Stacking monitoring and excess material calculation method for multi-dimensional data linkage processing

Abstract

The invention relates to a stacking monitoring and excess material calculation method for multi-dimensional data linkage processing, and the method comprises the following steps: environment evaluation and demand definition: carrying out the evaluation of a stacking region, and carrying out the customization design of a sensor layout and a data collection scheme; sensor network deployment and calibration: deploying a sensor, an RFID tag reader and a 3D scanner according to a design scheme; building a data interface and a fusion platform; constructing a three-dimensional stacking model; development and testing of an intelligent analysis algorithm: developing the intelligent analysis algorithm based on machine learning, training and testing the algorithm, and optimizing the performance of the algorithm; a dynamic monitoring and anomaly detection mechanism is established, specifically, a dynamic monitoring system is implemented, stacking state changes are continuously tracked, an anomaly detection algorithm is developed, potential safety hazards are recognized in real time, and an early warning mechanism is triggered; excess material calculation and optimization suggestion generation: fine excess material calculation is executed, and accurate material excess information is provided; and integrating the system and debugging on site.

Description

Technical Field

[0001] The present application relates to the technical field of material management, and in particular to a method for stacking monitoring and residual material calculation using multi-dimensional data linkage processing. Background Art

[0002] In the field of modern warehousing and logistics, efficient material management is crucial to improving operational efficiency, reducing waste and ensuring safety. Traditional manual inventory and management methods are not only time-consuming and labor-intensive, but also prone to errors, and it is difficult to meet the increasingly complex and dynamically changing inventory needs. With the development of the Internet of Things (IoT), big data, and artificial intelligence (AI) technologies, intelligent material management systems have gradually become the key to improving the level of warehouse management. However, existing technologies still have shortcomings in real-time monitoring, precise measurement, intelligent analysis, and space optimization, especially the lack of a comprehensive system that integrates multiple sensing technologies, can dynamically model, and provide optimization strategies. Although some automated systems have attempted to manage through sensors and computers, most of these systems are limited to single-dimensional data collection and analysis, and cannot fully consider the dynamic interaction of multiple factors such as spatial layout, material properties, and time series, resulting in low monitoring accuracy and inaccurate calculation of residual materials. Summary of the invention

[0003] The present application provides a stacking monitoring and remaining material calculation method with multi-dimensional data linkage processing. The stacking monitoring and remaining material calculation method can realize accurate monitoring of stacked materials and efficient calculation of remaining material quantity through comprehensive analysis of multi-dimensional data, thereby improving the automation level and accuracy of inventory management.

[0004] The stacking monitoring and residual material calculation method provided by the present application for multi-dimensional data linkage processing comprises the following steps:

[0005] S1, Environmental Assessment and Requirements Definition: Evaluate the stacking area and customize the sensor layout and data collection plan based on the assessment results;

[0006] S2, sensor network deployment and calibration: deploy sensors, RFID tag readers and 3D scanners according to the design plan. The deployment needs to ensure full coverage of the stacking area. After the deployment is completed, the sensors are accurately calibrated;

[0007] S3, data interface and fusion platform construction: Develop data receiving and processing interfaces to ensure that all types of sensor data can be seamlessly connected to the central processing system; build a data fusion platform to achieve standardized processing and integration of multi-source data to form a unified data model;

[0008] S4, 3D stacking model construction: 3D scanning and image recognition technology are used to dynamically build a 3D model of stacked materials in combination with weight and size data. The 3D model must be able to reflect changes in the stacking status in real time, including material addition, removal and position adjustment;

[0009] S5, Intelligent analysis algorithm development and testing: Develop intelligent analysis algorithms based on machine learning to predict material consumption trends, optimize stacking layouts, and calculate remaining material quantities. Train and test the algorithms through historical data to continuously optimize algorithm performance.

[0010] S6, Dynamic monitoring and anomaly detection mechanism establishment: Implement a dynamic monitoring system, continuously track changes in stacking status, develop anomaly detection algorithms, identify safety hazards in real time, and trigger early warning mechanisms;

[0011] S7, residual material calculation and optimization suggestion generation: Combine the real-time stacking model and material attributes to perform detailed residual material calculation and provide accurate material remaining quantity information. Based on the calculation results, the system automatically generates stacking optimization suggestions to guide inventory management and space utilization;

[0012] S8, system integration and on-site debugging: Integrate all modules into a unified management system, perform on-site installation and debugging, ensure the smooth operation of the entire system, and ensure that data processing and feedback are timely and accurate.

[0013] In an optional solution, in step S2, a genetic algorithm may be used to optimize the sensor position when deploying the sensor. The formula is briefly described as follows:

[0014] F = α × C - β × T + γ × R;

[0015] In the formula, F is the fitness function, C is the coverage ratio, T is the total cost, R is the redundancy index, α, β, and γ are weight factors adjusted according to actual conditions, C measures the completeness of the monitoring range, T reflects economic considerations, and R ensures the reliability of the system; find the optimal sensor deployment strategy through iterative optimization;

[0016] During deployment, the data collection frequency and accuracy requirements of various sensors are dynamically adjusted by intelligent algorithms, and the Kalman filter is used to fuse and reduce noise on sensor data. The formula is abbreviated as follows:

[0017]

[0018] P k|k-1 =AP k-1|k-1 A T +Q;

[0019] K k =P k|k-1 H T (HPk|k-1 H T +R) -1 ;

[0020]

[0021] P k|k =(IK k H)P k|k-1 ;

[0022] In the formula, represents state estimation, P is the covariance matrix, K is the Kalman gain, A and B are the system transfer matrices, Q and R are the covariance matrices of process noise and measurement noise respectively, z is the measurement value, and H is the observation model matrix.

[0023] In an optional solution, in step S2, the deployed sensor includes a weight sensor, which is installed on a supporting structure or a handling device to directly measure the weight change of the stacked material;

[0024] RFID tag readers can attach RFID tags to each batch or important single item of materials, deploy fixed or mobile readers at the entrance and exit channels and the edge of the stacking area to realize material identity recognition and location tracking, test the reading range and recognition rate, adjust the antenna angle and power, and ensure that all tags are stably read within the specified area;

[0025] 3D scanners can scan the stacking area periodically or on demand, build and update the three-dimensional model of the stacked materials, and accurately record the shape, volume and stacking structure.

[0026] In an optional solution, step S3 specifically includes the following steps:

[0027] Design and implement data receiving services, using a high-concurrency processing framework to ensure low-latency data transmission and buffering;

[0028] Perform preliminary cleaning and format verification on the received data to remove invalid or erroneous data. If measurement values ​​outside the normal range are detected, they are processed through data smoothing or outlier detection algorithms;

[0029] Standardize the preprocessed data to ensure that data from different sensors are comparable in the same dimension;

[0030] A data fusion algorithm is used to integrate time series analysis and spatial information to form a comprehensive and accurate description of the stacked materials. For the fusion of stack height and weight, the following weighted fusion formula can be used:

[0031] F(x)=ω1·H(x)+ω2·W(x);

[0032] Where F(x) is the comprehensive index after fusion, H(x) and W(x) are the measured values ​​of height and weight, respectively, and ω1 and ω2 are weight coefficients set according to the application scenario.

[0033] In an optional solution, step S4 specifically includes the following steps:

[0034] Scan the stacking area regularly or on demand to obtain 3D point cloud data of the material stacking, and use high-resolution cameras with image recognition algorithms to identify material types, boundaries and identification information to enhance the accuracy of model details;

[0035] Integrate the material location information read by weight sensors and RFID tags, as well as data from environmental sensors, to form a comprehensive material attribute database;

[0036] The original point cloud data is denoised and downsampled, and then converted into a triangular mesh model through a 3D reconstruction algorithm to form a preliminary 3D stacking model;

[0037] Based on image recognition technology, each batch or single piece of material is identified, and combined with RFID tag data, the correct material properties and weight information are assigned to the model elements;

[0038] Associating the real-time data from the weight sensor with the materials in the model, and adjusting the model in real time through interpolation or shape deformation algorithms to ensure that the model can reflect the weight changes, additions, removals, and position adjustments of the stacked materials;

[0039] Based on the dynamically updated 3D model, the mechanical analysis algorithm is used to evaluate the stability of the stacking structure, calculate parameters such as the center of gravity offset and stacking pressure distribution, and compare them with the preset safety threshold. When the threshold is exceeded, an early warning is triggered.

[0040] The dynamically updated 3D model is displayed through a graphical user interface and interactive browsing is enabled.

[0041] In an optional solution, the step S5 specifically includes the following steps:

[0042] Cross-validation is used to evaluate the generalization ability of the model, and hyperparameters are tuned through grid search and random search methods to find the best model configuration; the material consumption trend is predicted, and the model formula can be abstractly expressed as:

[0043] h t =f(W h [x t ,h t-1 ]+b h );

[0044] y t =g(W y ht +b y );

[0045] In the formula, h t is the hidden state, x t is the current input, W h , W y is the weight matrix, b h , b y is the bias term, f and g are activation functions, and the parameters are adjusted through back propagation and optimization algorithms.

[0046] In an optional solution, step S6 specifically includes the following steps:

[0047] Thresholds are set based on the normal state of material stacking. Any situation exceeding these thresholds is considered abnormal. Models are trained on historical data using supervised or unsupervised learning algorithms to identify normal and abnormal patterns. Multimodal anomaly detection methods are used to comprehensively judge abnormal states by combining 3D model data, weight data, and environmental parameters to improve detection accuracy.

[0048] Threshold setting and dynamic adjustment: Based on historical data and business logic, set the warning threshold of key indicators, use adaptive threshold algorithm, and automatically adjust the threshold according to real-time supply and demand changes to avoid false alarms and missed alarms. The conditions for triggering the warning are:

[0049]

[0050] In the formula, I(t) represents the current index value; represents the historical mean; σ represents the standard deviation; k represents the threshold coefficient;

[0051] Anomaly detection algorithm: Introduce machine learning methods to analyze data streams in real time, identify abnormal changes in stacking patterns, and issue early warning signals in a timely manner;

[0052] The anomaly score S output by the model is expressed as: S = f(x);

[0053] In the formula, f is the prediction function of the anomaly detection model, x is the input feature vector, and the higher the output anomaly score, the more the current state deviates from the normal mode.

[0054] In an optional solution, the step S7 specifically includes the following steps:

[0055] Real-time stacking model update: Based on the latest stacking status data obtained by the dynamic monitoring system, including stacking length, width, depth and distribution of materials in each layer, the three-dimensional stacking model is updated. If this process involves three-dimensional reconstruction, the ray casting method or voxelization algorithm is used for three-dimensional reconstruction. The formula is simplified as follows:

[0056]

[0057] In the formula, V represents the stacking volume, and L i represents the length of the i-th layer, and W i represents the width of the i-th layer, and H i represents the height of the i-th layer;

[0058] Precise calculation of leftover materials: Apply density mapping technology and volume-weight conversion, combined with the density information of each material, to precisely calculate the total volume and total weight of the remaining materials. The calculation formula is as follows:

[0059]

[0060] In the formula, W 剩 represents the weight of the leftover materials; V represents the stacking volume; ρ is the average density, considering the mixed average of each material in the stacking, represents the cumulative weight of the materials removed from the initial state to the current time point.

[0061] In an alternative solution, in step S7, the algorithm for generating stacking optimization suggestions includes:

[0062] Maximizing space utilization: Adopt a heuristic algorithm of genetic algorithm or particle swarm optimization to search for the optimal stacking layout plan for the current stacking state and the remaining materials. The objective function can be defined as minimizing the unused space volume, while considering the convenience of material removal. The formula is simplified as:

[0063] minZ = V_empty - α × C oper ;

[0064] In the formula, Z is the objective function, V_empty is the total idle volume, α is an adjustment factor to balance space utilization and operation efficiency, and C oper is the operation convenience;

[0065] Inventory management suggestion: Based on the calculation results of the leftover materials, use a prediction model to analyze the future demand trend and give suggestions for replenishment or allocation to reduce inventory costs. The formula is expressed as:

[0066] Q 补 = D 预测 - R 当前 + β × S;

[0067] In the formula, Q 补 represents the replenishment quantity, D 预测 represents the predicted demand quantity, R 当前 represents the current leftover material quantity, β is the safety stock coefficient, reflecting the buffer demand for market fluctuations, and S represents the safety stock quantity.

[0068] In an optional solution, in step S1, the evaluation of the stacking area includes: measuring the size of the stacking area, evaluating the bearing capacity of the ground, analyzing the lighting conditions, and considering ventilation and fire protection requirements; carefully classifying the types of materials, and recording the physical and chemical properties of each material to ensure the pertinence and safety of subsequent management measures;

[0069] The requirements definition includes: covering basic inventory tracking and security management, while going deep into the advanced goals of improving operational efficiency, energy conservation and emission reduction, and cost control; if flammable items are stored, the system needs to be highly sensitive to changes in temperature and humidity; if materials are time-sensitive, the focus is on the automated execution of the first-in-first-out principle.

[0070] The beneficial effects of this application are:

[0071] The stacking monitoring and residual material calculation method in this application realizes all-round and refined management of material stacking by integrating sensors, RFID technology, 3D scanning and data analysis algorithms. It can automatically monitor the stacking status of materials, calculate the residual material quantity in real time, predict demand trends, and propose stacking layout adjustment suggestions based on the optimization algorithm, effectively avoiding overloading, reducing space waste, and enhancing safety and response speed.

[0072] It should be understood that the foregoing general description and the following detailed description are exemplary only and are not restrictive of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0073] Figure 1 A schematic diagram of the flow of the stacking monitoring and residual material calculation method provided for this application.

[0074] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application. DETAILED DESCRIPTION

[0075] In order to better understand the technical solution of the present application, the embodiments of the present application are described in detail below with reference to the accompanying drawings.

[0076] It should be clear that the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in the field without creative work are within the scope of protection of the present application.

[0077] The terms used in the embodiments of the present application are only for the purpose of describing specific embodiments, and are not intended to limit the present application. The singular forms "a", "said" and "the" used in the embodiments of the present application and the appended claims are also intended to include plural forms, unless the context clearly indicates other meanings.

[0078] It should be understood that the term "and / or" used in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship.

[0079] like Figure 1 As shown, the present application provides a stacking monitoring and residual material calculation method for multi-dimensional data linkage processing, and the stacking monitoring and residual material calculation method includes the following steps:

[0080] S1, Environmental Assessment and Requirements Definition: Evaluate the stacking area and customize the sensor layout and data collection plan based on the assessment results.

[0081] Specifically, in this step, an in-depth environmental assessment is first conducted. This process is detailed and systematic, and aims to fully understand the specific conditions of the stacking area, for example, including but not limited to measuring the size of the stacking area, assessing the bearing capacity of the ground, analyzing the lighting conditions, considering ventilation and fire protection requirements, etc. At the same time, the types of materials are carefully classified, and the physical properties (such as density, shape, stacking stability, whether it is flammable and explosive, etc.) and chemical properties (such as corrosiveness and reactivity) of each material are recorded to ensure the pertinence and safety of subsequent management measures.

[0082] Based on this information, requirements are defined. The requirements analysis not only covers basic inventory tracking and safety management, but also goes deep into advanced goals such as improving operational efficiency, energy conservation and emission reduction, and cost control. For example, if flammable items are stored, the system needs to be highly sensitive to changes in temperature and humidity; if the materials are time-sensitive, it is necessary to focus on the automated execution of the first-in-first-out principle.

[0083] S2, sensor network deployment and calibration: Deploy sensors, RFID tag readers and 3D scanners according to the design plan. The deployment must ensure full coverage of the stacking area. After the deployment is completed, the sensors must be accurately calibrated.

[0084] Specifically, this stage introduces the design concept of intelligent early warning algorithm to ensure efficient and accurate data collection. The sensor layout strategy combines coverage, detection accuracy and cost-benefit analysis, and uses optimization algorithms to determine the best layout plan. For example, a genetic algorithm (GA) can be used to optimize the sensor location layout. The objective function may include indicators such as minimizing monitoring blind spots and maximizing information acquisition efficiency. The GA formula is briefly described as follows:

[0085] F = α × C - β × T + γ × R;

[0086] Where F is the fitness function, C is the coverage ratio, T is the total cost, R is the redundancy index, α, β, and γ are weight factors adjusted according to actual conditions, C measures the completeness of the monitoring range, T reflects economic considerations, and R ensures the reliability of the system; the optimal sensor deployment strategy is found through iterative optimization.

[0087] In this step, the deployment and calibration of the sensor network is the basis for ensuring the efficient operation of the system. The deployment process requires careful planning. According to the results of the environmental assessment and demand definition in the first step, the appropriate sensor type and quantity are selected and scientifically distributed in key locations in the stacking area. The specific deployment is as follows:

[0088] 1.Weight sensor: Installed on the supporting structure or handling equipment, directly measure the weight change of the stacked materials. For dynamic stacking areas, ensure that the sensor layout covers all possible loading and unloading points.

[0089] Use a standard object of known mass for calibration, record the zero offset without load and the linearity and repeatability after loading, and use the least squares method for linear fitting correction:

[0090] y=mx+b;

[0091] In the formula, y is the measured value, x is the true value, m is the slope, b is the intercept, and the correction coefficient is adjusted by fitting.

[0092] 2. RFID tags and readers: Attach RFID tags to each batch or important single item of materials, and deploy fixed or mobile readers at the entrance and exit channels and the edge of the stacking area to realize material identity recognition and location tracking.

[0093] Test the reading range and recognition rate, adjust the antenna angle and power, and ensure that all tags can be stably read within the specified area.

[0094] 3.3D scanner: Use a fixed or drone-mounted 3D laser scanner to scan the stacking area periodically or on demand, build and update a three-dimensional model of the stacked materials, and accurately record the shape, volume and stacking structure.

[0095] Standard geometric bodies are used for spatial calibration, and the scan data and model data are aligned through the iterative closest point (ICP) algorithm to eliminate system deviations and ensure the accuracy of 3D reconstruction.

[0096] The data collection solution is customized according to material characteristics and management requirements, integrating multiple technologies such as weight sensors, RFID tags, infrared thermal imagers, and 3D scanners. The data collection frequency and accuracy requirements of each type of sensor are dynamically adjusted by intelligent algorithms to ensure that unnecessary resource consumption is reduced while meeting monitoring requirements. For example, the Kalman filter is used to fuse and reduce noise on sensor data. The formula is abbreviated as follows:

[0097]

[0098] P k|k-1 =AP k-1|k-1 A T +Q;

[0099] K k =P k|k-1 H T (HP k|k-1 H T +R) -1 ;

[0100]

[0101] P k|k =(IK k H)P k|k-1 ;

[0102] In the formula, represents state estimation, P is the covariance matrix, K is the Kalman gain, A and B are system transfer matrices, Q and R are the covariance matrices of process noise and measurement noise respectively, z is the measurement value, and H is the observation model matrix. Through this algorithm, the material state can be accurately estimated even in a noisy environment, laying a solid foundation for subsequent intelligent early warning and decision support.

[0103] S3, data interface and fusion platform construction: develop data receiving and processing interfaces to ensure that all types of sensor data can be seamlessly connected to the central processing system; build a data fusion platform to achieve standardized processing and integration of multi-source data to form a unified data model.

[0104] In this step, the construction of data interface and fusion platform is the key link to realize multi-dimensional data linkage processing, which ensures that a large amount of heterogeneous data collected from different sensors can be efficiently integrated, processed, and converted into valuable information that can be used for stacking monitoring and residual material calculation. The specific implementation details are as follows:

[0105] Define a set of common data transmission protocols (such as MQTT, HTTP / RESTful, etc.) for all sensors to ensure the consistency of data format and communication method. The output data of each sensor is encapsulated in the format specified by the protocol, including basic information such as timestamp, sensor ID, and measurement value.

[0106] Design and implement a data receiving service that monitors sensor data streams and uses a high-concurrency processing framework (such as Apache Kafka, RabbitMQ) to ensure low-latency transmission and buffering of data to prevent data loss or congestion.

[0107] The received data is preliminarily cleaned and formatted to remove invalid or erroneous data. If a measurement value outside the normal range is detected, it is processed through data smoothing or outlier detection algorithms (such as Grubbs' Test).

[0108] The preprocessed data is standardized and transformed to ensure that data from different sensors are comparable in the same dimension. For example, weight data is standardized into tons, location coordinates are unified into the same coordinate system, and timestamps are synchronized to server time.

[0109] Using data fusion algorithms, such as weighted averaging, Kalman filtering, or deep learning fusion networks, integrate time series analysis with spatial information to form a comprehensive and accurate description of the stacked materials. For example, for the fusion of stack height and weight, the following simple weighted fusion formula can be used:

[0110] F(x)=ω1·H(x)+ω2·W(x);

[0111] Where F(x) is the comprehensive index after fusion, H(x) and W(x) are the measured values ​​of height and weight, respectively, and ω1 and ω2 are weight coefficients set according to the application scenario.

[0112] Based on the fused data, a three-dimensional stacking model is constructed, and a dynamically updated material stacking database is formed by combining material properties (such as density and stacking restrictions). The model includes not only static stacking layout, but also dynamically changing information such as remaining material quantity and stacking safety status, providing a basis for subsequent intelligent analysis.

[0113] S4, 3D stacking model construction: Utilize 3D scanning and image recognition technology, combined with weight and size data, to dynamically construct a 3D model of stacked materials. The 3D model must be able to reflect changes in the stacking status in real time, including material addition, removal, and position adjustment.

[0114] Specifically, in the process of building a three-dimensional stacking model, 3D scanning, image recognition technology and real-time sensor data are used to dynamically reproduce the actual situation of the stacking site, and to achieve accurate management and early warning response of stacked materials. The specific implementation details are as follows:

[0115] Use fixed or mobile 3D laser scanners to scan the stacking area regularly or on demand to obtain three-dimensional point cloud data of the material stacking; at the same time, use high-resolution cameras with image recognition algorithms to identify material types, boundaries and identification information to enhance the accuracy of model details.

[0116] Integrate the material location information read by the weight sensor and RFID tag, as well as the data from the environmental sensor, to form a comprehensive material attribute database, which will serve as an important supplement to the model construction to ensure that the physical properties of the model are consistent with the actual situation. Use point cloud processing algorithms (such as Voxel Grid Downsampling and Statistical Outlier Removal in the PCL library) to denoise and downsample the original point cloud data, and then use 3D reconstruction algorithms (such as the Marching Cubes algorithm for surface reconstruction) to convert it into a triangular mesh model to form a preliminary 3D stacking model.

[0117] Based on image recognition technology, each batch or single piece of material is identified, and combined with RFID tag data, the correct material properties and weight information are assigned to each model element. The real-time data of the weight sensor is associated with the material in the model, and the model is adjusted in real time through interpolation or shape deformation algorithms (such as Free-Form Deformation, FFD) to ensure that the model can reflect the weight changes, additions, removals and position adjustments of the stacked materials.

[0118] Based on the dynamically updated three-dimensional model, mechanical analysis algorithms (such as finite element analysis FEM) are used to evaluate the stability of the stacking structure, calculate parameters such as the center of gravity offset and stacking pressure distribution, and compare them with the preset safety thresholds. When the threshold is exceeded, an early warning is triggered.

[0119] Fuzzy Logic Controller (FLC) is used to evaluate the stacking risk level. Fuzzy sets and membership functions are set according to factors such as stacking height, weight distribution, and support conditions. The risk level is output through fuzzy rule reasoning. Formula example:

[0120] R=μ HIGH (H)∧μ MID (W)∨μ LOW (S);

[0121] Where μ represents the fuzzy membership function, R is the risk level, and H, W, and S represent the input variables such as stacking height, weight distribution uniformity, and support stability, respectively.

[0122] The dynamically updated 3D model is displayed through a graphical user interface, and interactive browsing is achieved using technologies such as OpenGL or Unity, allowing users to rotate and zoom the model to view stacking details and warning information. When the intelligent warning algorithm identifies potential risks, the system not only provides visual (such as color marking of unstable areas) and auditory (alarm sound) prompts, but also provides detailed warning reasons and recommended measures through pop-up windows on the interface, supporting users to intervene in real-time operations.

[0123] S5, Intelligent analysis algorithm development and testing: Develop intelligent analysis algorithms based on machine learning to predict material consumption trends, optimize stacking layouts, and calculate remaining material quantities. Train and test the algorithms through historical data to continuously optimize algorithm performance.

[0124] In this step, we will take a closer look at the development and testing process of intelligent analysis algorithms based on machine learning, with a particular emphasis on how to use these algorithms to predict material consumption trends, optimize stacking layouts, and accurately calculate remaining material. We will also use historical data to continuously optimize algorithm performance and ensure the accuracy and effectiveness of intelligent early warnings.

[0125] Based on the needs, choose the appropriate machine learning model, such as time series analysis (ARIMA, LSTM), cluster analysis (K-means, DBSCAN), regression analysis (random forest, gradient boosting tree), optimization algorithm (genetic algorithm, particle swarm optimization), etc.

[0126] Based on business understanding, extract key features that are helpful for prediction and optimization, such as daily consumption, stacking density, material category, seasonal factors, etc. Use Python's scikit-learn, TensorFlow, PyTorch and other libraries to build corresponding machine learning models based on the selection.

[0127] Cross-validation (such as K-fold cross-validation) is used to evaluate the generalization ability of the model, and hyperparameters are tuned through grid search, random search and other methods to find the best model configuration. Taking LSTM as an example to predict material consumption trends, the model formula can be abstractly expressed as:

[0128] h t =f(W h [x t ,h t-1 ]+b h );

[0129] y t =g(W y h t+b y );

[0130] In the formula, h t is the hidden state, x t is the current input, W h , W y is the weight matrix, b h , b y is the bias term, f and g are activation functions, and the parameters are adjusted through back propagation and optimization algorithms (such as Adam).

[0131] S6, establishment of dynamic monitoring and anomaly detection mechanism: implement a dynamic monitoring system, continuously track changes in stacking status, develop anomaly detection algorithms, identify safety hazards in real time, and trigger early warning mechanisms.

[0132] In this step, dynamic monitoring and anomaly detection mechanisms are key to ensuring safe and efficient stacking operations. Through real-time data collection and intelligent algorithm analysis, the system can quickly identify potential risks and take action. The specific implementation details are as follows:

[0133] Set thresholds based on the normal state of material stacking, such as the maximum load weight and tilt angle thresholds. Any situation exceeding these thresholds is considered abnormal. Use supervised or unsupervised learning algorithms, such as One-Class SVM, IsolationForest, or Autoencoder, to train models on historical data to identify normal and abnormal patterns. An example of anomaly score calculation formula (taking One-Class SVM as an example):

[0134] f(x)=d(x,v);

[0135] In the formula, x is the input data point, v is the support vector, and d is the distance from the data point to the support vector. The higher the anomaly score, the more the data point deviates from the normal range, and an early warning is triggered when the preset threshold is reached.

[0136] Combining 3D model data, weight data, environmental parameters, etc., multimodal anomaly detection methods, such as deep neural network fusion models, are used to comprehensively judge abnormal conditions and improve detection accuracy.

[0137] Threshold setting and dynamic adjustment: Based on historical data and business logic, set warning thresholds for key indicators such as material inventory and stacking safety factor. Adopt adaptive threshold algorithm to automatically adjust thresholds according to real-time supply and demand changes to avoid false alarms and missed alarms.

[0138] The conditions for triggering the warning are:

[0139]

[0140] In the formula, I(t) represents the current index value, represents the historical mean, σ represents the standard deviation, and k represents the threshold coefficient.

[0141] Anomaly detection algorithm: Introduce machine learning methods, such as Isolation Forest or One-ClassSVM, to analyze data streams in real time, identify abnormal changes in stacking patterns, and issue early warning signals in a timely manner.

[0142] S = f(x);

[0143] In the formula, f is the prediction function of the anomaly detection model, x is the input feature vector, and S is the anomaly score output by the model. The higher the output anomaly score, the more the current state deviates from the normal mode.

[0144] S7, residual material calculation and optimization suggestion generation: Combine the real-time stacking model with material attributes to perform detailed residual material calculation and provide accurate material remaining quantity information. Based on the calculation results, the system automatically generates stacking optimization suggestions to guide inventory management and space utilization.

[0145] In this step, the system uses advanced algorithm technology to conduct in-depth analysis of real-time stacking data to achieve accurate residual material calculation and put forward efficient stacking optimization suggestions. The specific operation process is as follows:

[0146] 1. Real-time stacking model update: First, based on the latest stacking status data obtained by the dynamic monitoring system, including but not limited to the stacking length, width, depth and the distribution of each layer of materials, the three-dimensional stacking model is updated. This process may involve three-dimensional reconstruction technology, using algorithms such as ray casting or voxelization. The formula is simplified as follows:

[0147]

[0148] In the formula, V represents the stacking volume, L i represents the length of the i-th layer, W i represents the width of the i-th layer, H i Indicates the height of the i-th layer.

[0149] 2. Accurate calculation of remaining materials: Apply density mapping technology and volume weight conversion, combine the density information of each material, and accurately calculate the total volume and total weight of the remaining materials. The calculation formula is as follows:

[0150]

[0151] Where W 剩 represents the weight of the remaining material, V represents the stacking volume, and ρ is the average density, taking into account the mixed average value of each material in the stacking; Represents the cumulative weight of materials removed from the initial state to the current time point.

[0152] 3. Optimization Suggestion Generation Algorithm: Maximize Space Utilization: Adopt heuristic algorithms such as genetic algorithms or Particle Swarm Optimization (PSO) to search for the optimal stacking layout plan for the current stacking state and remaining materials. The objective function may be defined as minimizing the volume of unused space while considering the convenience of material removal. The formula is simplified as follows:

[0153] minZ = V_empty - α × C oper ;

[0154] In the formula, Z is the objective function, V 空 is the total idle volume, α is the adjustment factor to balance space utilization and operation efficiency; C oper is the operation convenience.

[0155] Inventory Management Suggestion: Based on the remaining material calculation results, use prediction models (such as ARIMA, random forest) to analyze future demand trends, and give replenishment or allocation suggestions to reduce inventory costs. The formula example is:

[0156] Q 补 = D 预测 - R 当前 + β × S;

[0157] In the formula, Q 补 represents the replenishment quantity, D 预测 represents the predicted demand quantity, R 当前 represents the current remaining material quantity, β is the safety stock coefficient, reflecting the buffer demand for market fluctuations, and S represents the safety stock quantity.

[0158] S8, System Integration and On-site Commissioning: Integrate all modules into a unified management system, conduct on-site installation and commissioning to ensure the smooth operation of the entire system, and timely and accurately process and feedback data. Officially run the system, continuously collect feedback data, and monitor system performance. According to the performance in actual applications, continuously adjust algorithm parameters and optimize the data processing process to improve the accuracy and practicality of the overall system. At the same time, establish a regular maintenance and upgrade mechanism to ensure the long-term stable and efficient operation of the system.

[0159] The above are only the preferred embodiments of the present application and are not used to limit the present application. For those skilled in the art, the present application can have various changes and modifications. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method for stacking monitoring and residual material calculation by multi-dimensional data linkage processing, characterized in that: The following steps are involved: S1, Environmental Assessment and Requirements Definition: Evaluate the stacking area and customize the sensor layout and data collection plan based on the assessment results; S2, sensor network deployment and calibration: deploy sensors, RFID tag readers and 3D scanners according to the design plan. The deployment needs to ensure full coverage of the stacking area. After the deployment is completed, the sensors are accurately calibrated; S3, data interface and fusion platform construction: Develop data receiving and processing interfaces to ensure that all types of sensor data can be seamlessly connected to the central processing system; build a data fusion platform to achieve standardized processing and integration of multi-source data to form a unified data model; S4, 3D stacking model construction: 3D scanning and image recognition technology are used to dynamically build a 3D model of stacked materials in combination with weight and size data. The 3D model must be able to reflect changes in the stacking status in real time, including material addition, removal and position adjustment; S5, Intelligent analysis algorithm development and testing: Develop intelligent analysis algorithms based on machine learning to predict material consumption trends, optimize stacking layouts, and calculate remaining material quantities. Train and test the algorithms through historical data to continuously optimize algorithm performance. S6, Dynamic monitoring and anomaly detection mechanism establishment: Implement a dynamic monitoring system, continuously track changes in stacking status, develop anomaly detection algorithms, identify safety hazards in real time, and trigger early warning mechanisms; S7, residual material calculation and optimization suggestion generation: Combine the real-time stacking model and material attributes to perform detailed residual material calculation and provide accurate material remaining quantity information. Based on the calculation results, the system automatically generates stacking optimization suggestions to guide inventory management and space utilization; S8, system integration and on-site debugging: Integrate all modules into a unified management system, perform on-site installation and debugging, ensure the smooth operation of the entire system, and ensure that data processing and feedback are timely and accurate.

2. The method for stacking monitoring and residual material calculation based on multi-dimensional data linkage processing according to claim 1 is characterized in that: In step S2, a genetic algorithm may be used to optimize the sensor position when deploying the sensor. The formula is briefly described as follows: F = α × C - β × T + γ × R; In the formula, F is the fitness function, C is the coverage ratio, T is the total cost, R is the redundancy index, α, β, and γ are weight factors adjusted according to actual conditions, C measures the completeness of the monitoring range, T reflects economic considerations, and R ensures the reliability of the system; find the optimal sensor deployment strategy through iterative optimization; During deployment, the data collection frequency and accuracy requirements of various sensors are dynamically adjusted by intelligent algorithms, and the Kalman filter is used to fuse and reduce noise on sensor data. The formula is abbreviated as follows: P k|k-1 =AP k-1|k-1 From T +Q; K k =P k|k-1 H T (HP k|k-1 H T +R) -1 ; P k|k =(I-K k H)P k|k-1 ; In the formula, represents state estimation, P is the covariance matrix, K is the Kalman gain, A and B are the system transfer matrices, Q and R are the covariance matrices of process noise and measurement noise respectively, z is the measurement value, and H is the observation model matrix.

3. The stacking monitoring and residual material calculation method of multi-dimensional data linkage processing according to claim 2 is characterized in that: In step S2, the deployed sensors include weight sensors, which are installed on the supporting structure or handling equipment to directly measure the weight change of the stacked materials; RFID tag readers can attach RFID tags to each batch or important single item of materials, deploy fixed or mobile readers at the entrance and exit channels and the edge of the stacking area to realize material identity recognition and location tracking, test the reading range and recognition rate, adjust the antenna angle and power, and ensure that all tags are stably read within the specified area; 3D scanners can scan the stacking area periodically or on demand, build and update the three-dimensional model of the stacked materials, and accurately record the shape, volume and stacking structure.

4. The method for stacking monitoring and residual material calculation according to any one of claims 1 to 3, characterized in that: The step S3 specifically comprises the following steps: Design and implement data receiving services, using a high-concurrency processing framework to ensure low-latency data transmission and buffering; Perform preliminary cleaning and format verification on the received data to remove invalid or erroneous data. If a measurement value outside the normal range is detected, it will be processed through data smoothing or outlier detection algorithms; Standardize the preprocessed data to ensure that data from different sensors are comparable in the same dimension; A data fusion algorithm is used to integrate time series analysis and spatial information to form a comprehensive and accurate description of the stacked materials. For the fusion of stack height and weight, the following weighted fusion formula can be used: F(x)=ω1·H(x)+ω2·W(x); Where F(x) is the comprehensive index after fusion, H(x) and W(x) are the measured values ​​of height and weight, respectively, and ω1 and ω2 are weight coefficients set according to the application scenario.

5. The method for stacking monitoring and residual material calculation based on multi-dimensional data linkage processing according to claim 4 is characterized in that: The step S4 specifically comprises the following steps: Scan the stacking area regularly or on demand to obtain 3D point cloud data of the material stacking, and use high-resolution cameras with image recognition algorithms to identify material types, boundaries and identification information to enhance the accuracy of model details; Integrate the material location information read by weight sensors and RFID tags, as well as data from environmental sensors, to form a comprehensive material attribute database; The original point cloud data is denoised and downsampled, and then converted into a triangular mesh model through a 3D reconstruction algorithm to form a preliminary 3D stacking model; Based on image recognition technology, each batch or single piece of material is identified, and combined with RFID tag data, the correct material properties and weight information are assigned to the model elements; Associating the real-time data from the weight sensor with the materials in the model, and adjusting the model in real time through interpolation or shape deformation algorithms to ensure that the model can reflect the weight changes, additions, removals, and position adjustments of the stacked materials; Based on the dynamically updated 3D model, the mechanical analysis algorithm is used to evaluate the stability of the stacking structure, calculate parameters such as the center of gravity offset and stacking pressure distribution, and compare them with the preset safety threshold. When the threshold is exceeded, an early warning is triggered. The dynamically updated 3D model is displayed through a graphical user interface and interactive browsing is enabled.

6. The method for stacking monitoring and residual material calculation based on multi-dimensional data linkage processing according to claim 4 is characterized in that: The step S5 specifically comprises the following steps: Cross-validation is used to evaluate the generalization ability of the model, and hyperparameters are tuned through grid search and random search methods to find the best model configuration; the material consumption trend is predicted, and the model formula can be abstractly expressed as: h t =f(W h [x t ,h t-1 ]+b h ); y t =g(W y h t +b y ); In the formula, h t is the hidden state, x t is the current input, W h , W y is the weight matrix, b h 、b y is the bias term, f and g are activation functions, and the parameters are adjusted through back propagation and optimization algorithms.

7. The method for stacking monitoring and residual material calculation according to any one of claims 1-3 or 5-6, characterized in that: The step S6 specifically comprises the following steps: Thresholds are set based on the normal state of material stacking. Any situation exceeding these thresholds is considered abnormal. Models are trained on historical data using supervised or unsupervised learning algorithms to identify normal and abnormal patterns. Multimodal anomaly detection methods are used to comprehensively judge abnormal states by combining 3D model data, weight data, and environmental parameters to improve detection accuracy. Threshold setting and dynamic adjustment: Based on historical data and business logic, set the warning threshold of key indicators, use adaptive threshold algorithm, and automatically adjust the threshold according to real-time supply and demand changes to avoid false alarms and missed alarms. The conditions for triggering the warning are: In the formula, I(t) represents the current index value; represents the historical mean; σ represents the standard deviation; k represents the threshold coefficient; Anomaly detection algorithm: Introduce machine learning methods to analyze data streams in real time, identify abnormal changes in stacking patterns, and issue early warning signals in a timely manner; The anomaly score S output by the model is expressed as: S = f(x); In the formula, f is the prediction function of the anomaly detection model, x is the input feature vector, and the higher the output anomaly score, the more the current state deviates from the normal mode.

8. The method for stacking monitoring and residual material calculation according to any one of claims 1-3 or 5-6, characterized in that: The step S7 specifically includes the following steps: Real-time stacking model update: Based on the latest stacking status data obtained by the dynamic monitoring system, including stacking length, width, depth and distribution of materials in each layer, the three-dimensional stacking model is updated. If this process involves three-dimensional reconstruction, the ray casting method or voxelization algorithm is used for three-dimensional reconstruction. The formula is simplified as follows: In the formula, V represents the stacking volume, L i represents the length of the i-th layer, W i represents the width of the i-th layer, H i represents the height of the i-th layer; Accurate calculation of remaining materials: Apply density mapping technology and volume weight conversion, combine the density information of each material, and accurately calculate the total volume and total weight of the remaining materials. The calculation formula is as follows: Where W 剩 represents the weight of the remaining material; V represents the stacking volume; ρ is the average density, which takes into account the mixed average value of each material in the stacking. Represents the cumulative weight of materials removed from the initial state to the current time point.

9. The method for stacking monitoring and residual material calculation based on multi-dimensional data linkage processing according to claim 8, characterized in that: In step S7, the algorithm for generating stacking optimization suggestions includes: Maximizing space utilization: Using genetic algorithms or particle swarm optimization heuristic algorithms, we search for the optimal stacking layout for the current stacking status and remaining materials. The objective function can be defined as minimizing the unused space volume while taking into account the convenience of taking out materials. The simplified formula is: minZ = V_empty - α × C oper ; In the formula, Z is the objective function, V is the total vacant volume, α is the adjustment factor, balancing space utilization and operating efficiency, and C oper For ease of operation; Inventory management suggestions: Based on the remaining material calculation results, use the forecasting model to analyze future demand trends, give replenishment or allocation suggestions, and reduce inventory costs. The formula is expressed as: Q 补 =D 预测 -R 当前 +β×S; In the formula, Q 补 represents the replenishment quantity, D 预测 represents the predicted demand, R 当前 It represents the current surplus material, β is the safety stock coefficient, which reflects the buffer demand for market fluctuations, and S represents the safety stock amount.

10. The method for stacking monitoring and residual material calculation according to any one of claims 1-3 or 5-6 or 9 for multi-dimensional data linkage processing, characterized in that: In step S1, the evaluation of the stacking area includes: measuring the size of the stacking area, evaluating the bearing capacity of the ground, analyzing the lighting conditions, and considering ventilation and fire protection requirements; carefully classifying the types of materials, and recording the physical and chemical properties of each material to ensure the pertinence and safety of subsequent management measures; The requirements definition includes: covering basic inventory tracking and security management, while going deep into the advanced goals of improving operational efficiency, energy conservation and emission reduction, and cost control; if flammable items are stored, the system needs to be highly sensitive to changes in temperature and humidity; if materials are time-sensitive, the focus is on the automated execution of the first-in-first-out principle.

Images