Brick unit-based process intelligence improvement method and manufacturing industry engineering improvement system

Abstract

The application discloses a process intelligent improvement method based on building block units and a manufacturing industry engineering promotion system. The method comprises the following steps: constructing a three-layer knowledge base comprising a general knowledge base, a factory-specific knowledge base and an improvement case result base; adapting the model to the factory through two-stage training and continuous learning; segmenting the process video shot by the on-site personnel into at least one building block unit, wherein the building block unit is the smallest indivisible independent operation segment; extracting the action features of each building block unit and comparing them with the standard template to calculate the deviation quantization value; when the deviation exceeds the preset threshold, matching the waste type and retrieving similar historical cases to generate an improvement scheme; adopting at least one level of human-computer interaction, wherein the first level is confirmed or interactively modified by the on-site personnel and then submitted, and the second level is audited and approved by the industrial engineering department and then executed; after execution, the video is shot again to verify the effect, and the case is automatically stored in the warehouse; the stored case is used for pushing similar problems, training new people and introducing new products. The application realizes fine-grained quantization, low-cost adaptation, case reuse and controllable closed loop of process improvement.

Description

Technical Field

[0001] Intelligent manufacturing, industrial engineering (IE). Background Technology

[0002] Existing video analytics can only identify coarse actions such as "bagging," but cannot break them down into the smallest units such as "reaching out → pinching → translating → aligning → releasing," and cannot quantify minute deviations. General-purpose AI models are difficult to adapt to a factory's own BOM, SOP, and equipment parameters at low cost, and customization is expensive. Historical improvement cases lack structured storage and automatic retrieval, requiring repeated analysis of similar issues, and lacking available experience for new employee training and new product introduction. AI-generated improvement solutions lack human-computer interaction verification and professional review, making the solutions uncontrollable and untraceable. Summary of the Invention

[0003] Purpose of the invention Low-cost, fine-grained, and reusable process improvement methods and systems.

[0004] Technical solution 1. Knowledge base (at least three layers) General IE Knowledge Base: Seven Wastes, Motion Economy Principles, Time Equation Formula, Cross-Industry Case Templates (System Pre-built). Factory-specific knowledge base: BOM, SOP, standard operation videos, videos of malfunctions, equipment parameters, and historical improvement records (uploaded by the factory). Improvement Case Study Library: Automatically stores verified cases (including issues, deviations, waste types, solutions, before-and-after comparison videos, effects, and tags). 2. Building Block Unit Definition and Video Breakdown Building block unit: The smallest, indivisible, independent operation segment in a process video. Taking button packaging as an example, "bagging" can be broken down into six building block units: "reaching into the hopper → pinching the button → moving it to the bag opening → aligning it with the bag opening → releasing the button into the bag → returning the hand to its original position". Disassembly method: Use a mobile phone to shoot a video. The system automatically divides the video into multiple block units according to the start and end points of the action, and extracts the action trajectory and time of each unit. 3. Two-stage training and continuous learning Phase 1: Injecting general IE knowledge to build a search index. Phase Two: Inject factory-specific data (videos, BOMs, etc.) and perform incremental fine-tuning. Continuous learning: Automatically scans for new data at preset intervals, performs incremental training in the background, and continuously evolves the model. 4. Intelligent Analysis Process For new videos: break them down into building blocks → compare each block with a standard template and calculate the deviation → if the deviation exceeds a threshold, match the waste type → retrieve similar historical cases from the improvement case library → generate at least one improvement solution. 5. At least Level 1 human-computer interaction Level 1 (On-site personnel): Review the plan; if it's unreasonable, provide feedback for modification; if it's reasonable, submit it. It cannot be implemented directly. Level 2 (IE Department): Review the plan; it can only be implemented after approval. After execution: Record another video for verification, and the case will be automatically added to the database. 6. Closed Loop and Knowledge Distribution The case studies in the database are used for: automatic push notifications for similar issues, new employee training (directly viewing before-and-after comparisons of real-world cases), and new product introduction (searching for historical improvements to similar processes to avoid pitfalls in advance).

[0005] Beneficial effects It can detect micro-level waste by analyzing atomic actions. General knowledge combined with factory data allows for low-cost adaptation without customization. Case studies are automatically retrieved and reused, eliminating the need to reinvent the wheel. Human and machine-level oversight ensures controllable and traceable solutions. The more you use it, the more content your knowledge base accumulates, becoming your company's exclusive knowledge asset.

[0006] Detailed Implementation Method (Button Packaging) On-site personnel filmed a video on their mobile phones showing the buttons being loaded from the hopper into packaging bags. The system automatically disassembles into 6 building block units: ① Reach into the hopper, ② Pinch the button, ③ Move it to the bag opening, ④ Align it with the bag opening, ⑤ Release the button into the bag, ⑥ Return your hand to the original position. Comparing with the standard, it was found that the "alignment" time in Unit ④ exceeded the standard by 50%. The system identified "motion waste" and found a similar case in the case library: "An electronics factory adds a guide funnel". Solution: Add a guide funnel, estimated to save 2 seconds per cycle. On-site personnel select a solution and submit it. Approved by IE engineer. After adding the funnel, another video was taken to verify that the "alignment" time was reduced back to the standard value, actually saving 1.8 seconds per attempt. This case was automatically entered into the database and tagged "button packaging / alignment action". This case will be automatically pushed to similar issues in the future; new employees can view it directly during training; and it will be automatically recommended when introducing similar processes to new products. Attached Figure Description Figure 1 This is a flowchart of the method of the present invention.

Claims

1. A method for intelligent process improvement based on building block units, characterized in that: The video of the process operation is divided into at least one building block unit, and the building block unit is the smallest indivisible independent operation segment; Extract the motion features of each building block unit, compare them with the standard template, and calculate the deviation quantification value; When the deviation exceeds the preset threshold, the waste type is matched from the industrial engineering knowledge base, and similar historical cases are retrieved from the improvement case results database. At least one improvement plan is generated and output to the on-site personnel terminal.

2. The method as described in claim 1, characterized in that: The industrial engineering knowledge base is constructed through a two-stage training process: the first stage injects general industrial engineering knowledge, and the second stage injects factory-specific knowledge bases, and incremental training is performed by periodically scanning new data according to a preset cycle.

3. The method as described in claim 1, characterized in that: It also includes at least Level 1 human-computer interaction— Level 1: On-site personnel determine whether the plan is acceptable. If not, they will make interactive modifications. If accepted, the plan will be submitted (but they do not have the authority to execute it directly). Level 2: The industrial engineering department reviews the plan and approves it before it can be implemented; After execution, a video is taken again to verify the effect. Cases that pass the verification are automatically saved into the improvement case results library.

Images