Goods dispatching and distribution management system suitable for stocker vertical warehouse

Abstract

The invention discloses a cargo scheduling and distribution management system suitable for a stocker three-dimensional warehouse, and relates to the technical field of logistics warehousing, and the system comprises an intelligent scheduling module which employs a multi-algorithm fusion and dynamic switching mechanism, plans the path of a stocker according to an order and the condition in the warehouse, and distributes tasks; the accurate inventory management module is used for guaranteeing the accuracy of inventory data by applying a multi-sensor fusion and intelligent identification technology and combining environment self-adaption and intelligent auditing; the system integration module is used for realizing multi-system integration through a standardized interface and middleware and has a cluster deployment, fault processing and interface management mechanism; an equipment fault handling module; the energy management module is used for optimizing energy utilization based on an energy-efficiency balance model in combination with hybrid energy and an intelligent switching technology; according to the invention, through close cooperation of a plurality of modules and comprehensive application of innovative technical means, efficient, intelligent and reliable operation of the stacker three-dimensional warehouse cargo scheduling and distribution management system is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of logistics warehousing, and in particular to a cargo dispatching and distribution management system suitable for a stacker warehouse. Background Art

[0002] In the field of modern logistics and warehousing, stacker warehouses occupy an important position. However, traditional management methods rely on manual experience and have many defects. In terms of path planning, it is difficult for humans to determine the optimal path, which makes the operation of the stacker time-consuming and energy-consuming. In terms of inventory management, there is a lack of accurate real-time performance, which is prone to backlogs or out-of-stock goods, affecting business operations. The system has low integration and poor data interaction with external systems, forming information islands. Insufficient response to equipment failures, lack of intelligent detection and early warning, and easy chaos in operations when failures occur. Energy management is extensive, unable to be flexibly adjusted, and wastes energy. The personnel operation and training system is poor, the interface is unfriendly, and training updates are lagging behind. There is also a lack of effective artificial intelligence prediction, low efficiency in collaborative operations, unstable communications, and poor task allocation and traffic management.

[0003] In view of this, this application is hereby filed. Summary of the invention

[0004] The object of the present invention is to provide a cargo dispatching and distribution management system suitable for stacker vertical warehouses, so as to solve the problem of low efficiency of cargo dispatching and distribution in existing stacker vertical warehouses mentioned in the above background technology.

[0005] In order to solve the above technical problems, the present invention provides a cargo dispatching and distribution management system suitable for stacker warehouses, including an intelligent dispatching module: adopting multi-algorithm fusion and dynamic switching mechanism, planning the stacker path and assigning tasks according to orders and warehouse conditions, and having special cargo handling strategies; a precise inventory management module: using multi-sensor fusion and intelligent recognition technology, combined with environmental adaptation and intelligent auditing to ensure accurate inventory data; a system integration module: realizing multi-system integration with standardized interfaces and middleware, and having cluster deployment, fault handling and interface management mechanisms; an equipment fault response module: realizing fault detection and early warning with the help of multi-sensors and machine learning algorithms, and optimizing emergency dispatch using distributed computing and real-time monitoring; an energy management module: optimizing energy utilization based on an energy-efficiency balance model combined with hybrid energy and intelligent switching technology; a personnel operation and training assistance module: providing a personalized, safe and customized interface and a high-quality training system; an artificial intelligence prediction module: integrating multi-source data and assisting decision-making with visual multi-dimensional interactive tools; an automated three-dimensional warehouse and robot collaborative operation module: adopting redundant communication networks and dynamic task management strategies to ensure efficient collaboration.

[0006] Furthermore, the multi-algorithm fusion of the intelligent scheduling module includes a combination of path optimization algorithms such as the A* algorithm or the Dijkstra algorithm and a task allocation algorithm, and a dynamic switching mechanism switches between conventional and fast scheduling algorithms based on the order volume and computing resource conditions. At the same time, the special cargo handling strategy includes a special cargo database and its real-time update and targeted scheduling rule formulation.

[0007] Furthermore, the multi-sensor fusion of the precise inventory management module adopts Kalman filtering or fuzzy logic algorithm to process sensor data such as RFID and laser scanning, and the environmental adaptation adjusts the data fusion weight according to regional differences. The intelligent review includes a combination of artificial intelligence pre-screening and manual review.

[0008] Furthermore, the middleware of the system integration module adopts an enterprise service bus, the cluster deployment has a node fault isolation and self-healing function, and the interface management includes a version compatibility mechanism and a multi-system compatibility dynamic monitoring platform.

[0009] Furthermore, the machine learning algorithm of the equipment fault response module adopts a hybrid intelligent scheduling algorithm that integrates genetic algorithm and A* algorithm, the distributed computing architecture accelerates the data processing of the incremental learning algorithm, and the real-time monitoring adopts event-driven architecture and intelligent caching technology to optimize the generation of scheduling solutions.

[0010] Furthermore, the energy-efficiency balance model of the energy management module is combined with a real-time order correction mechanism to adjust the stacker energy strategy, and the hybrid energy intelligent switching system includes voltage stabilization filtering and pre-switching buffer functions.

[0011] Furthermore, the personalized safety customization interface of the personnel operation and training auxiliary module is provided with a safety audit mechanism, and the automatic update mechanism of the training system includes a manual audit link to ensure the quality of training.

[0012] Furthermore, the multi-source data fusion of the artificial intelligence prediction module includes an external data quality assessment and cleaning system, and the visual multi-dimensional interactive tool uses technologies such as pivot tables and interactive dashboards to achieve multi-dimensional analysis and display.

[0013] Furthermore, the redundant communication network of the automated stereoscopic warehouse and the robot collaborative operation module adopts double buffering technology and seamless switching mechanism to ensure data transmission, and the dynamic task management strategy adjusts the robot task priority and path planning in real time according to the task urgency and equipment status.

[0014] Compared with the prior art, the present invention has the following beneficial effects: 1. The intelligent scheduling module uses the A* algorithm to plan paths and assign tasks. Based on comprehensive considerations of multiple factors, it accurately determines the path of the stacker and reduces the running distance and time. In the face of a large number of orders, reasonable arrangements can be made based on the degree of urgency, such as being able to quickly process orders during e-commerce promotions. The automated three-dimensional warehouse and robot collaborative operation module optimizes the communication and management system, using algorithms to reduce robot conflicts and waiting, ensuring smooth operations during busy periods, accelerating cargo handling, and improving operational efficiency.

[0015] 2. The system integration module uses ESB middleware to achieve efficient interactive integration with external systems, clearly define interfaces and specifications, and realize information sharing and collaboration, such as synchronizing data between inventory and ERP systems to optimize resource allocation. Its perfect interface management and testing mechanism can ensure compatibility during upgrades, reduce maintenance cost risks, and enhance system performance; 3. The energy management module uses EMS to monitor and analyze the energy consumption of the stacker, adjusts parameters according to the operation and equipment status, combines renewable energy and builds a balance model, and adopts flexible strategies based on order urgency and energy reserves, such as slowing down to use solar energy during off-peak hours, and appropriately relaxing energy consumption restrictions in emergencies, to achieve efficient energy utilization and reasonable allocation; optimize resource utilization; 4. The artificial intelligence prediction module uses a variety of algorithms to predict the demand for goods in and out of the warehouse. The artificial intelligence prediction module integrates multi-source data and uses visual multi-dimensional interactive tools to assist in decision-making. The process of integrating multi-source data involves mining and analysis of industrial data, such as order data, inventory data, etc. It mines the relationships and rules between data and then uses visualization tools to display them to assist decision-making. This is a typical industrial data mining and analysis application; it makes predictions accurate and helps scientific decision-making. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 The schematic diagram is a functional block diagram of a cargo dispatching and distribution management system suitable for stacker warehouses. DETAILED DESCRIPTION

[0017] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0018] Embodiment 1: Suppose there is a large food and beverage production company with a wide range of products and a large demand for storage and distribution of raw materials and finished products. The company has a modern stacker warehouse with an area of ​​8,000 square meters, a shelf height of 12 meters, and a total of 8,000 cargo spaces. The company handles an average of 150 orders per day, each of which contains 5-100 different goods, covering various types of food raw materials, packaging materials, and finished beverages and foods.

[0019] See also Figure 1 The present invention provides a technical solution: a cargo dispatching and distribution management system suitable for stacker warehouses, including an intelligent dispatching module: the real-time path optimization algorithm of the intelligent dispatching module adopts the A* algorithm or the Dijkstra algorithm, and reduces the algorithm complexity through pruning operations and heuristic function optimization, and adopts high-performance servers or distributed computing architectures to improve computing resources to cope with large-scale data processing; and the intelligent dispatching module also includes a dynamic algorithm switching mechanism, which automatically switches to a simplified version of the fast dispatching algorithm when it is detected that the order volume exceeds a certain threshold and computing delays begin to occur, first preliminarily allocates urgent orders to ensure that the goods can be moved as soon as possible, and then uses idle computing resources to gradually optimize the path, and establishes a task hierarchy. The transition algorithm takes into account the initial position of the stacker assigned the task, the completed handling progress and the optimized path planning when the algorithm switches, to ensure a smooth task transition and reduce delays. It also includes a special cargo database. Before assigning tasks, the type of special cargo is identified first, and then a special scheduling strategy is formulated for it based on its special requirements. For fragile goods, bumpy sections are avoided during path planning, the acceleration and deceleration rates of the stacker are limited, and equipment corresponding to stacker operators with rich operating experience is given priority. A real-time monitoring and update mechanism for the special cargo database is established, and information sharing channels are established with suppliers, R&D departments, etc. Once there is new special cargo or special requirement change information, the database is updated immediately, and the scheduling strategy is adjusted and optimized simultaneously.

[0020] 1. Path planning and task allocation: Stacker A is located at coordinates XA=5, YA=3, ZA=4; Stacker B is located at XB=10, YB=6, ZB=5; the target location coordinates are XT=15, YT=8, ZT=7. The cost of moving one grid in the horizontal direction is 1.5, and the cost of moving one grid in the vertical direction is 3.

[0021] According to the A* algorithm formula A= , first calculate the distance from stacker A to the target location: ; Assume that we move from level A to ,but Increase by 1.5, and then move vertically to ,but Increase ,at this time , (This is just a preliminary calculation; the actual calculation needs to consider every step of the entire path.) Through such calculations, the system can compare the value, select the optimal path.

[0022] Assume there are three stackers, the straight-line distance between stacker A and the target cargo is: rice, The distance of stacker C is: rice, The distance of stacker C is: rice, Based on the principle of proximity, it is allocated to stacker B first.

[0023] Assume that the order urgency is divided into high (0.8), medium (0.5), and low (0.2). The current order urgency is high. There are 5 stackers, and the load conditions are as follows: Stacker 1 has been assigned task workload , Stacker 2 workload , Stacker 3 workload , Stacker 4 workload , Stacker 5 workload , average workload Calculate the deviation , , , , , preferred The corresponding stacker is stacker 3, but because stacker B is closer and more urgent, it is still given priority after comprehensive consideration.

[0024] 2. Special cargo handling: A special cargo scheduling priority model can be established, assuming that the fragility of special cargo is (value ranges from 0 to 1, the closer to 1, the more fragile), the degree of deterioration is (value ranges from 0 to 1, the closer to 1, the more likely it is to deteriorate) The fragility of a particular cargo , the degree of perishability , set the priority of special cargo dispatch (The weight coefficient is set according to the enterprise experience), then ,according to The special cargo tasks are prioritized. The special cargo will be handled first and assigned to the corresponding equipment of the operator with rich operation experience. The bumpy sections are avoided during the path planning. Assuming that the bumpy sections occupy 8% of the warehouse floor area, they are mainly distributed in the passages near the cargo loading and unloading area at the edge of the warehouse. The acceleration and deceleration rate of the stacker is limited to no more than 0.3m / s. 2 , deceleration does not exceed 0.2m / s 2 .

[0025] A cargo dispatching and distribution management system suitable for stacker warehouses includes a precise inventory management module: including RFID radio frequency identification and laser scanning sensors, and adopting a data fusion algorithm based on Kalman filtering or fuzzy logic to fuse data from different sensors, and establishing a sensor calibration mechanism to regularly calibrate sensors to reduce data errors; and, adding an environmental adaptation module to the data fusion algorithm, through additional environmental sensors, such as dust concentration sensors, metal detectors, to monitor environmental changes in real time, and automatically adjust the weights or parameters of the data fusion algorithm when environmental interference exceeds a certain range, and optimize the threshold setting and judgment logic of the environmental adaptation module, combined with the warehouse Layout and division of operating areas, setting differentiated tolerance for environmental changes for different areas; also includes intensive training of the intelligent recognition system, collecting more subtle feature data of goods with similar appearance, and adopting a more sophisticated deep learning model structure, such as introducing an attention mechanism to enable the model to focus on the key difference features of the goods. At the same time, a manual review mechanism is established. When the similarity of the recognition results is high and there are doubts, the manual review process is automatically triggered to ensure recognition accuracy, and artificial intelligence is introduced to assist in manual review. Machine learning models are used to pre-screen and classify goods suspected of being misidentified, giving priority to high-risk goods, such as those with high value and small quantity, to improve review efficiency.

[0026] 1. Data collection and fusion: The dimensions of a batch of beverage boxes are length Meter, width Meter, height Meter. Laser scanning sensor measures cargo weight Kilograms, the theoretical weight recorded in the database associated with the RFID sensor is kilogram.

[0027] Assume that the state equation of the measurement system is (in , , , is zero-mean Gaussian noise, the covariance matrix ), the measurement equation is ( is the measurement matrix, is zero-mean Gaussian noise, and the covariance matrix is ), first initialize the state estimate and the covariance matrix , according to the Kalman filter formula: ; , Get a more accurate weight estimate kilogram.

[0028] 2. Environmental adaptation and recognition optimization: When dust concentration (Exceeding the threshold ), metal interference intensity Micro Tesla (above threshold Micro Tesla), set the environmental impact factor (The coefficient is determined based on experimental data), then .according to The value adjusts the sensor weight. Normally, the laser scanning sensor data weight is , the RFID sensor data weight is , the adjusted laser scanning sensor data weight is , the RFID sensor data weight is .

[0029] Assume that a convolutional neural network (CNN) is used for cargo identification. The network structure includes 3 convolutional layers, 2 pooling layers, and 2 fully connected layers. The input cargo image is extracted through the convolutional layer, the pooling layer is used for dimensionality reduction, and the fully connected layer is used for classification. The cosine similarity formula is used to calculate the recognition similarity. , assuming that the feature vector of a certain product and the target product , ,but , , , , If the similarity exceeds , automatically triggering the manual review process.

[0030] A cargo dispatching and distribution management system suitable for stacker warehouses includes a system integration module: the middleware of the system integration module adopts an enterprise service bus (ESB) middleware, and follows the low-coupling design principle in modular design, strictly defines the interface standards and data interaction specifications between modules, and establishes a communication mechanism and coordination framework between modules; in addition, the middleware is clustered and deployed, and load balancing and caching technologies are used. When the network is unstable, the cache is used to temporarily store data, and the data is automatically reissued after the network is restored. At the same time, the middleware performance is regularly stress-tested, the cluster configuration is dynamically adjusted according to the test results, and a node fault isolation and self-healing mechanism of the middleware cluster is established. When a node fails, it is automatically isolated and its Tasks are dynamically assigned to other healthy nodes, and self-repair programs are started to diagnose and repair faulty nodes. It also includes establishing an interface version management mechanism to ensure that the new interface and the old interface can coexist and be compatible for a period of time when the module function is upgraded. When modifying the interface standard, the relevant module development team is notified in advance, and detailed interface change documents and adaptation sample code are provided. At the same time, an automated interface compatibility testing tool is developed to conduct comprehensive compatibility testing after the module upgrade, to promptly discover and solve problems, and to establish a multi-system interface version compatibility dynamic monitoring platform to monitor the changes in each system interface version and data interaction in real time, to provide early warning of possible compatibility issues, and to provide automated compatibility testing scripts and repair suggestions.

[0031] 1. Data interaction and modular design: The inventory management module exchanges data such as inventory quantity and inventory cost with the ERP system. Assume that within a certain period of time, the inventory management module transmits data to the ERP system. Successfully transferred times, the data interaction reliability , through monitoring The value ensures the stability of data interaction. The value is lower than the set threshold (such as ), check the data transmission line, interface and other issues.

[0032] 2. Middleware deployment and management: Initially each middleware node can handle requests / second, using Nginx as a load balancer to evenly distribute data requests to four middleware nodes. In the stress test, the simulation requests / second. If the average response time exceeds 400 milliseconds, add a node. The load is Requests / second, load balancing For example, the actual loads of the four nodes are , , , ,but .like If the value is too large, it means that the load is unbalanced and the load balancing strategy needs to be adjusted.

[0033] Establish an interface version management mechanism. Assuming that the inventory management module function is upgraded, the new interface and the old interface can coexist and be compatible for 4 months. When modifying the interface standard, notify the relevant module development team 3 weeks in advance and provide detailed interface change documents and adaptation sample code. Develop an automated interface compatibility testing tool to conduct a comprehensive compatibility test after the module is upgraded. Assume that the interface compatibility index is , based on comprehensive evaluation of factors such as interface parameter matching and data format consistency, if The value is lower than the set threshold (such as ), it indicates a compatibility issue that requires further inspection and modification.

[0034] A cargo dispatching and distribution management system suitable for stacker warehouses includes an equipment failure response module: the machine learning algorithm of the equipment failure response module adopts a hybrid intelligent dispatching algorithm that integrates a genetic algorithm and an A* algorithm to improve the adaptability and stability of the fault emergency dispatching strategy, and establishes an algorithm adaptive adjustment mechanism to automatically adjust algorithm parameters or switch algorithm strategies according to changes in the warehouse environment; and, an incremental learning algorithm is used to update the fault prediction model. When a new fault case occurs, its data is promptly included in the training set so that the model can continuously learn new fault characteristics. At the same time, a fault case library is established to record and analyze new faults in detail so as to adjust the model in time and formulate corresponding response measures, and a distributed computing architecture is adopted. Accelerate the data processing of the incremental learning algorithm, distribute new fault data to multiple computing nodes for parallel processing, and improve the model update speed; it also includes adding real-time monitoring and dynamic adjustment mechanisms to the fault emergency dispatch strategy. During the emergency dispatch process, continuously monitor the real-time changes of the equipment and environment in the warehouse. When unexpected situations such as backup equipment failure or channel obstruction are found, immediately re-plan the dispatch plan, use other available resources or adjust the transportation route to ensure that the goods can continue to be dispatched smoothly, and optimize the real-time monitoring information processing process. The event-driven architecture and intelligent caching technology are used. When key information in the warehouse, such as channel status, location, etc., changes, the corresponding dispatch plan adjustment program is immediately triggered, and the cached data is used to quickly generate a new dispatch path.

[0035] 1. Data collection and algorithm application: The motor temperature data collected by the temperature sensor is , , The vibration amplitude data collected by the vibration sensor is mm, Millimeters, etc. The current data collected by the current sensor is , wait.

[0036] Assume that the fitness function of the genetic algorithm is , assuming , during the training process, if the accuracy , false alarm rate ,but: , by continuously iteratively updating the parameter vector Make To achieve the optimum, the parameters of the fault detection model are optimized.

[0037] 2. Model update and emergency handling: The incremental learning algorithm is used to update the fault prediction model. Assuming that the accuracy of the fault prediction model evaluation index is , in the initial model, , , , ,but When a new fault case appears and its data is included in the training set, the accuracy is recalculated. If the accuracy is improved to , it means that the model performance is improved.

[0038] During the emergency dispatch process, it is assumed that the normal dispatch time Minutes, emergency dispatch completion time minutes, the emergency dispatch efficiency , through monitoring The value is used to evaluate the effectiveness of the emergency dispatch strategy. If the value is low, the emergency dispatch plan is adjusted, such as optimizing resource allocation and improving path planning.

[0039] A cargo dispatching and distribution management system suitable for stacker warehouses includes an energy management module: the energy management module also includes establishing an energy-efficiency balance model, dynamically adjusting the energy optimization strategy of the stacker according to the urgency of the order and the energy reserve situation, appropriately relaxing the energy consumption restrictions when the order is urgent to ensure that the goods can be quickly put in and out of the warehouse, and increasing the energy optimization efforts and reducing the operating speed and power when the order volume is small. At the same time, by predicting the order volume and energy demand, the stacker's operation plan is arranged in advance to make energy utilization more reasonable; and a hybrid energy storage and intelligent switching system is used to combine renewable energy such as solar energy and wind energy with traditional electricity , and equipped with energy storage equipment, such as battery packs. When renewable energy is sufficient, it will give priority to renewable energy and store excess electricity. When renewable energy is insufficient, it will automatically switch to traditional power supply and dynamically adjust the power use strategy according to the energy storage situation and energy prices. A real-time order correction mechanism is introduced to supplement the energy-efficiency balance model. Combined with the actual operation situation at the warehouse site and the order processing progress, the order urgency prediction value is corrected in real time to ensure accurate and reasonable energy allocation; it also includes optimizing the control logic and hardware performance of the hybrid energy switching device, adding voltage stabilization and filtering functions, and setting up pre-switching detection and buffer mechanisms to ensure a smooth transition during the energy switching process.

[0040] 1. Monitoring and parameter adjustment: The power consumption data of the stacker during a certain period of time is kWh, kWh, etc., and the operating speed is m / s, load weight is kilogram.

[0041] When the order volume , , Meters per second, according to the formula , m / s, that is, the adjusted speed is about Meters / second.

[0042] Assume that the energy consumption and workload relationship model , obtained by fitting the experimental data , , , then when , , , Hours (minutes) Kilowatt-hours, the model can be used to estimate energy consumption in different operating situations and adjust operating parameters.

[0043] Assuming the area of ​​solar panels on the roof of the warehouse Square meter, photoelectric conversion efficiency , local sunshine intensity kW / m2, time hours, the solar power generation is Kilowatt-hours, when solar energy is sufficient, it can be used first and excess electricity can be stored.

[0044] 2. Energy strategy and switching optimization: When the order is urgent , kWh / hour, When, according to the formula: , kWh / hour, appropriately relax energy consumption limits to ensure that goods can be quickly put in and out of the warehouse.

[0045] Assume that the ARIMA model is used to predict the order volume for the next week. Let the order quantity sequence be , establish an ARIMA(2,1,2) model (the order here is just an example), that is ,in , , estimated from historical data , , , Assuming the current order volume , then the next order quantity is predicted: (Here we also need to base our calculations on the residual sequence The historical data is calculated, and for simplicity, it is assumed etc. are all 0), then, (Assumption ), and the predicted order quantity for the next period is The required energy amount is calculated based on the prediction results and the energy consumption characteristics of the stacker, and the energy supply strategy is adjusted in advance.

[0046] Assume that energy price threshold is set Yuan / kWh, when the price of traditional electricity is higher than When the energy storage device has sufficient power, it will be used first. Yuan / kWh, traditional electricity cost Yuan / kWh, assuming that the decision variable for using renewable energy is , the decision variables for using traditional electricity are , the objective function is to minimize the cost , while meeting energy demand constraints ( is the total energy demand, The amount of renewable energy available, is the amount that conventional electricity can provide). For example, if kWh, kWh, kWh, a linear programming model can be established to solve it. If the price of renewable energy is low and the electricity is sufficient, renewable energy will be used first, that is, Will be larger.

[0047] Assume that during the energy switching process, the voltage value sequence before the switching is collected , average voltage value , then the RMS value of the voltage fluctuation amplitude is .

[0048] like The value exceeds the set threshold (such as ), it is necessary to further optimize the control logic and hardware performance of the hybrid energy switching device, such as adjusting the parameters of the voltage stabilization circuit or increasing the capacity of the filter capacitor.

[0049] A cargo dispatching and distribution management system suitable for stacker warehouses includes a personnel operation and training auxiliary module: the personnel operation and training auxiliary module provides a personalized customization function for a human-machine interface (HMI), allowing operators to adjust the interface layout, operation button position and function display mode according to their own habits and needs. At the same time, a user feedback mechanism is established to regularly collect operators' opinions and suggestions on the interface, and optimize and improve according to the feedback; and an automatic update mechanism for training content is established. When the system function is updated, the training content update process is automatically triggered, and the training materials for the new function are quickly generated by using the automatic script technology, including online courses, simulated operation scenarios, etc., and are pushed to the operators in time. At the same time, on-site training activities are regularly organized, and system developers are invited to give detailed explanations and demonstrations of new functions to ensure that operators can master the latest functions of the system in time, and a security review mechanism for personalized customization of the human-machine interface is established. When the operator customizes the interface, the system automatically evaluates the customized content and prompts or prohibits the settings that may affect the safety; and it also includes adding a manual review link in the automatic update mechanism of training content, and professional training personnel review and optimize the automatically generated training materials to ensure the accuracy, completeness and understandability of the training content.

[0050] 1. Human-computer interaction and feedback: Assume that the convenience of operation of the human-computer interaction interface is evaluated, and 10 operation tasks are recorded, and the operation time is Seconds, number of operation errors , set the convenience score (Here we add 1 to avoid the denominator being zero.) For example, the first operation time Seconds, number of errors ,but Calculate the average convenience score ,like If the value is low, the button position or function will be redesigned based on operator feedback to improve operational convenience.

[0051] 2. Training system and audit: Assume that operators are assessed after training. For the new inventory counting process, the accuracy of operation (i.e. the ratio of correct operations to total operations), operation speed improvement rate (Assuming the average operation time before training is seconds, and the average operation time after training is seconds, then ), set the training effect score , assuming , ,but By monitoring The value of evaluating the training effect is If the value does not meet the expected target, optimize the training content and methods, such as increasing the difficulty of the simulated operation scenario or providing more case explanations.

[0052] A cargo dispatching and distribution management system suitable for stacker warehouses includes an artificial intelligence prediction module: the artificial intelligence prediction module introduces an external data monitoring and fusion mechanism, collects external market data in real time, such as social media hot spots, policy and regulatory changes, etc., and combines it with internal business data as the input of the prediction model. At the same time, a model adaptive adjustment mechanism is established. When a major change in external data is detected, the model parameters or structure are automatically adjusted to improve the response speed and accuracy of the model to market changes; and a visual prediction result display tool is developed to present the prediction results and key indicator change trends in an intuitive manner such as charts and graphs. At the same time, data interpretation and decision-making suggestion functions are provided to automatically generate easy-to-understand analysis reports and decision-making suggestions based on the prediction results, so as to help enterprise managers quickly understand the prediction information and make reasonable decisions, and establish an external data quality assessment and cleaning system to conduct credibility assessment, data cleaning and format conversion on external data from different sources to ensure data reliability and consistency; and also includes expanding the function of the visual prediction result display tool, adding multi-dimensional data interactive display and analysis functions, such as using pivot tables, interactive dashboards and other technologies, so that enterprise managers can flexibly adjust the display dimensions and analysis angles according to their own decision-making needs.

[0053] 1. Prediction algorithm and data fusion: For the ARIMA model to predict the number of goods entering the warehouse, assume that the monthly data of the number of goods entering the warehouse in the past two years is , determine the autoregressive order of the ARIMA model using the AIC criterion , moving average order , the difference order After establishing the ARIMA (2,1,2) model and performing parameter estimation and model fitting, the inventory quantity for the next half month is predicted. Assume that the inventory quantity for the first month in the future is , the second month is .

[0054] For the neural network model, assume that the input layer has 4 nodes representing the month (values ​​1-12), season (spring, summer, autumn, winter, represented by 0-3), the average order volume in the past two months, and whether there is a promotion. The output layer is the predicted number of goods in and out of the warehouse this month. For example, the current month is May (corresponding input node value is 5), it is spring (input node value is 0), and the average order volume in the past two months is , and there is no promotional activity in the near future (the input node value is 0). Assume that the weight matrix and threshold vector obtained by neural network training are used to calculate the predicted value through forward propagation. Assume that the input vector , hidden layer calculation ( is an activation function, such as the sigmoid function), the output layer calculates , and get the predicted inventory quantity for this month. This is just an example of the calculation framework. The actual weights and thresholds need to be obtained through a large amount of data training.

[0055] Suppose that social media mentions of a key beverage increase by ,Right now Internal order data shows that the beverage's recent order volume has increased ,Right now , let the external data influence coefficient , assuming , ,but .according to The prediction model parameters can be adjusted based on the value, such as increasing the weight of features related to social media mentions.

[0056] 2. Decision support and visualization: Assume that the work efficiency is analyzed in different time periods in the morning and afternoon. tasks, the total transport distance is m, spend time Hour( minutes), then the morning work efficiency m / min; processed from 2-4 pm tasks, the total transport distance is m, spend time Hour( minutes), then the afternoon work efficiency m / min. Based on this result, it is recommended to arrange urgent order processing or complex cargo handling tasks in the morning, and simple inventory sorting work or equipment maintenance tasks in the afternoon.

[0057] Assume that the credibility of social media data is evaluated and the credibility score of the data source is scored (assuming the social media platform has high visibility and credibility), data integrity score (The data contains key information such as the number of mentions, sentiment, etc.), data accuracy score (After preliminary verification, the data has no obvious errors), set the data credibility , assuming , , ,but .according to Value filtering and processing of external data, if The value is lower than the set threshold (such as ), the data were further reviewed or discarded.

[0058] Suppose that the company manager uses a pivot table to view the forecast data of goods in and out of the warehouse for different regions, different order types, and different product categories, and finds that the forecast quantity of goods in warehouse for order type Y of product category X in region A will increase in the next week. , while inventory costs will also rise Through the interactive dashboard, the time range is dynamically adjusted to the next half month, and the data screening condition is the products of a specific supplier. In-depth analysis shows that there is a strong positive correlation between the number of inbound and outbound goods of the supplier's products in a certain period of time and the inventory cost (correlation coefficient ). Assume that the decision accuracy improvement rate (Assume that the accuracy of decision-making after data analysis is Promote to ,but , decision-making speed improvement rate (Assuming that the average time taken for previous decisions is Hours, now takes Hours, then ), let the validity score , assuming , ,but By monitoring Value optimization visualization tool functions and display methods, if If the value is low, more data display dimensions are added or interactive operations are optimized.

[0059] A cargo dispatching and distribution management system suitable for stacker warehouses includes an automated warehouse and robot collaborative operation module: the automated warehouse and robot collaborative operation module adopts a wired and wireless hybrid communication network and establishes a communication redundancy mechanism. When wireless communication is normal, wireless communication is used for data transmission. When wireless communication is interfered or interrupted, it automatically switches to a wired communication backup line to ensure the stability of data transmission. At the same time, the wireless communication protocol and signal enhancement technology are optimized to reduce the impact of the metal environment on the signal; and a unified robot task allocation and traffic management system is established to centrally dispatch and plan the tasks of AGVs and shuttle vehicles. When planning the path, algorithms such as time window allocation and priority queuing are used to avoid robots in channels and crossroads. For example, a specific time window is allocated to each robot to enter a narrow passage, and the priority is determined according to the urgency of the task. When the paths of a high-priority robot conflict with those of a low-priority robot, the low-priority robot actively avoids or waits. The system also includes optimizing the switching protocols and hardware equipment of wired and wireless hybrid communication networks, using double buffering technology and seamless switching mechanism to temporarily store data to be transmitted in a buffer area before switching, and immediately transmit buffered data after the switching is completed to ensure that data is not lost. The system also introduces dynamic priority adjustment and emergency handling modules in the robot task allocation and traffic management system. When an emergency task or equipment failure occurs, the robot's task priority and path planning are dynamically adjusted according to factors such as the urgency of the task and the remaining resources of the equipment, and emergency tasks are handled first.

[0060] 1. Communication and task allocation: Assuming the total number of wireless communication transmissions , number of successful transmissions , then the communication reliability .like The value is lower than the set threshold (such as ), optimize the wireless communication protocol, such as adjusting the transmission power, optimizing the data packet size, or changing the communication frequency band, and check the effect of the signal enhancement technology, such as whether the layout of the antenna array is reasonable.

[0061] 2. Traffic management and emergency response: Assume that a channel busyness coefficient (calculated based on the number of robots or tasks passing through the channel per unit time), robot operation efficiency (calculated based on the amount of tasks completed by the robot per unit time), initial time window Minutes, set time window , assuming , ,but Minutes, that is, the adjusted time window is minutes to improve channel utilization.

[0062] Assuming the total number of data transfers , number of data transmission interruptions , then the data transmission continuity score .like If the value is low, the buffer size of the double buffering technology or the response speed of the switching mechanism is optimized to ensure the continuity of data transmission and avoid task interruption or data errors caused by switching.

[0063] Assume the urgency of an urgent task , remaining resources of the device (For example, the remaining power is , the remaining load capacity is etc.), assuming that the task priority adjustment formula is , assuming the initial priority , , ,but (The priority can be normalized according to the actual situation). The priority of this task is increased, and the relevant robots will handle this task first and coordinate other robots to give way. Assuming the emergency task completion time Minutes, normal task completion time minutes, the emergency response efficiency By monitoring The value is used to evaluate the effectiveness of the emergency response strategy. The value did not reach the expected target (such as less than ), the emergency resource allocation plan will be adjusted or optimized.

[0064] Path planning algorithm. For example, if it is found that a commonly used path is often congested, resulting in inefficient emergency handling, you can consider increasing the weight of the backup path or replanning the overall path strategy to prioritize guiding the robot to choose a more unobstructed path to perform tasks.

[0065] Assume that during an emergency response to an equipment failure, a stacker suddenly fails and the AGV task that was originally waiting for the stacker service is reassigned. Suppose the estimated completion time of the original AGV task is minutes. After replanning the path and assigning it to the new stacker to complete the task, the actual time spent is minutes. Then the time saving rate of this emergency treatment is , indicating that the task completion time is saved by emergency handling strategy Through statistics of multiple emergency response situations If the average value is low, further analyze the reasons, such as whether the robot scheduling algorithm is not flexible enough, communication delay leads to untimely information transmission, etc., and make targeted improvements and optimizations.

[0066] For the running conflicts of robots in the channel, suppose the number of potential conflicts detected within an hour is After adjusting the time window allocation and priority queuing algorithm, the number of potential conflicts in the next hour is reduced to times. Conflict reduction rate , that is, the conflict is reduced .like The value does not reach the ideal level (such as less than ), the time window allocation strategy is further refined. For example, the time window allowed for each robot to pass through the channel can be dynamically adjusted according to factors such as the speed difference of the robots and the urgency of the task types, or the priority judgment factors in the priority queuing algorithm can be optimized. Not only the urgency of the task but also the remaining power of the robot, the distance to the target point and other factors can be comprehensively considered to avoid conflicts more accurately and improve the fluency and efficiency of warehouse operations.

[0067] From the above, we can see that: Intelligent scheduling module: adopts a dynamic algorithm switching mechanism combined with a task connection transition algorithm. When faced with high order volumes and complex task situations, it can quickly switch algorithms and transition smoothly, avoiding task delays and confusion that may occur when traditional scheduling systems switch algorithms. The special cargo database and its real-time update and targeted scheduling rule formulation provide refined management of special needs of special cargo, which is different from the general scheduling system's unified handling of conventional cargo, and improves the accuracy and safety of special cargo scheduling.

[0068] Precise inventory management module: The environmental adaptation module in the data fusion algorithm adopts regional differentiated threshold setting and judgment logic, fully considering the environmental characteristics of different areas of the warehouse, and can process sensor data more accurately, reducing inventory data errors caused by environmental interference. The artificial intelligence-assisted manual review mechanism uses machine learning models to pre-screen and classify goods, improving the efficiency and accuracy of the review, which is in sharp contrast to the traditional inventory management system that relies solely on manual review or simple automatic review.

[0069] System integration module: The node fault isolation and self-healing mechanism of the middleware cluster and the dynamic monitoring platform for the compatibility of multiple system interface versions are the innovations of this solution. During the system integration process, it can proactively respond to middleware node failures, ensure the stability of data transmission, and monitor interface version compatibility in real time, and promptly discover and solve potential problems. This is a relatively advanced function in existing system integration technologies, which improves the reliability and scalability of system integration.

[0070] Equipment failure response module: It is innovative in that it uses a distributed computing architecture to accelerate the data processing of incremental learning algorithms and optimizes the real-time monitoring information processing process (event-driven architecture and intelligent caching technology). It can quickly update the fault prediction model and adjust the scheduling plan in a timely manner, effectively reducing the interruption and delay of operations caused by equipment failures, improving the system's ability to respond to equipment failures and overall operating efficiency, and has obvious advantages over traditional equipment failure response systems.

[0071] Energy management module: The real-time order correction mechanism supplements the energy-efficiency balance model and the optimization control logic and hardware performance improvement of the hybrid energy switching device are innovative. The energy allocation strategy is dynamically corrected by real-time order information, making energy management more accurate and flexible. The optimization of the hybrid energy switching device ensures a smooth transition of the energy switching process, improves energy utilization efficiency and system operation stability, and is a significant improvement over existing energy management technologies.

[0072] Personnel operation and training assistance module: The safety audit mechanism of the personalized human-machine interaction interface and the manual audit link in the automatic update mechanism of training content are innovative. It provides an effective solution to ensure operational safety and improve training quality. Compared with general personnel operation training assistance systems, it pays more attention to the personalized needs of operators and the accuracy of training content, which helps to reduce human errors and improve the overall operation effect of the system.

[0073] Artificial Intelligence Forecasting Module: The external data quality assessment and cleaning system and the multi-dimensional data interaction function of the extended visualization prediction result display tool are the key points that distinguish it from other artificial intelligence prediction technologies. Through reliable processing of external data and interactive display of multi-dimensional data, it can provide more comprehensive and in-depth decision support for enterprise managers, and it is innovative in data fusion and visualization display, which can better meet complex decision-making needs.

[0074] Automated warehouse and robot collaborative operation module: The dual buffering technology and seamless switching mechanism of the wired and wireless hybrid communication network, as well as the dynamic priority adjustment and emergency handling module in the robot task allocation and traffic management system are unique. It can effectively ensure communication stability and respond to emergencies, improve the efficiency and reliability of the automated warehouse and robot collaborative operation, and compared with the general automated warehouse collaborative operation technology, it has obvious technical progress in communication and task management.

[0075] In summary: This technical solution realizes the efficient, intelligent and reliable operation of the stacker warehouse cargo dispatching and distribution management system through the close coordination of multiple modules and the comprehensive application of innovative technical means. Compared with the existing technology, it does not simply piece together various functional modules, but optimizes all aspects of the system as a whole to form an organic whole. For example, the dynamic algorithm switching mechanism of the intelligent scheduling module cooperates with the environmental adaptive data fusion of the precise inventory management module and the interface compatibility management of the system integration module, which can quickly adapt to order changes, environmental interference and system integration requirements in the complex and changeable warehouse operation environment, and improve the overall adaptability and stability of the system.

[0076] The data sharing and interaction mechanism between modules is also a manifestation of technological innovation. The unified data platform built through the system integration module realizes the real-time and accurate transmission of data between different modules, enabling each module to make more reasonable decisions based on global information. For example, the prediction results of the artificial intelligence prediction module can be transmitted to multiple modules such as intelligent scheduling and energy management in a timely manner, allowing these modules to prepare for response in advance. This cross-module data-driven decision-making mechanism is rare in existing technologies.

Claims

1. A cargo dispatching and distribution management system suitable for stacker warehouses, characterized by: include: Intelligent scheduling module: It adopts multi-algorithm fusion and dynamic switching mechanism to plan the stacker path and assign tasks according to orders and warehouse conditions, and has special cargo handling strategies; Accurate inventory management module: Use multi-sensor fusion and intelligent recognition technology, combined with environmental adaptation and intelligent auditing to ensure accurate inventory data; System integration module: realizes multi-system integration with standardized interfaces and middleware, and has cluster deployment, fault handling and interface management mechanisms; Equipment failure response module: Use multiple sensors and machine learning algorithms to achieve fault detection and early warning, and use distributed computing and real-time monitoring to optimize emergency dispatch; Energy management module: Optimize energy utilization based on energy-efficiency balance model combined with hybrid energy and intelligent switching technology; Personnel operation and training auxiliary module: providing personalized safety customization interface and high-quality training system; Artificial intelligence prediction module: integrates multi-source data and uses visual multi-dimensional interactive tools to assist decision-making; Automated three-dimensional warehouse and robot collaborative operation module: adopt redundant communication network and dynamic task management strategy to ensure efficient collaboration.

2. A cargo dispatching and distribution management system suitable for stacker warehouses as claimed in claim 1, characterized in that: The multi-algorithm fusion of the intelligent scheduling module includes the combination of path optimization algorithms such as A* algorithm or Dijkstra algorithm and task allocation algorithm, and the dynamic switching mechanism switches between conventional and fast scheduling algorithms according to the order volume and computing resource conditions. At the same time, the special cargo handling strategy includes a special cargo database and its real-time update and targeted scheduling rule formulation.

3. A cargo dispatching and distribution management system suitable for stacker warehouses as claimed in claim 1, characterized in that: The multi-sensor fusion of the precise inventory management module adopts Kalman filtering or fuzzy logic algorithm to process sensor data such as RFID and laser scanning. The environmental adaptation adjusts the data fusion weight according to regional differences. The intelligent review includes a combination of artificial intelligence pre-screening and manual review.

4. A cargo dispatching and distribution management system suitable for stacker warehouses as claimed in claim 1, characterized in that: The middleware of the system integration module adopts an enterprise service bus, the cluster deployment has a node fault isolation and self-healing function, and the interface management includes a version compatibility mechanism and a multi-system compatibility dynamic monitoring platform.

5. The cargo dispatching and distribution management system suitable for stacker warehouse according to claim 1, characterized in that: The machine learning algorithm of the equipment fault response module adopts a hybrid intelligent scheduling algorithm that integrates the genetic algorithm and the A* algorithm. The distributed computing architecture accelerates the data processing of the incremental learning algorithm. The real-time monitoring adopts the event-driven architecture and intelligent caching technology to optimize the generation of scheduling solutions.

6. A cargo dispatching and distribution management system suitable for stacker warehouses as claimed in claim 1, characterized in that: The energy-efficiency balance model of the energy management module is combined with a real-time order correction mechanism to adjust the stacker energy strategy, and the hybrid energy intelligent switching system includes voltage stabilization filtering and pre-switching buffer functions.

7. The cargo dispatching and distribution management system applicable to a stacker warehouse as claimed in claim 1, characterized in that: The personalized safety customization interface of the personnel operation and training auxiliary module is equipped with a safety audit mechanism, and the automatic update mechanism of the training system includes a manual audit link to ensure the quality of training.

8. A cargo dispatching and distribution management system suitable for stacker warehouses as claimed in claim 1, characterized in that: The multi-source data fusion of the artificial intelligence prediction module includes an external data quality assessment and cleaning system, and the visual multi-dimensional interactive tool uses technologies such as pivot tables and interactive dashboards to achieve multi-dimensional analysis and display.

9. A cargo dispatching and distribution management system suitable for stacker warehouses as claimed in claim 1, characterized in that: The redundant communication network of the automated stereoscopic warehouse and the robot collaborative operation module adopts double buffering technology and seamless switching mechanism to ensure data transmission, and the dynamic task management strategy adjusts the robot task priority and path planning in real time according to the task urgency and equipment status.

Images