Electric fitting processing intelligent manufacturing production efficiency evaluation method

Abstract

The invention belongs to the technical field of aeronautical manufacturing, and provides an electric fitting processing intelligent manufacturing production efficiency evaluation method, which comprises the following steps of: according to production efficiency attributes, refining production quality, production cycle and production cost into production efficiency sub-characteristics; taking the production efficiency sub-characteristics as metric elements to form a metric element set; respectively establishing production efficiency sub-characteristic evaluation models for production quality, production cycle and production cost by using the pigeon inspired improved RBF network; and evaluating the intelligent manufacturing production line of the airborne unit based on the evaluation model, and predicting the production efficiency after the intelligent reconstruction of the production line. According to the electric fitting processing intelligent manufacturing production efficiency evaluation method provided by the invention, the production efficiency of the production line can be predicted before the construction of the production line is successful, so that the design of the scheme is prevented from being inconsistent with the expectation.

Description

Technical Field

[0001] The present invention relates to the field of aviation manufacturing technology, and in particular to a method for evaluating production efficiency of intelligent manufacturing of electrical equipment processing. Background Art

[0002] Intelligent manufacturing is a general term for advanced manufacturing processes, systems and models that are based on new-generation information technology and run through all aspects of manufacturing activities, including design, production, management and service. It has functions such as deep information self-perception, intelligent optimization and self-decision-making, and precise control and self-execution.

[0003] The advancement of intelligent manufacturing is reflected in the digital design and intelligent full life cycle management of products, the full digital control of production processes, and the great improvement of production efficiency. Among them, the most obvious benefit, the most direct impact, and the most concerned by enterprises is the improvement of production efficiency.

[0004] Production efficiency refers to the effective output per unit time under fixed input. The most important influencing factors are production quality, production cycle, and production cost. For electrical processing in the aviation manufacturing system, production quality is the most important of the three influencing factors. Quality is the lifeline of the product, and it also affects the production cycle and production cost.

[0005] The existing production quality, production cycle, and production cost are calculated after the production line is built by measuring and counting the quality data, cycle data, and cost data generated in actual production, and then calculating the various indicators of the current production line. The method of calculating production efficiency based on the actual production data of the electric production line has a certain degree of accuracy, but it lacks foresight and evaluation capabilities. The reality is that after the production line is built and trial production is carried out, it is often found that some production efficiency indicators do not match the design of the plan. However, at this time, the production line has been put into production and major modifications cannot be made, so it can only be accepted with concessions. Summary of the invention

[0006] In view of the defects in the prior art, the present invention provides a method for evaluating the production efficiency of intelligent manufacturing of electrical equipment processing, so as to realize the evaluation of the production efficiency of the electrical equipment production line at the production line design stage.

[0007] The present invention provides a method for evaluating production efficiency of intelligent manufacturing of electrical equipment processing, comprising: According to the production efficiency attributes, production quality, production cycle, and production cost are determined to be refined into production efficiency sub-characteristics; Taking the production efficiency sub-characteristic as a metric element to form a metric element set; Using pigeon flocks to improve the RBF network, production efficiency sub-characteristic evaluation models for production quality, production cycle, and production cost were established respectively; Based on the evaluation model, the intelligent manufacturing production line of the airborne unit is evaluated, and the production efficiency after the intelligent transformation of the production line is predicted.

[0008] It can be seen from the above technical scheme that the method for evaluating the production efficiency of intelligent manufacturing of electrical equipment processing provided by the present invention can predict the production efficiency of the production line before the production line is successfully built, thereby avoiding the design of the scheme not being in line with expectations.

[0009] Optionally, the production efficiency attributes are refined for production quality, production cycle, and production cost respectively according to evaluation technical specifications, production efficiency improvement methods, and equipment / production line characteristics.

[0010] Optionally, the production efficiency characteristics are formed by refining the production quality and include production efficiency sub-characteristics: adhesion, hardness, wear resistance, weather resistance, chemical resistance, etching rate, selectivity, etching uniformity, etching deviation, etching angle, coating thickness, adhesion, corrosion resistance, surface quality, pressing thickness, bonding, dimensional stability, raw material quality, quality inspection method, process maturity and production line digitalization degree; The production efficiency characteristics are refined according to the production cycle to form production efficiency sub-characteristics including: average quality inspection time, maximum mounting speed, average mounting speed, preheating speed, cooling speed, plug-in speed, no-load speed, production line balance rate, mean time between failures, and mean time to repair; The production efficiency characteristics are formed by refining the production costs, and the production efficiency sub-characteristics include: defective product cost, raw material cost, management cost, comprehensive labor cost, fixed asset cost, and energy consumption cost.

[0011] Optionally, the production efficiency sub-characteristics include quantitative characteristics and non-quantitative characteristics, wherein the quantitative characteristics are directly used as measurement elements, and the non-quantitative characteristics are used as measurement elements after being evaluated by preset indicators.

[0012] Optionally, the use of pigeon flocks to improve the RBF network to establish evaluation models respectively includes: The weight, center position and standard deviation of the RBF network prediction model are used as the quasi-optimal pigeon position vector of the pigeon algorithm; the fitness function of the pigeon , is the jth ideal output value, is the jth actual output value, is a constant gain, and m is the number of nodes in the output layer.

[0013] Optionally, the training method of the evaluation model includes: Step 1: RBF network initialization: Initialize the RBF network structure, select n production efficiency measurement data samples as input data, initialize the number of hidden layer neurons h, and the center of RBF , Standard Deviation And the weight values , is the connection weight from the i-th hidden layer neuron to the j-th output layer neuron; Step 2: Data preprocessing: Perform principal component analysis on the production efficiency measurement metadata using the PCA method, and perform screening preprocessing of production efficiency measurement elements and measurement metadata using the fuzzy clustering method; Step 3: Calculate the standard deviation: Calculate the standard deviation of RBF , Represents the maximum distance between the center points of each selected basis function; Step 4: Calculate the connection weight between the hidden layer and the output layer: Connection weight , To input sample data; Step 5: Use the pigeon flock algorithm to optimize the RBF network parameters: optimize the weight, center position and standard deviation parameter values ​​of the RBF neural network; Step 6: Assign the weights optimized in Step 5 to the RBF neural network, and then calculate the predicted value of production efficiency; Step 7: Update the parameters of the RBF network; Step 8: Learning termination condition: Determine whether the preset indicators of the map and compass operators are less than the preset thresholds; if so, stop RBF network learning; if not, jump to Step 2; Optionally, the RBF network parameters updated in Step 7 include the center of the RBF , Standard Deviation , the connection weights between the hidden layer and the output layer , the update equation is: , is the activation function, is the learning rate; .

[0014] By adopting the above technical solution, the present application has the following beneficial effects: (1) The method for evaluating the production efficiency of intelligent manufacturing of electrical equipment processing provided by the present invention can predict the production efficiency of a production line before the production line is successfully built, thereby avoiding the design of the solution being inconsistent with expectations; (2) The improved RBF network algorithm of pigeon flock can perform nonlinear fitting, which makes the evaluation of production efficiency more accurate. Compared with other commonly used artificial neural network algorithms, the fitting results are more accurate and the fitting speed is faster; (3) By selecting multiple types of sub-features, production efficiency can be analyzed and calculated from multiple dimensions, making the final prediction results more accurate. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following is a brief introduction to the drawings required for the specific embodiments or the description of the prior art. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, the elements or parts are not necessarily drawn according to the actual scale.

[0016] Figure 1 A flow chart of a method for evaluating production efficiency of intelligent manufacturing of electrical equipment processing provided by an embodiment of the present invention is shown; Figure 2 A schematic diagram showing a design route of a method for evaluating production efficiency of intelligent manufacturing of electrical equipment processing provided by an embodiment of the present invention; Figure 3 The topological structure diagram of the RBF neural network is shown; Figure 4 A schematic diagram showing the principle of the pigeon flock algorithm. DETAILED DESCRIPTION

[0017] The following embodiments of the technical solution of the present invention are described in detail in conjunction with the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and are therefore only used as examples, and cannot be used to limit the protection scope of the present invention. It should be noted that, unless otherwise specified, the technical terms or scientific terms used in this application should have the common meanings understood by those skilled in the art to which the present invention belongs.

[0018] like Figure 1 As shown, this embodiment provides a method for evaluating production efficiency of intelligent manufacturing of electrical equipment processing, including: S1. According to the production efficiency attributes, determine the production quality, production cycle, and production cost to be refined into production efficiency sub-characteristics. Figure 2 A schematic diagram showing the design route of the production efficiency evaluation method for intelligent manufacturing of electrical equipment processing.

[0019] The production efficiency attribute layer research includes typical production processes and related quality characteristics based on SJ / T11924-2023 Printed Circuit Board Production Life Cycle Assessment Technical Specifications (Product Category Rules); Improved characteristics included in common production efficiency improvement methods; and the characteristics of the processing and testing equipment itself. In addition, the production efficiency attribute layer can also be changed and expanded according to different products, production processes and actual needs. The production efficiency attributes are refined for production quality, production cycle, and production cost according to the evaluation technical specifications, production efficiency improvement methods, and equipment / production line characteristics.

[0020] The production efficiency characteristic layer research includes the production process based on SJ / T11924-2023 Technical Specification for Life Cycle Assessment of Printed Circuit Board Production (Product Category Rules), including: cutting, coating / laminating, inner layer exposure, development, etching, film stripping, blackening / browning, typesetting, lamination, drilling, chemical copper deposition and full-board electroplating, outer layer pattern transfer, pattern electroplating, outer layer etching, solder mask and characters, gold finger electroplating, surface treatment, shape processing, testing and inspection; as well as related quality characteristics, such as matter, energy, information, materials, resources, products, risks, monitoring, measurement, analysis, evaluation, etc. Improvement characteristics in common production efficiency improvement methods: Lean Six Sigma system and production efficiency improvement characteristics in ECRS production line analysis methods; The processing characteristics of the processing equipment itself: SMT production capacity characteristics, AOI detection capability characteristics, SPI detection capability characteristics, reflow oven temperature control capability characteristics, etc.

[0021] See Table 1 for specific influencing characteristics: Table 1

[0022] After analyzing and refining the above production efficiency related characteristics, a series of sub-characteristics are further obtained, specifically, the production efficiency sub-characteristic layer.

[0023] Production efficiency characteristics The production efficiency sub-characteristics formed by the refinement of production quality include: adhesion, hardness, wear resistance, weather resistance, chemical resistance, etching rate, selectivity, etching uniformity, etching deviation, etching angle, coating thickness, adhesion, corrosion resistance, surface quality, pressing thickness, bonding, dimensional stability, raw material quality, quality inspection method, process maturity and production line digitalization degree; The production efficiency characteristics are formed by the refinement of the production cycle. The sub-characteristics of production efficiency include: average quality inspection time, maximum mounting speed, average mounting speed, preheating speed, cooling speed, plug-in speed, no-load speed, production line balance rate, mean time between failures, and mean time to repair; The production efficiency characteristics are formed by the refinement of production costs, and the sub-characteristics of production efficiency include: defective product cost, raw material cost, management cost, comprehensive labor cost, fixed asset cost, and energy consumption cost.

[0024] See Table 2 for specific sub-characteristics: Table 2

[0025] The neutron characteristics in Table 2 are explained as follows: 1. Production quality characteristics 1. Adhesion Adhesion refers to the bonding strength between the coating and the substrate. Common testing methods include stretching, peeling, and cross-cutting. 2. Hardness Hardness refers to the ability of the coating to resist scratching or cutting. Currently, there are mainly pencil hardness test method, Taber scratch resistance test method, Konig hardness test method, etc. 3. Wear resistance Wear resistance refers to the tolerance of the coating to mechanical wear or friction. Currently, there are mainly grinding wheel wear test method, Crockmeter friction test method, etc. 4. Weather resistance Weather resistance refers to the tolerance of the coating film under long-term exposure to the external natural environment, such as ultraviolet rays, humidity, high temperature, low temperature, etc. The main test methods include Xenon lamp weather resistance test, humidity and heat cycle test, salt spray test, etc. 5. Chemical resistance Chemical resistance means that the coating will not undergo chemical reactions or physical damage after contact with chemicals, and has the effect of protecting objects. The main testing methods currently include solvent wipe test, acid and alkali resistance test, etc. 6. Etching rate Etch rate refers to the rate at which material is removed during the etching process, usually expressed as the amount of thickness reduction per unit time, and the unit is usually nanometers per second (nm / s) or angstroms per second (Å / s). The thickness can be characterized by a film thickness meter, step profiler or SEM.

[0026] 7. Etching selectivity During the etching process, the resist (such as photoresist) on the upper layer of the etched material or the lower layer of the material that does not need to be etched will be etched at the same time. Therefore, there needs to be a definition between the resist and the etching material, namely the etching selectivity ratio.

[0027] Etching selectivity = rate of etched material / rate of mask material.

[0028] In micro-nano processing, etching selectivity is an important parameter that determines the removal rate of the target material relative to the mask material during the etching process. Generally speaking, the higher the etching selectivity, the faster the removal rate of the target material relative to the mask material, and the more selective the etching process. High etching selectivity is crucial for the manufacture of micro-nano structures and devices because it ensures that the desired structure is only etched in specific areas without affecting other areas.

[0029] 8. Etching uniformity Uniformity is usually used to measure the consistency of etching rate across the entire wafer. It is a parameter that measures the etching ability of the etching process across the entire wafer, or across a batch, or between batches. (Within the wafer, between wafers, and between batches).

[0030] 9. Etching deviation Changes in line width or key dimension spacing after etching. Changes in line width, pattern height, depth and other dimensions after etching. As long as chemical reactions are involved, there must be more or less side etching. Effective use of side etching can achieve some special structures.

[0031] Etching bias = Wb - Wa, Wb = line width of the mask before etching, Wa = line width of the etched material after the mask is removed.

[0032] 10. Etching Angle The etching angle usually refers to the angle between the etching direction perpendicular to the surface and the horizontal line of the surface during the etching process. The etching angle can be achieved by controlling the ratio of the etching rate.

[0033] In the etching process, the control of the etching angle is very important for preparing devices with specific shapes or structures. By adjusting the etching angle, the etching rate of the material in a specific direction can be controlled to achieve the desired shape or structure. For example, if a device or structure with a bevel needs to be prepared, it can be achieved by adjusting the etching angle.

[0034] 11. Coating thickness The most basic quality standard of electroplating process is the thickness of the coating. If the coating is too thin, it is not only easy to cause oxidation, corrosion, peeling and other problems, but also cannot achieve the expected decorative or protective effect. Therefore, the quality standard of coating thickness is a very critical indicator.

[0035] 12. Adhesion Adhesion is one of the most important quality indicators of electroplating process, which determines the bonding strength between the coating and the substrate. If the adhesion is not strong enough, it is easy to peel off and fall off, which will have a great impact on the protection and decorative effects of the product.

[0036] 13. Corrosion resistance Corrosion resistance is one of the important indicators to measure the quality of electroplating process. Good corrosion resistance can ensure that electroplated parts are not easy to rust and corrode in special environments such as moisture, acid and alkali, so that electroplated parts can maintain good appearance and performance for a longer time.

[0037] 14. Surface quality Surface quality is one of the important indicators to measure the quality of electroplating process. It refers to the performance of electroplating surface finish, flatness, etc. Good surface quality can improve the aesthetics and decorative performance of electroplated parts, reduce the friction between products and the environment, and also have better corrosion resistance.

[0038] 15. Lamination thickness Provides related electrical insulation, impedance control, and filling between inner layer circuits.

[0039] 16. Combinability Provides bonding with inner layer black (brown) and outer layer copper foil.

[0040] 17. Dimensional stability The consistency of the size changes of each inner layer plate ensures the alignment of the hole rings of each layer.

[0041] 18. Raw material quality refers to whether there are defects such as fatigue, stress, micro cracks, etc. in the raw materials. When the raw materials have defects and are not detected, they are likely to form irreparable waste, which has a great impact on quality. The quality of the fixture affects the positioning accuracy of the raw materials, and therefore affects the processing accuracy and production quality. The quality of the tool directly affects the surface accuracy and feed capacity of the processing, and is also one of the important factors affecting production quality.

[0042] 19. Quality inspection method (automatic / manual) represents the quality inspection method of the production line, whether it is online automated inspection (such as AOI visual inspection, infrared sensor inspection, ultrasonic flaw detection, three-coordinate dimension inspection, etc.), or manual inspection of semi-finished products and finished products. Quality inspection frequency (full inspection / spot inspection) represents the frequency of product inspection. Online inspection is generally full inspection, and manual inspection is mostly spot inspection, which only conducts full inspection on key features of key products. Due to the manual errors of manual inspection and the low accuracy of manual inspection tools, the accuracy of online automated inspection is better than that of manual inspection. At the same time, online automated inspection can achieve full inspection, so the quality control is stronger than manual inspection.

[0043] 20. Process Readiness Level (PRL) mainly evaluates the level of process capability from product design and development to final product realization. The object of process readiness level evaluation is technology, that is, the level of product manufacturing technology. The focus of process readiness level evaluation is the manufacturability of the product, whether the product can be mass-produced, whether the product quality and cost can meet the target requirements, and whether the product can be put on the market to meet customer needs. The maturity of the production process represents the rationality and advancement of the product's production process and procedures. Processing technology that has been verified in practice is less likely to produce waste, while taking into account production cycle and production cost considerations. The quantitative method of process maturity will be described in the next section.

[0044] 21. The degree of digitalization of the production line represents whether the various production data of the production line can be collected, transmitted, recorded and applied in a timely manner. An automated production line with a high degree of digitalization can predict changes in production efficiency through changes in indicators of various production links and equipment parameters, and can further reversely guide production from the control system. The quantitative method of the degree of digitalization of the production line will be described in the next section.

[0045] Mounting accuracy, component error rate, material rejection rate, mounting size range, mounting height, mounting pressure, temperature control accuracy, temperature control range, temperature uniformity, and plug-in accuracy are common sub-characteristics of electronic assembly equipment related to production quality. Among them, mounting accuracy, component error rate, material rejection rate, and plug-in accuracy have a particularly significant impact on production quality. The mean time between failures (MTBF) is the average time that the equipment can work normally between two failures, that is, the ratio of the total working time to the total number of failures within the life span, that is, MTBF = total working time / total number of failures. The mean repair time is the average time required to repair the equipment or production line after a failure and restore it to normal production. The two are used to evaluate the stability of the equipment, and the stability of the equipment will affect both production quality and production cycle.

[0046] 2. Production cycle characteristics Maximum placement speed, average placement speed, preheating speed, cooling speed, plug-in speed, no-load speed, line balance rate, mean time between failures, and mean time to repair are common sub-characteristics of electronic assembly equipment related to production cycle. Among them, the average placement speed and plug-in speed have a particularly significant impact on the production cycle.

[0047] The maturity of the production process not only affects the quality of the product, but also the production cycle of each product. The rationality of the production schedule is used to evaluate the generation method of the production schedule (manual / system), whether there are conflicts between multiple production schedules, and the rationality of various KPIs of the production schedule results. The quantitative method of the rationality of the production schedule will be described in the next section.

[0048] Before the raw materials are machined, they need to be clamped on the fixture; similarly, after the processing, the finished product / semi-finished product needs to be removed from the fixture. This process is generally manual, and when the product is heavy, it requires overhead cranes and other tooling to assist in the operation. Therefore, the fixture installation / disassembly time often accounts for a large proportion of the production cycle and needs to be fully taken into account.

[0049] The time it takes to load and unload materials on the production line, the automation rate of logistics transfer, the delivery distance from the warehouse to the workstation, the number of bottlenecks that hinder logistics, the automation rate of transportation between processes, and the average distance of transportation between processes are all specific sub-characteristics that affect the logistics operation time. Shortening the logistics cycle is also crucial to the optimization of the entire production cycle.

[0050] Line Balance is a technical means and method to analyze the load of all processes in the production line and adjust the load distribution between processes to achieve capacity balance (operation time is as close as possible) for each process, ultimately eliminating various waiting waste phenomena and improving the overall efficiency of the production line. This method of improving the capacity between processes to balance them is also called "bottleneck improvement". This sub-characteristic can be quantified and evaluated using the line balance rate. See the next section for a detailed description.

[0051] Spindle speed, feed speed, acceleration, coordinate travel, tool magazine capacity, tool change time, etc. are important indicators that affect the processing speed and production cycle of CNC machine tools, and have a direct impact on the production cycle.

[0052] 3. Production cost characteristics The production equipment sub-characteristics include various cost-related characteristic factors. The cost of quality inspection equipment, the cost of repairing and handling defective products are the costs brought by the quality management system. The comprehensive labor cost includes the labor costs of production personnel, quality inspection personnel, management personnel, technical personnel and other production-related personnel.

[0053] The average production cycle of a product affects the production cost. Raw material cost, logistics cost, management cost, fixed asset cost and energy consumption cost are all direct factors affecting the production cost.

[0054] S2. Use production efficiency sub-characteristics as measurement elements to form a measurement element set.

[0055] The production efficiency sub-characteristics determined in step S1 include quantitative characteristics and non-quantitative characteristics, wherein the quantitative characteristics are directly used as measurement elements, and the non-quantitative characteristics are used as measurement elements after being evaluated by preset indicators. The sub-characteristics are explained in detail as follows: 1. Process Maturity Level Process maturity level is a quantitative expression of process maturity. Process maturity is interrelated with technology maturity, and both are intended to identify related risks. Immature standard technology and unstable design will affect the manufacturing process of the product. If the product technology is immature or the design is unstable, process maturity is out of the question. Therefore, the definition of each process maturity level includes the corresponding technology maturity level, that is, when the process maturity reaches a certain level, its technology maturity must reach the corresponding technology maturity level, which is defined as: PRL1: Discover the fundamental principles of the process; PRL2: Confirm the manufacturing process concept; PRL3: Confirm the process plan and verify the feasibility of the process plan; PRL4: Ability to produce prototypes, and the prototypes have been verified in the laboratory environment; PRL5: Capable of producing samples, installing them on subsystem-level prototypes, and passing verification in simulated usage environments; PRL6: Capable of producing pilot production parts and installing them into system-level equipment, and verified in typical use environments; PRL7: Ability to produce small batches and verify production line capabilities; PRL8: Capable of mass production on production lines; PRL9: Possessing efficient, fast and flexible production capabilities.

[0056] 2. Degree of digitalization of production lines This article uses the following conditions to judge the degree of digitalization of a production line. The more conditions are met, the higher the degree of digitalization.

[0057] 1) Intelligent equipment data collection capabilities: Complete intelligent manufacturing equipment should have complete archival information, including product number, product description, product status, product timestamp, etc. Key data in the production process can be collected through various sensors, barcode scanners and other equipment.

[0058] 2) Smart device interconnection capability: With informatization as the foundation, by connecting smart devices to the communication network, the collected information can be uploaded to central management systems such as MES.

[0059] 3) Upper-level system management and control capabilities: Establish an MES manufacturing execution system and use modules such as production scheduling management, process execution and management, process control management, scheduling management, quality management, and equipment management to manage and display the entire production and manufacturing process at all times.

[0060] 4) Intelligence of the entire production line: After achieving automation, CNC machine tools, robots and other production equipment use management software such as ERP and MES as the central management system, visual cameras, RFID tags, barcode scanners, barcodes, sensors and other components, NC numerical control systems or PLCs as control units, and field bus PROFIBUS, industrial Ethernet PROFINET, MODUBUS and other communication technologies as transmission networks to form a complete intelligent production line layout.

[0061] 3. Reasonableness of production scheduling The rationality of production scheduling has the greatest impact on the production cycle, and also indirectly affects the production cost. This paper evaluates the rationality of production scheduling with the following indicators.

[0062] 1) Occupancy capacity, which indicates the time that a work order is scheduled to occupy on the equipment within the planned scope; 2) Equipment costs: the equipment usage costs are calculated based on the equipment standard rate; 3) Idle time, which means the working time when there are no work orders on the equipment within the planned scope; 4) Idle cost: the idle cost of the equipment is calculated based on the equipment standard rate; 5) Process delay, the sum of the delay time of all processes; 6) Line change cost, including static and dynamic line change time of all processes; 7) Labor costs, calculated based on the standard rate for workers; 8) Work order delay, the sum of all work order delay times; 9) Waiting time, the sum of transition time between processes of each work order; 10) Work order cycle: the sum of the time between the start time of the first process and the end time of the last process of each work order.

[0063] The more indicators a production schedule covers and optimizes when it is formulated, the more reasonable it is.

[0064] 4. Production line balance rate Production line balancing is to average all production processes and adjust the load of each operation to make the operation time as close as possible. It is the most important issue that must be considered in production process design and operation standardization. The purpose of production line process balancing is to: reduce the preparation time between processes and shorten the production cycle; reduce the work-in-progress between processes and reduce the site occupation; shorten the product assembly time, increase the output per unit time, and reduce the production cost; improve the overall production efficiency of the factory and reduce various waste phenomena on site.

[0065] The production line balance rate is a metric of the above sub-characteristics, and its calculation formula is: Balance rate = (total operation time of each process / (number of processes*bottleneck process time))*100%=(∑ti / (number of processes*CT))*100%.

[0066] S3. Use pigeon flocks to improve the RBF network to establish production efficiency sub-characteristic evaluation models for production quality, production cycle, and production cost.

[0067] Specifically in Figure 2 The measurement method layer shown uses the above measurement elements to evaluate production efficiency. The evaluation models are established using the pigeon group to improve the RBF network; the historical production data of the intelligent manufacturing production line of the airborne demonstration verification unit are then used to train the three evaluation models and calculate the weight indicators corresponding to each measurement element.

[0068] RBF neural network is a three-layer feedforward neural network that can approximate any nonlinear function with arbitrary precision and has global approximation capability. Its specific principle and topological structure are as follows: Figure 3 shown.

[0069] The RBF neural network consists of three layers: input layer, single hidden layer and output layer. It uses radial basis function neural network as the activation function of hidden layer neurons. The transformation from input layer to hidden layer space is nonlinear, while the transformation from hidden layer space to output layer is linear.

[0070] This embodiment uses the pigeon flock algorithm to update and optimize the parameters of the RBF neural network. The pigeon flock optimization algorithm is a cluster intelligence optimization algorithm that learns and simulates the autonomous homing behavior of pigeons. Compared with bionic optimization algorithms such as particle swarms that simulate birds, the pigeon flock algorithm has a faster convergence speed and alleviates the local optimal problem to a certain extent. It can be regarded as an improvement on the particle swarm algorithm.

[0071] The pigeon flock algorithm contains two bionic mathematical tools: map and compass operator and landmark operator. It is a high-level summary of using the sun, earth's magnetic field and landmarks to find nests. Its principle is as follows Figure 4 shown.

[0072] The map and compass operators use the operator form to update the speed and position of each pigeon. The mathematical expression of the formula is: , In the formula, and Respectively represent the position and speed of the kth pigeon; is the updated algebra; vector , ; r represents the map and compass factor, and its value range is 0~1; rand() function represents a random number from 0 to 1; yes After the iteration cycle is completed, the global optimal position of the pigeon flock is obtained by comparing the positions of all pigeons. When the number of cycles of the update reaches the required maximum number of iterations, or the preset indicator is less than a certain threshold, the update ends and the landmark operator is used to continue the calculation.

[0073] Based on the above method, this paper proposes to use the pigeon flock optimization algorithm to improve the weight, center position and standard deviation of the RBF network. The weight, center position and standard deviation value of the RBF network prediction model are used as the quasi-optimized pigeon flock position vector of the pigeon flock algorithm; the fitness function of the pigeon flock , is the jth ideal output value, is the jth actual output value, is a constant gain, and m is the number of nodes in the output layer.

[0074] Evaluation model training methods, including: Step 1: RBF network initialization: Initialize the RBF network structure, select n production efficiency measurement data samples as input data, initialize the number of hidden layer neurons h, and the center of RBF , Standard Deviation And the weight values , is the connection weight from the i-th hidden layer neuron to the j-th output layer neuron; Step 2: Data preprocessing: Perform principal component analysis on the production efficiency measurement metadata using the PCA method, and perform screening preprocessing of production efficiency measurement elements and measurement metadata using the fuzzy clustering method; Step 3: Calculate the standard deviation: Calculate the standard deviation of RBF , Represents the maximum distance between the center points of each selected basis function; Step 4: Calculate the connection weight between the hidden layer and the output layer: Connection weight , To input sample data; Step 5: Use the pigeon flock algorithm to optimize the RBF network parameters: optimize the weight, center position and standard deviation parameter values ​​of the RBF neural network; Step 6: Assign the weights optimized in Step 5 to the RBF neural network, and then calculate the predicted value of production efficiency; Step 7: Update the parameters of the RBF network; the updated RBF network parameters include the center of the RBF , Standard Deviation , the connection weights between the hidden layer and the output layer , the update equation is: , is the activation function, is the learning rate; ; Step 8: Learning termination condition: Determine whether the preset indicators of the map and compass operators are less than the preset thresholds; if so, stop RBF network learning; if not, jump to Step 2.

[0075] Through the above learning and optimization steps, the optimal neural network computing model can be trained.

[0076] S4. Evaluate the intelligent manufacturing production line of the airborne unit based on the evaluation model and predict the production efficiency after the intelligent transformation of the production line. Figure 2 At the production efficiency evaluation solution level, the measurement method is applied to evaluate the intelligent manufacturing production line of the airborne unit. Conduct on-site surveys of the application verification unit to collect the required production data and equipment parameters. Predict the production efficiency after the intelligent transformation of the production line and predict whether the various indicators meet the requirements of the production line transformation plan.

[0077] The above embodiments are only used to introduce the technical solutions of the present application in detail, but the description of the above embodiments is only used to help understand the methods of the embodiments of the present invention and should not be understood as limiting the embodiments of the present invention. Any changes or substitutions that can be easily thought of by those skilled in the art should be included in the protection scope of the embodiments of the present invention.

Claims

1. A method for evaluating production efficiency of intelligent manufacturing of electrical equipment processing, characterized in that: include: According to the production efficiency attributes, production quality, production cycle, and production cost are determined to be refined into production efficiency sub-characteristics; Taking the production efficiency sub-characteristic as a metric element to form a metric element set; Using pigeon flocks to improve the RBF network, production efficiency sub-characteristic evaluation models for production quality, production cycle, and production cost were established respectively; Based on the evaluation model, the intelligent manufacturing production line of the airborne unit is evaluated, and the production efficiency after the intelligent transformation of the production line is predicted.

2. The method according to claim 1, characterized in that The production efficiency attributes are refined for production quality, production cycle and production cost according to evaluation technical specifications, production efficiency improvement methods and equipment / production line characteristics.

3. The method according to claim 2, characterized in that The production efficiency characteristics are formed by the production quality refinement, including: adhesion, hardness, wear resistance, weather resistance, chemical resistance, etching rate, selectivity, etching uniformity, etching deviation, etching angle, coating thickness, adhesion, corrosion resistance, surface quality, pressing thickness, bonding, dimensional stability, raw material quality, quality inspection method, process maturity and production line digitalization degree; The production efficiency characteristics are refined according to the production cycle to form production efficiency sub-characteristics including: average quality inspection time, maximum mounting speed, average mounting speed, preheating speed, cooling speed, plug-in speed, no-load speed, production line balance rate, mean time between failures, and mean time to repair; The production efficiency characteristics are formed by refining the production costs, and the production efficiency sub-characteristics include: defective product cost, raw material cost, management cost, comprehensive labor cost, fixed asset cost, and energy consumption cost.

4. The method according to claim 3, characterized in that The production efficiency sub-characteristics include quantitative characteristics and non-quantitative characteristics, wherein the quantitative characteristics are directly used as measurement elements, and the non-quantitative characteristics are used as measurement elements after being evaluated by preset indicators.

5. The method according to claim 4, characterized in that The evaluation models are established by using the pigeon group to improve the RBF network, including: The weight, center position and standard deviation of the RBF network prediction model are used as the quasi-optimal pigeon position vector of the pigeon algorithm; the fitness function of the pigeon , is the jth ideal output value, is the jth actual output value, is a constant gain, and m is the number of nodes in the output layer.

6. The method according to claim 2, characterized in that The training method of the evaluation model comprises: Step 1: RBF network initialization: Initialize the RBF network structure, select n production efficiency measurement data samples as input data, initialize the number of hidden layer neurons h, and the center of RBF , Standard Deviation And the weight values , is the connection weight from the i-th hidden layer neuron to the j-th output layer neuron; Step 2: Data preprocessing: Perform principal component analysis on the production efficiency measurement metadata using the PCA method, and perform screening preprocessing of production efficiency measurement elements and measurement metadata using the fuzzy clustering method; Step 3: Calculate the standard deviation: Calculate the standard deviation of RBF , Represents the maximum distance between the center points of each selected basis function; Step 4: Calculate the connection weight between the hidden layer and the output layer: Connection weight , To input sample data; Step 5: Use the pigeon flock algorithm to optimize the RBF network parameters: optimize the weight, center position and standard deviation parameter values ​​of the RBF neural network; Step 6: Assign the weights optimized in Step 5 to the RBF neural network, and then calculate the predicted value of production efficiency; Step 7: Update the parameters of the RBF network; Step 8: Learning termination condition: Determine whether the preset indicators of the map and compass operators are less than the preset thresholds; if so, stop RBF network learning; if not, jump to Step 2.

7. The method according to claim 6, characterized in that The RBF network parameters updated in Step 7 include the center of the RBF , Standard Deviation , the connection weights between the hidden layer and the output layer , the update equation is: , is the activation function, is the learning rate; 。

Images