initial stage of goods sorting. The identification code on the goods is identified within the initial target recognition area range to determine the goods information. The goods order volume data collection and analysis module acquires and analyzes the current goods order volume data in real time during goods sorting. The data analysis and feature extraction module establishes an analysis set based on real-time order volume data according to different time periods and cargo flow, and extracts feature data related to surges in order volume from it under the monitoring window; The Long Short-Term Memory Network (LSTM) prediction module inputs the extracted features into a pre-trained LTM network model. Based on the prediction results of the LTM network model, it accurately predicts the changing trend of goods order volume. The intelligent target recognition area adjustment module intelligently adjusts the initial target recognition area based on the prediction results of the long short-term memory network model. Specifically, when the order volume surges, the recognition area is automatically reduced to more efficiently focus on label recognition in dense areas. Conversely, when the order volume decreases, the target recognition area is expanded to efficiently handle goods in low-density areas.

2. The IoT-based intelligent logistics and warehousing management system according to claim 1, characterized in that, The target recognition area range refers to the initial scanning area set by the machine vision system during the goods sorting process, which is the size and location of the area used to identify the identification code on the goods.

3. The IoT-based intelligent logistics and warehousing management system according to claim 1, characterized in that, Feature data related to order surges are extracted from the analysis set. The extracted features include the time interval between order peaks and the density of each order occupying the logistics channel per unit time. Within the monitoring window, after analyzing the extracted features, order peak interval time factor and order density factor are generated respectively. The order peak interval time factor quantifies the suddenness and density of order traffic, reflecting whether the time interval between order peaks shortens sharply when order volume surges. The order density factor quantifies the concentration of orders in the logistics channel per unit time, reflecting the congestion of the goods storage and processing area when orders surge.

4. The IoT-based intelligent logistics and warehousing management system according to claim 3, characterized in that, The peak order interval time factor and order density factor extracted from the analysis set under the monitoring window are input into a pre-trained long short-term memory network model. The long short-term memory network model generates an order increment reference value, and the change trend of goods order volume is predicted based on the order increment reference value.

5. The IoT-based intelligent logistics and warehousing management system according to claim 3, characterized in that, Within the monitoring window, the specific steps for generating the order peak interval time factor after analyzing the time interval characteristics between extracted order volume peaks are as follows: Based on real-time order data, peak order volume moments are identified. The peak order volume is the point in time when order volume reaches its highest point. By calculating the time interval between adjacent peak order volumes, a series of peak interval time series is obtained. The expression for the peak interval time series is as follows: ,in, This represents the peak interval time for the i-th order. This represents the time point of the i-th order peak, and n represents the total number of all order peaks; By weighting the impact of different time intervals, the urgency of order surges is reflected. The weighted expression is: ,in, Indicates the weighted peak interval time. These are weighting coefficients, which depend on the short-term volatility of the time interval. It is the sum of all weighted coefficients.

6. The IoT-based intelligent logistics and warehousing management system according to claim 3, characterized in that, Within the monitoring window, the specific steps for generating an order density factor after analyzing the density characteristics of each order occupying the logistics channel within a unit of time are as follows: Within the monitoring window, first calculate the space occupied by each order within the logistics channel, assuming the space occupied by each order is... , Let be the space occupied by the u-th order within the monitoring window, where u is the order number, representing different orders. Then, calculate the total space occupied by all orders within a unit of time. The expression for this calculation is: ,in, This represents the total space occupied by all orders within a unit of time, where N is the total number of orders processed within a unit of time. Based on the layout and volume design of the logistics channel, the maximum capacity is determined. Then, the ratio of the total space occupied by all orders per unit time to the maximum capacity is used to generate a load factor, which quantifies the load level of the logistics channel area. The load factor represents the degree to which goods are currently occupying the logistics channel. The calculation expression is as follows: Where C is the maximum capacity of the logistics channel area. It is the load factor.

7. The IoT-based intelligent logistics and warehousing management system according to claim 4, characterized in that, Based on the prediction results of the Long Short-Term Memory network model, the initial target recognition region is intelligently adjusted. The specific steps are as follows: First, based on the output of the Long Short-Term Memory network model, we obtain the order increment reference value, which reflects the future trend of the order volume. Based on the generated order increment reference value, the adjustment ratio of the initial target recognition area is calculated. The adjustment ratio will affect the size of the initial target recognition area. The formula for calculating the adjustment ratio is: Where k is an adjustment factor used to control the adjustment range of the target recognition region, and its value ranges from (0, 1). This serves as a reference value for order increments predicted by the Long Short-Term Memory (LSTM) network model. γ is a constant representing the sensitivity of changes in the order increment reference value, controlling the degree of exponential decay; γ is the adjustment coefficient.

8. The IoT-based intelligent logistics and warehousing management system according to claim 7, characterized in that, The initial target recognition area is adjusted according to the calculated adjustment ratio. Specifically, based on changes in the order increment reference value, the system will intelligently adjust the size of the recognition area. The expression for the target recognition area adjustment is: ,in, The area of ​​the initial target recognition region. The adjusted target recognition area area. Further increase or decrease the adjustment range.

9. The IoT-based intelligent logistics and warehousing management system according to claim 5, characterized in that, The order peak interval time factor is generated by weighting the peak interval time to quantify the degree of shortening of the order peak interval time during a surge in order traffic, thus measuring the suddenness and density of order traffic. The calculation expression for the order peak interval time factor is as follows: ,in, This is the peak order interval time factor. This is the average of the weighted peak interval times. The maximum interval time, The minimum interval time, The standard deviation of the weighted peak interval time. and This is an adjustment coefficient used to control sensitivity to suddenness and density.

10. The IoT-based intelligent logistics and warehousing management system according to claim 6, characterized in that, The order density factor is generated by combining the total space occupancy of all orders per unit time with the load factor. This factor reflects the density of goods in the logistics channel. The calculation expression for the order density factor is as follows: ,in, Indicates the order density factor. This indicates the space occupied by all orders within a unit of time under the monitoring window, reflecting the total demand for orders.

Images