Capacitor production control method based on AI

Abstract

The invention belongs to the technical field of electronic component production, and particularly relates to an AI-based capacitor production control method, which comprises a data acquisition module, a data processing module, an index generation module, an intelligent analysis module, an intelligent decision module and an intelligent optimization module. Millisecond-level dynamic acquisition of process parameters such as temperature and tension is realized through the distributed sensor network, the hysteresis problem of traditional manual sampling inspection is solved, 12 + indexes such as personnel safety, quality and cost are fused, the defect prediction accuracy is improved by adopting the AI algorithm, and the defect prediction accuracy is improved. The sintering temperature and the winding tension are automatically adjusted through an AI model according to the electrical performance quality index, the medium thickness uniformity is improved, the equipment load is dynamically adjusted based on the production cost energy efficiency index, the energy consumption per unit yield is reduced, the foil utilization rate is improved, and finally a full-closed-loop intelligent control system of perception-decision-optimization is formed.

Description

Technical Field

[0001] The present invention belongs to the field of electronic component production, and particularly relates to capacitor production. Specifically disclosed is an AI-based capacitor production control method. Background Art

[0002] Capacitors play a decisive role in the performance of basic components in the electronics industry, and must simultaneously ensure product quality, production safety, and economic benefits.

[0003] Existing capacitor production control mainly uses a semi-automated model that combines manual experience with automated equipment. The system uses fixed values ​​to control basic process parameters such as temperature and tension. Capacity deviations or appearance defects are detected through regular manual inspections. Quality issues are often handled with a lag. Energy efficiency management sets static thresholds based on historical data, and safety protection relies primarily on passive alarms. This has the following drawbacks:

[0004] Process lag defects: Traditional production lines rely on manual sampling, with large delays in parameter adjustment, and are unable to capture instantaneous anomalies such as sintering temperature fluctuations.

[0005] Limitations of single-point analysis: Existing methods only focus on failure modes and lack collaborative evaluation of 12+ indicators such as personnel safety and cost, resulting in limited defect prediction accuracy.

[0006] Parameter curing problem: Parameters such as winding tension use fixed thresholds, and medium thickness deviation leads to a high delamination defect rate.

[0007] Energy waste: The motor runs at peak load, the unit energy consumption exceeds the standard, and the foil cutting utilization rate is insufficient.

[0008] Therefore, full-process real-time monitoring, multi-dimensional intelligent decision-making, dynamic process optimization and energy efficiency closed-loop control methods are needed to solve the above problems. Summary of the Invention

[0009] In view of this, the present invention proposes an AI-based capacitor production control method. This method collects multi-source data of the production line in real time through a distributed sensor network, extracts core indicators such as personnel safety, process quality, and cost through a processing module, and generates quantitative evaluation parameters such as a safety index and a comprehensive production index. After the intelligent analysis module grades the index, the decision module executes a differentiated control strategy. At the same time, the optimization module dynamically adjusts the process parameters and realizes self-correction of defects, ultimately forming a fully closed-loop intelligent control system of "perception-decision-making-optimization".

[0010] The purpose of the present invention can be achieved by the following technical solution: A capacitor production control method based on AI, specifically comprising the following steps:

[0011] S1. Install a distributed sensor network to collect multi-source data of the production line in real time through the data acquisition module;

[0012] S2. Process the data through the data processing module to obtain indicators related to personnel operation safety, production process, product quality and cost control;

[0013] S3. Calculate the personnel operation safety index, capacitor production comprehensive index, electrical performance quality index and production cost energy efficiency index through the index generation module based on the relevant indicators obtained by the data processing module;

[0014] S4. Analyze each index through the intelligent analysis module and classify each index into grades;

[0015] S5. Based on the analysis results, the intelligent decision-making module is used to control production and implement response measures corresponding to different levels;

[0016] S6. Dynamically optimize process parameters and energy efficiency costs through intelligent optimization modules and perform defect protection and self-correction.

[0017] Combining all the above technical solutions, the present invention has the following positive effects:

[0018] 1. The present invention realizes millisecond-level dynamic acquisition of process parameters such as temperature and tension through a distributed sensor network, solving the hysteresis problem of traditional manual sampling.

[0019] 2. This invention integrates 12+ indicators such as personnel safety, quality, and cost, and uses AI algorithms to improve the accuracy of defect prediction.

[0020] 3. The AI ​​model of the present invention automatically adjusts the sintering temperature and winding tension according to the electrical performance quality index, thereby improving the uniformity of the dielectric thickness.

[0021] 4. The present invention dynamically adjusts the equipment load based on the production cost energy efficiency index, reduces unit energy consumption, and improves foil utilization. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0023] Attachment Figure 1 This is a system block diagram of the present invention.

[0024] Attachment Figure 2 Flowchart of the present invention. DETAILED DESCRIPTION

[0025] The technical solutions of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0026] See also Figure 1 As shown, the present invention proposes an AI-based capacitor production control method, which includes a data acquisition module, a data processing module, an index generation module, an intelligent analysis module, an intelligent decision-making module and an intelligent optimization module.

[0027] The specific implementation steps of the present invention include the following steps:

[0028] S1. Install a distributed sensor network to collect multi-source data of the production line in real time through the data acquisition module.

[0029] It should be specified that the multi-source data include data used for multiple indicators such as computing skill matching degree, operating standard degree, safety protection compliance rate, dielectric thickness compliance rate, winding tension fluctuation rate, sintering temperature stability, capacity accuracy, loss tangent, insulation resistance, voltage resistance, single unit energy consumption, production cycle and raw material utilization rate.

[0030] S2. Process the data through the data processing module to obtain indicators related to personnel operation safety, production process, product quality and cost control.

[0031] The personnel operation-related indicators include skill matching degree, operation standardization degree, and safety protection compliance rate.

[0032] The production process related indicators include dielectric thickness compliance rate, winding tension fluctuation rate and sintering temperature stability.

[0033] The product quality related indicators include capacity accuracy, loss tangent, insulation resistance and voltage withstand performance.

[0034] The cost control related indicators include unit capacity consumption, production cycle and raw material utilization rate.

[0035] S3. Calculate the personnel operation safety index, capacitor production comprehensive index, electrical performance quality index and production cost energy efficiency index through the index generation module based on the relevant indicators obtained by the data processing module;

[0036] It should be specified that the personnel operation safety index is specifically:

[0037] R=ω1*G+ω2*A+ω3*J;

[0038] R is the personnel operation safety index, which quantifies the core risk dimension of personnel operations and realizes scientific and dynamic safety management.

[0039] It should be explained that G is the score of the degree of operation standardization, where the degree of operation standardization = (number of operations that comply with technical regulations / total number of operations) * 100%, which is determined by breaking down the key steps in the operation standard book;

[0040] The degree of operational standardization is directly related to the quality of implementation of operational procedures, and has the highest risk weight. The value range of weight δ1 is between 0.45 and 0.5.

[0041] It should be explained that A is the safety protection compliance rate score, where safety protection compliance rate = (number of inspections for wearing protective equipment / total number of inspections) * 100%. It clarifies the required protective equipment and judges its completeness and correctness.

[0042] Failure of labor protection can lead to serious accidents, and the value range of δ2 is between 0.3-0.35.

[0043] It should be explained that J is the skill matching score, where skill matching = (number of people who have passed job skill certification / total number of people on the job) * 100. Job skill certification requires passing theoretical examinations and practical assessments, and retesting every six months.

[0044] The degree of skill matching is the basic ability guarantee and affects long-term stability. The value range of δ3 is between 0.2-0.25.

[0045] It should be noted that the scores of the three indicators are calculated through standardized values, where the standardized value score = (actual measurement value / benchmark value) * 100, and the benchmark value is the compliance requirement set by the enterprise.

[0046] It should be noted that the capacitor production comprehensive index is:

[0047] S=υ1*H+υ2*T+υ3*F;

[0048] Where S is the comprehensive index of capacitor production, which quantifies the stability of key process parameters and provides a systematic quality evaluation for the production process.

[0049] It should be explained that H is the dielectric thickness compliance score, where thickness deviation = |measured thickness - designed thickness|. The measured thickness is obtained by sampling at least n sheets from a single batch, and the average value is obtained by measuring n points on each sheet. The designed thickness is determined by reverse calculation using the capacitance formula: capacitance = dielectric constant * electrode area * number of layers / designed thickness. The dielectric thickness compliance rate = 100% * number of samples with deviations less than the threshold / total number of samples.

[0050] The dielectric thickness compliance rate directly determines the capacity accuracy and voltage resistance performance, and the value range of weight υ1 is between 0.4-0.45.

[0051] It should be explained that T is the sintering temperature stability score, where the real-time temperature deviation = |measured temperature - set temperature|. The curve stability is determined by the maximum deviation at each time point that is less than the threshold. If the deviation of all temperature measurement points throughout the process is less than the threshold, it is considered to meet the standard.

[0052] The sintering temperature stability affects the structural integrity and reliability of the ceramic body, and the value range of the weight υ2 is between 0.3-0.35.

[0053] It should be explained that F is the winding tension fluctuation score, where instantaneous tension fluctuation = |real-time tension - standard tension| * 100%, and overall fluctuation = 100% * (maximum tension - minimum tension) / standard tension. The fluctuation of each winding mandrel throughout the entire process must be less than the threshold, otherwise the correction mechanism is triggered.

[0054] The winding tension fluctuation rate is related to the contact quality between the film capacitor layers, and the weight υ3 ranges from 0.2 to 0.25.

[0055] It should be noted that the individual scores of the three indicators = (1-|actual value-target value| / upper limit of allowable deviation)*100, where the target value is the center value of the process standard and the upper limit of allowable deviation is the process tolerance.

[0056] It should be noted that the electrical performance quality index is specifically:

[0057] Z=δ1*L+δ2*tanθ+δ3*E+δ4*N;

[0058] Where Z is the electrical performance quality index, which quantifies key electrical parameters to achieve visual assessment of capacitor quality, risk grading and supply chain management.

[0059] It should be explained that L is the capacity accuracy score, which is converted by the capacity deviation rate. Full marks are awarded within a%, and 10 points are deducted for every 1% exceeding the limit.

[0060] Capacitance accuracy reflects the deviation between the energy storage capacity of the capacitor and the design value, which directly affects the accuracy of circuit functions such as filtering and tuning;

[0061] Capacity deviation rate = |(measured capacity - nominal capacity) / nominal capacity| * 100%, where the nominal capacity is the rated capacitance marked on the capacitor, and the measured capacity is the actual value measured in the laboratory using an LCR bridge under standard conditions;

[0062] Capacity deviation directly affects circuit performance, and the value range of weight δ1 is between 0.35-0.4.

[0063] It should be explained that tanθ is the loss tangent score, which is converted by the difference between the measured value and the standard value. If it is within b%, full marks will be awarded, and 5 points will be deducted for every 0.1% exceeding the standard value.

[0064] The loss tangent tanδ represents the dielectric loss efficiency. In high-frequency scenarios, excessively high tanδ can lead to thermal failure.

[0065] Wherein, loss tangent = equivalent series resistance / capacitive reactance, capacitive reactance = 1 / (2π * test frequency * measured capacitance);

[0066] The loss tangent is a key indicator of high-frequency loss, and the value range of the weight δ2 is between 0.25 and 0.3.

[0067] It should be explained that E is the insulation resistance score. If the measured value meets E≥c, full marks will be awarded. If it does not meet the standard, points will be deducted proportionally.

[0068] Insulation resistance determines the leakage current level, which is related to charge retention and system energy consumption;

[0069] Insulation resistance = test voltage / leakage current. The test condition is to read the leakage current value after applying the rated DC voltage for 1 minute. The test voltage is 80% to 100% of the rated voltage.

[0070] Insulation resistance is the core of leakage current control, and the value range of weight δ3 is between 0.15-0.2.

[0071] It should be explained that N is the voltage withstand performance score. If the measured breakdown voltage is greater than d times the rated voltage, the full score will be awarded. If it is less than d times, the linear deduction will be made.

[0072] The withstand voltage performance verifies the insulation strength of the dielectric layer to avoid short circuit accidents caused by breakdown;

[0073] The withstand voltage performance = measured breakdown voltage / rated voltage. The test method is the step-up method, increasing the voltage at a rate of 3% to 5% per second until breakdown.

[0074] The withstand voltage performance is the core of the dielectric layer reliability, and the value range of the weight δ4 is between 0.15 and 0.2.

[0075] It should be noted that the full score is 100 points, and the minimum score is 0 when a single item exceeds the standard.

[0076] It should be noted that the production cost energy efficiency index is specifically:

[0077] C=ε1*D+ε2*Q+ε3*Y;

[0078] Where C is the production cost energy efficiency index, which quantifies the raw material utilization rate, unit capacity consumption and production cycle deviation, accurately locates resource waste points and drives down unit costs.

[0079] It should be explained that D is the unit energy consumption score, which is an inverse indicator. The lower the value, the better. It reflects energy efficiency and is strongly correlated with equipment performance.

[0080] Energy consumption score = (1-(actual value-baseline value) / baseline value)*100. If the energy consumption score is negative, it is 0. The base value is the industry advanced level.

[0081] Unit product energy consumption = total energy / product output, where total energy includes raw material consumption, operating activity energy consumption and heat loss, and product output is the actual output of qualified products per unit time;

[0082] Energy costs account for 15% to 25% of total production costs, so the weight ε1 is between 0.3 and 0.35.

[0083] It should be explained that Q is the production cycle score. The production cycle is a reverse indicator. The lower the value, the better. It measures time efficiency and relies on process optimization.

[0084] Cycle score = (baseline cycle / actual cycle) * 100, where the base cycle is the theoretical minimum value of the process;

[0085] Production cycle = processing time + waiting time + inspection time + transportation time, where processing time refers to the actual processing time of the product on the equipment, and non-processing time refers to the time spent on material queuing, quality inspection, and flow between processes;

[0086] Time efficiency affects equipment depreciation and labor costs, so the weight ε2 is between 0.25 and 0.3.

[0087] It should be explained that Y is the raw material utilization score. Raw material utilization is a positive indicator. The higher the value, the better. It represents the material conversion rate and is dominated by the process level.

[0088] Utilization score = (actual utilization / target utilization) * 100, where the target utilization is the designed maximum value;

[0089] There are two ways to calculate the raw material utilization rate. According to the net weight ratio method, the utilization rate = 100% * net weight of materials in qualified products / total material consumption; according to the output method, the utilization rate = 100% * amount of materials contained in qualified products / total material input;

[0090] Raw materials account for 50% to 70% of production costs, so the weight ε3 is between 0.4 and 0.45.

[0091] S4. Analyze each index through the intelligent analysis module and classify each index into levels.

[0092] It should be noted that the classification of personnel operation safety index is as follows:

[0093] When R>r1, it is excellent;

[0094] When r2≤R<r1, it is good;

[0095] When r3≤R<r2, it is qualified;

[0096] When R<r3, it is high risk.

[0097] It should be explained that the value range of r1 is 89-90, the value range of r2 is 79-80, and the value range of r3 is 69-70.

[0098] It should be noted that the classification of the comprehensive index of capacitor production is as follows:

[0099] When S>s1, it is excellent;

[0100] When s2≤S<s1, it is good;

[0101] When s3≤S<s2, it is qualified;

[0102] When S<s3, it is high risk.

[0103] It should be explained that the value range of s1 is 89-90, the value range of s2 is 79-80, and the value range of s3 is 69-70.

[0104] It should be noted that the electrical performance quality index is graded as follows:

[0105] When Z>z1, it is excellent;

[0106] When z2≤Z<z1, it is good;

[0107] When z3≤Z<z2, it is qualified;

[0108] When Z<z3, it is high risk.

[0109] It should be explained that the value range of z1 is 89-90, the value range of z2 is 79-80, and the value range of z3 is 69-70.

[0110] It should be noted that the production cost energy efficiency index is classified as follows:

[0111] When C>c1, it is excellent;

[0112] When c2≤C<c1, it is good;

[0113] When c3≤C<c2, it is qualified;

[0114] When C<c3, it is high risk.

[0115] It should be explained that the value range of c1 is 95-96, the value range of c2 is 89-90, and the value range of c3 is 79-80.

[0116] S5. Based on the analysis results, the intelligent decision-making module is used to control production and implement response measures corresponding to different levels.

[0117] It should be noted that the specific implementation measures for the personnel operation safety index grading are as follows:

[0118] When the rating is excellent, an AI video analysis system is deployed to monitor operational compliance in real time, such as the accuracy of identifying the wearing of labor protection equipment.

[0119] Select 3-5 standardized operation cases each month and create VR training modules for all employees to learn;

[0120] Prioritize the allocation of budget for automation transformation, such as increasing the robot replacement rate for high-risk processes.

[0121] When the grade is good, courses are customized according to the type of defect. For example, those who violate safety protection regulations must complete 8 hours of immersive accident simulation training, and one outstanding employee will teach multiple weak employees.

[0122] Two-person mutual inspection links are added to key processes, such as full mutual inspection coverage of capacitor winding processes;

[0123] Some operation videos are randomly selected every week for expert review.

[0124] When the level is qualified, high-risk operation permissions are restricted, such as prohibiting independent operation of high-voltage test equipment;

[0125] Mandatory wearing of smart bracelets to monitor physiological status, and automatic suspension of work if heart rate is abnormal;

[0126] Part of the monthly safety bonus will be deducted and used for the rectification fund, and the annual promotion qualification will be frozen until the standards are met continuously.

[0127] If the risk level is high, work will be stopped immediately and 72 hours of off-site training will be initiated, including VR experience of accident cases.

[0128] Re-certify job qualifications, and improve the passing score through theory + practice;

[0129] Trace back accountability, check all operation records within 3 months to trace the responsibility chain, and jointly assess the performance of direct superiors.

[0130] It should be noted that the implementation measures for the comprehensive index classification of capacitor production are as follows:

[0131] When the grade is excellent, the frequency of daily spot checks on the dielectric thickness is reduced, an AI prediction model is used to dynamically compensate for fluctuations in the viscosity of the tape casting slurry, and the sintering temperature allows for a looser temperature control accuracy, reducing energy consumption.

[0132] Equipment maintenance extends the predictive maintenance cycle and gives priority to trialing the new thermal expansion bolt die head.

[0133] When the grade is good, the winding tension sensor is calibrated every 2 hours, buffer dampers are installed on machines with out-of-tolerance conditions, infrared thermal imaging cameras are added to assist in monitoring the sintering curve, and the heating rate is adjusted;

[0134] Carry out special training on dielectric material characteristics to improve the ability to identify anomalies.

[0135] When the grade is qualified, the medium thickness switches to full inspection mode, using an interferometer instead of a micrometer to improve accuracy. The winding process is forced to stop to replace worn guide rollers, and automatic correction is triggered when the tension fluctuation exceeds the standard.

[0136] Suspend the entry of problematic batches of raw materials into the warehouse and trace the supplier's batch data.

[0137] When the risk level is high, the furnace is emptied and purged during sintering. Production can only resume after the oxygen content is verified. The entire process is traceable, and production data is frozen for 72 hours. X-ray tomography is used to locate defects in the dielectric layer.

[0138] Initiate root cause analysis of quality incidents and submit improvement reports within 48 hours.

[0139] It should be noted that the implementation measures for the electrical performance quality index grading are as follows:

[0140] When the grade is excellent, the fully automated release system automatically binds the green mark and uploads it to the customer's quality traceability platform, giving priority to automotive electronics / aerospace-grade orders.

[0141] Benchmark data is archived and process parameter combinations are recorded in the knowledge base as training cases for new employees. Excellent batch analysis reports are generated every month, including related parameters such as ambient temperature and humidity, equipment vibration values, etc.

[0142] If the grade is good, enhanced testing is carried out, including temperature drift test, testing the capacity change rate in 10°C steps between -55°C and 125°C, and DC bias test, applying a load of 120% of the rated voltage for 2 hours and then retesting the capacity attenuation.

[0143] Fine-tune the process, calibrate the slurry viscosity in the casting process to ±2%, and extend the holding time of the sintering process by 5 to 8 minutes.

[0144] When the grade is qualified, it is only applicable to consumer electronics orders, and customers are clearly informed that it is not suitable for high-frequency scenarios. A yellow warning label is added to the outer packaging, and the product is stored in a segregated warehouse.

[0145] Launch the 8D reporting process, covering four dimensions: human operation records, equipment inspection sheets, incoming material inspection sheets, and work instruction versions;

[0146] Focus on checking the CV value of the dielectric layer thickness and the burrs on the electrode edge.

[0147] When the risk level is high, a three-level interception mechanism is implemented. The first level automatically triggers an audible and visual alarm to lock materials from the same batch. The second level completes failure analysis within 48 hours, uses X-ray inspection for delamination and thermal imaging to locate hot spots. Waste materials must be signed by the quality, process, and equipment departments.

[0148] Revise the control chart, tighten the upper limit of sintering temperature fluctuation, enforce equipment maintenance, and calibrate the scraper accuracy of the casting machine.

[0149] It should be noted that the implementation measures for the production cost energy efficiency index classification are as follows:

[0150] When the rating is excellent, a lean production case report is generated weekly, focusing on analyzing equipment operating parameter combinations with low energy consumption fluctuations and supplier lists with small standard deviations in raw material batch utilization;

[0151] Establish a golden parameter library to lock in the optimal process window;

[0152] Set up production line rewards, and teams that meet the standards will receive a bonus pool equal to a certain percentage of the labor cost for that month;

[0153] Raw material suppliers whose utilization rate reaches a certain value for three consecutive months will be given a purchase volume bonus.

[0154] When the grade is good, a two-color warning dashboard is implemented. When the daily energy consumption exceeds the benchmark value by a certain percentage, a yellow warning is triggered and automatically sent to the equipment section chief. If the cycle is extended by a certain percentage for several consecutive days, a red warning is triggered and a process review is carried out.

[0155] Conduct monthly analysis meetings to analyze the abnormal increase in scrap and scrap rate;

[0156] Install smart meters to quickly monitor energy consumption and implement timely mold replacement during the mold replacement process that accounts for a long period of downtime.

[0157] When the grade is qualified, the technical department completes the raw material substitution verification, and the production department enforces the four-shift three-shift operation to ensure delivery;

[0158] Launch a cost optimization project, optimize raw materials and negotiate with suppliers to recycle scraps, start tiered bargaining for materials with low utilization rates, give discounts based on the compliance rate, and dispatch engineers to verify the supplier's slitting process.

[0159] When the risk level is high, the work will be stopped immediately for rectification. The maintenance team will complete a full inspection of the equipment, and the process / quality management will jointly sign a resumption review form.

[0160] Audit energy consumption, carry out equipment transformation, and reorganize process routes.

[0161] S6. Dynamically optimize process parameters and energy efficiency costs through intelligent optimization modules and perform defect protection and self-correction.

[0162] It should be noted that the dynamic process parameter optimization is specifically as follows:

[0163] Through reinforcement learning algorithms, core parameters such as winding tension and sintering temperature stability are adjusted in real time to improve the compliance rate of medium thickness.

[0164] Predict the impact of humidity on film capacitor capacitance based on historical data and automatically compensate for environmental parameters.

[0165] It should be noted that defect prevention and self-correction are specifically:

[0166] Use convolutional neural networks to analyze production images to identify defects such as core bulging and metallized film offset in advance, triggering adjustments to the pre-bake / pre-pressing process.

[0167] The anomaly detection model is used to predict the insulation resistance decrease trend and dynamically correct the electrolyte injection amount or oxide film formation process.

[0168] It should be noted that energy efficiency and cost optimization are specifically:

[0169] Establish a digital twin model of energy consumption, optimize motor power and cooling system operation strategy, and reduce unit energy consumption.

[0170] Genetic algorithms are used to optimize the raw material cutting path and improve foil utilization.

[0171] It should be noted that multi-objective collaborative optimization is specifically as follows:

[0172] Build cutting-edge models, balance capacity accuracy and production costs, and output the optimal process parameter combination.

[0173] Through the description of the above embodiments, those skilled in the art can clearly understand that the various embodiments of the present application can be implemented by means of software or software combined with a necessary general hardware platform, and of course can also be implemented by hardware functions; based on such understanding, the technical solution of the present application can essentially be embodied in the form of a software product or the part that contributes to the prior art. The software product is stored in a storage medium and includes a number of instructions for enabling a computer device, such as but not limited to a personal computer, a server, or a network device, to execute all or part of the steps of the method described in any embodiment of the present application.

[0174] The above describes exemplary embodiments of the present application. It should be understood that the above exemplary embodiments are not restrictive but illustrative, and the scope of protection of the present application is not limited thereto. It should be understood that those skilled in the art can modify and vary the embodiments of the present application without departing from the spirit and scope of the present application, and these modifications and variations should be within the scope of protection of the present application.

Claims

1. A capacitor production control method based on AI, characterized in that: The specific steps include: S1. Install a distributed sensor network to collect multi-source data of the production line in real time through the data acquisition module; S2. Process the data through the data processing module to obtain indicators related to personnel operation safety, production process, product quality and cost control; S3. Calculate the personnel operation safety index, capacitor production comprehensive index, electrical performance quality index and production cost energy efficiency index through the index generation module based on the relevant indicators obtained by the data processing module; S4. Analyze each index through the intelligent analysis module and classify each index into grades; S5. Based on the analysis results, the intelligent decision-making module is used to control production and implement response measures corresponding to different levels; S6. Dynamically optimize process parameters and energy efficiency costs through intelligent optimization modules and perform defect protection and self-correction.

2. The AI-based capacitor production control method according to claim 1, characterized in that: The personnel operation-related indicators include skill matching degree, operation standardization degree, and safety protection compliance rate; The production process related indicators include dielectric thickness compliance rate, winding tension fluctuation rate and sintering temperature stability; The product quality related indicators include capacity accuracy, loss tangent, insulation resistance and voltage resistance; The cost control related indicators include unit capacity consumption, production cycle and raw material utilization rate.

3. The AI-based capacitor production control method according to claim 1, characterized in that: The personnel operation safety index is specifically: R=ω1*G+ω2*A+ω3*J; Where R is the personnel operation safety index and ω is the weight; G is the score for the degree of operational standardization, A is the score for the safety protection compliance rate, and J is the score for the degree of skill matching; the scores of the three indicators are all calculated through standardized values, where the standardized value score = (actual measurement value / benchmark value) * 100, and the benchmark value is the compliance requirement set by the enterprise.

4. The AI-based capacitor production control method according to claim 1, wherein: The capacitor production comprehensive index is specifically: S=υ1*H+υ2*T+υ3*F; Where S is the comprehensive index of capacitor production and υ is the weight; H is the medium thickness compliance score, T is the sintering temperature stability score, and F is the winding tension fluctuation score. The individual score of the three indicators = (1-|actual value-target value| / allowable deviation upper limit)*100, where the target value is the center value of the process standard and the allowable deviation upper limit is the process tolerance.

5. The AI-based capacitor production control method according to claim 1, wherein: The electrical performance quality index is specifically: Z=δ1*L+δ2*tanθ+δ3*E+δ4*N; Where Z is the electrical performance quality index and δ is the weight; L is the capacity accuracy score, which is calculated by dividing the actual capacity by the nominal capacity. If the deviation is within a%, full marks will be awarded, and 10 points will be deducted for every 1% exceeding the limit. Tanθ is the loss tangent score, which is calculated by the difference between the measured value and the standard value. Full marks are awarded if the value is within b%, and 5 points are deducted for every 0.1% exceeding the standard value. E is the insulation resistance score. If the measured value meets E≥c, full marks will be awarded. If it does not meet the standard, points will be deducted proportionally. N is the withstand voltage performance score. The full score is when the measured breakdown voltage is greater than d times the rated voltage. If it is less than d times, the linear deduction will be made. The full score is 100 points, and the minimum score is 0 when a single item exceeds the standard.

6. The AI-based capacitor production control method according to claim 1, wherein: The production cost energy efficiency index is specifically: C=ε1*D+ε2*Q+ε3*Y; Where C is the production cost energy efficiency index and ε is the weight; D is the energy consumption score per unit, energy consumption score = (1-(actual value-benchmark value) / benchmark value)*100. If the energy consumption score is negative, it is 0. The benchmark value is the advanced level of the industry. Q is the production cycle score, cycle score = (benchmark cycle / actual cycle) * 100, the benchmark cycle is the theoretical minimum value of the process; Y is the raw material utilization score, utilization score = (actual utilization / target utilization) * 100, and the target utilization is the designed maximum value.

7. The AI-based capacitor production control method according to claim 1, wherein: The specific classification of each index is as follows: The specific classification of personnel operation safety index is as follows: When R>r1, it is excellent; when r2≤R<r1, it is good; when r3≤R<r2, it is qualified; when R<r3, it is high risk; The classification of capacitor production comprehensive index is as follows: When S>s1, it is excellent; when s2≤S<s1, it is good; when s3≤S<s2, it is qualified; when S<s3, it is high risk; The electrical performance quality index is graded as follows: When Z>z1, it is excellent; when z2≤Z<z1, it is good; when z3≤Z<z2, it is qualified; when Z<z3, it is high risk; The production cost energy efficiency index is classified as follows: When C>c1, it is excellent; when c2≤C<c1, it is good; when c3≤C<c2, it is qualified; when C<c3, it is high risk.

Images