Automatic station optimization algorithm of product assembly line

Abstract

The invention discloses an automatic station optimization algorithm for a product assembly line, and relates to the technical field of product assembly line management. The algorithm provided by the invention comprises the following steps: S1, collecting image / video data containing each working step, and obtaining working step time through working step identification; s2, periodically calculating average time consumption and time consumption fluctuation standard deviation of each process step; s3, dynamic station optimization; and S4, calculating the change of the balance rate of the production line before and after dynamic adjustment. According to the invention, the technical problem of online automatic station scheduling optimization of the product assembly line is solved. The algorithm is applied to a product assembly line, automatic station scheduling optimization can be carried out on line, equalization, flexibility and intelligentization of the production process are achieved by eliminating the bottleneck, then the production efficiency and quality are improved, and finally cost reduction and efficiency improvement of enterprises are promoted.

Description

Technical Field

[0001] The present invention relates to the technical field of product assembly line management, in particular to an automatic optimization algorithm for workstations on a product assembly line. Background Art

[0002] During the product assembly process on an assembly line, once a bottleneck workstation appears, it will affect the overall beat control of the assembly line, and further affect the production (assembly) efficiency and quality.

[0003] In the prior art, in view of the above situation, it is generally necessary for on-site assembly managers to perform workstation scheduling online, and then cancel the bottleneck workstation to restore the normal beat of the assembly line. For the workstation optimization method of the assembly line, due to insufficient intelligence, the analysis of real-time production data is not rigorous enough and the pertinence is insufficient, resulting in insufficient accuracy of data analysis and inaccurate control of the production process.

[0004] Therefore, the applicant proposes the present invention. Summary of the Invention

[0005] The purpose of the present invention is to provide an automatic optimization algorithm for workstations on a product assembly line to solve the above-mentioned deficiencies of the prior art. Through online automatic workstation scheduling optimization, it can improve production efficiency and quality.

[0006] To achieve the above purpose, an automatic optimization algorithm for workstations on a product assembly line designed by the present invention includes the following steps:

[0007] S1. Collect image / video data containing each work step, and obtain the work step time through work step recognition;

[0008] S2. Periodically calculate the average time consumption and the standard deviation of time consumption fluctuation of each work step, specifically:

[0009] S2.1. Work step Average time consumption formula:

[0010]

[0011] Wherein, : The average time consumption of the work step within the statistical period; : The total number of workers; : The number of workers who execute the work step within the statistical period; : The worker executes The average time consumption of the work step;

[0012] Worker Executes the work step The average time consumption calculation formula:

[0013]

[0014] Among them, : The time consumption for a single execution of a working step; : Worker Execute The number of times of executing the working step; assuming the worker has no execution record, then = 0;

[0015] S2.2. Standard deviation formula for time consumption fluctuation:

[0016]

[0017] Among them, : Standard deviation of time consumption fluctuation, reflecting the degree of data dispersion; 、 、 : Defined as above;

[0018] S3. Dynamic workstation optimization, specifically:

[0019] S3.1. Time axis construction

[0020] Accumulate the time consumptions of each working step in sequence to generate the total time axis. The formula is:

[0021] ( is the time consumption of the th working step)

[0022] S3.2. Equally divided point calculation

[0023] According to the number of workers , divide the total time axis into segments. The coordinates of the equally divided points are:

[0024] ( )

[0025] S3.3. Process node matching

[0026] Traverse the equally divided points and select the end time of the working step with the closest distance as the process node. The matching rule is:

[0027]

[0028] S3.4. Coincidence coefficient evaluation

[0029] Calculate the coincidence coefficient of each process node , where ;

[0030] When merging adjacent process segments to generate a special process by adding workstations to ensure that the time consumption of this process is reduced below Tavg.

[0031] Furthermore, an automatic workstation optimization algorithm for a product assembly line preferably further includes:

[0032] S4. Calculating the change in the production line balance rate before and after dynamic adjustment, the formula is:

[0033] .

[0034] Still further, in the above-mentioned automatic workstation optimization algorithm for a product assembly line, preferably, in step S2, abnormal work steps with time consumption exceeding the threshold are also dynamically marked by comparing the current time consumption with the historical average value, and the formula is:

[0035]

[0036] wherein, : the average time consumption within the current statistical period; : the long-term historical average time consumption; : the threshold coefficient.

[0037] Compared with the prior art, the algorithm of the present invention is applied to a product assembly line, which can perform automatic workstation scheduling optimization online. By eliminating bottleneck workstations, the production process is realized to be balanced, flexible and intelligent, thereby improving production efficiency and quality, and finally promoting the enterprise to achieve cost reduction and efficiency increase. Specifically:

[0038] 1. Improve the overall production efficiency

[0039] Eliminate inconsistent production beats: By balancing the workload of each workstation, avoid some workstations from slowing down the speed of the whole production line due to task overload (bottleneck), and make the production beats tend to be consistent.

[0040] Increase production capacity: After eliminating the bottleneck, the overall output rate of the production line changes from being determined by the slowest workstation to a higher level coordinated by each workstation, and the output per unit time increases significantly.

[0041] 2. Optimize resource utilization

[0042] Reduce waiting waste: Avoid downstream workstations from being idle due to upstream bottlenecks, and reduce the waiting time of equipment or personnel.

[0043] Balance the load of manpower and equipment: By reallocating tasks, avoid some workstations from being over-fatigued or resources being idle, and reduce operating costs.

[0044] 3. Reduce production costs

[0045] Shorten the production cycle: Reduce the accumulation and stagnation of semi-finished products caused by bottlenecks and accelerate the product delivery speed.

[0046] Reduce inventory backlog: After eliminating bottlenecks, the in-process inventory of semi-finished products decreases, and the costs of warehousing and capital occupation are reduced.

[0047] 4. Improve product quality

[0048] Reduce human errors: Balancing the workload at workstations can reduce the error rate of workers caused by rushing to complete tasks.

[0049] Standardize processes: The optimized workstation tasks are more ergonomic, reducing the impact of fatigue operations on quality.

[0050] 5. Data-driven continuous improvement

[0051] Real-time monitoring and feedback: The algorithm combines the Internet of Things (IoT) to collect data in real time, continuously identify potential bottlenecks and dynamically optimize.

[0052] Predictive maintenance: By analyzing the data of workstation equipment, potential failure risks are detected in advance, reducing unexpected downtimes.

[0053] 6. Improve employee satisfaction

[0054] Reasonably allocate the workload: Avoid some workstations being overloaded for a long time, reducing employee fatigue and turnover rate.

[0055] Clarify responsibilities: After optimization, the task division of labor is clearer, reducing collaboration problems caused by process chaos. Specific implementation manners

[0056] Next, the technical solution of the present invention will be clearly and completely described in conjunction with the embodiments. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art belong to the scope of protection of the present invention.

[0057] As an implementation manner of the present invention, in the specific implementation process of an automatic workstation optimization algorithm for a product assembly line provided by the present invention, it includes the following specific steps:

[0058] Step 1: Image acquisition and preprocessing

[0059] Hardware deployment:

[0060] Install the industrial camera Basler ace 2 directly above each workstation on the assembly line, and configure a ring-shaped LED fill light to eliminate shadow interference.

[0061] The camera is connected to the edge computing device NVIDIA Jetson AGX Xavier via Gigabit Ethernet to ensure that the video stream transmission rate is ≥ 60 frames per second.

[0062] Image preprocessing:

[0063] Grayscale conversion and noise reduction: Convert the captured RGB image to a grayscale image and apply Gaussian filtering (kernel size 5×5) to eliminate noise;

[0064] ROI extraction: Define the worker's operation area (Region of Interest, ROI) to reduce background interference;

[0065] Timestamp marking: Add a millisecond-level timestamp to each frame of the image with a precision error ≤ 10ms.

[0066] Step 2: Workstep recognition and time recording

[0067] Action recognition model training:

[0068] Data annotation: Collect video data containing different worksteps (such as B1 - part grasping, B2 - assembly operation), and use the LabelImg annotation tool to mark the bounding boxes of the worker's hand, tools, and parts;

[0069] Model selection: Adopt the YOLOv8s model and adjust the input resolution to 640×640 to balance precision and speed;

[0070] Training configuration: Perform transfer learning based on the COCO pre-trained model, set the initial learning rate to 0.01, batch size to 16, and the number of training epochs to 100.

[0071] Real-time workstep detection:

[0072] Input the preprocessed image into the trained YOLOv8 model, and the output detection results include:

[0073] Tool category (such as electric screwdriver, clamp);

[0074] Part type (such as L1 - bolt, L2 - bearing);

[0075] Action status (such as "grasping", "assembly completed").

[0076] Workstep switching determination: When a change in the combination of specific tools and parts is detected (for example, the disappearance of an electric screwdriver and the appearance of a clamp), it is determined that a workstep switch has occurred.

[0077] Timestamp recording:

[0078] When the workstep starts, record the current system time as start_time;

[0079] When the end of a work step is detected, record end_time and calculate the duration: duration = end_time - start_time;

[0080] Example of data format:

[0081] {

[0082] "worker_id": "G1",

[0083] "step_id": "B2",

[0084] "part_id": "L3",

[0085] "start_time": "2023-10-05 14:23:45.678",

[0086] "end_time": "2023-10-05 14:23:48.912",

[0087] "duration": 3.234

[0088] }

[0089] Step 3: Data storage and dynamic analysis

[0090] The data storage and dynamic analysis module of the present invention realizes the persistent storage and intelligent analysis of assembly line data through a structured database and a real-time computing mechanism. The specific implementation is as follows:

[0091] 1. Design of data storage structure

[0092] The work step operation record table is used to store the detailed time information of workers completing each work step, including the following core fields:

[0093] Record identifier: A unique auto-incrementing integer, used as the primary key for data retrieval and association;

[0094] Worker number: A string with a length limit of 20 characters, following the "G + number" coding rule (such as G1, G2), used to identify the specific operating worker;

[0095] Work step number: A string with a length limit of 20 characters, following the "B + number" coding rule (such as B1, B2), identifying the current operating work step;

[0096] Part number: A string with a length limit of 20 characters, following the "L + number" coding rule (such as L1, L2), identifying the type of part being operated;

[0097] Start and end time: A date and time type with millisecond precision, recording the start and end times of each work step respectively, in the format of "year-month-day hour:minute:second.millisecond";

[0098] Duration: A decimal numerical type, accurate to three decimal places, with the unit of seconds, automatically calculated by the difference between the end time and the start time.

[0099] The work station optimization record form is used to store the historical data of the dynamic adjustment plan, including the following core fields:

[0100] Optimization identifier: A unique auto-incrementing integer, used as the primary key to identify each optimization operation;

[0101] Timestamp: A standard date and time type, recording the generation time of the optimization plan;

[0102] Original and new work station numbers: A string with a length limit of 20 characters, recording the worker assignments before and after the adjustment respectively;

[0103] Associated work step number: A string with a length limit of 20 characters, identifying the work step range involved in the optimization (such as B5 - B7);

[0104] Efficiency improvement rate: A decimal numerical type, accurate to two decimal places, representing the expected percentage of efficiency improvement.

[0105] 2. Data processing flow

[0106] (1) Real-time data writing

[0107] Through the transaction batch submission mechanism, write the work step records generated by the visual recognition module into the database every fixed period (such as 5 seconds) to ensure data integrity and writing efficiency;

[0108] Perform integrity verification before data writing: Verify whether the duration is positive, and whether the worker and work step numbers conform to the coding specifications. Illegal data is transferred to the exception log table for manual review.

[0109] (2) Dynamic aggregation analysis

[0110] Periodic statistics: Automatically execute an aggregation query every hour to calculate the average duration and the standard deviation of duration fluctuations for each work step, used to identify efficiency bottlenecks;

[0111] Work step Average duration formula:

[0112]

[0113] Among them, : The average duration of the work step within the statistical period; : The total number of workers; : The number of workers who execute the work steps within the statistical period; : Workers execute the average time consumption of the work steps;

[0114] Workers execute the work steps The calculation formula for the average time consumption:

[0115]

[0116] Among them, : The time consumption of a single execution of the work step (unit: second); : Workers execute the number of times of the work steps; Assuming that the worker has no execution record, then = 0;

[0117] The formula for the standard deviation of time consumption fluctuation:

[0118]

[0119] Among them, : The standard deviation of time consumption fluctuation, reflecting the degree of data dispersion; , , : Defined as above;

[0120] Anomaly detection: By comparing the current time consumption with the historical average value, dynamically mark the abnormal work steps whose time consumption exceeds the threshold (such as 120% of the average value), trigger real-time warnings, the formula:

[0121]

[0122] Among them, : The average time consumption within the current statistical period; : The historical long-term average time consumption; : The threshold coefficient (default );

[0123] Optimization effect tracking: Associate the optimization record table with the work step operation record table, calculate the actual efficiency improvement rate, verify the effectiveness of the algorithm and support model iteration.

[0124] (3) Data fault tolerance mechanism

[0125] Local cache: When the network is interrupted, the edge device temporarily stores the operation records for up to 24 hours and automatically synchronizes them to the central database after the connection is restored;

[0126] Resume interrupted transfer: By recording the identifier of the last successfully written data, it ensures that only incremental data is synchronized after an interruption, avoiding duplication or omission.

[0127] 3. Data Security and Permission Control

[0128] Hierarchical access permissions:

[0129] Only the database insertion permission is open at the data writing end to prevent accidental operations from tampering with historical records;

[0130] The data analysis end has read-only permissions, supporting aggregation queries but prohibiting modification of the original data;

[0131] Transmission encryption: Encrypt the database connection through the Transport Layer Security Protocol (TLS 1.3) to prevent data from being intercepted or tampered with during transmission;

[0132] Storage encryption: Sensitive fields such as the mapping relationship between worker numbers and production line locations are encrypted and stored using the Advanced Encryption Standard (AES-256).

[0133] Step Four: Dynamic Workstation Optimization

[0134] Implementation of the optimization algorithm:

[0135] Construction of the time axis: The durations of each work step are accumulated in sequence to generate the total time axis. The formula is:

[0136] ( is the duration of the th work step)

[0137] Calculation of equally divided points: According to the number of workers , the total time axis is equally divided into segments, and the coordinates of the equally divided points are:

[0138] ( )

[0139] Matching of process nodes: Traverse the equally divided points and select the end time of the work step closest in distance as the process node. The matching rule is:

[0140]

[0141] Dynamic adjustment strategy:

[0142] Evaluation of the coincidence coefficient: Calculate the coincidence coefficient of each process node, where ;

[0143] Exception handling: When occurs, trigger the following operations:

[0144] Merge adjacent process segments to generate special processes ;

[0145] Allocate additional workstations (workers) and have multiple people collaborate to ensure that the time consumption of this process is reduced to T avg or less;

[0146] Instruction issuance: Send the new workstation allocation plan to the production line PLC controller via the MQTT protocol. The format example is:

[0147] {

[0148] "command": "reassign",

[0149] "workers": ["G3", "G4"],

[0150] "steps": ["B5 - B7"],

[0151] "effective_time": "2023-10-05 15:00:00"

[0152] }

[0153] Step Five: Visualization and Abnormality Monitoring

[0154] Real-time dashboard construction:

[0155] Use Grafana to connect to the MySQL data source and create the following monitoring panels:

[0156] Workstation efficiency heat map: Display the time consumption distribution according to worker ID and process step ID;

[0157] Bottleneck process warning: When the time consumption of a certain process step exceeds 20% of the historical average for 3 consecutive times, trigger a red warning;

[0158] Optimization effect comparison: Display the change in the production line balance rate before and after dynamic adjustment. The calculation formula is:

[0159]

[0160] Abnormality handling process:

[0161] Local alarm: Trigger an audible and visual alarm on the edge computing device and push a notification to the workshop supervisor's mobile phone;

[0162] Automatic logging: Record abnormal events in the database, including time, workstation, over-standard range, and handling measures;

[0163] Root cause analysis: Train a decision tree model based on historical data and automatically recommend potential improvement measures (such as tool replacement, skills training).

[0164] The present invention is not limited to the above-mentioned optimal embodiments, and any person can obtain other various forms of products under the inspiration of the present invention. However, no matter what changes are made in its shape or structure, as long as it has a technical solution identical or similar to the present application, it falls within the protection scope of the present invention.

Claims

1. An automatic optimization algorithm for workstations on a product assembly line, characterized in that Including the following steps: S1. Collect image / video data containing each working step, and obtain the working step time through working step recognition; S2. Periodically calculate the average duration and the standard deviation of duration fluctuations of each working step, specifically: S2.

1. Working Step Average Time-consuming Formula: ； Among them, : The average time consumption of the work step within the statistical period; : The total number of workers; : The number of workers who execute the work step within the statistical period; : The average time consumption of the worker executing the work step; Worker Execute the working step The calculation formula for the average time consumption is: ； Among them, : The time consumed for a single execution of a working step; : The worker Executes The number of times of the working step; Assume that the worker Has no execution record, then = 0; S2.

2. Formula for the standard deviation of duration fluctuations: ； Among them, : Standard deviation of time-consuming fluctuation, reflecting the degree of data dispersion; , , : The definition is the same as above; S3. Dynamically optimize the workstations, specifically: S3.

1. Construction of the time axis Accumulate the durations of each working step in sequence to generate the total time axis, and the formula is: ( For the consumption time of the nth working step) S3.

2. Calculation of equally divided points According to the number of workers , divide the total time axis into segments, and the coordinates of the equal division points are: ( ) S3.

3. Matching of process nodes Traverse the equally divided points, and select the end time of the working step with the closest distance as the process node. The matching rule is: ； S3.

4. Evaluation of the coincidence coefficient Calculate the coincidence coefficient of each process node , where ; When merging adjacent process segments to generate a special process and ensuring that the time taken for this process is reduced to below Tavg by adding workstations.

2. The automatic optimization algorithm for the workstations of a product assembly line according to claim 1, characterized in that It also includes: S4. Calculate the change in the production line balance rate before and after dynamic adjustment, and the formula is: 。 3. The automatic optimization algorithm for workstations of a product assembly line according to claim 1 or 2, characterized in that: In step S2, abnormal working steps with durations exceeding the threshold are also dynamically marked by comparing the current duration with the historical average value. The formula: Among them, : The average time consumption in the current statistical period; : The long-term historical average time consumption; : The threshold coefficient.

Images