An intelligent warehousing space optimization method based on digital twin technology

Abstract

The present invention relates to the technical field of data processing, and particularly to an intelligent warehousing space optimization method based on digital twin technology. The content includes: collecting original data and preprocessing the original data to obtain preprocessed data; constructing a digital twin model based on the preprocessed data; introducing an adaptive probability space distribution optimization algorithm guided by multiple factors based on the digital twin model, designing the warehousing space layout, and dynamically adjusting the warehousing space layout in combination with the original goods data. It solves the technical problems of low utilization rate of warehousing space and unbalanced resource allocation caused by the low accuracy of processing and analysis of warehousing environment, goods information, etc. in the traditional warehousing space layout method.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and particularly to an intelligent warehousing space optimization method based on digital twin technology. Background Art

[0002] With the rapid development of the logistics industry and the continuous growth of warehousing business, warehousing space management has become the core link in modern logistics systems. Most traditional warehousing space layout methods are designed based on manual rules or static configuration methods, which rely on preset goods classification, historical transfer experience, and fixed warehouse structure parameters, lacking dynamic perception and adaptive adjustment capabilities, and are difficult to cope with real application scenarios such as strong goods liquidity, high operation frequency, and variable demand structure. In warehousing operations with multiple categories, multiple batches, and high frequencies, static layout methods often lead to problems such as low space utilization rate, congested channels, long operation paths, low handling efficiency, and frequent operation conflicts, seriously restricting the operation efficiency and service capacity of the warehousing system.

[0003] Therefore, the above technologies have technical problems such as low warehousing space utilization rate and unbalanced resource allocation due to low accuracy in processing and analyzing warehousing environments, goods information, etc. Summary of the Invention

[0004] The present invention provides an intelligent warehousing space optimization method based on digital twin technology to solve the technical problems of low warehousing space utilization rate and unbalanced resource allocation caused by the low accuracy in processing and analyzing warehousing environments, goods information, etc. in traditional warehousing space layout methods.

[0005] An intelligent warehousing space optimization method based on digital twin technology of the present invention specifically includes the following technical solutions:

[0006] An intelligent warehousing space optimization method based on digital twin technology includes the following steps:

[0007] S1. Collect raw data, preprocess the raw data to obtain preprocessed data; based on the preprocessed data, construct a digital twin model;

[0008] S2. Based on the digital twin model, introduce an adaptive probability space distribution optimization algorithm guided by multiple factors, design the warehousing space layout, and dynamically adjust the warehousing space layout in combination with the original goods data.

[0009] Preferably, the S1 specifically includes:

[0010] In the process of constructing the digital twin model, geometric modeling of the warehousing physical space is carried out based on the physical components in the preprocessed data; on the basis of the completion of geometric modeling, topological modeling logic is introduced to construct a topological graph structure, and through topological mapping operations, the constructed topological graph structure is bound to the preprocessed data.

[0011] Preferably, the S1 specifically includes:

[0012] Define real-time state variables for each spatial node in the topological graph structure, and construct a state vector based on the real-time state variables; adopt a vector field modeling method to bind each state vector to the corresponding spatial node to construct a multi-dimensional state field to express the distribution characteristics of the state vector.

[0013] Preferably, the S1 specifically includes:

[0014] In the process of constructing the digital twin model, a state update mechanism is introduced, and a warehousing state transition function and a reward function are set to enable the digital twin model to have time-varying response capabilities.

[0015] Preferably, the S1 specifically includes:

[0016] In the state update mechanism, based on the warehousing operation data retrieved in real time from the warehousing inbound and outbound records, the action function is dynamically called to update the real-time state variables of each spatial node.

[0017] Preferably, the S2 specifically includes:

[0018] Based on the digital twin model, the warehousing space is discretely divided to generate spatial sub-regions; the original cargo data is collected by sensors deployed in each spatial sub-region, the original cargo data is constituted into an attribute vector, and normalization processing is carried out; the normalized attribute vectors are weighted and superimposed to obtain the priority factor of the cargo.

[0019] Preferably, the S2 specifically includes:

[0020] In the implementation process of the adaptive probability space distribution optimization algorithm under the guidance of multiple factors, based on the position characteristics in the preprocessed data, the spatial environment characteristics in the warehousing operation data, and combined with the historical task density, the attraction weight of the spatial sub-region is obtained.

[0021] Preferably, the S2 specifically includes:

[0022] Based on the priority factor of the cargo and the attraction weight of the spatial sub-region, calculate the sub-region allocation probability of the cargo; based on the sub-region allocation probability of the cargo, carry out the warehousing space layout to obtain the distribution state of the warehousing space.

[0023] Preferably, S2 specifically includes:

[0024] In the implementation process of the adaptive probability space distribution optimization algorithm guided by multiple factors, based on the sub-region allocation probability of goods, calculate the usage density of spatial sub-regions; by calculating the gradient of the usage density of spatial sub-regions and combining the attraction weight of spatial sub-regions, adjust the sub-region allocation probability of goods.

[0025] Preferably, S2 specifically includes:

[0026] In the implementation process of the adaptive probability space distribution optimization algorithm guided by multiple factors, introduce a prediction error feedback mechanism. Based on historical goods data, construct a prediction model through a neural network to obtain the expected inbound and outbound frequency of goods; based on the expected inbound and outbound frequency of goods, combine with the actual inbound and outbound frequency of goods to define the prediction error; based on the prediction error, correct the priority factor of goods, and use the corrected priority factor to calculate the sub-region allocation probability of goods in the subsequent round, forming a closed-loop optimization process.

[0027] The beneficial effects of the technical solution of the present invention are:

[0028] 1. By constructing a digital twin model, the equipment status, goods distribution, task flow, etc. in the warehousing space are mapped in real time, accurately expressing the warehousing geometric structure and reachability topological relationship, maximizing the space utilization rate, avoiding the concentration and congestion of goods or the vacancy of resources, and improving the space scheduling accuracy and flexibility of the warehouse.

[0029] 2. By constructing an adaptive probability space distribution optimization algorithm guided by multiple factors, taking the goods data of each spatial sub-region as input, and measuring the imbalance of the current space density by calculating the usage density of spatial sub-regions, automatically optimizing the space distribution structure through the gradient descent method, realizing the intelligent migration of goods flow from dense regions to sparse regions, adaptively adjusting the space distribution strategy, and enhancing the layout robustness and intelligent level under complex tasks and changing environments.

[0030] 3. By establishing a neural network prediction model driven by historical data, predicting the future inbound and outbound frequency of goods, comparing it with the actual inbound and outbound frequency, and dynamically adjusting the priority factor of goods according to the prediction error, so as to improve the response ability to external demand changes, enhance the forward-looking and anti-disturbance ability of the layout strategy, and avoid resource misallocation caused by inaccurate demand prediction. Description of the Drawings

[0031] Figure 1 It is a flowchart of an intelligent warehousing space optimization method based on digital twin technology described in the present invention. Detailed Embodiment

[0032] To further elaborate on the technical means and effects adopted by the present invention to achieve the intended invention purpose, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without making creative efforts fall within the protection scope of the present invention.

[0033] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the technical field to which the present invention belongs.

[0034] The following specifically describes the specific solution of an intelligent warehouse space optimization method provided by the present invention in conjunction with the accompanying drawings.

[0035] Refer to the attached Figure 1 , which shows a flowchart of an intelligent warehouse space optimization method provided by an embodiment of the present invention. The method includes the following steps:

[0036] S1. Collect the original data, preprocess the original data to obtain the preprocessed data; based on the preprocessed data, construct a digital twin model;

[0037] The original data is obtained by deploying data collection devices in the warehouse environment. The original data includes information data such as goods information, equipment status, environmental changes, item storage locations, access frequencies, physical composition elements, etc.; the data collection devices such as environmental sensors, RFID tags and scanners, barcode scanners, video monitoring devices, robots or automated devices, environmental monitoring devices, etc.; preprocess the original data to obtain the preprocessed data; the preprocessing process includes data cleaning (such as removing duplicate data, filling missing values by methods such as interpolation, mean filling, nearest neighbor interpolation, etc.), data transformation and standardization (such as converting the data of different devices into the same time scale, ensuring that all data has a consistent time stamp, and converting the physical quantities of different units), data smoothing and noise reduction (such as moving average, Gaussian filtering), data aggregation (such as aggregating data according to time periods), data feature extraction, etc. In the preprocessing process, the existing technical means well-known to those skilled in the art are adopted, and will not be elaborated here;

[0038] Further, based on the preprocessed data, construct a digital twin model; the digital twin model can real-time map the operating state of the warehouse space and provide a basis for subsequent space optimization; the specific process is as follows:

[0039] First, the storage space needs to be geometrically modeled and topologically modeled; specifically, geometric modeling uses three-dimensional modeling technology (such as building information modeling BIM modeling technology or open source modeling platform Open3D) to accurately restore the key components in the storage space based on the physical components in the preprocessed data (such as the location and size of the shelves, the width and length of the aisles, the occupied volume of fixed equipment, the boundary range of the working area, etc.); the physical components are expressed in a three-dimensional coordinate system. Its spatial position and geometric properties make the digital twin model highly restored in the structural dimension. The spatial scene output by geometric modeling has a complete structural boundary and spatial constraint relationship, which is the physical basis for building a highly simulated virtual storage environment;

[0040] On the basis of geometric modeling, topological modeling logic is introduced to give the digital twin model semantic connection and accessibility; each geometric entity (such as cargo compartments, entrances and exits, and cargo loading points) is abstracted as a node (i.e., spatial node) in the topological map structure, and the physical accessibility relationship or operation path between the geometric entities is defined as an edge, and attributes (such as traffic status, traffic time, channel capacity, constraints, etc.) are added to the edge to establish a complete topological map structure. The above process is topological modeling, which can convert the logic of "shelf A can reach exit C through channel B" in reality into a path sequence of A→B→C in the topological map structure; further through the topological mapping operation, the constructed topological map structure is bound to the preprocessed data, so that the digital twin model has core capabilities such as path search, connectivity analysis, area division, and connectivity verification;

[0041] After the geometric modeling and topological modeling are completed in coordination, the state expression stage is entered, that is, the real-time state variables are defined for each spatial node in the topological spectrum structure, mainly including the current volume occupancy rate, number of cargo stacking layers, access frequency, cargo type coding, operation task density, regional temperature and humidity, vibration amplitude, equipment workload and other multi-dimensional indicators; based on the real-time state variables, a state vector is constructed. In order to express the distribution characteristics of the state vector in the entire space, a vector field modeling method is used to bind each state vector to its corresponding spatial node, and a multi-dimensional state field is constructed to describe the attribute set presented by any point in the storage space at a certain moment;

[0042] Furthermore, to endow the digital twin model with the ability of dynamic evolution and continuously maintain the synchronous mapping of the actual warehousing state during actual operation, a state update mechanism is introduced to enable the digital twin model to have time-varying response capabilities. The state update mechanism is based on the Markov decision process (MDP), and a warehousing state transition function and a reward function are set. The warehousing state transition function defines how the warehousing state transfers from the current state to the next state after the execution of a certain warehousing operation (such as goods shelving, picking, replenishment, etc.). The factors considered during the state transition include changes in space occupancy, aisle accessibility, goods turnover priority, etc.; while the reward function takes factors such as improved space utilization rate, goods turnover rate, response time, etc. as evaluation indicators and is used to guide the evolution of the warehousing space towards the optimal target state; the data of the considered factors involved in the warehousing state transition function and the reward function all come from the preprocessed data; the warehousing state transition function and the reward function are both well-known technical means to those skilled in the art and will not be elaborated here;

[0043] In the state update mechanism, the warehousing operation data (including goods inbound records, outbound behaviors, path usage frequencies, etc.) is retrieved in real time from the warehousing inbound and outbound records and used as the basis for updating the state of the digital twin model. The real-time state variables of each spatial node are updated by dynamically invoking the action function. For example, when goods are transferred from shelf A to shelf B, the action function will automatically modify the volume occupancy rate, access frequency and other parameters of these two spatial nodes, and at the same time adjust the traffic load distribution of the entire topology map to ensure that the digital twin model can respond in real time to actual operations. For variables that are difficult to directly observe, such as equipment failure risks, operation efficiency trends, path blockage probabilities, etc., a particle filter is introduced for state estimation, a hypothetical trajectory is established, and the likelihood of the hypothetical trajectory is evaluated based on the preprocessed data, so as to infer the warehousing potential state closest to the actual situation. The particle filter and the action function are both well-known technical means to those skilled in the art and will not be elaborated here;

[0044] Through the above process, the digital twin model is obtained.

[0045] S2. Based on the digital twin model, an adaptive probability space distribution optimization algorithm guided by multiple factors is introduced to design the warehousing space layout, and the warehousing space layout is dynamically adjusted in combination with the original goods data.

[0046] Based on the digital twin model, an adaptive probability space distribution optimization algorithm guided by multiple factors is introduced, and the warehousing layout is dynamically optimized by integrating real-time data input, a prediction model and a prediction error feedback mechanism. The specific implementation process is as follows:

[0047] First, based on the digital twin model, the entire storage space is discretely partitioned through existing automated geometric partitioning techniques to generate spatial sub-regions. Raw cargo data is collected by sensors deployed in each spatial sub-region, including attributes such as the basic properties of all goods, storage locations, current task scheduling status, and historical storage trajectories. The basic properties of the goods include volume, mass, inbound and outbound frequency, temperature and humidity sensitivity (calculated by modeling the variation range of temperature and humidity at different temperatures and humidities), and handling cost. The above raw cargo data constitutes an attribute vector. To unify the magnitude differences between different attributes, the attribute vector is normalized to eliminate the influence of dimensions. The normalization method uses the range normalization method, whose purpose is to make the variables in all dimensions be on the same numerical scale in subsequent weight calculations to prevent the variables in a certain dimension from having a dominant influence on the calculation results due to large values.

[0048] After normalization, to quantify the sensitivity or priority usage demand of each cargo for the storage space, the priority factor of the cargo is defined. , representing the -th cargo's spatial scheduling weight in the entire warehouse, is generated by weighted superposition of multiple attribute vectors of the cargo.

[0049] Meanwhile, the concept of regional attractiveness is introduced into the storage space. The attractiveness weight of each spatial sub-region is obtained by experimental fitting according to parameters such as location characteristics (such as access to aisles), spatial environmental characteristics (such as temperature and humidity control ability, equipment support level), and historical task density. Among them, the location characteristics come from preprocessed data, the spatial environmental characteristics come from storage operation data, and the historical task density comes from the existing database.

[0050] Furthermore, based on the priority factor of the cargo and the attractiveness weight of the spatial sub-region, the sub-region allocation probability of the cargo is calculated. The specific formula is:

[0051] ,

[0052] where is the probability that the -th cargo is allocated to the -th spatial sub-region, describing the distribution tendency of each cargo in different spatial sub-regions, that is, the sub-region allocation probability of the cargo. is the priority factor of the -th cargo; represents the attractiveness weight of the -th spatial sub-region; is the number of spatial sub-regions obtained after partitioning the storage space;

[0053] Specifically, based on the sub-region allocation probability of goods, the layout of the storage space is carried out to obtain the distribution state of the storage space; each good tends to be allocated to an area with higher attractiveness (such as an area close to the main passage or with better temperature and humidity control capabilities);

[0054] Further, traverse the goods included in the scheduling optimization in the storage space, multiply the sub-region allocation probability of each good by the three-dimensional space volume occupied by the good, and sum them up to calculate the usage density of each space sub-region, so as to measure whether a certain space sub-region is in a high-density, oversaturated or resource-waste state at present; by calculating the gradient of the usage density of the space sub-region, adjust the sub-region allocation probability of the goods; the specific adjustment formula is:

[0055] ,

[0056] wherein, is the probability that the th good is allocated to the th space sub-region at time ; is the probability that the th good is allocated to the th space sub-region at time ; is the probability adjustment amplitude coefficient, determined by the expert experience method, and the value range is ; is the gradient of the usage density of the space sub-region at time , and the calculation of the gradient is a well-known technical means to those skilled in the art and will not be elaborated here; is the three-dimensional space volume occupied by the th good;

[0057] Finally, introduce a prediction error feedback mechanism to correct the layout error caused by inaccurate estimation of external task requirements. Based on the historical goods data called from the existing database, construct a prediction model through the existing neural network to predict the expected inbound and outbound frequency of each good; the neural network is a well-known technical means to those skilled in the art and will not be elaborated here; based on the expected inbound and outbound frequency of the goods, combined with the actual inbound and outbound frequency of the goods, define the prediction error, and based on the prediction error, correct the priority factor of the goods to obtain the corrected priority factor; the specific formula is:

[0058] ,

[0059] wherein, is a small positive real number used to prevent the denominator from being zero, and the value is ; is at time the The expected frequency of goods entering and leaving the warehouse; is Moment The actual frequency of goods entering and leaving the warehouse comes from the existing database; is the revised priority factor, which replaces the priority factor before the revision , and is used to calculate the probability of sub-area allocation of goods in subsequent rounds, thus forming a closed-loop optimization process of "prediction-execution-feedback-adjustment". The continuous correction of the priority factor of goods realizes dynamic adaptive layout, so that the storage space not only responds to static data, but also continuously learns and corrects strategies from the operation history.

[0060] In summary, an intelligent storage space optimization method based on digital twin technology was completed.

[0061] The order of the embodiments of the invention is for description only and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0062] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

[0063] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that the technical solutions described in the above embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the protection scope of the present invention.

Claims

1. An intelligent warehousing space optimization method based on digital twin technology, characterized in that, It includes the following steps: S1. Collect the original data, preprocess the original data to obtain the preprocessed data; Based on the preprocessed data, construct a digital twin model; S2. Based on the digital twin model, discretely divide the storage space to generate spatial sub-regions; collect the original cargo data through sensors deployed in each spatial sub-region, form an attribute vector from the original cargo data, and perform normalization processing; perform weighted superposition on the normalized attribute vectors to calculate the spatial scheduling weight of the cargo in the entire storage as the priority factor of the cargo; introduce an adaptive probability space distribution optimization algorithm guided by multiple factors, based on the position features in the preprocessed data and the spatial environment characteristics in the storage operation data, and combine with the historical task density to obtain the attraction weight of the spatial sub-region; based on the priority factor of the cargo and the attraction weight of the spatial sub-region, calculate the sub-region allocation probability of the cargo; based on the sub-region allocation probability of the cargo, design the storage space layout, and dynamically adjust the storage space layout in combination with the original cargo data.

2. The intelligent warehousing space optimization method based on digital twin technology according to claim 1, wherein, The S1 specifically includes: In the process of constructing the digital twin model, based on the physical composition elements in the preprocessed data, perform geometric modeling on the storage physical space; on the basis of the completion of the geometric modeling, introduce the topological modeling logic, construct a topological graph structure, and through the topological mapping operation, bind the constructed topological graph structure to the preprocessed data.

3. An intelligent warehousing space optimization method based on digital twin technology according to claim 2, characterized in that, The S1 specifically includes: Define real-time state variables for each spatial node in the topological graph structure, and based on the real-time state variables, construct a state vector; adopt a vector field modeling method to bind each state vector to the corresponding spatial node to construct a multi-dimensional state field to express the distribution characteristics of the state vector.

4. The intelligent warehousing space optimization method based on digital twin technology according to claim 3, wherein, The S1 specifically includes: In the process of constructing the digital twin model, introduce a state update mechanism, set a storage state transition function and a reward function to enable the digital twin model to have time-varying response capabilities.

5. The intelligent warehousing space optimization method based on digital twin technology according to claim 4, characterized in that The S1 specifically includes: In the state update mechanism, based on the storage operation data retrieved in real time from the storage inbound and outbound records, dynamically call the action function to update the real-time state variables of each spatial node.

6. The intelligent warehouse space optimization method based on digital twin technology according to claim 1, wherein, The S2 specifically includes: The specific calculation formula for the sub-region allocation probability of the cargo is: ， in, It is Goods are assigned to The probability of a spatial sub-region represents the probability of sub-region allocation of goods; It is Priority factor for each shipment; Indicates The attraction weight of each spatial sub-region; It is the number of spatial sub-areas obtained after dividing the storage space.

7. An intelligent warehousing space optimization method based on digital twin technology according to claim 6, characterized in that, The S2 specifically includes: In the implementation process of the adaptive probability space distribution optimization algorithm guided by multiple factors, based on the sub-region allocation probability of the cargo, calculate the usage density of the spatial sub-region; by calculating the gradient of the usage density of the spatial sub-region and combining with the attraction weight of the spatial sub-region, adjust the sub-region allocation probability of the cargo.

8. An intelligent warehousing space optimization method based on digital twin technology according to claim 7, characterized in that, The S2 specifically includes: In the implementation process of the adaptive probability space distribution optimization algorithm guided by multiple factors, a prediction error feedback mechanism is introduced. Based on historical cargo data, a prediction model is constructed through a neural network to obtain the expected inbound and outbound frequencies of the cargo. Based on the expected inbound and outbound frequencies of the cargo and combined with the actual inbound and outbound frequencies of the cargo, the prediction error is defined. Based on the prediction error, the priority factor of the cargo is corrected, and the corrected priority factor is used to calculate the sub-region allocation probability of the cargo in the subsequent round, forming a closed-loop optimization process.

Images