Production workshop bottleneck identification method

Abstract

The invention provides a production workshop bottleneck identification method, which comprises the following steps of: firstly, configuring a proper sensor and a sensing network for a production workshop through an Internet of Things technology, and acquiring state data and position data of a product in the production workshop and position data of an AGV (Automatic Guided Vehicle) in real time; secondly, data preprocessing operation is carried out so that the data can be directly used, and then needed production parameters are calculated through the processed data; and finally, establishing a bottleneck identification model based on a DNN neural network model, outputting bottleneck indexes of different links according to the state data of the work-in-process, the position data of the work-in-process and the position data of the AGV in the production workshop, and identifying the position of the bottleneck. According to the invention, the bottleneck of the production workshop at the current moment can be dynamically identified, and an important reference value is provided for the management of the production workshop.

Description

Technical Field

[0001] The present invention relates to the field of industrial production technology, and in particular to a real-time identification method for production bottlenecks. Based on this method, production managers can accurately identify the bottleneck of the production system at the current moment in real time, providing important reference information for active control of the production process. Background Art

[0002] Dr. Goldratt pointed out in TOC (Theory of Constraint) that the production bottleneck is the part with the greatest impact on the entire manufacturing system, so the key to improving system efficiency is to find and eliminate the bottleneck. However, the manufacturing system has the characteristics of high coupling between various production links, complex manufacturing details, and frequent random disturbances. The production bottlenecks in different environments are different. Therefore, accurate and rapid identification of bottlenecks is the basis and prerequisite for optimizing workshop production line performance and improving production capacity. Traditional production bottleneck identification has the phenomenon of poor real-time performance, and the identified bottlenecks often lag behind the actual bottlenecks. Summary of the invention

[0003] The present invention designs a production workshop bottleneck identification method, the purpose of which is to dynamically identify the bottleneck of the production workshop at the current moment and provide important reference information for the management of the production workshop.

[0004] First, the Internet of Things technology is used to configure appropriate sensors and sensor networks in the production workshop to collect real-time data on the status of work-in-progress, the location of work-in-progress, and the location of AGVs in the production workshop. Secondly, data preprocessing operations are performed to make it directly usable, and the required production parameters are calculated through the processed data. Finally, a bottleneck identification model based on the DNN neural network model is established. According to the status of work-in-progress, the location of work-in-progress, and the location of AGVs in the production workshop, the bottleneck index of different links is output and the location of the bottleneck is identified.

[0005] The technical solution adopted by the present invention is:

[0006] A method for identifying bottlenecks in a production workshop comprises the following steps:

[0007] Step 1: Configure the IoT network in the production workshop, build intelligent manufacturing resources that can perceive the surrounding environment and its own state changes in real time, and collect multi-source manufacturing data in the production workshop;

[0008] Step 2: Obtain work-in-progress production information data;

[0009] Step 3: Preprocess the multi-source manufacturing data and production information data to build a multivariate correlation data model of workshop production status;

[0010] Step 4: Establish a production workshop bottleneck identification model, and identify the current bottleneck of the actual production workshop based on the identification model.

[0011] The configuration of the Internet of Things network includes the following steps:

[0012] Equip WIP and AGV with corresponding electronic tags, and obtain real-time location and status information data through high-precision positioning modules;

[0013] Set up a corresponding sensor network to transmit the collected location and status information data in real time.

[0014] The production information data is obtained by loading from MES, ERP and other systems based on the electronic tag information.

[0015] The production information data includes the current process number, production plan, and production priority.

[0016] The data preprocessing is to smooth the data using a median filtering algorithm.

[0017] The multivariate association data model of the workshop production status is a {work-in-process status data model, AGV status data model} that uses sampling time as the primary key to associate and represent the production status.

[0018] The WIP status data model is constructed by using the WIP electronic tag number as the primary key, and associating the WIP's current process number, the WIP's corresponding production plan, and the WIP's production priority information with the WIP's real-time location information.

[0019] The AGV status data model is constructed by using the electronic tag number of the AGV as the primary key and associating the work-in-progress number information of the AGV that is performing the transfer task in the AGV with the real-time location information of the AGV.

[0020] The establishment of the production workshop bottleneck identification model comprises the following steps:

[0021] According to the historical production data of the workshop, calculate the bottleneck index of all workstations in the workshop;

[0022] The DNN deep neural network is used as the recognition model, and the multivariate correlation data model of the historical bottleneck index and the historical workshop production status is used for training;

[0023] In the training sample, the input is a multivariate association data model of the workshop production status at time θ, and the output is the bottleneck index of all workstations in the workshop at time θ;

[0024] The workstation with the highest bottleneck index is identified as the bottleneck of the production workshop.

[0025] The bottleneck index is calculated for each workstation and each process at the current time. Bottleneck Index for:

[0026]

[0027] Among them, m p Indicates the number of workstations included in process p, l pi Indicates S pi The length of the queue in the buffer area, Indicates S pi Enter the kth workpiece to be processed in the cache queue; traverse the workpiece Enter workstation S pi The reasonableness of Indicates the workpiece at time θ When entering process p, if entering station S pj , which is at workstation S pj Such as the waiting time in the buffer area.

[0028] The present invention has the following beneficial effects and advantages:

[0029] 1. All algorithm formulas mentioned in the present invention do not involve integral calculation, making the present invention simple to implement;

[0030] 2. The present invention applies the existing Internet of Things technology to the production workshop, realizes active perception of the production process, and collects multi-source manufacturing data in the production process in real time. At the same time, the neural network used is small in scale, high in calculation efficiency, and good in real time, thus avoiding the phenomenon that the identified bottleneck lags behind the actual bottleneck; BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 It is the overall flow chart of the present invention; DETAILED DESCRIPTION

[0032] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation method of the present invention is described in detail below in conjunction with the accompanying drawings. In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention can be implemented in many other ways different from those described herein, and those skilled in the art can make similar improvements without violating the connotation of the invention, so the present invention is not limited by the specific implementation disclosed below.

[0033] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art of the present invention. The terms used in the specification of the invention herein are only for the purpose of describing specific embodiments and are not intended to limit the present invention.

[0034] like Figure 1As shown, the present invention designs a production workshop bottleneck identification method, the purpose of which is to dynamically identify the bottleneck of the production workshop at the current moment, and provide important reference information for the active control of the production process. The higher the bottleneck index, the greater the production link of the workstation and the process will restrict the entire production process, and once a problem occurs, it should be solved first. The specific implementation steps are as follows:

[0035] Step 1: Configure the existing IoT network in the production workshop, build intelligent manufacturing resources that can perceive the surrounding environment and its own state changes in real time, and collect multi-source manufacturing data in the production workshop, including the following steps:

[0036] Step 1.1: Equip the WIP and AGV with corresponding electronic tags, and obtain real-time location and status information data through a high-precision positioning module; the electronic tag is a chip (such as an eSIM card);

[0037] Step 1.2: Select the corresponding sensor network to transmit the collected location and status information data in real time;

[0038] Step 2: Based on the electronic tag information, obtain the current process number, production plan, production priority and other information data of the work-in-progress from the MES, ERP and other systems;

[0039] Step 3: Preprocess the multi-source manufacturing data and production information data collected in Step 1 and Step 2 to make them into usable data sets, including the following steps:

[0040] Step 3.1: Preprocess the collected location information data and use the median filter algorithm to complete the data smoothing;

[0041] Step 3.2: The processed real-time location information data of the work-in-process is used as the primary key, the real-time location information of the work-in-process is used as the main information, and the current process number of the work-in-process, the corresponding production plan of the work-in-process and the production priority information of the work-in-process are used to construct a work-in-process status data model;

[0042] Step 3.3: The processed AGV real-time location information data is used with the AGV electronic tag number as the primary key, the AGV real-time location information as the main information, and the work-in-progress number information of the AGV that is performing the transfer task, to construct the AGV status data model;

[0043] Step 3.4: Using the sampling time as the primary key, the WIP status data model and the AGV status data model as the main information, construct a multivariate association data model of the workshop production status;

[0044] Step 3.5: Using the sampling time as the primary key, the WIP status data model and the AGV status data model as the main information, construct a multivariate association data model of the workshop production status;

[0045] Step 4: Establish a production workshop bottleneck identification model model, and identify the current bottleneck of the production workshop based on the identification model; including the following steps:

[0046] Step 4.1: If the production workshop bottleneck identification model has been established, directly execute step 4.4, otherwise continue to execute;

[0047] Step 4.2: Based on the historical production data of the workshop, calculate the bottleneck index of all workstations in the historical workshop.

[0048] Step 4.2.1: Initialize the process number p=1, the workstation number i=1 and the corresponding work-in-progress queue number k=0 in the buffer area of ​​the workstation;

[0049] Step 4.2.2 traverses all processes and executes step 4.2.3; the program terminates after traversal;

[0050] Step 4.2.3 traverses all workstations of the current process and executes step 4.2.4; after traversal, set p = p + 1 and return to step 4.2.2;

[0051] Step 4.2.4 traverses all the WIPs in the buffer area waiting for processing in the current station, and executes step 4.2.5; after traversal, calculate the current station Bottleneck Index Let i=i+1 and return to step 4.2.3.

[0052] in, m p Indicates the number of workstations included in process p, l pi Indicates S pi The length of the queue in the buffer area, Indicates S pi Enter the kth workpiece to be processed in the buffer queue;

[0053] Step 4.2.5 Based on the unique number of the current WIP, search the workshop production status from time θ until the current WIP appears in the current workstation for the first time, and record the starting time t 1 ;

[0054] Step 4.2.6 Based on the unique number of the current WIP, search the workshop production status from time θ until the status of the current WIP at the current workstation changes to being processed, and record the end time t 2 ;

[0055] Step 4.2.7 Calculate the actual waiting time of the current work in process at the current station:

[0056] Step 4.2.8 is to calculate the waiting time for the current work-in-progress to enter other workstations when it is at its nearest transfer decision point, and initialize the workstation number j = 1;

[0057] Step 4.2.9 traverses all workstations of the current process again and executes step 4.2.10; after traversing, the workpiece Enter workstation S pi The reasonableness of And return to step 4.2.4;

[0058] in, Indicates the workpiece at time θ When entering process p, if entering station S pj , which is at workstation S pj Such as the waiting time in the buffer area.

[0059] If the new station number in step 4.2.10 is the same as the original station number, return to step 4.2.9; otherwise, execute step 4.2.11;

[0060] Step 4.2.11 Assume that t 1 At the moment when the current work-in-progress enters the new workstation, according to the relevant settings of the workshop production priority, the current work-in-progress should appear at a certain position in the queue to be processed, starting from t 1 Start searching the workshop production status backwards from time t' until all the workpieces in the queue before the current work-in-progress have left the buffer area and record the time t';

[0061] Step 4.2.12 Calculate the hypothetical waiting time of the current WIP at other workstations And return to step 4.2.9;

[0062] Step 4.3: Use the bottleneck index of all workstations in the historical workshop and the multivariate correlation data model of the historical workshop production status to train the DNN deep neural network recognition model; in the training sample, the input is the multivariate correlation data model of the workshop production status at time θ, and the output is the bottleneck index of all workstations in the workshop at time θ; RMSE (Root Mean Squared Error) is used as an indicator, and the training is terminated when the indicator is close to 0 and almost no longer changes.

[0063] Step 4.4: Use the multivariate correlation data model of the current workshop production status as the input of the DNN deep neural network recognition model, and output the bottleneck index of all workstations in the workshop at the current moment;

[0064] Step 4.5: Based on the bottleneck indexes of all workstations in the workshop, identify the workstation with the highest bottleneck index as the bottleneck of the production workshop.

[0065] The above description is only a preferred embodiment of the present invention and does not limit the present invention in any way. Any simple modification, change and equivalent structural change made to the above embodiment based on the technical essence of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. A method for identifying bottlenecks in a production workshop, characterized in that: The following steps are involved: Step 1: Configure the IoT network in the production workshop, build intelligent manufacturing resources that can perceive the surrounding environment and its own state changes in real time, and collect multi-source manufacturing data in the production workshop; Step 2: Obtain work-in-progress production information data; Step 3: Preprocess the multi-source manufacturing data and production information data to build a multivariate correlation data model of workshop production status; Step 4: Establish a production workshop bottleneck identification model, and identify the current bottleneck of the actual production workshop based on the identification model.

2. A method for identifying bottlenecks in a production workshop according to claim 1, characterized in that: The configuration of the Internet of Things network includes the following steps: Equip WIP and AGV with corresponding electronic tags, and obtain real-time location and status information data through high-precision positioning modules; Set up a corresponding sensor network to transmit the collected location and status information data in real time.

3. A method for identifying bottlenecks in a production workshop according to claim 1, characterized in that: The production information data is obtained by loading from MES, ERP and other systems based on the electronic tag information.

4. A method for identifying bottlenecks in a production workshop according to claim 3, characterized in that: The production information data includes the current process number, production plan, and production priority.

5. A method for identifying bottlenecks in a production workshop according to claim 1, characterized in that: The data preprocessing is to smooth the data using a median filtering algorithm.

6. A method for identifying bottlenecks in a production workshop according to claim 1, characterized in that: The multivariate association data model of the workshop production status is a {work-in-process status data model, AGV status data model} that uses sampling time as the primary key to associate and represent the production status.

7. A method for identifying bottlenecks in a production workshop according to claim 1, characterized in that: The WIP status data model is constructed with the WIP electronic tag number as the primary key, and the WIP real-time location information is associated with the WIP's current process number, the WIP's corresponding production plan, and the WIP production priority information.

8. A method for identifying bottlenecks in a production workshop according to claim 1, characterized in that: The AGV status data model is constructed by using the electronic tag number of the AGV as the primary key and associating the work-in-progress number information of the AGV that is performing the transfer task in the AGV with the real-time location information of the AGV.

9. A method for identifying bottlenecks in a production workshop according to claim 1, characterized in that: The establishment of the production workshop bottleneck identification model comprises the following steps: According to the historical production data of the workshop, calculate the bottleneck index of all workstations in the workshop; The DNN deep neural network is used as the recognition model, and the multivariate correlation data model of the historical bottleneck index and the historical workshop production status is used for training; In the training sample, the input is a multivariate association data model of the workshop production status at time θ, and the output is the bottleneck index of all workstations in the workshop at time θ; The workstation with the highest bottleneck index is identified as the bottleneck of the production workshop.

10. A method for identifying bottlenecks in a production workshop according to claim 1, characterized in that: The bottleneck index is calculated for each workstation and each process at the current time. Bottleneck Index for: Among them, m p Indicates the number of workstations included in process p, l pi Indicates S pi The length of the queue in the buffer area, Indicates S pi Enter the kth workpiece to be processed in the cache queue; traverse the workpiece Enter workstation S pi The reasonableness of Indicates the workpiece at time θ When entering process p, if entering station S pj , which is at workstation S pj Such as the waiting time in the buffer area.

Images