CN117371916B

CN117371916B - Data processing method based on digital maintenance and intelligent management system for measuring tool

Info

Publication number: CN117371916B
Application number: CN202311653751.9A
Authority: CN
Inventors: 陈志雄; 文心红; 朱慧元
Original assignee: Zhiyue Railway Equipment Co ltd
Current assignee: Zhiyue Railway Equipment Co ltd
Priority date: 2023-12-05
Filing date: 2023-12-05
Publication date: 2024-02-23
Anticipated expiration: 2043-12-05
Also published as: CN117371916A

Abstract

The method obtains a usage record tracking tag of a usage record of a to-be-analyzed tool, obtains a target preference view added corresponding to the usage record tracking tag, stores a first tool usage preference vector into a set relational database as a past tool usage preference vector in subsequent tool usage record analysis, and enables tool management to be more digital and intelligent, thereby not only improving management efficiency and reducing error rate, but also helping to find potential problems and opportunities, and further improving management efficiency. So design, this application can improve the management efficiency and the low problem of intelligent degree that traditional industry measuring tool intelligent management technique exists.

Description

Data processing method based on digital maintenance and intelligent management system for measuring tool

Technical Field

The application relates to the technical field of data analysis, in particular to a data processing method based on digital maintenance and an intelligent management system for a tool.

Background

The tooling scale is a term that generally refers to various tools and measuring equipment used in production, manufacturing, and maintenance operations. Such tools may include hand tools such as screwdrivers, hammers, wrenches, etc.; and may also include power tools such as drills, grinders, etc.; and various measuring devices such as calipers, micrometer gauges, and the like. These tools play an important role in daily production and maintenance work. To ensure that these tools are used properly and effectively, periodic inspection and maintenance is required. At the same time, because some high-precision tools and equipment are high in value, the management of the tools and equipment is also important.

The intelligent tool management technology is an advanced tool management technology, and uses modern information technology (such as cloud computing, internet of things, big data and the like) to manage and control the service condition of tools. The system mainly tracks, monitors and records the use condition of the tool, thereby improving the utilization rate of the tool, reducing the risk of losing or damaging and reducing the maintenance and replacement cost.

The intelligent management technology of the tool and the gauge mainly comprises the technical functions of tool inventory management, usage record tracking, alarm system, data analysis and the like. By utilizing the technology, the production efficiency can be improved, and the operation cost of enterprises can be reduced.

Disclosure of Invention

In order to improve the technical problems in the related art, the application provides a data processing method based on digital maintenance and an intelligent management system for the tool.

In a first aspect, an embodiment of the present application provides a data processing method based on digital maintenance, which is applied to an intelligent management system of an engineering tool, where the method includes:

mining the usage preference vector of the to-be-analyzed measuring tool usage record to obtain a first measuring tool usage preference vector;

inquiring X past work tool use preference vectors obtained in a target past use period from a set relational database; the time sequence span between the target past use period and the current use period meets the set span requirement, and X is a positive integer;

the first tool use preference vector and the X past tool use preference vectors are sorted according to the record acquisition priority of the corresponding tool use record, and a corresponding tool use preference vector queue is obtained;

identifying a target contextual preference description of the first workload usage preference vector included in the workload usage preference vector queue; wherein the target contextual preference description reflects a usage relationship between the first gauge usage preference vector and the X past gauge usage preference vectors;

And according to the target fore-and-aft preference description, obtaining a use record tracking label of the to-be-analyzed tool use record, obtaining a target preference viewpoint added corresponding to the use record tracking label, and storing the first tool use preference vector into a set relational database to serve as a past tool use preference vector in the subsequent tool use record analysis.

In some aspects, the identifying the target-precursor-and-subsequent-preference description of the first tool usage preference vector that the tool usage preference vector queue includes:

and loading the workload to a depth residual model comprising Y-level coding branches by using a preference vector queue, wherein in the u-level coding branches, u is more than or equal to 1 and less than or equal to Y, and the following steps are implemented:

responding u is 1, and based on a 1 st-stage coding branch, mining the first-stage front-and-back preference description of the first-stage preference vector for the first-stage preference vector and the X past-stage preference vectors for the first-stage preference vector;

responding to u being more than or equal to 2 and less than or equal to Y, inquiring to obtain X u-1 level past intermediate preference vectors of a u-1 level coding branch in the target past use period, and mining the u-1 level front and rear preference description generated by the u-1 level coding branch by combining the X u-1 level past intermediate preference vectors based on the u-1 level coding branch to obtain u level front and rear preference description; the X u-1 level past intermediate preference vectors are obtained by the X past workload using preference vectors through u-1 level coding branches in the Y level coding branches;

And taking the finally generated Y-level front and rear preference description as the target front and rear preference description.

In some aspects, further comprising:

storing the u-level fore-and-aft preference description into a relational database corresponding to the u-level coding branch, taking the u-level fore-and-aft preference description as a u-level intermediate preference vector in subsequent tool usage record analysis, and removing the cached fore-and-aft preference description of the u-level in the relational database corresponding to the u-level coding branch when the first tool usage record analysis, so that the number of members in the tool usage preference vector set in the relational database corresponding to the u-level coding branch is X.

In some aspects, before identifying the target-front-and-back preference description of the first tool usage preference vector included in the tool usage preference vector queue, the method further includes:

if the first tool use preference vector and the X past tool use preference vectors are characterized by double-precision variable states, changing target double-precision variables in the first tool use preference vector and the X past tool use preference vectors into quantized variable states so as to obtain original changing weights;

Based on the original change weight, carrying out first variable change processing on the first measuring tool use preference vector and the X past measuring tool use preference vectors to obtain a first measuring tool use preference vector and X past measuring tool use preference vectors of quantized variable states;

then after the obtaining the first level of the first-order contextual preference description of the first gauge usage preference vector, further comprising: and performing second variable change processing on the first-stage front-and-rear preference description of the quantized variable state according to the original change weight to obtain the first-stage front-and-rear preference description of the double-precision variable state.

In some aspects, further comprising:

in response to u being greater than or equal to 2 and less than or equal to Y, if the u-1 level front-back preference description and the X u-1 level past intermediate preference vectors are characterized by a double precision variable state, changing target double precision variables in the u-1 level front-back preference description and the X u-1 level past intermediate preference vectors to a quantization variable state to obtain u-level change weights;

based on the u-level changing weight, performing first variable changing processing on the u-1 level front and rear preference description and the X u-1 level past intermediate preference vectors of the double-precision variable state to obtain a u-1 level front and rear preference description and X u-1 level past intermediate preference vectors of the quantized variable state;

Then after the u-level contextual preference description is obtained, further comprising: and carrying out second variable change processing on the u-level front-rear preference description of the quantized variable state according to the u-level change weight to obtain the u-level front-rear preference description of the double-precision variable state.

In some aspects, before the mining of the usage preference vector for the first gauge usage preference vector and the X past gauge usage preference vectors based on the level 1 coding branch, the method further includes: according to the preference vector used by the first measuring tool and the preference vectors used by the X past measuring tools of the quantized variable states, the neural network parameters of the 1 st-stage coding branch are adjusted;

the method for mining the u-1 level front and back preference description generated by the u-1 level coding branch based on the u-1 level coding branch and combining the X u-1 level past intermediate preference vectors further comprises the following steps: and adjusting the neural network parameters of the coding branch of the u th level according to the u-1 level front and back preference description of the quantized variable state and X u-1 level past intermediate preference vectors.

In some aspects, the obtaining the usage record tracking tag of the usage record of the tool to be analyzed according to the target contextual preference description includes:

And adopting a multi-layer perceptron to perform feature conversion on the target fore-and-aft preference description to obtain a usage record tracking label of the usage record of the measuring tool to be analyzed.

In some aspects, the method is implemented by using a log analysis algorithm for a target tool, and a debugging process of the target tool using the log analysis algorithm is as follows:

based on a set algorithm debugging case set, performing a plurality of times of cyclic debugging on a to-be-debugged tool using a record analysis algorithm, wherein each algorithm debugging case comprises a tool using record case to be analyzed, X past tool using record cases and a priori using record tracking label, and the tool using record case to be analyzed and the X past tool using record cases are arranged according to record acquisition priorities;

wherein, in the course of one cycle, the following steps are implemented:

the method comprises the steps of mining use preference vectors of to-be-analyzed tool use record cases and X past tool use record cases in an algorithm debugging case respectively to obtain a predicted tool use preference vector and X tool use preference vector cases;

identifying a front-to-back preference description case of the predicted workload using preference vector in the predicted workload using preference vector and the X workload using preference vector cases;

Wherein the contextual model preference description case reflects a usage relationship between the predicted workload usage preference vector and the X workload usage preference vector cases;

obtaining a predicted usage record tracking tag according to the contextual descriptive cases, and improving the algorithm configuration variables of the tool usage record analysis algorithm based on the difference between the predicted usage record tracking tag and the prior usage record tracking tag of the one algorithm debugging case.

In a second aspect, the present application further provides an intelligent management system for an engineering tool, including a processor and a memory; the processor is in communication with the memory, and the processor is configured to read and execute a computer program from the memory to implement the method described above.

In a third aspect, the present application also provides a computer readable storage medium having stored thereon a program which, when executed by a processor, implements the method described above.

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 schematic flow chart of a data processing method based on digital maintenance according to an embodiment of the present application.

Detailed Description

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 are not representative of all implementations consistent with the present application. Rather, they are merely examples of apparatus and methods consistent with aspects of the present application.

It should be noted that the terms "first," "second," and the like in the description of the present application and the above-described figures 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 implemented in an engineering tool intelligent management system, a computer device, or a similar computing device. Taking the example of running on the tool intelligent management system, the tool intelligent management system may include one or more processors (the processor may include, but is not limited to, a microprocessor MCU or a processing device such as a programmable logic device FPGA) and a memory for storing data, and optionally, the tool intelligent management system may further include a transmission device 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 tool intelligent management system. For example, the tool intelligent management system may also include more or fewer components than those shown above, or have a different configuration than those shown above.

The memory may be used to store a computer program, for example, a software program of application software and a module, such as a computer program corresponding to a data processing method based on digital maintenance in the embodiments 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 tool intelligent management 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 tool intelligent management 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 schematic flow chart of a data processing method based on digital maintenance according to an embodiment of the present application, where the method is applied to an intelligent management system for an engineering tool, and further may include steps 110 to 150.

Step 110, mining the usage preference vector of the to-be-analyzed measuring tool usage record to obtain a first measuring tool usage preference vector.

In step 110, the usage record of the tool to be analyzed refers to the data that needs to be processed and analyzed, and these data generally include information such as the frequency of use, the user, the time of use, and the environment of use of the specific tool. For example, a worker uses a wrench every day on a specific project, and always uses the wrench in the morning, and this action forms a work tool usage record.

The usage preference vector mining is an analysis method with the goal of extracting the user's usage preferences from a large number of tool usage records. Such a preference may be that a worker prefers to use a certain brand of tool, or that he is more accustomed to using a certain tool for a certain period of time.

Finally, the first tool usage preference vector is the result of the usage preference vector mining, which represents the usage preference of the tool in the form of a vector. For example, if a worker prefers to use a wrench over a drill, then his tool usage preference vector may be [1,0], where 1 represents a wrench and 0 represents a drill.

For example, there is a log of the use of the worker's tool, and the data shows that he uses the wrench between 8 and 9 a.m. each day during the past month. By using preference vector mining, it can be derived that his tool uses preference vectors [1, 0], where 1 represents the 8 o 'clock to 9 o' clock time period in the morning, and the following 0 represents the other time periods, respectively. This means that the worker is more inclined to use the wrench in the morning from 8 to 9.

Step 120, inquiring X past work tool use preference vectors obtained in a target past use period from the set relational database; the time sequence span between the target past use period and the current use period meets the set span requirement, and X is a positive integer.

In step 120, the terms involved are explained as follows.

Relational database: this is a widely used type of database in which data is stored in the form of tables, each table consisting of rows and columns. Each row represents a record and each column represents a field. For example, the tool usage record may be stored in a table, where each row corresponds to a single tool usage, and the columns may include tool names, users, usage times, etc.

Target past use period: this refers to the period of time of the historical data that is desired to be analyzed. For example, if one wants to know what the tool most commonly used by workers in the past year, then that year is the target past use period.

Past gauges used preference vectors: this is a tool usage preference vector that has been previously derived by using log mining, reflecting tool usage preferences for a period of time in the past.

Current usage period: this refers to the period of time that the tool is currently in progress or is about to be used.

Time sequence span: this refers to the time interval between the target past usage period and the current usage period.

Setting span requirements: this is a constraint of the set timing span. For example, data over the last three months may only be of interest, and then the span requirement set is three months.

There is a task that needs to find the tool that is most suitable for use in the 8 o 'clock to 9 o' clock time period in the morning. First, all of the tool usage records within a target past usage period (e.g., the past year) are queried in a relational database to obtain past tool usage preference vectors. Then, it is checked whether the timing span meets the set span requirement (e.g., the gap between the last year and the current usage period does not exceed three months). If so, these past tools can be used to predict the tool that is best suited for the 8 o 'clock to 9 o' clock time period in the morning using the preference vector.

And 130, sorting the first tool use preference vector and the X past tool use preference vectors according to the record acquisition priority of the corresponding tool use record to obtain a corresponding tool use preference vector queue.

In step 130, the terms involved are explained as follows.

Recording acquisition priority: this refers to the priority that is given to each record when collecting and analyzing the records of the tool's use. The priority may be determined based on various factors such as the degree of freshness, relevance, credibility, etc. of the records. For example, recent records may have a higher priority than older records because they may be more reflective of current usage preferences.

The tool uses a preference vector queue: this means that all relevant tools are arranged in a certain order using the preference vector to form a queue. In this queue, each gauge usage preference vector corresponds to a particular gauge usage record, and its location (i.e., order in the queue) depends on its corresponding record acquisition priority.

Assuming that the query in step 120 resulted in 10 gauge usage preference vectors over the past year, and that there is also a new gauge usage preference vector (i.e., the first gauge usage preference vector), then there are now a total of 11 gauge usage preference vectors.

Then, it is necessary to determine a recording acquisition priority of each gauge using the preference vector. For example, the most recent record may be considered more important, thus giving the highest priority to new gauge use preference vectors and the lowest priority to gauge use preference vectors one year ago.

Finally, the 11 tool usage preference vectors are arranged into a queue in the order of recording acquisition priorities. This is the gauge usage preference vector queue, where the first bit is the new gauge usage preference vector and the last bit is the gauge usage preference vector one year ago.

Step 140, identifying a target fore-aft preference description of the first gauge usage preference vector included in the gauge usage preference vector queue.

Wherein the target contextual preference description reflects a usage relationship between the first gauge usage preference vector and the X past gauge usage preference vectors.

In step 140, the terms involved are explained as follows.

Target contextual preference description: this is an expression of the order preference for use of the tool. For example, if a worker is always using a wrench and then a drill when he is performing a task, he may be said to have a wrench-then-drill preference.

The usage condition is linked: this refers to the correlation or link between different gauges using preference vectors. For example, if it is found that all workers using the wrench will use the drill later, it can be said that there is a link in use between the wrench and the drill.

It is assumed that the tool measure obtained in step 130 uses the preference vector queue [ wrench, drill, hammer ], [ drill, wrench, hammer ], [ hammer, wrench, drill ], [ wrench, hammer, drill ]. By analysing this queue it can be found that in any case the wrench is always used prior to the drill and hammer. Thus, it can be considered that there is a fore-aft preference for a priority wrench. This is the target contextual preference description.

At the same time, it can also be seen that both electric drills and hammers are often used after wrenches. This means that there is a certain link between the use of the wrench and the use of the drill and hammer, i.e. the use situation. It is possible that when performing a task, workers are used to perform preliminary operations using a wrench and then further processing with an electric drill or hammer.

And 150, obtaining a usage record tracking tag of the usage record of the measuring tool to be analyzed according to the target fore-and-aft preference description, obtaining a target preference viewpoint added corresponding to the usage record tracking tag, and storing the usage preference vector of the first measuring tool into a set relational database to serve as a past measuring tool usage preference vector in the subsequent measuring tool usage record analysis.

In step 150, the terms involved are explained as follows.

Usage record tracking tag: this is a label for marking a tool usage record. For example, if a worker is found to always use a wrench from 8 to 9 a.m., his record of usage may be added with a tracking tag of the early user.

Target preference point of view: this is a description or evaluation of the usage preference of a particular gauge. For example, if it is found that most workers prefer to use a wrench in the morning, then the idea that the morning is the best time for using a wrench can be derived.

Assuming a description of the tandem preference of a wrench-then-drill is obtained in step 140, a record tracking tag of the preferred wrench use may be added to the corresponding tool use record based on this description.

At the same time, it has been found that workers always use the drill after using the wrench whenever and wherever, which may indicate that the drill is an indispensable tool for performing work tasks. Thus, the target preference perspective that is derived may be that the drill is a requisite tool.

The resulting first gauge usage preference vector is then stored in a database along with this new usage record tracking tag. In this way, this information can be obtained directly from the database in a subsequent analysis without the need for complex calculations and analyses to be re-performed.

From the first aspect, during the execution of steps 110-150, the frequency and order of use of the tool can be clearly understood by mining the use preference vector for the tool use record. This helps the enterprise or team more accurately predict which tools may be frequently used in the future, thereby leading to preventive maintenance in advance or reasonable purchasing decisions based on use, avoiding affecting production efficiency due to damage or loss of tools.

For example, if a tool is found to be very frequently used, periodic inspection and maintenance may be required to prevent premature failure; if a tool is found to be rarely used, its quantity can be reduced appropriately at the next purchase to save costs.

From the second aspect, through the execution of steps 120-150, not only is a large amount of data collected regarding tool usage, but also this data is analyzed in depth, resulting in the tool usage preference vector and usage record tracking label. Such information may help improve the efficiency of use and management of the tool.

For example, by analyzing the tool usage preference vector, it may be found that a worker always uses a wrench and then uses a drill when performing a particular task. Such findings may help to better understand worker's work inertia, thereby optimizing workflow and improving tool use efficiency. Meanwhile, the use record tracking label can help to track and manage the use condition of the tool better, and the management effect is improved.

In summary, the use condition of the tool can be better known by implementing steps 110-150, and maintenance and purchasing decisions can be performed in advance, so that the working flow is improved, the use efficiency of the tool is improved, and the management effect of the tool is optimized.

In addition, from a digital maintenance perspective, steps 110-150 provide a method for tool management.

First, by mining the tool usage record for the usage preference vector at step 110, a more efficient way to understand and record the tool usage is obtained. These preference vectors can reflect not only the frequency of use of each tool in detail, but also more information about the user, time of use, etc. The data information is more accurate and comprehensive than the traditional manual recording mode. Next, in step 120, past tool usage preference vectors are stored in a relational database, which facilitates querying and analysis of historical data and also provides the possibility of long-term tracking and comparison of tool usage. In step 130, by sorting the tool usage preference vectors according to the recording acquisition priority, understanding of the tool usage situation can be further deepened, tools with high usage frequency can be found, and tool combinations commonly used in specific tasks can be found, which helps to optimize the allocation and usage of tools. In steps 140 and 150, the contextual preference of the gauge usage preference vector is identified and a corresponding usage record tracking tag is generated. The labels can be used for conveniently and quickly acquiring the use state of the tool, and can also provide deeper data analysis, such as finding out the priority use sequence of the tool in certain specific tasks, so as to optimize the workflow.

It can be seen that the implementation of steps 110-150 allows for more digital and intelligent management of the tool, which not only increases management efficiency, reduces error rates, but also helps to find potential problems and opportunities, thereby further increasing production efficiency. By the design, the problems of low management efficiency and low intelligent degree existing in the traditional engineering tool intelligent management technology can be solved.

In some alternative embodiments, identifying the target-front-back preference description of the first tool usage preference vector that the tool usage preference vector queue includes as described in step 140 includes steps 141-142.

Step 141, loading the tool to a depth residual model including a Y-level coding branch by using a preference vector queue, wherein in the u-level coding branch, u is greater than or equal to 1 and less than or equal to Y, and performing the following step 1411 or step 1412.

Step 1411, responsive to u being 1, performing usage preference vector mining on the first gauge usage preference vector and the X past gauge usage preference vectors based on the 1 st level coding branch, to obtain a first level front-back preference description of the first gauge usage preference vector.

And 1412, inquiring to obtain the u-1 level coding branch in the target past use period in response to u being more than or equal to 2 and less than or equal to Y, caching X u-1 level past intermediate preference vectors, and mining the u-1 level front and rear preference description generated by the u-1 level coding branch by combining the X u-1 level past intermediate preference vectors based on the u-1 level coding branch, so as to obtain the u level front and rear preference description.

The X past intermediate preference vectors are obtained by the X past workload using preference vectors through a u-1 stage coding branch in the Y stage coding branches.

Step 142, using the finally generated Y-level front and rear preference description as the target front and rear preference description.

In the embodiment of the application, the depth residual model is a machine learning model, and is particularly commonly used in the field of deep learning. The method is mainly characterized in that by introducing residual connection, a deeper neural network can be effectively trained, and the performance of the model is improved. The Y-level coding branch is an important structure in the depth residual model. The coding branches of each level represent a level or stage in the model, and the information complexity of the processing of the coding branches of different levels is different, and the higher the level is, the more abstract or complex the processed information is. The u-1 level past intermediate preference vector refers to the gauge usage preference vector generated during the processing of the previous u-1 level encoding branch, which reflects the gauge usage preference so far.

For example, there is one depth residual model, which contains 3-level coding branches (i.e., y=3). In step 141, the tool is loaded into this model using a preference vector queue.

When the level 1 coding branch (i.e., u=1), step 1411 is executed to perform usage preference vector mining on the first gauge usage preference vector and the X past gauge usage preference vectors, so as to obtain a first level context preference description. This may include information of which tools are used first, which tools are used later, etc.

Then, step 1412 is performed when the level 2 and level 3 coding branches (i.e., u=2 or 3). First, the X u-1 level past intermediate preference vectors of the u-1 level coding branch cache in the target past use period are inquired. Then, based on the u-th coding branch, combining the u-1 level past intermediate preference vectors, mining the u-1 level front and rear preference description by using the preference vectors to obtain the u-level front and rear preference description. This may include more detailed or complex tool usage sequence preferences, such as how often workers would use various tools when completing a complex task.

Through the process, the use preference of the tool can be deeply understood step by step, and more accurate and more detailed fore-and-aft preference description is generated, so that the working efficiency is improved, and the resource allocation is optimized.

In further embodiments, the method further comprises step 210.

Step 210, storing the u-level front-rear preference description in a relational database corresponding to the u-level coding branch, using the u-level front-rear preference vector in the record analysis as a subsequent tool, and removing the cached front-rear preference description of the u-level in the relational database corresponding to the u-level coding branch when the first tool uses the record analysis, so that the number of members in the set of the u-level coding branch corresponding relational database is X.

In the present embodiment, the u-level past intermediate preference vector refers to a tool usage preference vector generated during the processing of the u-th level coding branch, which reflects the tool usage preference so far. The set of tool usage preference vectors is a set containing a plurality of tool usage preference vectors. In this scheme, there is one such set in the database corresponding to each coding branch, and the number of members in the set is X.

For example, a u-level contextual preference description is obtained in step 1412. Then, in step 210, the u-level previous and subsequent preference description is stored in a relational database corresponding to the u-th level coding branch, and the u-level past intermediate preference vector in the record analysis is used as a subsequent tool.

For example, if a new context preference description is obtained in the level 2 coding branch (i.e., u=2), then this description is stored in the database corresponding to the level 2 coding branch. Thus, when the tool usage record analysis is performed next time, the past 2-level front and rear preference description can be directly obtained from the database.

Meanwhile, in order to keep the number of the members of the tool usage preference vector set in the database as X, the past u-level front-back preference description cached in the first tool usage record analysis in the database needs to be removed. For example, if a level 2 contextual preference description is cached at the first analysis, then this old description needs to be removed when a new level 2 contextual preference description is added.

Through the process, the latest and fixed-number working tool use preference vectors in the database corresponding to each coding branch can be ensured, so that the subsequent use record analysis and prediction are convenient.

Under some optional design considerations, the method further includes steps 310-330 before identifying the target-front-back preference description of the first workload tool usage preference vector included in the workload tool usage preference vector queue described in step 140.

Step 310, if the first tool usage preference vector and the X past tool usage preference vectors are characterized by a double-precision variable state, changing the target double-precision variables in the first tool usage preference vector and the X past tool usage preference vectors to quantized variable states to obtain original changing weights.

Step 320, based on the original changing weight, performing a first variable changing process on the first tool use preference vector and the X past tool use preference vectors, to obtain a first tool use preference vector and X past tool use preference vectors of the quantized variable state.

Based on this, after obtaining the first level of the first-order contextual preference description of the first-gauge usage preference vector described in step 1411, the method further includes: and performing second variable change processing on the first-stage front-and-rear preference description of the quantized variable state according to the original change weight to obtain the first-stage front-and-rear preference description of the double-precision variable state.

Related terms under the above design concept are explained as follows.

Double precision variable state: this is a type of data used to store numbers with high accuracy. When the process tool uses preference vectors, it is possible to use double precision variables to record more detailed and accurate information.

Quantization variable state: this refers to converting double precision variables into a more manageable form of data, such as an integer or fixed length binary representation. This may reduce computational complexity and memory requirements.

Original change weights: this is a weight determined according to the characteristics (e.g., range, distribution, etc.) of the raw data when transitioning from the double-precision variable state to the quantized variable state. It reflects the importance of each variable in the course of the change.

For example, if the first tool usage preference vector and the X past tool usage preference vectors are both found to be characterized by a double precision variable state in step 310, then they need to be converted to quantized variable states. For example, if the original double-precision variable is a probability value between 0 and 1, it can be multiplied by 100 and rounded to convert to an integer between 0 and 100, which is the quantized variable state. In this process, an original change weight is also needed to be obtained, which can be calculated according to the characteristics of the range, the distribution and the like of the double-precision variable.

Then, in step 320, based on the original change weight, a first variable change process is performed on the first gauge use preference vector and the X past gauge use preference vectors, so as to obtain a first gauge use preference vector and X past gauge use preference vectors for quantizing the variable state.

Next, after step 1411, a second variable change process is performed on the obtained first-stage front-and-back preference description according to the original change weight, and it is converted from the quantized variable state back to the double-precision variable state. In this way, a first level of contextual preference description of the state of the double-precision variable is obtained, providing more accurate data for subsequent analysis and prediction.

In some other possible embodiments, the method further comprises steps 410-420.

Step 410, in response to u being greater than or equal to 2 and less than or equal to Y, if the u-1 level front-back preference description and the X u-1 level past intermediate preference vectors are characterized by a double precision variable state, changing a target double precision variable in the u-1 level front-back preference description and the X u-1 level past intermediate preference vectors to a quantized variable state to obtain a u-level change weight.

And step 420, based on the u-level changing weight, performing a first variable changing process on the u-1 level front and rear preference description and the X u-1 level past intermediate preference vectors of the double-precision variable state to obtain a u-1 level front and rear preference description and X u-1 level past intermediate preference vectors of the quantized variable state.

Based on this, after obtaining the u-level contextual preference description as described in step 1412, the method further comprises: and carrying out second variable change processing on the u-level front-rear preference description of the quantized variable state according to the u-level change weight to obtain the u-level front-rear preference description of the double-precision variable state.

In the above embodiment, the related nouns are explained as follows.

u-level variable weight: this refers to a weight determined according to the characteristics (e.g., range, distribution, etc.) of the original data when transitioning from the double-precision variable state to the quantized variable state during the u-th level encoding branch processing. It reflects the importance of each variable in the course of the change.

For example, it is first assumed that in step 410 u=3, i.e. the 3 rd level coding branch is being processed. If both the 3-1 level (i.e., level 2) front and back preference descriptions and the X level 2 past intermediate preference vectors are found to be characterized by a double precision variable state, then they need to be converted to quantized variable states. For example, if the original double-precision variable is a probability value between 0 and 1, it can be multiplied by 100 and rounded to convert to an integer between 0 and 100, which is the quantized variable state. In this process, a 3-level variable weight is also required, which may be calculated from the characteristics of the range and distribution of the double-precision variable.

Then, in step 420, based on the 3-level change weight, a first variable change process is performed on the 2 nd-level front-rear preference description and the X2-level past intermediate preference vectors of the double-precision variable state, so as to obtain the 2 nd-level front-rear preference description and the X2-level past intermediate preference vectors of the quantized variable state.

Next, after step 1412, a second variable change process is performed on the resulting 3 rd level front and rear preference description according to the 3 rd level change weight, which is converted from the quantized variable state back to the double precision variable state. In this way, a level 3 contextual preference description of the state of the double precision variable is obtained, providing more accurate data for subsequent analysis and prediction.

In some alternative embodiments, before mining the first tool usage preference vector and the X past tool usage preference vectors based on the level 1 coding branch described in step 1411, further comprising: and adjusting the neural network parameters of the 1 st stage coding branch according to the preference vector used by the first measuring tool of the quantized variable state and the preference vectors used by the X past measuring tools.

Based on this, the u-1 level front and rear preference description generated by the u-1 level coding branch is mined by combining the X u-1 level past intermediate preference vectors based on the u-level coding branch described in step 1412, and further includes: and adjusting the neural network parameters of the coding branch of the u th level according to the u-1 level front and back preference description of the quantized variable state and X u-1 level past intermediate preference vectors.

In the above embodiment, the neural network parameters are basic components of the neural network model, including weights, biases, and the like. These parameters are optimized through a training process, which determines the performance of the model.

Before step 1411, the neural network parameters of the level 1 coding branch need to be adjusted according to the first gauge usage preference vector and the X past gauges usage preference vectors of the quantization variable states. For example, the weights or biases of the neural network may be adjusted to better adapt to the data based on characteristics (e.g., range, distribution, etc.) of the usage preference vector, thereby improving the performance of the model.

Likewise, prior to step 1412, the neural network parameters of the u-th encoding branch need to be adjusted according to the u-1 level front-to-back preference description of the quantized variable state and the X u-1 level past intermediate preference vectors. For example, if certain patterns or trends are found in these preference descriptions and the intermediate preference vectors, the parameters of the neural network may be adjusted accordingly to capture these patterns or trends, improving the accuracy of the model.

In this way, the neural network model of each coding branch can be better adapted to the corresponding data, so that the identification and prediction capabilities of the whole depth residual error model on the use preference of the engineering tool are improved.

In some preferred embodiments, the obtaining of the usage record tracking tag of the usage record of the tool to be analyzed according to the target contextual preference description described in step 150 comprises: and adopting a multi-layer perceptron to perform feature conversion on the target fore-and-aft preference description to obtain a usage record tracking label of the usage record of the measuring tool to be analyzed.

Related terms of the above embodiments are explained as follows.

Multilayer perceptron: multilayer perceptrons (MLPs) are a common model of neural networks that include an input layer, one or more hidden layers, and an output layer. Each layer contains several neurons that non-linearly transform the input through an activation function.

Feature conversion: this is a common technique in machine learning for converting raw data into a more tractable or meaningful form. For example, the original data may be feature-converted by dimension reduction, normalization, or the like.

For example, in step 150, a usage record tracking tag of the usage record of the tool to be analyzed is obtained according to the target contextual preference description. For this purpose, a multi-layer perceptron is used to perform feature transformation on the target contextual preference descriptions.

For example, the target context preference description may be input to the input layer of the multi-layer sensor, then non-linearly transformed by the hidden layer, and finally transformed features are obtained at the output layer. These converted features can capture deep information in the target's contextual preference descriptions, thereby acting as usage record tracking tags for subsequent tool usage record analysis.

Thus, not only can the tracking label recorded by the use of each tool be obtained, but also useful information can be further extracted through feature conversion, so that the use condition of the tool can be better understood and predicted.

In some independently implementable embodiments, the method is implemented by a target gauge using a log analysis algorithm, the debugging step of the target gauge using the log analysis algorithm comprising step 500.

Step 500, based on a set algorithm debugging case set, performing a plurality of times of cyclic debugging on the to-be-debugged tool using a record analysis algorithm, wherein each algorithm debugging case comprises a tool using record case to be analyzed, X past tool using record cases and a priori using record tracking label, and the tool using record case to be analyzed and the X past tool using record cases are sorted according to record acquisition priorities.

Wherein, during one cycle, the following steps 510-530 are performed.

And 510, mining the use preference vectors of the to-be-analyzed tool use record cases and X past tool use record cases in one algorithm debugging case to obtain a predicted tool use preference vector and X tool use preference vector cases.

Step 520, identifying a case of the predicted workload using preference vector and the X workload using preference vector cases, where the predicted workload using preference vector is described by a sequential preference of the predicted workload using preference vector.

Wherein the contextual model preference description case reflects a usage relationship between the predicted workload usage preference vector and the X workload usage preference vector cases.

Step 530, obtaining a predicted usage record tracking label according to the contextual descriptive cases, and improving the algorithm configuration variables of the tool usage record analysis algorithm based on the difference between the predicted usage record tracking label and the prior usage record tracking label of the one algorithm debugging case.

In the embodiment of the present application, related terms are explained as follows.

Algorithm debug case set: this is a specific set of data for testing and optimizing the tool using a log analysis algorithm. Each algorithm debug case may include one gauge use log case to be analyzed, X past gauge use log cases, and an a priori use log tracking tag.

The predictive gauge uses a preference vector: this is the result of mining the work implement usage record cases to be analyzed by the work implement usage record analysis algorithm.

Algorithm configuration variables: this refers to some parameter or variable that controls or affects the behavior of the algorithm. For example, in a neural network model, the learning rate, the optimizer type, etc. can all be considered as algorithm configuration variables.

For example, first, a case set is debugged based on a set algorithm, and a work tool to be debugged is circularly debugged for a plurality of times by using a log analysis algorithm. In each cycle, first, the usage preference vector mining is performed on the to-be-analyzed tool usage record case and the X past tool usage record cases in one algorithm debugging case, so as to obtain a predicted tool usage preference vector and X tool usage preference vector cases (step 510). Then, among the predicted workload usage preference vector and the X workload usage preference vector cases, the predicted workload usage preference vector is identified as a proceeding and proceeding preference description case (step 520).

After the contextual preference description case is obtained, a predictive usage record tracking tag is obtained therefrom and the algorithm configuration variables of the tool usage record analysis algorithm are modified based on the difference between the predictive usage record tracking tag and the a priori usage record tracking tag of the algorithm debug case (step 530). For example, if the predictive usage record tracking tag differs significantly from the prior usage record tracking tag, it may be necessary to adjust algorithm configuration variables, such as modifying the learning rate or replacing the optimizer, to improve the performance of the algorithm.

In this way, the tool usage log analysis algorithm can be continuously optimized and improved so that it can obtain more accurate and useful results when processing actual data.

In addition, in some independent embodiments, after obtaining the usage record tracking tag of the usage record of the tool to be analyzed according to the target contextual preference description described in step 150, and obtaining the target preference viewpoint added corresponding to the usage record tracking tag, the method further includes step 160 after storing the usage preference vector of the first tool in the set relational database as the usage preference vector of the past tool in the subsequent tool usage record analysis.

Step 160, generating a target tool maintenance policy according to the usage record tracking tag and the corresponding target preference point.

In the above embodiments, the target gauge maintenance strategy is a maintenance plan or method that is formulated for a specific gauge, including the time, frequency, manner, etc. of maintenance. This strategy is typically adjusted according to the use and status of the tool.

For example, in step 160, a maintenance policy for the target tool is generated from the usage record tracking tag and its corresponding target preference perspective. For example, if the usage record tracking tag shows that a tool is frequently malfunctioning in past usage, and the corresponding target preference perspective also confirms this, a more frequent maintenance strategy may be formulated, such as checking and maintenance once per week, to ensure proper operation of the tool.

In this way, the most appropriate maintenance strategy can be made for each tool according to the actual use condition and the requirements, thereby improving the use efficiency and the service life of the tool.

In addition, in other, but independent embodiments, the method further includes step 170.

And step 170, responding to the disaster recovery instruction of the to-be-analyzed tool usage record, and carrying out structural disaster recovery processing on the to-be-analyzed tool usage record.

In the above embodiment, the disaster recovery instruction refers to an instruction that the system will issue to protect and resume the normal use of the tool if a fault or other unpredictable situation occurs during the use of the tool. The structured disaster recovery processing refers to performing systematic error repair, data recovery and other operations on the tool usage record according to a preset rule or flow.

For example, in step 170, in response to the disaster recovery instructions for the gauge usage record to be analyzed, the gauge usage record to be analyzed is subjected to a structured disaster recovery process. For example, if a disaster recovery command is received, the disaster recovery command indicates that the tool fails, the relevant information in the tool usage record, such as the usage status of the tool, the failure time, etc., may be checked first. Then, according to a preset rule, if the operation of the tool is stopped, then the fault is checked and repaired, and finally the normal use of the tool is resumed. In this process, the gauge usage record may also need to be updated to reflect the latest status of the gauge.

By the mode, the normal use of the tool can be quickly and effectively protected and restored when faults or other problems occur, so that the use efficiency and stability of the tool are improved.

In addition, in other independent embodiments, the structured disaster recovery process is performed on the usage record of the tool to be analyzed in step 170, including steps 171-177.

Step 171, obtaining the tool use record structured text.

Step 172, acquiring a work tool use event analysis result in the work tool use record to be analyzed and a work tool use event analysis result in the structured document of the work tool use record.

And step 173, determining the event commonality factors of the to-be-analyzed tool use record and the tool use of the structured text of the tool use record according to the tool use event analysis result in the to-be-analyzed tool use record and the tool use event analysis result in the structured text of the tool use record.

Step 174, determining that the tool to be analyzed uses the recorded structural semantic description and that the tool uses the recorded structural semantic description of the structured text.

Step 175, determining a first structural semantic description commonality factor of the to-be-analyzed measuring tool use record and the measuring tool use record structured text according to the structural semantic description of the to-be-analyzed measuring tool use record and the structural semantic description of the measuring tool use record structured text.

Step 176, determining a global commonality factor between the to-be-analyzed tool usage record and the tool usage record structured text according to the tool usage record to be-analyzed tool usage record and the tool usage record structured text and the first structured semantic description commonality factor.

And step 177, determining a target disaster recovery template text from the structured text of the tool usage record according to the global commonality factor between the tool usage record to be analyzed and the structured text of the tool usage record, and carrying out structured disaster recovery processing on the tool usage record to be analyzed according to the target disaster recovery template text.

In the above embodiment, the related nouns are explained as follows.

The tooling is used for recording structured texts: this may refer to the result of the tool usage record being processed to reveal the tool usage information in a structured manner (e.g., a table or database) that allows the information to be more easily understood and manipulated.

The tool uses the event parsing result: this refers to information obtained after analysis of the gauge usage record, including details of various usage events of the gauge (e.g., start of use, end of use, failure occurrence, etc.).

Commonality factor: this refers to similar features or attributes that are present in both or more tool usage records, such as the same usage pattern, the same fault type, etc.

Structured semantic description: the description obtained after the logic arrangement and semantic analysis of the use record content of the tool can be used for describing the use state, the use effect and the like of the tool.

Target disaster recovery template text: in this context, this refers to the gauge usage record that is most similar to the gauge usage record to be analyzed, which has completed the structured disaster recovery process (i.e., structured backup).

For example, the objective of steps 171 to 177 is to find a similar tool usage record to the tool usage record to be analyzed, and to guide the structured disaster recovery process for the tool usage record to be analyzed according to the structured disaster recovery (structured backup) result of the similar record. First, a structured text of the tool usage record is acquired (step 171), and then a tool usage event parsing result in the tool usage record and the structured text to be analyzed is acquired (step 172). Next, both sex factors are determined (step 173). Thereafter, the structured semantic descriptions of the tool to be analyzed using the record and the structured text are determined (step 174), and the commonality factor of the structured semantic descriptions of both is further determined (step 175). Then, based on the commonality factor, a global commonality factor between the tool usage record and the structured text to be analyzed is determined (step 176). Finally, according to the global commonality factor, the tool usage record which is most similar to the tool usage record to be analyzed is found from the structured text, namely the target disaster recovery template text, and the structured disaster recovery processing is carried out on the tool usage record to be analyzed according to the structured disaster recovery (structured backup) result (step 177).

Therefore, the existing successful experience can be used for effectively carrying out structured backup on the new tool use record, and the efficiency and the success rate of data recovery are improved. Further, by structurally storing the tool usage record, the ease of use and readability of the data can be greatly improved. Meanwhile, the structured processing mode is also beneficial to inquiring, analyzing and mining the data, so that the application value of the data is further improved. By analyzing the global commonality factor of the utilization record of the tool, key information such as the utilization mode, the utilization effect and the like of the tool can be deeply known, which is helpful for more reasonably configuring and utilizing resources, such as more accurately predicting the maintenance requirement of the tool or more reasonably scheduling the utilization plan of the tool.

Further, there is also provided a computer-readable storage medium having stored thereon a program which, when executed by a processor, implements 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, the 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 such 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 methods described in the embodiments of the present application. And the aforementioned storage medium includes: a U-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 foregoing description is only of the preferred embodiments of the present application and is not intended to limit the same, but rather, various modifications and variations may be made by those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the present application should be included in the protection scope of the present application.

Claims

1. A data processing method based on digital maintenance, which is characterized by being applied to an intelligent management system of an engineering tool, the method comprising:

Obtaining a usage record tracking tag of the to-be-analyzed tool usage record according to the target fore-and-aft preference description, obtaining a target preference viewpoint added corresponding to the usage record tracking tag, and storing the first tool usage preference vector into a set relational database to serve as a past tool usage preference vector in subsequent tool usage record analysis;

wherein the identifying the target contextual preference description of the first gauge usage preference vector included in the gauge usage preference vector queue includes:

2. The method as recited in claim 1, further comprising:

3. The method of claim 1, wherein the identifying the target-front-and-back preference description of the first gauge usage preference vector included in the gauge usage preference vector queue further comprises:

4. The method as recited in claim 1, further comprising:

5. The method of claim 4, wherein prior to mining the first tool usage preference vector and the X past tool usage preference vectors based on the level 1 coding branch, further comprising: according to the preference vector used by the first measuring tool and the preference vectors used by the X past measuring tools of the quantized variable states, the neural network parameters of the 1 st-stage coding branch are adjusted;

6. The method of claim 1, wherein the obtaining a usage record tracking tag of the usage record of the tool to be analyzed according to the target contextual preference description comprises:

7. The method of claim 1, wherein the method is implemented by a target tool using a log analysis algorithm, and wherein the debugging process of the target tool using the log analysis algorithm is as follows:

wherein, in the course of one cycle, the following steps are implemented:

8. The intelligent management system for the tool is characterized by comprising a processor and a memory; the processor is communicatively connected to the memory, the processor being configured to read a computer program from the memory and execute the computer program to implement the method of any of claims 1-7.

9. A computer readable storage medium, characterized in that a program is stored thereon, which program, when being executed by a processor, implements the method of any of claims 1-7.