CN117910906A

CN117910906A - Data visualization method and system applied to intelligent logistics

Info

Publication number: CN117910906A
Application number: CN202410184387.4A
Authority: CN
Inventors: 白小波
Original assignee: Guangdong Kanglida Internet Of Things Technology Co ltd
Current assignee: Guangdong Kanglida Internet Of Things Technology Co ltd
Priority date: 2024-02-19
Filing date: 2024-02-19
Publication date: 2024-04-19
Anticipated expiration: 2044-02-19
Also published as: CN117910906B

Abstract

The embodiment of the application relates to the technical field of data processing, in particular to a data visualization method and a data visualization system applied to intelligent logistics, which can realize high-efficiency visual conversion of logistics system data, thereby providing powerful decision support for logistics management and planning. According to the technical scheme, key data features and indexes are identified by mining state knowledge in logistics system data, so that pertinence and accuracy of visual contents are ensured. And then, by judging the visualized connection weights, the weights of different data features in the visualization can be reasonably distributed according to the actual demands and focus points of a logistics system, so that key information is highlighted, and information overload is avoided.

Description

Data visualization method and system applied to intelligent logistics

Technical Field

The application relates to the technical field of data processing, in particular to a data visualization method and system applied to intelligent logistics.

Background

Along with the rapid development and digital transformation of the logistics industry, massive logistics data are continuously generated, how to efficiently process and analyze the data and present the result to a manager in an intuitive and clear manner becomes a problem to be solved urgently. The conventional data visualization method often lacks in-depth consideration of the characteristics of the logistics system, so that the visualization result is not strong in pertinence, information overload or key information is ignored, and the like. Therefore, the development of the high-efficiency and intelligent data visualization method capable of being applied to intelligent logistics has important significance in improving logistics management and planning level. In recent years, although some researches try to apply techniques such as data mining and machine learning to physical distribution data visualization, many challenges still exist. .

Disclosure of Invention

In order to improve the technical problems in the related art, the application provides a data visualization method and a system applied to intelligent logistics.

In a first aspect, an embodiment of the present application provides a data visualization method applied to a smart logistics, and the method is applied to a data visualization system, where the method includes:

Carrying out state knowledge mining on logistics system data to be subjected to visual conversion to obtain an original logistics system state vector of the logistics system data to be subjected to visual conversion;

According to the original logistics system state vector, carrying out visual engagement weight discrimination on the image output characteristics of the logistics system data to be subjected to visual conversion in at least one characteristic channel to obtain visual engagement weight discrimination results of the logistics system data to be subjected to visual conversion in each characteristic channel respectively;

When the visual engagement weight discrimination result characterizes that the logistic system data to be subjected to visual conversion does not have the visual engagement weight under the corresponding characteristic channel, acquiring a knowledge optimization indication vector of the original logistic system state vector under the corresponding characteristic channel;

carrying out knowledge optimization processing on the original logistics system state vector according to the knowledge optimization indication vector to obtain a target logistics system state vector corresponding to the original logistics system state vector;

and performing visual mapping on the target logistics system state vector to obtain a target visual result corresponding to the logistics system data to be subjected to visual conversion, wherein the target visual result has the visual linking weight under each characteristic channel.

In a second aspect, the present application also provides a data visualization system, comprising: a memory for storing program instructions and data; and a processor coupled to the memory for executing instructions in the memory to implement the method as described above.

In a third aspect, the present application also provides a computer storage medium containing instructions which, when executed on a processor, implement the above-described method.

Drawings

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

Fig. 1 is a flow chart of a data visualization method applied to an intelligent logistics according to an embodiment of the present application.

FIG. 2 is a block diagram of a data visualization system 300 according to an embodiment of the present application

Detailed Description

The technical scheme of the application will be described below with reference to the accompanying drawings.

Reference will now be made in detail to exemplary embodiments, examples of which are illustrated in the accompanying drawings. When the following description refers to the accompanying drawings, the same numbers in different drawings refer to the same or similar elements, unless otherwise indicated. The implementations described in the following exemplary examples do not represent all implementations consistent with the application. Rather, they are merely examples of apparatus and methods consistent with aspects of the application.

It should be noted that the terms "first," "second," and the like in the description of the present application and the above-described drawings are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order.

The method embodiments provided by the embodiments of the present application may be performed in a data visualization system, a computer device, or similar computing device. Taking as an example operation on a data visualization system, the data visualization system may comprise one or more processors (which may include, but is not limited to, a processing means such as a microprocessor MCU or a programmable logic device FPGA) and a memory for storing data, and optionally the data visualization system may further comprise transmission means for communication functions. It will be appreciated by those of ordinary skill in the art that the above-described structure is merely illustrative and is not intended to limit the structure of the data visualization system described above. For example, the data visualization system may also include more or fewer components than shown above, or have a different configuration than shown above.

The memory may be used to store a computer program, for example, a software program of application software and a module, for example, a computer program corresponding to a data visualization method applied to a smart stream in an embodiment of the present application, and the processor executes the computer program stored in the memory, thereby performing various functional applications and data processing, that is, implementing the method described above. The memory may include high speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid state memory. In some examples, the memory may further include memory remotely located with respect to the processor, the remote memory being connectable to the data visualization system through a network. Examples of such networks include, but are not limited to, the internet, intranets, local area networks, mobile communication networks, and combinations thereof.

The transmission means is used for receiving or transmitting data via a network. Specific examples of the network described above may include a wireless network provided by a communication provider of the data visualization system. In one example, the transmission means comprises a network adapter (Network Interface Controller, simply referred to as NIC) that can be connected to other network devices via a base station to communicate with the internet. In one example, the transmission device may be a Radio Frequency (RF) module, which is used to communicate with the internet wirelessly.

Referring to fig. 1, fig. 1 is a flow chart of a data visualization method applied to a smart stream according to an embodiment of the present application, where the method is applied to a data visualization system, and further includes steps 110 to 150.

And 110, carrying out state knowledge mining on the logistics system data to be subjected to visual conversion to obtain an original logistics system state vector of the logistics system data to be subjected to visual conversion.

And 120, according to the state vector of the original logistics system, performing visual engagement weight discrimination on the image output characteristics of the logistics system data to be subjected to visual conversion in at least one characteristic channel to obtain visual engagement weight discrimination results of the logistics system data to be subjected to visual conversion in each characteristic channel.

And 130, when the visual engagement weight discrimination result characterizes that the logistic system data to be subjected to visual conversion does not have the visual engagement weight under the corresponding characteristic channel, acquiring a knowledge optimization indication vector of the original logistic system state vector under the corresponding characteristic channel.

And 140, carrying out knowledge optimization processing on the original logistics system state vector according to the knowledge optimization indication vector to obtain a target logistics system state vector corresponding to the original logistics system state vector.

And 150, performing visual mapping on the target logistics system state vector to obtain a target visual result corresponding to the logistics system data to be subjected to visual conversion, wherein the target visual result has the visual linking weight under each characteristic channel.

In the above embodiment, the logistic system data to be subjected to the visualization conversion means those logistic system data which have not been subjected to the visualization processing and exist in an original or complex form. Examples: such as order data, stock quantity, and records of goods in and out of a logistics center.

State knowledge mining is the process of extracting hidden state information and knowledge from raw data. For example, by analyzing historical order data, peak and valley periods of the order volume in a certain period, and the most popular commodity category in each period are mined.

Original logistics system state vector: the numerical vector representing the current state of the logistics system generally comprises a plurality of dimensions, each dimension representing a particular system parameter or characteristic. For example, a three-dimensional vector (1200, 80, 5), where 1200 represents the current stock quantity, 80 represents the average daily amount of orders, and 5 represents the current number of transportation vehicles.

The characteristic channel is as follows: in data visualization, channels, such as color, shape, size, etc., are used to distinguish between different data types or attributes. For example, in the logistics path visualization, different colors represent different transportation modes (red for land transportation and blue for sea transportation).

Image output characteristics refer to image attributes such as brightness, contrast, texture, etc. exhibited by data during the visualization process. For example, in a stock quantity visualization chart, the height and color of the histogram may represent stock quantity and stock status, respectively (e.g., green for adequate, red for insufficient).

Visual engagement weight discrimination is a process of judging the importance and applicability of visual expression of data under a specific characteristic channel. For example, for transportation path data, in map visualization, the width of the road may be more important (higher weight) to show traffic flow, while in path optimization analysis, the color of the road may be more important (higher weight) to show the cost of different road segments.

And the weight distribution result of the visual expression of the data under each characteristic channel is obtained after the visual engagement weight judgment result is subjected to weight judgment. For example, the discrimination results may indicate that the size of the node (representing warehouse capacity) has a higher visualization weight than the color of the node (representing warehouse type) when the logistics network is presented.

The knowledge optimization instruction vector is a vector for guiding data optimization processing, and is used for indicating which data features need to be optimized and how to optimize based on the visual engagement weight discrimination result. If the discrimination results show that the data under a certain characteristic channel is not good in visualization effect, the indication vector may suggest to smooth or resample the data under the channel.

The knowledge optimization processing is processing of data according to the knowledge optimization instruction vector, and aims to improve the visual effect of the data or improve the accuracy of data analysis. For example, for time of transportation data in the presence of noise, a moving average method may be used for smoothing to reduce the effect of data fluctuations on the visualization effect.

The target logistics system state vector is obtained after knowledge optimization treatment, and is more suitable for visual expression or analysis. For example, the original vector (1200, 80, 5) may be optimized to become (1200, 85,5) in which the amount of orders is predictively adjusted to reflect the impact of the upcoming promotional event on the amount of orders.

The visual mapping is a process of mapping the optimized logistics system state vector into a visual chart or image so as to intuitively show the logistics state or trend. For example, the target logistics system state vector (1200, 85,5) is mapped into a histogram, where the height of each column represents a corresponding value.

The target visualization result is finally obtained, and the logistics data after optimization treatment and visualization mapping is visualized into a chart or an image. For example, a dynamic bar graph showing stock quantity, predicted amount of orders and number of transport vehicles, and visual expression under each characteristic channel are clear and effective.

The description of steps 110-150 is presented below by way of an example.

The logistics system of a certain e-commerce platform accumulates a large amount of data about order processing, inventory management, distribution timeliness and the like. These data include the amount of orders processed per day, the real-time stock amount per SKU (stock level unit), the average delivery time of each delivery area, and the like. In step 110, by analyzing and mining these historical data, some hidden status knowledge, such as which merchandise sales increase during a certain period of time, which areas have higher distribution pressures, etc., can be obtained. This information is abstracted into an original logistics system state vector, where each dimension may represent a different logistics index, such as order processing speed, inventory turnover, distribution timeliness, etc.

Continuing with the e-commerce platform, in step 120, it is necessary to determine which data features are more important for visualization based on the original logistics system state vector. For example, the overall order processing efficiency and inventory status may be of greater concern to the management layer, as this directly affects customer satisfaction and operating costs. Therefore, in the visual engagement weight determination, the order processing speed and the inventory turnover rate may be given a high weight. For logistics planning departments, the coverage and timeliness of the distribution network may be more of a concern, and thus these features will be highlighted when visualized.

If it is found in step 120 that certain data characteristics are not suitable for visual presentation in the current logistics system state (e.g., because the data fluctuates too much or lacks sufficient information), then step 130 is entered. In this step, the system may generate a knowledge-optimized indicator vector suggesting further processing or optimization of these data features that are not suitable for direct visualization. For example, for highly fluctuating delivery time data, the system may suggest smoothing by moving average or exponential smoothing methods, etc., to more clearly show trends and patterns when visualized.

In step 140, the original logistics system state vector is correspondingly processed and optimized according to the knowledge optimization instruction vector generated in step 130. Continuing to take the e-commerce platform as an example, if the system suggests to carry out smooth processing on the delivery time data, a corresponding algorithm is adopted to process the original delivery time data in the step, so that a target logistics system state vector which is more stable and easy to visually display is obtained. This new vector will more accurately reflect the actual state of the logistics system and help provide clearer, more insight information in the subsequent visualization step.

Finally, in step 150, the optimized target logistics system state vector is mapped into a specific visualization chart or image. For example, a histogram may be used to show the distribution of the order amounts in each region, a line graph may be used to show the trend of the change in the stock amounts over a period of time, a thermodynamic diagram may be used to show the coverage and timeliness of the distribution network, etc. Through the visual charts or images, the management layer and the logistics planning department can intuitively know the current state and future trend of the logistics system, so that more intelligent decision making and optimization measures can be made. Meanwhile, since the visualized results are generated on the basis of the state vector of the target logistics system after the optimization treatment, the visualized results can reflect the actual running condition of the logistics system more accurately and provide more valuable information and insight.

In other examples, the original logistics system state vector, the image output features, and the knowledge optimization designation vector will be described in detail by specific numerical feature vectors.

Original logistics system state vector: there is a simplified logistics system, the state of which can be described by the following three key indicators: order quantity (O), stock level (S) and number of transport vehicles (V). In this system, each index may be represented by a specific numerical value. Thus, the raw logistics system state vector may be a three-dimensional vector, where each dimension corresponds to an index.

For example, at a particular point in time, the state vector for the logistics system may be: original logistics system state vector= (120, 850, 20).

This vector indicates that there are currently 120 orders waiting for processing, inventory levels of 850 units, and 20 transport vehicles are available.

Image output characteristics: each dimension (i.e., order quantity, inventory level, and number of transportation vehicles) may be mapped to an output feature of the image when converting the raw logistics system state vector to a visualization chart. These output features may be colors, shapes, sizes, etc. for distinguishing and representing different data dimensions in the visualization chart.

For example, in one bar graph, the height of the columns may be used to represent the order volume, the color of the columns to represent the level of inventory (e.g., green for adequate, red for insufficient), and the width or texture of the columns may be used to represent the number of transportation vehicles.

In this example, the image output features may include:

Column height: corresponding order quantity (O);

Column color: status (sufficient/insufficient) of the corresponding stock level (S);

Pillar width or texture: corresponding to the number of transport vehicles (V).

The knowledge optimization indicating vector is a vector guiding the data optimization process and indicates which data features need to be optimized and how to optimize based on the visual engagement weight discrimination results. This vector may contain information about how to adjust or transform the original logistics system state vector in order to obtain better results when visualized.

For example, in the visual engagement weight discrimination process, the direct visual effect of the stock level (S) is found to be not ideal, because the numerical range is too large to be clearly shown in the graph. At this point, the knowledge optimization indicating vector may suggest a logarithmic transformation or normalization process on inventory levels to better show their relative changes and trends when visualized.

Knowledge optimization indicates that the vector may be shaped as: (0, log (S), 0).

This vector indicates that the order volume and the number of transport vehicles are kept unchanged, while a logarithmic transformation is applied to the stock level. In this way, an optimized logistics system state vector can be obtained, and the stock level dimension of the logistics system state vector is more suitable for visual display.

In practical applications, knowledge optimization indicating vectors may be more complex, including optimization suggestions and weight assignments for multiple dimensions, to ensure that status information of the logistics system can be accurately and clearly conveyed in the visualization process.

Based on the above examples of vectors, the target logistics system state vector is obtained through a series of processing and optimization, and is more suitable for visually expressing or analyzing the current state of the logistics system. This vector typically contains key indicators that describe the state of the logistics system in its entirety, and these indicators have been optimized to more clearly show the characteristics and trends of the logistics system when visualized.

The following is a specific example of a numerical feature vector for developing and introducing a target logistics system state vector: consider, for example, a logistics system that includes order processing, inventory management, and shipping delivery. In this system, three key indicators are selected to describe its state: order quantity (O), inventory turnover (T), and transport efficiency (E). These three indicators represent the order handling, inventory liquidity, and shipping efficiency in the logistics system, respectively.

After a series of data processing and optimization, the following target logistics system state vectors are obtained:

Target logistics system state vector= (100,0.8, 90).

The specific numerical features in this vector are as follows:

Order quantity (O) =100: indicating that the current volume of orders waiting to be processed in the logistics system is 100. The value is the result after optimization processing, and may include adjustment of the predicted order quantity or the actual order quantity so as to more accurately reflect the actual load condition of the system;

Inventory turnover rate (T) =0.8: indicating the liquidity of the inventory, i.e., the ratio of sales of the inventory items over a period of time to the average inventory. The closer this value is to 1, the faster the inventory turnover, and the less inventory backlog. In this example, 0.8 indicates a higher inventory turnover rate and good inventory liquidity;

Transport efficiency (E) =90: indicating the efficiency level of shipping delivery in a logistics system. This value may be based on a comprehensive assessment of several factors such as shipping timeliness, cost effectiveness, etc. In this example, 90 indicates that the transportation efficiency is high, and the cargo can be efficiently delivered to the destination.

This target logistics system state vector provides a comprehensive description of the current state of the logistics system and makes these metrics more suitable for visual expression and analysis through optimization processing. By mapping these indicators into visual charts or images, the operational condition and performance of the logistics system can be intuitively understood, thereby making more intelligent decisions to optimize logistics operations and management.

Therefore, by applying the technical scheme, the high-efficiency visual conversion of the logistics system data can be realized, so that powerful decision support is provided for logistics management and planning. According to the technical scheme, key data features and indexes are identified by mining state knowledge in logistics system data, so that pertinence and accuracy of visual contents are ensured. And then, by judging the visualized connection weights, the weights of different data features in the visualization can be reasonably distributed according to the actual demands and focus points of a logistics system, so that key information is highlighted, and information overload is avoided.

When the discrimination result shows that certain data features are not suitable for direct visualization, the technical scheme provides a knowledge optimization indication vector to guide necessary optimization processing on the state vector of the original logistics system, and further improves the quality and effect of data visualization. Finally, through mapping the optimized target logistics system state vector to a visual chart or an image, the state characteristics and the trend of the logistics system can be intuitively and clearly displayed, a manager is helped to quickly grasp the overall operation condition of the logistics system, potential problems are found in time, the resource allocation is optimized, and the logistics efficiency and the customer satisfaction are improved.

Therefore, the technical scheme not only improves the visual intelligent level of the logistics data, but also provides powerful technical support for digital transformation and upgrading in the logistics industry, and promotes the efficient, intelligent and sustainable development of the logistics system.

In other words, by integrating and processing various data from the logistics system, the complex logistics data are converted into visual and easily understood graphics or images by using advanced data analysis and visualization technology, so that management personnel can conveniently and quickly understand logistics states and trends. Meanwhile, the system also combines artificial intelligence and machine learning algorithm to realize intelligent analysis and prediction of data and help decision makers to make more accurate decisions. The intelligent logistics management system has remarkable effects in the aspects of improving logistics efficiency, optimizing resource allocation and the like, and provides a powerful support tool for intelligent logistics management and operation.

In some possible examples, the state knowledge mining is implemented by at least one state knowledge mining model, and when the number of state knowledge mining models is plural, the state knowledge mining is performed on the logistics system data to be subjected to the visual conversion in step 110, to obtain an original logistics system state vector of the logistics system data to be subjected to the visual conversion, including: carrying out state knowledge mining on the logistics system data to be subjected to visual conversion through a 1 st state knowledge mining model to obtain a 1 st original logistics system state vector; a self-add 1 loop is performed on x to achieve the following: carrying out state knowledge mining on the logistics system data to be subjected to visual conversion according to the x-1 original logistics system state vector through an x-th state knowledge mining model to obtain the x-th original logistics system state vector; wherein x is 1 or more and N or less.

In the extended technical scheme, a plurality of state knowledge mining models are introduced to mine the logistics system data to be subjected to visual conversion in a deep and multi-angle mode. These models may be based on different algorithms or theories, such as machine learning models, deep learning models, statistical models, etc., that are capable of extracting state knowledge of different levels, different dimensions, etc., from the data.

For example, there are three state knowledge mining models: model a, model B, and model C. Firstly, the original logistics data is mined through a model A to obtain an initial original logistics system state vector, and the vector possibly contains basic information such as order quantity, stock quantity, transportation speed and the like. Then, the model B is utilized, the original logistics system state vector obtained in the previous step is used as input, excavation is carried out again, a second original logistics system state vector is obtained, and the vector may further reveal deeper information such as association relation between order quantity and stock quantity, fluctuation trend of transportation speed and the like. Finally, through the model C, a second original logistics system state vector is taken as input, and final state knowledge mining is carried out to obtain a third original logistics system state vector, wherein the third original logistics system state vector possibly contains predictive information, such as prediction of future order quantity, risk assessment of stock exhaustion and the like.

In this process, each model works based on the output of the previous model, forming a kind of chained excavation process. This design may allow each model to focus on extracting state knowledge of a particular type or hierarchy, thereby improving the accuracy and efficiency of mining.

The state knowledge mining model is a mathematical model or computational tool that is specialized to extract state knowledge from the data of the physical flow system. These models may be based on different algorithms or theories for identifying, analyzing, and interpreting patterns, trends, and associations in the data.

Self-imposed 1-cycle is a programming technique for implementing a cyclic operation. Here, it means that from model 1, state knowledge mining is performed using each model in turn until all models are used. After each cycle, the number (x) of the model is automatically incremented by 1 so that the next model is used in the next cycle.

By introducing a plurality of state knowledge mining models and sequentially mining in a chained mode, the technical scheme can realize more comprehensive and deeper state knowledge extraction of logistics system data. Each model can further refine and deepen state knowledge based on the previous model, so that more hidden information and association relations in the data are revealed.

The multi-level and multi-angle mining mode not only improves the accuracy and the integrity of state knowledge mining, but also provides a richer and more valuable data base for subsequent visual conversion. At the same time, this design also helps to improve the efficiency and computational performance of the mining, since each model is focused on extracting state knowledge of a particular type or hierarchy.

Therefore, the technical scheme can further improve the quality and effect of the visualization of the logistics data, help a manager to know the running state and potential problems of the logistics system more deeply, and provide more accurate and comprehensive data support for logistics decision.

In some optional embodiments, the method further includes, by the xth state knowledge mining model, performing state knowledge mining on the logistics system data to be subjected to visual conversion according to the xth-1 original logistics system state vector, and before obtaining the xth original logistics system state vector: according to the x-1 th original logistics system state vector, carrying out visual engagement weight discrimination on the image output characteristics of the logistics system data to be subjected to visual conversion in each characteristic channel to obtain x-1 th visual engagement weight discrimination results of the logistics system data to be subjected to visual conversion in each characteristic channel respectively; and carrying out state identification on the x-1 original logistics system state vector according to the x-1 visualized engagement weight discrimination result to obtain the x-1 target logistics system state vector. Based on this, the state knowledge mining is performed on the logistics system data to be subjected to visual conversion according to the x-1 th original logistics system state vector through the x-th state knowledge mining model to obtain the x-th original logistics system state vector, which includes: and carrying out state knowledge mining on the logistics system data to be subjected to visual conversion according to the (x-1) target logistics system state vector through an (x) state knowledge mining model to obtain the (x) original logistics system state vector.

In this alternative embodiment, the solution adds two steps before state knowledge mining through the xth state knowledge mining model. Firstly, according to the x-1 original logistics system state vector, carrying out visual connection weight judgment on the image output characteristics of the logistics system data to be subjected to visual conversion in each characteristic channel. The purpose of this step is to determine the importance and contribution of each feature channel in the visual representation so that the subsequent state knowledge mining can focus more accurately on the key features.

For example, there are three characteristic channels: order quantity channel, stock quantity channel and transport speed channel. The weight value of each channel under the x-1 original logistics system state vector can be obtained through visual engagement weight discrimination, such as an order quantity channel 0.6, a stock quantity channel 0.3 and a transportation speed channel 0.1. This means that in the current state, the characteristics of the order quantity channel are most important for visual representation, next to the stock quantity channel and finally to the transport speed channel.

And then, carrying out state identification on the x-1 original logistics system state vector according to the x-1 visual engagement weight discrimination result to obtain the x-1 target logistics system state vector. The purpose of this step is to adjust or optimize the original state vector according to the weight discrimination results to better adapt to the requirements of visual expression. The specific state recognition method can be based on rules, models or algorithms, and can correspondingly process or convert the value of each characteristic channel according to the weight discrimination result.

And finally, carrying out state knowledge mining on the logistics system data to be subjected to visual conversion according to the (x-1) th target logistics system state vector through an x-th state knowledge mining model to obtain an x-th original logistics system state vector. This step is similar to the previous step, except that it is mined based on the optimized target logistics system state vector, so that state knowledge related to the visual expression can be extracted more accurately.

The characteristic channel is as follows: in the logistics system data, different data characteristics can be expressed and visualized through different channels. Each characteristic channel corresponds to a particular aspect or dimension in the logistics system, such as order quantity, inventory quantity, transport speed, etc.

The visual engagement weight judgment is a method or a process used for determining the weight or importance of the logistics system data when the logistics system data are visually expressed under each characteristic channel. It can help identify which feature channels are most critical to visual expression in the current state.

The state identification is a process of adjusting or optimizing the state vector of the original logistics system based on the visual engagement weight discrimination result. The method aims at enabling the state vector to better adapt to the requirements of visual expression, highlighting key features and improving visual effects.

By adding two steps of visual engagement weight discrimination and state recognition in the alternative embodiment, the technical scheme can more accurately extract the state knowledge related to the visual expression. The visual engagement weight discrimination helps to determine the importance and contribution degree of each feature channel in visual expression, so that the key features can be focused more on by subsequent state knowledge mining. And the state identification is carried out on the original state vector according to the weight discrimination result, so that the accuracy and the effect of state knowledge mining are further improved.

Therefore, the technical scheme can improve the quality and effect of the visualization of the logistics data, so that the visual expression can reflect the actual state and trend of the logistics system more accurately and intuitively. This is of great significance to the logistics manager, as it can help them to grasp the running status of the logistics system more quickly, discover and solve potential problems in time, thereby improving logistics efficiency and reducing costs.

In some preferred embodiments, according to the x-1 th visual engagement weight discrimination result, performing state identification on the x-1 st original logistics system state vector to obtain an x-1 st target logistics system state vector, including: when the x-1 th visual engagement weight discrimination result indicates that the logistic system data to be subjected to visual conversion does not have the visual engagement weight under the corresponding characteristic channel, carrying out knowledge optimization processing on the x-1 th original logistic system state vector to obtain an x-1 st target logistic system state vector; when each x-1 th visual engagement weight discrimination result represents that the logistic system data to be subjected to visual conversion has the visual engagement weight under the corresponding characteristic channel, determining the x-1 st original logistic system state vector as the x-1 st target logistic system state vector.

In these preferred embodiments, the process of status recognition of the x-1 th raw logistics system status vector becomes more specific and detailed. According to the x-1 th visual engagement weight discrimination result, the scheme distinguishes two different situations to process the original logistics system state vector.

First, when the x-1 th visual engagement weight discrimination result indicates that the logistic system data to be subjected to visual conversion does not have the visual engagement weight under the corresponding characteristic channel, it means that some data characteristics may not be applicable or effective in direct visualization. In this case, the solution will perform knowledge optimization on the x-1 th raw logistics system state vector. Such processing may include conversion, scaling, normalization, or even complete replacement of certain unsuitable feature values of the data to ensure that the final target logistics system state vector is more suitable for visual presentation.

For example, assume that under the inventory feature channel, one inventory value in the original logistics system state vector is unsuitable for direct visualization for some reason (e.g., outliers, unit mismatch, etc.). Through knowledge optimization processing, this value can be converted into a form more suitable for visual presentation, such as a relative value or a normalized index.

Secondly, when the x-1 th visual engagement weight discrimination results all represent that the logistic system data to be subjected to visual conversion has the visual engagement weights under the corresponding characteristic channels, the method means that all data characteristics in the state vector of the original logistic system can be directly used for visualization without additional optimization processing. In this case, the scenario would determine the x-1 th original logistics system state vector directly as the x-1 st target logistics system state vector.

The knowledge optimization processing is a data processing method and is used for optimizing or converting data characteristics which are not suitable for direct visualization in the original logistics system state vector so as to obtain a target logistics system state vector which is more suitable for visual display.

These preferred embodiments further enhance the accuracy and effectiveness of the visualization of the logistics data by introducing a finer state identification process. According to the visual engagement weight judging result, the scheme can intelligently determine whether knowledge optimization processing is needed for the state vector of the original logistics system, so that the final target logistics system state vector is ensured to be more suitable for visual display.

When certain data features are not suitable for direct visualization, knowledge optimization processing can perform necessary conversion or optimization on the features, so that misleading or inaccurate information possibly caused by direct visualization is avoided. When all the data features are suitable for direct visualization, the scheme can maintain the integrity and authenticity of the original data, and unnecessary processing steps and errors possibly introduced are avoided.

Thus, these preferred embodiments can provide more accurate, reliable visualization of logistic data, helping a decision maker to better understand and analyze the actual state and trends of the logistic system, and thus make more informed decisions.

Under some possible design ideas, in step 120, according to the state vector of the original logistics system, performing visual engagement weight discrimination on the image output feature of the logistics system data to be subjected to visual conversion in at least one feature channel to obtain visual engagement weight discrimination results of the logistics system data to be subjected to visual conversion in each feature channel, where the steps include: the visual engagement weight judging model corresponding to each characteristic channel is obtained, and the following operations are sequentially carried out for each characteristic channel: according to the original logistics system state vector, carrying out visual engagement weight discrimination on the image output characteristics of the logistics system data to be subjected to visual conversion in the characteristic channel through the corresponding visual engagement weight discrimination model to obtain the visual engagement weight coefficient of the logistics system data to be subjected to visual conversion in the characteristic channel; when the visual engagement weight coefficient is greater than or equal to a weight coefficient threshold, determining a visual engagement weight discrimination result of the characteristic channel as a first discrimination viewpoint, wherein the first discrimination viewpoint is used for representing that the visual engagement weight exists in the logistics system data to be subjected to visual conversion under the characteristic channel; and when the visual engagement weight coefficient is smaller than the weight coefficient threshold, determining a visual engagement weight judgment result of the characteristic channel as a second judgment view, wherein the second judgment view is used for representing that the visual engagement weight does not exist in the logistics system data to be subjected to visual conversion under the characteristic channel.

Under these possible design considerations, the process of determining the visual engagement weights based on the original logistics system state vector in step 120 is developed in detail. In particular, this step involves a plurality of feature channels and their respective corresponding visual engagement weight discriminant models.

Firstly, the system acquires a visual engagement weight discrimination model corresponding to each characteristic channel. These models may be pre-trained and are specifically used to evaluate whether the image output features of the logistics system data to be visually transformed have sufficient visual engagement weights under a particular feature channel.

Next, for each characteristic channel, the system will perform the following operations in sequence: and carrying out visual engagement weight discrimination on the image output characteristics of the logistics system data to be subjected to visual conversion under the characteristic channel by utilizing a corresponding visual engagement weight discrimination model and an original logistics system state vector. The discriminating process outputs a visual engagement weight coefficient reflecting the importance or contribution of the data feature to the visual representation under the feature channel.

The system then compares the visual engagement weight coefficient to a predetermined weight coefficient threshold. If the visualized engagement weight coefficient is greater than or equal to the weight coefficient threshold, the system considers that the logistic system data to be subjected to visualized conversion has enough visualized engagement weight under the characteristic channel, and therefore the visualized engagement weight judgment result of the characteristic channel is determined to be a first judgment view. Otherwise, if the visualized engagement weight coefficient is smaller than the weight coefficient threshold, the system considers that the importance of the data feature for the visualized expression is insufficient under the feature channel, and therefore the visualized engagement weight judgment result of the feature channel is determined to be a second judgment view.

The visual engagement weight judging model is specially used for evaluating whether the image output characteristics of the logistics system data to be subjected to visual conversion have enough visual engagement weights or not under the specific characteristic channel.

The visualized engagement weight coefficients reflect the coefficients of importance or contribution of the data features to the visualized expression under a particular feature channel.

The weight coefficient threshold is a preset threshold value, and is used for comparing the visualized connection weight coefficients to judge whether the logistics system data to be subjected to visualized conversion has enough visualized connection weights or not under the specific characteristic channel.

First discrimination point: and when the visual engagement weight coefficient is larger than or equal to the weight coefficient threshold, characterizing that the logistic system data to be subjected to visual conversion has a discrimination result of the visual engagement weight under a specific characteristic channel.

A second discrimination point: and when the visual engagement weight coefficient is smaller than the weight coefficient threshold, characterizing that the logistic system data to be subjected to visual conversion does not have a discrimination result of the visual engagement weight under the specific characteristic channel.

The possible design ideas lead the process of the visual engagement weight discrimination to be more accurate and flexible by introducing the concepts of the visual engagement weight discrimination model, the weight coefficient threshold and the like. The method can adaptively adjust the weight distribution of the visual expression according to the characteristics of each characteristic channel and the importance of the data characteristics, thereby improving the quality and accuracy of the visual effect.

In addition, the design ideas provide more clear guidance for subsequent data processing and visual expression by distinguishing the first discrimination viewpoint from the second discrimination viewpoint. This helps to reduce unnecessary data processing and visualization operations, improving the efficiency and stability of the overall visualization process.

In some possible examples, when the number of the feature channels is one, the obtaining the visual engagement weight discrimination model corresponding to each feature channel includes steps 121 to 123.

Step 121, acquiring an original discrimination model, and acquiring a plurality of logistics system state vector training examples corresponding to the logistics system data training examples, and visual engagement weight priori annotations of the logistics system state vector training examples.

Step 122, for each logistics system state vector training example, performing visual engagement weight discrimination on the image output features of the logistics system data training examples in the feature channel according to the logistics system state vector training examples through the original discrimination model to obtain visual engagement weight coefficients corresponding to the logistics system state vector training examples, and determining training cost variables corresponding to the logistics system state vector training examples by combining the visual engagement weight coefficients and the corresponding visual engagement weight priori annotations.

And step 123, debugging the original discrimination model according to training cost variables corresponding to the logistics system state vector training examples to obtain the visualized engagement weight discrimination model corresponding to the characteristic channel.

In some possible examples, when only one feature channel is involved, the process of obtaining the visual engagement weight discrimination model corresponding to the feature channel includes three detailed steps: step 121 to step 123.

In step 121, an original discriminant model is first obtained. This model may be constructed based on some machine learning algorithm for preliminary determination of visual engagement weights for logistic system data. Meanwhile, a plurality of logistics system state vector training examples corresponding to the logistics system data training examples and visual engagement weight priori annotations of each training example are required to be obtained. These prior annotations are noted in advance by an expert or according to some criteria, for guiding the training of the model.

In step 122, for each logistic system state vector training example, the original discrimination model is used to perform visual engagement weight discrimination on the image output features of the logistic system data training example on the feature channel. This process outputs a visual engagement weight coefficient that reflects the importance of the data feature to the visual representation under the feature channel. And then, combining the visual engagement weight coefficient with the corresponding visual engagement weight priori annotation to determine the training cost variable corresponding to the logistics system state vector training example. This cost variable is used to quantify the difference between the model discrimination results and the prior annotation.

In step 123, the original discriminant model is debugged according to training cost variables corresponding to all the logistic system state vector training examples. This process may include adjusting parameters of the model, optimizing the structure of the model, or employing other tuning techniques to reduce the difference between the model discrimination results and the prior annotations. Finally, a debugged visual engagement weight judging model suitable for the characteristic channel is obtained.

The original judging model is an initial model which is not debugged in the training process and is used for primarily judging the visual connection weight of the logistics system data.

The logistics system state vector training examples are used for training logistics system data samples of the visual engagement weight discrimination model, and each sample comprises an example of a logistics system state vector.

The visual engagement weight priori annotation is annotation information of visual engagement weights of logistics system data, which are annotated in advance by an expert or according to a certain standard and are used for guiding model training.

The training cost variable is a variable for quantifying the difference between the model discrimination result and the prior annotation, and is usually calculated through a certain loss function.

These possible examples provide a specific and operable method for obtaining the visual engagement weight discrimination model corresponding to the feature channel by introducing detailed model training steps and related technical terms. The method combines the characteristics of logistics system data and the requirements of visual expression, and improves the accuracy and reliability of visual engagement weight discrimination by training and adjusting model parameters.

Meanwhile, the method fully utilizes priori annotation information, integrates expert knowledge and experience into model training, and further enhances the discrimination capability and generalization performance of the model. The method has important significance for improving the visual quality of the logistics data and promoting the intellectualization and the high efficiency of the logistics system.

Under some preferred design considerations, the acquiring a plurality of physical distribution system state vector training examples corresponding to the physical distribution system data training examples in step 121 includes: acquiring a logistics system data training example, and carrying out state knowledge mining on the logistics system data training example to obtain an original logistics system state vector of the logistics system data training example; and carrying out vector segmentation on the original logistics system state vector of the logistics system data training example to obtain a plurality of logistics system state vector training examples corresponding to the logistics system data training example.

Under some preferred design considerations, the process of obtaining a plurality of logistic system state vector training examples corresponding to the logistic system data training examples in step 121 is subdivided into two sub-steps.

First, the system will obtain a logistic system data training example. These training examples may be extracted from historical logistics data, including data in various logistics scenarios. The system then performs state knowledge mining on these logistic system data training examples. This process may involve techniques such as data analysis, pattern recognition, etc., in order to extract from the raw data characteristic information that reflects the state of the logistics system. Through the mining, the system can obtain the original logistics system state vector of the logistics system data training example.

Secondly, the system performs vector segmentation on the obtained original logistics system state vector. This process may be based on some predefined rule or algorithm in order to split the original state vector into a plurality of sub-vectors, each sub-vector containing a portion of the information in the original vector. In this way, the system can obtain a plurality of logistics system state vector training examples corresponding to the logistics system data training examples.

State knowledge mining is a process of extracting information capable of reflecting characteristics of a system state from original data. In the field of logistics, state knowledge mining may involve analysis and processing of logistics data to extract features that reflect the operational state of a logistics system.

Vector partitioning is a process of splitting an original vector into a plurality of sub-vectors. In the processing of state vectors for logistic systems, vector partitioning may be based on a rule or algorithm that aims to split an original state vector containing multidimensional information into a plurality of sub-vectors containing partial information.

The optimal design thought enables the obtained physical distribution system state vector training examples to be more in line with the running state of an actual physical distribution system by introducing processing technologies such as state knowledge mining and vector segmentation, and the diversity and representativeness of the training examples are improved. This helps to promote the training effect and discrimination accuracy of the subsequent visual engagement weight discrimination model.

Meanwhile, by carrying out vector segmentation on the original logistics system state vector, a plurality of sub-vectors containing partial information can be obtained. The sub-vectors can reflect the running state of the logistics system from different angles, and provide richer information input for subsequent model training and discrimination. The method is favorable for improving the perception and understanding capability of the model to the state of the complex logistics system, and further improving the visual engagement weight discrimination performance of the model.

In other possible embodiments, when the number of the feature channels is plural, the obtaining the visual engagement weight discrimination model corresponding to each feature channel includes: acquiring an original discrimination model, and acquiring a1 st logistics system state vector training example corresponding to a logistics system data training example of a1 st characteristic channel, and a1 st visual engagement weight priori annotation of the 1 st logistics system state vector training example; performing visual engagement weight discrimination on the image output characteristics of the logistics system data training examples of the 1 st characteristic channel according to the 1 st logistics system state vector training examples through the original discrimination model to obtain a1 st visual engagement weight coefficient, and debugging the original discrimination model by combining the 1 st visual engagement weight coefficient and the 1 st visual engagement weight priori annotation to obtain a visual engagement weight discrimination model corresponding to the 1 st characteristic channel; a self-add 1 loop is performed on y to achieve the following: acquiring a y-1 th visual engagement weight coefficient corresponding to a logistics system data training example of a y-1 th characteristic channel, and debugging the original judgment model according to the y-1 th visual engagement weight coefficient to obtain a visual engagement weight judgment model corresponding to the y characteristic channel; wherein y is greater than or equal to 2 and less than or equal to M, and M is used for representing the number of the characteristic channels.

In other possible embodiments, when the number of the feature channels is multiple, the process of obtaining the visual engagement weight discrimination model corresponding to each feature channel respectively adopts an iterative method. The specific steps are as follows.

First, a first logistics system state vector training example corresponding to the logistics system data training example of the original discrimination model and the first characteristic channel is obtained, and a first visual engagement weight priori annotation of the training example is obtained. The method is an initial step of an iterative process, and lays a foundation for the acquisition of a subsequent characteristic channel model.

And then, carrying out visual engagement weight judgment on the image output characteristics of the logistics system data training examples of the first characteristic channel according to the first logistics system state vector training examples through the original judgment model. This process outputs a first visual engagement weight coefficient that reflects the importance of the data feature to the visual representation under the feature channel.

And then, debugging the original discriminant model by combining the first visual engagement weight coefficient and the first visual engagement weight priori annotation. The purpose of debugging is to make the model adapt to the data characteristic of the current characteristic channel better and improve the accuracy of discrimination. And after debugging, obtaining a visual engagement weight judging model corresponding to the first characteristic channel.

Thereafter, an iterative loop process is entered. Performing a self-adding 1 operation on the variable y to achieve the following: and acquiring a y-1 visual engagement weight coefficient corresponding to the logistics system data training example of the y-1 characteristic channel, and debugging the original discrimination model according to the weight coefficient to obtain a visual engagement weight discrimination model corresponding to the y characteristic channel. This process is repeated until y is equal to the number M of characteristic channels.

Among them, the iterative method is a method for solving a problem by repeatedly performing a series of operation steps. In this technical solution, an iterative method is used to sequentially obtain the visual engagement weight discrimination model of each feature channel.

The possible embodiments realize the sequential acquisition of the visual engagement weight discrimination model of a plurality of characteristic channels by introducing technical means such as an iteration method, a self-adding 1 loop and the like. The method not only improves the efficiency and accuracy of model acquisition, but also enables each characteristic channel to be processed by a targeted model, thereby improving the flexibility and adaptability of the whole visualization process.

Meanwhile, the original discrimination model is continuously debugged, so that the visualized connection weight discrimination model corresponding to each characteristic channel can be better adapted to the data characteristics of the current characteristic channel. The method is favorable for improving the accuracy and reliability of the judgment of the visual engagement weight, and provides a firmer foundation for the subsequent data processing and visual expression.

In an alternative embodiment, the debugging the original discriminant model according to the y-1 th visualized engagement weight coefficient to obtain a visualized engagement weight discriminant model corresponding to the y-th feature channel includes: acquiring a y-th logistics system state vector training example corresponding to a logistics system data training example of a y-th characteristic channel, and a y-th visual engagement weight priori annotation of the y-th logistics system state vector training example; carrying out visual engagement weight discrimination on the image output characteristics of the logistics system data training examples of the y-th characteristic channel according to the y-th logistics system state vector training examples through the original discrimination model to obtain a y-th visual engagement weight coefficient; combining the y-th visual engagement weight coefficient and the y-1-th visual engagement weight coefficient to determine a first training cost variable, and combining the y-th visual engagement weight coefficient and the y-th visual engagement weight priori annotation to determine a second training cost variable; and debugging the original discrimination model by combining the first training cost variable and the second training cost variable to obtain the visualized engagement weight discrimination model corresponding to the y-th characteristic channel.

In an alternative embodiment, the process of debugging the original discriminant model to obtain the visual engagement weight discriminant model corresponding to the y-th feature channel is further refined. The specific steps are as follows.

Firstly, a y-th logistics system state vector training example corresponding to a logistics system data training example of a y-th characteristic channel is obtained, and a y-th visual engagement weight priori annotation of the training example is obtained. This information will be used for subsequent visual engagement weight discrimination and model debugging.

And then, carrying out visual engagement weight judgment on the image output characteristics of the logistics system data training examples of the y-th characteristic channel according to the y-th logistics system state vector training examples through the original judgment model. This process will output a y-th visual engagement weight coefficient reflecting the importance of the data feature to the visual representation under the current feature channel.

Then, a first training cost variable is determined by combining the y-th visual engagement weight coefficient and the y-1 th visual engagement weight coefficient. The training cost variable is used for quantifying the change of the visual linking weight coefficient between adjacent characteristic channels, and is helpful for maintaining the consistency of weight discrimination between different characteristic channels.

And simultaneously, determining a second training cost variable by combining the y-th visual engagement weight coefficient and the y-th visual engagement weight priori annotation. The training cost variable is used for quantifying the difference between the model discrimination result and the priori annotation, and is helpful for guiding the model to debug towards the direction more in line with the actual demand.

And finally, debugging the original discrimination model by combining the first training cost variable and the second training cost variable. The purpose of debugging is to enable the model to better meet the actual visual engagement weight requirement while maintaining the weight discrimination continuity between different characteristic channels. And after debugging, obtaining a visual engagement weight judging model corresponding to the y-th characteristic channel.

The first training cost variable is used for quantifying a variable of the change of the visual engagement weight coefficient between adjacent characteristic channels. The method combines the visualized connection weight coefficients of the current characteristic channel and the previous characteristic channel, and is helpful for maintaining the consistency of weight discrimination between different characteristic channels.

The second training cost variable is used for quantifying a variable that is a difference between the model discrimination result and the prior annotation. The visual engagement weight coefficient of the current characteristic channel and the corresponding priori annotation are combined, so that the model is guided to be debugged in a direction more in line with actual requirements.

According to the alternative embodiment, the first training cost variable and the second training cost variable are introduced, so that the model can simultaneously consider the consistency of weight discrimination between adjacent characteristic channels and the accuracy meeting the actual visual connection weight requirement in the debugging process. The method is favorable for improving the perception and understanding capability of the model to the state of the complex logistics system, and further improving the visual engagement weight discrimination performance of the model.

Meanwhile, through multiple times of debugging on the original discrimination model, each characteristic channel can be subjected to targeted model processing, so that the flexibility and adaptability of the whole visual flow are improved. The method is favorable for improving the quality and efficiency of the visualization of the logistics data and promoting the intellectualization and the high efficiency of the logistics system.

In other preferred embodiments, the method further includes, according to the original state vector of the logistics system, performing a visual engagement weight discrimination on an image output feature of the logistics system data to be subjected to visual conversion in at least one feature channel, and after obtaining a visual engagement weight discrimination result of the logistics system data to be subjected to visual conversion in each feature channel, respectively: when the visualized connection weight discrimination results of the characteristic channels represent that the visualized connection weights exist in the logistics system data to be subjected to the visualized conversion under the corresponding characteristic channels, the original logistics system state vector is subjected to visualized mapping, and the target visualized result corresponding to the logistics system data to be subjected to the visualized conversion is obtained.

In other preferred embodiments, when the image output feature of the logistic system data to be subjected to the visual conversion is subjected to the visual engagement weight discrimination in at least one feature channel according to the original logistic system state vector, and the visual engagement weight discrimination results of the logistic system data to be subjected to the visual conversion in each feature channel are obtained, the method further comprises a subsequent processing step.

Specifically, this subsequent processing step is: first, the visual connection weight discrimination result of each characteristic channel is checked. When the discrimination results represent that the logistics system data to be subjected to visual conversion has visual connection weights under the corresponding characteristic channels, the logistics system data has important visual information under each characteristic channel.

Next, the method will perform visual mapping on the original logistics system state vector. This process converts the information in the state vector into a visual form to more intuitively reveal the state of the logistics system. Through the mapping, a target visualization result corresponding to the logistics system data to be subjected to visualization conversion can be obtained.

Among other things, visualization mapping is a technique that converts state vectors or other forms of data into a visual form. In this process, the information in the data is converted into graphics, colors, or other visual elements so that one can more intuitively understand and analyze the data.

These preferred embodiments further enhance the effect of visualization of the logistics data by introducing the processing step of visualization mapping. When the logistics system data to be subjected to visual conversion are judged to have the visual linking weights under each characteristic channel, the important information of the data can be fully displayed in the visual result by performing visual mapping on the logistics system data.

The method not only improves the integrity and accuracy of the visualization of the logistics data, but also helps users to more comprehensively know the state of the logistics system. Meanwhile, by mapping the state vector into a visual form, a user can more intuitively perceive the dynamic change of the logistics system, so that a more timely and accurate decision is made. This has important significance for improving the operation efficiency and response speed of the logistics system.

In some examples, the knowledge optimization instruction vector corresponds to the target feature channel one by one, and the logistic system data to be subjected to the visual conversion does not have the visual engagement weight under the target feature channel, and the knowledge optimization processing is performed on the original logistic system state vector according to the knowledge optimization instruction vector described in step 140 to obtain a target logistic system state vector corresponding to the original logistic system state vector, where the method includes: obtaining the visual engagement weight coefficients of the logistics system data to be subjected to visual conversion under each target characteristic channel, and respectively determining the visual engagement weight coefficients as the confidence degrees of the corresponding knowledge optimization indication vectors; carrying out characteristic reinforcement on each knowledge optimization indicating vector according to the confidence coefficient of each knowledge optimization indicating vector to obtain the fused knowledge optimization indicating vector; and carrying out knowledge optimization processing on the original logistics system state vector according to the fused knowledge optimization indication vector to obtain a target logistics system state vector corresponding to the original logistics system state vector.

In some examples, the knowledge optimization process described in step 140 is developed in detail when the knowledge optimization designation vectors are in one-to-one correspondence with the target feature channels and the logistic system data to be visually transformed does not have a visual engagement weight under the target feature channels. This process aims to optimize the raw logistics system state vector to better accommodate visualization needs.

Firstly, obtaining the visualized connection weight coefficients of the logistics system data to be subjected to visualized conversion under each target characteristic channel. These weighting coefficients reflect the importance or influence of the data under each characteristic channel.

These visual engagement weight coefficients are then determined as the confidence levels of the corresponding knowledge optimization indicator vectors, respectively. The confidence reflects the reliability and accuracy of the knowledge optimization indicating vector, and provides a basis for subsequent feature reinforcement.

And then, carrying out characteristic reinforcement on the knowledge optimization indicating vectors according to the confidence level of the knowledge optimization indicating vectors. The purpose of feature reinforcement is to highlight important features in the vector and suppress unimportant features, so that a fusion knowledge optimization indicating vector is obtained. The fusion vector integrates information of a plurality of knowledge optimization indication vectors, and has stronger expression capability and generalization capability.

And finally, carrying out knowledge optimization processing on the state vector of the original logistics system according to the fused knowledge optimization indication vector. The structure and parameters of the original vector are adjusted through the guidance of the fusion vector, so that the structure and parameters are better adapted to the visualization requirements. And obtaining a target logistics system state vector corresponding to the original logistics system state vector after the optimization treatment.

Wherein the knowledge optimization indicating vector is a vector for guiding the optimization of the state vector of the logistics system. It contains knowledge and experience related to logistics systems and can instruct how to adjust the state vector to better meet the visualization needs.

Confidence is used to quantify knowledge to optimize an indicator that indicates the reliability and accuracy of the vector. It reflects the importance and influence of the weight coefficients in the knowledge optimization process.

Feature enhancement is a method of optimizing a vector by adjusting the importance and prominence of features in the vector. The method can make the important characteristics in the vector more remarkable, thereby improving the expression capability and generalization capability of the vector.

The knowledge optimization processing of the state vector of the original logistics system is realized by introducing technical means such as knowledge optimization indication vector, confidence level, feature reinforcement and the like. The method not only improves the adaptability of the state vector to the visual requirement, but also enhances the expression capability and generalization capability of the vector.

Meanwhile, by fusing information of a plurality of knowledge optimization indicating vectors, a fused knowledge optimization indicating vector with stronger expression capability is obtained. The fusion vector can more comprehensively reflect the state and the characteristics of the logistics system, and provides a more reliable data basis for subsequent visual processing.

In addition, the knowledge optimization processing is carried out on the original logistics system state vector, so that a target logistics system state vector which meets the visual requirement better is obtained. The target vector can more accurately reflect the actual state and the operation rule of the logistics system, thereby improving the accuracy and the reliability of the visualization of the logistics data.

In some optional embodiments, the performing knowledge optimization processing on the original logistics system state vector according to the fused knowledge optimization instruction vector to obtain a target logistics system state vector of the original logistics system state vector includes: acquiring the vector size of the state vector of the original logistics system and the vector size of the fusion knowledge optimization indication vector; when the vector size of the original logistics system state vector is different from the vector size of the fusion knowledge optimization indicating vector, updating the vector size of the fusion knowledge optimization indicating vector to obtain a target knowledge optimization indicating vector; when the vector size of the original logistics system state vector is the same as the vector size of the fused knowledge optimization indicating vector, determining the fused knowledge optimization indicating vector as the target knowledge optimization indicating vector; determining the optimization coefficient of the state vector of the original logistics system according to the number of the knowledge optimization indicating vectors, wherein the optimization coefficient and the number of the knowledge optimization indicating vectors have a set quantization relation; and determining a linkage involvement vector based on the product of the optimization coefficient and the target knowledge optimization indication vector, and summing the original logistics system state vector and the linkage involvement vector to obtain the target logistics system state vector.

In some alternative embodiments, the process of knowledge optimization processing of the original logistics system state vector in accordance with the fused knowledge optimization designation vector is further refined. The process aims at ensuring that the optimized logistics system state vector is matched with the target knowledge optimization indicating vector in size and structure, so that the optimization effect is improved.

Firstly, the vector size of the original logistics system state vector is obtained, and the vector size of the knowledge optimization instruction vector is fused. These two dimensions reflect the complexity of the original logistics system state and the target optimization designation in the feature dimension, respectively.

Then, it is determined whether the two vectors are the same size. If the vector sizes of the fusion knowledge optimization indication vectors are different, the vector sizes of the fusion knowledge optimization indication vectors need to be updated to match the sizes of the original logistics system state vectors. This updating process may include adding new feature dimensions, deleting redundant feature dimensions, or readjusting the order of feature dimensions, etc., to arrive at a target knowledge optimization designation vector.

When the vector size of the original logistics system state vector is the same as the vector size of the fused knowledge optimization indicating vector, the fused knowledge optimization indicating vector is directly determined to be the target knowledge optimization indicating vector, and the size updating is not needed.

And then, determining the optimization coefficient of the state vector of the original logistics system according to the number of the knowledge optimization indicating vectors. The optimization coefficient has a set quantization relation with the number of knowledge optimization indicating vectors, and can be a linear relation, an exponential relation or other complex functional relation. The determination of the optimization coefficients aims at quantifying the contribution degree of different amounts of knowledge optimization indicating vectors to the state vector optimization of the logistics system.

Finally, a linkage involvement vector is determined based on the product of the optimization coefficient and the target knowledge optimization indication vector. This linkage involvement vector reflects the direction and strength of optimization of the original logistics system state vector by the target knowledge optimization indication vector. And summing the original logistics system state vector and the linkage involving vector to obtain the target logistics system state vector. The target vector integrates the state of the original logistics system and information of target optimization indication, and can better meet the requirement of logistics system visualization.

The vector dimension refers to the size of the vector in the feature dimension, i.e. the number of features contained in the vector.

The optimization coefficients are used to quantify different amounts of knowledge optimization coefficients indicating the degree of contribution of the vector to the logistic system state vector optimization.

The linkage involvement vector reflects the vector of the target knowledge optimization indication vector to the optimization direction and strength of the original logistics system state vector.

The optional embodiments realize the fine knowledge optimization processing of the state vector of the original logistics system by introducing technical means such as vector size matching, optimization coefficient determination, linkage involved vector calculation and the like. The method ensures that the optimized logistics system state vector is matched with the target knowledge optimization indication vector in size and structure, and improves the accuracy and reliability of the optimization effect.

Meanwhile, the optimization coefficient is determined according to the number of the knowledge optimization indicating vectors, so that the contribution degree of the knowledge optimization indicating vectors with different numbers to the optimization is quantized, and the optimization process is more scientific and reasonable. In addition, the introduction of the linkage involving vector further enhances the optimization effect, so that the state vector of the target logistics system can better meet the requirement of the logistics system for visualization.

In an alternative embodiment, the step 150 of performing visualization mapping on the target logistics system state vector to obtain a target visualization result corresponding to the logistics system data to be subjected to visualization transformation includes steps 151-153.

Step 151, obtaining the display demand type of the logistics system data to be subjected to visual conversion, and obtaining a display demand discrimination model corresponding to the display demand type.

Step 152, when the display requirement type is a state trend deduction display requirement for outputting the logistics system data to be subjected to visual conversion, performing state trend deduction on the logistics system data to be subjected to visual conversion according to the target logistics system state vector by using a display requirement discrimination model corresponding to the state trend deduction display requirement, so as to obtain a state trend deduction image corresponding to the logistics system data to be subjected to visual conversion, wherein the state trend deduction image has the visual connection weight under each characteristic channel.

Step 153, when the display requirement type is a service scheduling display requirement for implementing service scheduling processing of the logistics system data to be subjected to visual conversion, performing service scheduling simulation on the logistics system data to be subjected to visual conversion according to the target logistics system state vector by using a display requirement discrimination model corresponding to the service scheduling display requirement, so as to obtain a service scheduling line simulation image corresponding to the logistics system data to be subjected to visual conversion, wherein the service scheduling line simulation image has the visual connection weight under each characteristic channel.

In an alternative embodiment, the process of performing the visual mapping on the target logistics system state vector described in step 150 is further refined into steps 151 to 153, and the steps generate corresponding visual results according to the display requirement types of the logistics system data to be subjected to the visual conversion.

In step 151, the display requirement type of the logistics system data to be visually converted is first obtained. The display requirement types may include state trend deduction display requirement, business scheduling display requirement and the like. Aiming at each display demand type, a corresponding display demand discrimination model is provided. These models are designed to identify and handle specific types of presentation requirements.

In step 152, the presentation requirements are deduced for status trends. When the display demand type is determined to be the state trend deduction, the corresponding display demand discrimination model is used for processing the state vector of the target logistics system. The model can carry out state trend deduction on the logistics system data to be subjected to visual conversion according to the state vector of the target logistics system. The result of the deduction is a state trend deduction image, and the image has visual connection weights under each characteristic channel, which means that the information in the image is effectively expressed and connected at each visual level.

In step 153, the requirements are illustrated for the traffic scheduling presentation. When the display demand type is service scheduling, the corresponding display demand discrimination model is utilized to process the state vector of the target logistics system. The model can simulate the business scheduling process of the logistics system data to be subjected to visual conversion according to the target logistics system state vector. The result of the simulation is a traffic dispatch line simulation image, which likewise has visual engagement weights under each characteristic channel.

The display requirement type refers to different types of information or function requirements, such as state trend deduction, service scheduling and the like, which need to be displayed by the logistics system data to be subjected to visual conversion.

The display demand discrimination model is a model designed for a specific display demand type and is used for processing and analyzing the state vector of the target logistics system to generate a visual result conforming to the display demand type.

This alternative embodiment further refines the visual mapping process for the target logistics system state vector by introducing concepts of the display demand category and the display demand discrimination model. According to different display demand types, targeted visual results such as state trend deduction images, business scheduling line simulation images and the like can be generated.

The visualized results have visualized connection weights under all characteristic channels, so that the integrity and continuity of information are guaranteed, the state trend and the service scheduling condition of the logistics system can be intuitively displayed, and a user can more accurately understand the running state of the logistics system and plan a scheduling scheme. In addition, the accuracy and efficiency of the visual mapping are improved by using a special display demand discrimination model to process the state vector of the target logistics system.

Fig. 2 shows a block diagram of a data visualization system 300, comprising: memory 310 for storing program instructions and data; a processor 320, coupled to the memory 310, executes instructions in the memory 310 to implement the methods described above.

Further, a computer storage medium is provided containing instructions which, when executed on a processor, implement the above-described method.

In the embodiments provided in the present application, it should be understood that the disclosed apparatus and method may be implemented in other manners. The apparatus and method embodiments described above are merely illustrative, for example, flow diagrams and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of apparatus, methods and computer program products according to various embodiments of the present application. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and/or flowchart illustration, and combinations of blocks in the block diagrams and/or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.

In addition, functional modules in the embodiments of the present application may be integrated together to form a single part, or each module may exist alone, or two or more modules may be integrated to form a single part.

The functions, if implemented in the form of software functional modules and sold or used as a stand-alone product, may be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application may be embodied essentially or in a part contributing to the prior art or in a part of the technical solution in the form of a software product stored in a storage medium, comprising several instructions for causing a computer device (which may be a personal computer, a network device, or the like) to perform all or part of the steps of the method according to the embodiments of the present application. And the aforementioned storage medium includes: a usb disk, a removable hard disk, a read-only memory (ROM), a random access memory (RAM, random Access Memory), a magnetic disk, or an optical disk, or other various media capable of storing program codes. It should be noted that, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element.

The above description is only of the preferred embodiments of the present application and is not intended to limit the present application, but various modifications and variations can be made to the present application by those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the protection scope of the present application.

Claims

1. A data visualization method applied to a smart logistics, characterized in that the method is applied to a data visualization system, and the method comprises:

2. The method according to claim 1, wherein the state knowledge mining is implemented by at least one state knowledge mining model, and when the number of state knowledge mining models is plural, the performing state knowledge mining on the logistic system data to be subjected to the visual conversion to obtain an original logistic system state vector of the logistic system data to be subjected to the visual conversion, including:

Carrying out state knowledge mining on the logistics system data to be subjected to visual conversion through a1 st state knowledge mining model to obtain a1 st original logistics system state vector;

a self-add 1 loop is performed on x to achieve the following:

Carrying out state knowledge mining on the logistics system data to be subjected to visual conversion according to the x-1 original logistics system state vector through an x-th state knowledge mining model to obtain the x-th original logistics system state vector;

Wherein x is 1 or more and N or less;

the method further comprises the steps of: according to the x-1 th original logistics system state vector, carrying out visual engagement weight discrimination on the image output characteristics of the logistics system data to be subjected to visual conversion in each characteristic channel to obtain x-1 th visual engagement weight discrimination results of the logistics system data to be subjected to visual conversion in each characteristic channel respectively; carrying out state identification on the x-1 original logistics system state vector according to the x-1 visualized engagement weight discrimination result to obtain an x-1 target logistics system state vector; the state knowledge mining is carried out on the logistics system data to be subjected to visual conversion according to the x-1 original logistics system state vector through the x-th state knowledge mining model to obtain the x original logistics system state vector, and the method comprises the following steps: carrying out state knowledge mining on the logistics system data to be subjected to visual conversion according to the (x-1) th target logistics system state vector through an (x) th state knowledge mining model to obtain the (x) th original logistics system state vector;

The step of carrying out state identification on the x-1 original logistics system state vector according to the x-1 visualized engagement weight discrimination result to obtain the x-1 target logistics system state vector comprises the following steps:

When the x-1 th visual engagement weight discrimination result indicates that the logistic system data to be subjected to visual conversion does not have the visual engagement weight under the corresponding characteristic channel, carrying out knowledge optimization processing on the x-1 th original logistic system state vector to obtain an x-1 st target logistic system state vector;

when each x-1 th visual engagement weight discrimination result represents that the logistic system data to be subjected to visual conversion has the visual engagement weight under the corresponding characteristic channel, determining the x-1 st original logistic system state vector as the x-1 st target logistic system state vector.

3. The method according to claim 1, wherein the step of performing a visual engagement weight discrimination on the image output feature of the logistic system data to be subjected to the visual conversion in at least one feature channel according to the original logistic system state vector to obtain a visual engagement weight discrimination result of the logistic system data to be subjected to the visual conversion in each feature channel, respectively, includes:

the visual engagement weight judging model corresponding to each characteristic channel is obtained, and the following operations are sequentially carried out for each characteristic channel:

According to the original logistics system state vector, carrying out visual engagement weight discrimination on the image output characteristics of the logistics system data to be subjected to visual conversion in the characteristic channel through the corresponding visual engagement weight discrimination model to obtain the visual engagement weight coefficient of the logistics system data to be subjected to visual conversion in the characteristic channel;

When the visual engagement weight coefficient is greater than or equal to a weight coefficient threshold, determining a visual engagement weight discrimination result of the characteristic channel as a first discrimination viewpoint, wherein the first discrimination viewpoint is used for representing that the visual engagement weight exists in the logistics system data to be subjected to visual conversion under the characteristic channel;

and when the visual engagement weight coefficient is smaller than the weight coefficient threshold, determining a visual engagement weight judgment result of the characteristic channel as a second judgment view, wherein the second judgment view is used for representing that the visual engagement weight does not exist in the logistics system data to be subjected to visual conversion under the characteristic channel.

4. The method of claim 3, wherein when the number of the feature channels is one, the obtaining the visual engagement weight discrimination model corresponding to each feature channel includes:

acquiring an original discrimination model, and acquiring a plurality of logistics system state vector training examples corresponding to the logistics system data training examples, and visual engagement weight priori annotations of the logistics system state vector training examples;

Aiming at each logistics system state vector training example, carrying out visual engagement weight discrimination on the image output characteristics of the logistics system data training examples in the characteristic channel according to the logistics system state vector training examples through the original discrimination model to obtain visual engagement weight coefficients corresponding to the logistics system state vector training examples, and determining training cost variables corresponding to the logistics system state vector training examples by combining the visual engagement weight coefficients and the corresponding visual engagement weight priori annotations;

Debugging the original discrimination model according to training cost variables corresponding to the state vector training examples of the logistics system to obtain a visualized engagement weight discrimination model corresponding to the characteristic channel;

The obtaining a plurality of logistics system state vector training examples corresponding to the logistics system data training examples comprises the following steps:

acquiring a logistics system data training example, and carrying out state knowledge mining on the logistics system data training example to obtain an original logistics system state vector of the logistics system data training example;

And carrying out vector segmentation on the original logistics system state vector of the logistics system data training example to obtain a plurality of logistics system state vector training examples corresponding to the logistics system data training example.

5. The method of claim 3, wherein when the number of the feature channels is plural, the obtaining the visual engagement weight discrimination model corresponding to each of the feature channels includes:

Acquiring an original discrimination model, and acquiring a 1 st logistics system state vector training example corresponding to a logistics system data training example of a 1 st characteristic channel, and a 1 st visual engagement weight priori annotation of the 1 st logistics system state vector training example;

Performing visual engagement weight discrimination on the image output characteristics of the logistics system data training examples of the 1 st characteristic channel according to the 1 st logistics system state vector training examples through the original discrimination model to obtain a 1 st visual engagement weight coefficient, and debugging the original discrimination model by combining the 1 st visual engagement weight coefficient and the 1 st visual engagement weight priori annotation to obtain a visual engagement weight discrimination model corresponding to the 1 st characteristic channel;

a self-add 1 loop is performed on y to achieve the following:

Acquiring a y-1 th visual engagement weight coefficient corresponding to a logistics system data training example of a y-1 th characteristic channel, and debugging the original judgment model according to the y-1 th visual engagement weight coefficient to obtain a visual engagement weight judgment model corresponding to the y characteristic channel;

wherein y is greater than or equal to 2 and less than or equal to M, and M is used for representing the number of the characteristic channels;

The debugging is carried out on the original discrimination model according to the y-1 th visualized engagement weight coefficient to obtain the visualized engagement weight discrimination model corresponding to the y-th characteristic channel, which comprises the following steps:

acquiring a y-th logistics system state vector training example corresponding to a logistics system data training example of a y-th characteristic channel, and a y-th visual engagement weight priori annotation of the y-th logistics system state vector training example;

Carrying out visual engagement weight discrimination on the image output characteristics of the logistics system data training examples of the y-th characteristic channel according to the y-th logistics system state vector training examples through the original discrimination model to obtain a y-th visual engagement weight coefficient;

Combining the y-th visual engagement weight coefficient and the y-1-th visual engagement weight coefficient to determine a first training cost variable, and combining the y-th visual engagement weight coefficient and the y-th visual engagement weight priori annotation to determine a second training cost variable;

And debugging the original discrimination model by combining the first training cost variable and the second training cost variable to obtain the visualized engagement weight discrimination model corresponding to the y-th characteristic channel.

6. The method according to claim 1, wherein the step of performing a visual engagement weight discrimination on the image output feature of the logistic system data to be subjected to the visual conversion in at least one feature channel according to the original logistic system state vector, and obtaining a visual engagement weight discrimination result of the logistic system data to be subjected to the visual conversion in each feature channel, the method further comprises: when the visualized connection weight discrimination results of the characteristic channels represent that the visualized connection weights exist in the logistics system data to be subjected to the visualized conversion under the corresponding characteristic channels, the original logistics system state vector is subjected to visualized mapping, and the target visualized result corresponding to the logistics system data to be subjected to the visualized conversion is obtained.

7. The method according to claim 1, wherein the knowledge optimization instruction vector corresponds to a target feature channel one by one, the logistic system data to be subjected to the visual conversion does not have the visual engagement weight under the target feature channel, and the knowledge optimization processing is performed on the original logistic system state vector according to the knowledge optimization instruction vector to obtain a target logistic system state vector corresponding to the original logistic system state vector, including:

Obtaining the visual engagement weight coefficients of the logistics system data to be subjected to visual conversion under each target characteristic channel, and respectively determining the visual engagement weight coefficients as the confidence degrees of the corresponding knowledge optimization indication vectors;

Carrying out feature reinforcement on each knowledge optimization indicating vector according to the confidence coefficient of each knowledge optimization indicating vector to obtain a fused knowledge optimization indicating vector;

Carrying out knowledge optimization processing on the original logistics system state vector according to the fused knowledge optimization indication vector to obtain a target logistics system state vector corresponding to the original logistics system state vector;

The knowledge optimization processing is performed on the original logistics system state vector according to the fused knowledge optimization instruction vector to obtain a target logistics system state vector of the original logistics system state vector, which comprises the following steps:

acquiring the vector size of the state vector of the original logistics system and the vector size of the fusion knowledge optimization indication vector;

When the vector size of the original logistics system state vector is different from the vector size of the fusion knowledge optimization indicating vector, updating the vector size of the fusion knowledge optimization indicating vector to obtain a target knowledge optimization indicating vector;

When the vector size of the original logistics system state vector is the same as the vector size of the fused knowledge optimization indicating vector, determining the fused knowledge optimization indicating vector as the target knowledge optimization indicating vector;

determining the optimization coefficient of the state vector of the original logistics system according to the number of the knowledge optimization indicating vectors, wherein the optimization coefficient and the number of the knowledge optimization indicating vectors have a set quantization relation;

And determining a linkage involvement vector based on the product of the optimization coefficient and the target knowledge optimization indication vector, and summing the original logistics system state vector and the linkage involvement vector to obtain the target logistics system state vector.

8. The method of claim 1, wherein the performing the visual mapping on the target logistics system state vector to obtain the target visual result corresponding to the logistics system data to be subjected to the visual conversion includes:

Acquiring the display demand type of the logistics system data to be subjected to visual conversion, and acquiring a display demand discrimination model corresponding to the display demand type;

When the display requirement type is a state trend deduction display requirement for outputting the logistics system data to be subjected to visual conversion, carrying out state trend deduction on the logistics system data to be subjected to visual conversion according to the target logistics system state vector by using a display requirement judging model corresponding to the state trend deduction display requirement to obtain a state trend deduction image corresponding to the logistics system data to be subjected to visual conversion, wherein the state trend deduction image has the visual connection weight under each characteristic channel;

When the display requirement type is a service scheduling display requirement for realizing service scheduling processing of the logistics system data to be subjected to visual conversion, performing service scheduling simulation on the logistics system data to be subjected to visual conversion according to the target logistics system state vector by using a display requirement judging model corresponding to the service scheduling display requirement, so as to obtain a service scheduling line simulation image corresponding to the logistics system data to be subjected to visual conversion, wherein the service scheduling line simulation image has the visual connection weight under each characteristic channel.

9. A data visualization system, comprising: a memory for storing program instructions and data; a processor coupled to a memory for executing instructions in the memory to implement the method of any of claims 1-8.

10. A computer storage medium containing instructions which, when executed on a processor, implement the method of any of claims 1-8.