Silver powder production detection method and system based on Internet of Things and storage medium

Abstract

The invention relates to the technical field of silver powder production detection control, and discloses a silver powder production detection method and system based on the Internet of Things and a storage medium. The method comprises the following steps: deploying a laser scattering and conductivity sensor to construct a silver powder detection Internet of Things to obtain real-time data; filtering the data to calculate feature points to obtain a parameter set; establishing a dual-threshold feedback control model based on the parameter set to realize purity monitoring and obtain an early warning signal; and generating a detection report according to the early warning signal to record silver powder characteristics to form a batch file. The technical problems of insufficient real-time monitoring capability, insufficient environmental factor interference compensation and insufficient quality data integrated management in traditional silver powder production detection are solved.

Description

Technical Field

[0001] The present application relates to the technical field of silver powder production detection and control, and in particular to a silver powder production detection method, system and storage medium based on the Internet of Things. Background Art

[0002] Traditional silver powder production quality testing relies primarily on manual sampling and offline analysis, typically employing spot checks to control production batch quality. In this testing model, operators periodically take samples from the production line and send them to the laboratory for particle size analysis, purity testing, and morphology observation. Particle size analysis is primarily performed using a laser particle analyzer, purity testing using atomic absorption spectroscopy or inductively coupled plasma mass spectrometry, and morphology observation using a scanning electron microscope. These testing methods have been used in silver powder production for decades, forming a relatively mature quality control system.

[0003] However, traditional detection methods have obvious limitations. First, due to the discontinuous detection process, only limited sampling point data can be obtained, and there is a lack of real-time monitoring capabilities for the entire production process, which can easily lead to missing short-term abnormal conditions in the production process. Secondly, the offline detection method leads to a significant time delay from sampling to obtaining results, and it is impossible to detect and correct quality problems in the production process in a timely manner, resulting in the continuous output of substandard products and waste of resources. In addition, the traditional detection process lacks an effective compensation mechanism for the impact of environmental factors such as temperature and humidity changes on the test results, which affects the accuracy and reliability of the test results. Most importantly, it is difficult for traditional methods to establish a correlation model between process parameters, environmental conditions and product quality, and it is impossible to achieve data-driven intelligent quality control.

[0004] The widespread application of silver powder in high-end fields such as electronic pastes, conductive adhesives, and multilayer ceramic capacitors has led to higher requirements for product quality consistency and traceability. Existing technologies urgently need to address issues such as real-time online monitoring, multi-dimensional quality data integration, and intelligent judgment. In particular, during the production process, how to continuously monitor the particle size, purity, and morphological characteristics of silver powder, how to process and analyze large amounts of real-time test data, and how to establish an impact model between environmental parameters and test results have become key technical challenges in improving the quality control level of silver powder production. At the same time, existing quality data management methods lack standardization and systematization, making it difficult to support quality traceability throughout the product life cycle, which is not conducive to further improving product quality and customer satisfaction. Summary of the Invention

[0005] The present application provides a silver powder production detection method, system and storage medium based on the Internet of Things, which are used to solve the technical problems of insufficient real-time monitoring capabilities, insufficient compensation for environmental interference factors and insufficient quality data integration management in traditional silver powder production detection.

[0006] In a first aspect, the present application provides a silver powder production detection method based on the Internet of Things, which includes: deploying laser scattering sensors and conductivity sensors on the chemical reaction kettle, separation device, drying system and packaging unit in the silver powder production line to form a silver powder production detection Internet of Things to obtain real-time silver powder particle size and conductivity data; performing Butterworth filtering and feature extraction on the real-time silver powder particle size and conductivity data, calculating the three particle size characteristic points D10, D50, and D90 and the conductive characteristic parameters, and obtaining a silver powder characteristic parameter set; constructing a real-time impurity content dual-threshold feedback control model based on the silver powder characteristic parameter set, setting the main threshold of the total impurity content and the secondary threshold of a single element, realizing silver powder purity graded monitoring, and obtaining a silver powder quality warning signal; generating a silver powder detection report based on the silver powder quality warning signal, recording the silver powder particle size distribution, purity level and morphological characteristics, and obtaining a silver powder production batch quality file.

[0007] Optionally, laser scattering sensors and conductivity sensors are deployed on the chemical reaction kettle, separation device, drying system and packaging unit in the silver powder production line to form a silver powder production detection Internet of Things to obtain real-time silver powder particle size and conductivity data, including:

[0008] A dual-wavelength laser scattering sensor is installed inside the chemical reactor, with visible light and near-infrared light wavelength ranges set to collect raw data on the particle size distribution of the silver powder to obtain real-time particle size information;

[0009] A four-electrode conductivity sensor is installed on each of the separation device and the drying system, and the maximum measurement range and measurement accuracy parameters are set to collect conductivity characteristic information of the silver powder to obtain raw conductivity data;

[0010] The real-time information of the particle size and the raw data of the conductivity are transmitted to the edge computing unit, and a data communication network is established using a star topology to obtain a data transmission channel;

[0011] The data in the edge computing unit is timestamped and synchronized, and the synchronization error is controlled using the network time protocol to obtain synchronized silver powder particle size and conductivity data.

[0012] Optionally, the real-time silver powder particle size and conductivity data are subjected to Butterworth filtering and feature extraction, and three particle size characteristic points D10, D50, and D90 and conductive characteristic parameters are calculated to obtain a silver powder characteristic parameter set, including:

[0013] Based on a Butterworth bandpass filter, a low-frequency cutoff value and a high-frequency cutoff value are set for the real-time silver powder particle size data to filter out high-frequency interference and low-frequency drift to obtain purified particle size data;

[0014] Performing particle cumulative distribution statistics on the purified particle size data, calculating the cumulative distribution function using a standard particle size analysis algorithm, and obtaining a particle size distribution curve;

[0015] Extracting a particle size value D10 at a cumulative distribution percentage of 10%, a particle size value D50 at a cumulative distribution percentage of 50%, and a particle size value D90 at a cumulative distribution percentage of 90% based on the particle size distribution curve to obtain a particle size characteristic vector;

[0016] The conductivity data is subjected to a four-layer decomposition process using the db4 wavelet transform method to eliminate electromagnetic interference signals in the production environment and obtain denoised conductivity data;

[0017] Executing a conductivity feature extraction algorithm based on the de-noised conductivity data to calculate the slope, peak point, and steady-state value of the conductivity curve to obtain conductivity characteristic parameters;

[0018] The particle size characteristic vector and the conductivity characteristic parameter are combined into a data structure according to a preset format to obtain a normalized silver powder characteristic parameter set.

[0019] Optionally, the real-time impurity content dual-threshold feedback control model is constructed based on the silver powder characteristic parameter set, a main threshold of total impurity content and a secondary threshold of a single element are set, and the silver powder purity graded monitoring is realized to obtain a silver powder quality early warning signal, including:

[0020] Inputting the silver powder characteristic parameter set into a multi-layer convolutional neural network structure, setting the number of convolution kernels to increase layer by layer, extracting the silver powder impurity feature correlation vector, and obtaining an impurity content feature model;

[0021] Based on the impurity content characteristic model, the main threshold of the total impurity content is set as the maximum allowable impurity content of the silver powder, and the secondary thresholds of the copper element, the iron element, and the lead element are set as the maximum allowable content of a single element, thereby obtaining a dual-threshold judgment condition;

[0022] According to the dual-threshold judgment condition, a sliding window mechanism is established for the real-time detected silver powder impurity data, the window width is set to the number of continuous sampling points, and the sliding step is set to the fixed sampling point interval, to obtain a dynamic monitoring sequence of impurity content;

[0023] Performing a threshold comparison operation on the impurity content dynamic monitoring sequence, when the detection value exceeds the single element secondary threshold but does not exceed the total impurity content primary threshold, triggering a first-level warning state mark; when the detection value exceeds the total impurity content primary threshold, triggering a second-level warning state mark, obtaining different levels of silver powder quality warning signals;

[0024] Based on the silver powder quality warning signals of different levels, an adaptive process drift estimation algorithm is executed to adjust the noise compensation parameters to obtain an accurate correction value for impurity content detection;

[0025] The precise correction value is combined with the silver powder quality warning signals of different levels to generate a warning data structure including the warning level, impurity type, content value and time stamp, and obtain a final silver powder quality warning signal.

[0026] Optionally, based on the impurity content characteristic model, the main threshold of the total impurity content is set as the maximum allowable impurity content of silver powder, and the secondary thresholds of copper element, iron element and lead element are set as the maximum allowable content of a single element, to obtain a dual-threshold determination condition, including:

[0027] Inputting historical silver powder purity data into the impurity content characteristic model, using a supervised learning method to train a threshold prediction function to obtain a threshold reference curve;

[0028] According to the threshold reference curve and the application field standards of silver powder products, three silver powder quality grades are divided into electronic paste grade, conductive glue grade and ordinary grade, and a grading standard table is obtained;

[0029] Based on the grading standard table, a main threshold value is set for the total impurity content, and secondary threshold values ​​are set for copper, iron, and lead, respectively, to obtain a threshold parameter matrix;

[0030] A detection environment compensation calculation is performed on the threshold parameter matrix, and the actual execution value of the threshold is adjusted to obtain a dual-threshold determination condition.

[0031] Optionally, performing detection environment compensation calculation on the threshold parameter matrix, adjusting actual execution values ​​of the thresholds, and obtaining dual-threshold determination conditions include:

[0032] The temperature, humidity, and pressure parameters of the silver powder production environment are collected in real time. The environmental parameter sensor network is used to obtain the production site environmental data and obtain the environmental parameter vector.

[0033] A test result influencing factor calculation model is constructed based on the environmental parameter vector, and the influence weight of each environmental parameter on the test result is determined by multivariate linear regression analysis to obtain an environmental influence coefficient matrix;

[0034] Performing matrix operations on the threshold parameter matrix and the environmental impact coefficient matrix to calculate the compensation coefficient under environmental conditions to obtain a threshold dynamic adjustment formula;

[0035] According to the threshold dynamic adjustment formula, the main threshold and the auxiliary threshold are corrected in real time, and the corrected values ​​are applied to the dual-threshold feedback control model to obtain the dual-threshold judgment condition.

[0036] Optionally, generating a silver powder test report based on the silver powder quality warning signal, recording the silver powder particle size distribution, purity level and morphological characteristics, and obtaining a silver powder production batch quality file includes:

[0037] Performing time series integration on the silver powder quality warning signals, grouping and integrating the warning data according to the production batch number, forming batch-level quality monitoring records, and obtaining a batch quality data set;

[0038] Generate particle size distribution statistical charts based on the batch quality data set, including a particle size distribution histogram, a D value curve graph, and a particle size uniformity index, to obtain a silver powder particle size distribution report;

[0039] Classify and summarize the impurity content data in the batch quality data set, generate a statistical table of the content of each impurity element and a purity grade determination result, and obtain a silver powder purity level report;

[0040] Based on the morphological feature data in the batch quality data set, generating a silver powder shape classification result and a morphological feature parameter table to obtain a silver powder morphological feature report;

[0041] Merging the silver powder particle size distribution report, the silver powder purity level report, and the silver powder morphology characteristic report into a test report document, adding the production time, batch number, and quality grade information to obtain a silver powder test report;

[0042] The silver powder test report is stored in a distributed database system, a unique identification code is generated and associated with the corresponding batch of silver powder, and a silver powder production batch quality file is obtained.

[0043] In a second aspect, the present application provides a silver powder production and detection system based on the Internet of Things, the silver powder production and detection system based on the Internet of Things comprising:

[0044] A deployment module is used to deploy laser scattering sensors and conductivity sensors on the chemical reactor, separation device, drying system, and packaging unit in the silver powder production line, forming a silver powder production detection Internet of Things and obtaining real-time silver powder particle size and conductivity data;

[0045] An extraction module is used to perform Butterworth filtering and feature extraction on the real-time silver powder particle size and conductivity data, calculate the three particle size characteristic points D10, D50, and D90 and the conductivity characteristic parameters, and obtain a silver powder characteristic parameter set;

[0046] A construction module is used to construct a real-time impurity content dual-threshold feedback control model based on the silver powder characteristic parameter set, set a main threshold of total impurity content and a secondary threshold of a single element, realize graded monitoring of silver powder purity, and obtain a silver powder quality early warning signal;

[0047] The recording module is used to generate a silver powder detection report based on the silver powder quality warning signal, record the silver powder particle size distribution, purity level and morphological characteristics, and obtain a silver powder production batch quality file.

[0048] In a third aspect, a computer-readable storage medium is provided, wherein instructions are stored in the computer-readable storage medium, which, when executed on a computer, enables the computer to execute the above-mentioned silver powder production detection method based on the Internet of Things.

[0049] In the technical solution provided by this application, a silver powder production detection Internet of Things is formed by deploying laser scattering sensors and conductivity sensors in the production line, which realizes the real-time collection of silver powder particle size and conductivity data. Compared with traditional offline sampling detection, the time resolution and coverage of data collection are greatly improved, fundamentally solving the problems of discontinuous data and limited sampling points in traditional methods. Butterworth filtering and feature extraction technology are used to process real-time data, effectively eliminating interference signals in the production environment and improving data quality. At the same time, by calculating the three particle size characteristic points D10, D50, and D90 and the conductive characteristic parameters, massive raw data is converted into a characteristic parameter set with clear physical meaning, greatly simplifying the complexity of subsequent analysis. Based on the characteristic parameter set, a real-time impurity content dual-threshold feedback control model is constructed, and a dual judgment mechanism of the total impurity content main threshold and the single element secondary threshold is innovatively introduced. Compared with the traditional single threshold judgment, the purity of the silver powder is more comprehensively evaluated, significantly improving the accuracy and reliability of quality monitoring. Based on the silver powder quality warning signals, a test report is generated and a batch quality file is established, realizing digital management of the entire process from data collection, processing and analysis to result output, providing a solid foundation for product quality traceability.

[0050] In this solution, the step of inputting the silver powder characteristic parameter set into a multi-layer convolutional neural network for processing fully reflects the innovative application of artificial intelligence algorithms in specific application fields. Unlike general convolutional neural networks, the present invention designs a layer-by-layer increasing convolution kernel structure based on the particularity of silver powder production, so that the network can gradually extract high-level impurity characteristic correlation vectors from low-level particle size and conductivity characteristics. This specific network structure design fully considers the complex relationship between physical properties and chemical composition in the silver powder production process, making the algorithm more suitable for the application scenario of silver powder detection. At the same time, the sliding window mechanism and adaptive process drift estimation algorithm in the dual-threshold feedback control model are also specially designed for the parameter fluctuation characteristics in the silver powder production process. The contribution of these algorithm features to the solution is reflected in: on the one hand, it improves the tolerance of the detection process to short-term fluctuations and avoids false alarms; on the other hand, it can accurately identify the systematic errors caused by process drift and achieve accurate correction of the detection results. In addition, the environmental parameter influencing factor calculation model in the scheme quantifies the impact of environmental conditions on the test results through multivariate linear regression analysis, and realizes real-time compensation through the dynamic threshold adjustment formula. This algorithm feature solves the technical difficulty of eliminating environmental interference in traditional detection methods, significantly improves the accuracy and consistency of test results, and is particularly suitable for the electronic paste and multilayer ceramic capacitor manufacturing fields with extremely high product quality requirements. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0052] Figure 1 This is a schematic diagram of an embodiment of a silver powder production detection method based on the Internet of Things in an embodiment of the present application;

[0053] Figure 2 This is a schematic diagram of an embodiment of a silver powder production and detection system based on the Internet of Things in the embodiment of this application. DETAILED DESCRIPTION

[0054] The embodiments of the present application provide a method, system and storage medium for silver powder production detection based on the Internet of Things. The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of this application and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or that are inherent to these processes, methods, products or devices.

[0055] For ease of understanding, the specific process of the embodiment of the present application is described below. Figure 1 In the embodiment of the present application, an embodiment of the silver powder production detection method based on the Internet of Things includes:

[0056] Step S101: deploying laser scattering sensors and conductivity sensors on the chemical reactor, separation device, drying system, and packaging unit in the silver powder production line to form a silver powder production detection Internet of Things to obtain real-time silver powder particle size and conductivity data;

[0057] Step S102: performing Butterworth filtering and feature extraction on the real-time silver powder particle size and conductivity data, calculating the three particle size characteristic points D10, D50, and D90 and the conductivity characteristic parameters, and obtaining a silver powder characteristic parameter set;

[0058] Step S103: constructing a real-time impurity content dual-threshold feedback control model based on the silver powder characteristic parameter set, setting a main threshold for total impurity content and a secondary threshold for a single element, realizing graded monitoring of silver powder purity, and obtaining a silver powder quality warning signal;

[0059] Step S104: Generate a silver powder test report based on the silver powder quality warning signal, record the silver powder particle size distribution, purity level and morphological characteristics, and obtain a silver powder production batch quality file.

[0060] It is understandable that the execution subject of this application can be a silver powder production and detection system based on the Internet of Things, or a terminal or a server, which is not limited here. The embodiment of this application is described by taking the server as the execution subject as an example.

[0061] Specifically, sensors are deployed throughout the silver powder production line. Laser scattering sensors are installed in the chemical reactor to monitor the size distribution of silver powder particles in the reaction solution in real time. Conductivity sensors are installed at the outlet of the separation device to monitor silver powder purity. Corresponding sensors are also deployed in the drying system and packaging unit, forming a comprehensive monitoring Internet of Things (IoT) system covering the entire process. These sensors are connected to a central data processing unit via industrial Ethernet, enabling real-time data transmission and centralized analysis. This generates a real-time data stream of particle size and conductivity data throughout the silver powder production process. The collected real-time data is processed using a Butterworth filter, a low-pass filter with maximum flatness that effectively removes high-frequency noise while preserving the original signal characteristics. After filtering, feature extraction is performed on the particle size data, calculating three key characteristic points in the particle size distribution: D10 (the particle size at the 10th percentile of the cumulative volume distribution), D50 (the median particle size at the 50th percentile of the cumulative volume distribution), and D90 (the particle size at the 90th percentile of the cumulative volume distribution). The mean and fluctuation range of the conductivity data are also calculated and converted into the conductive characteristic parameters of the silver powder. These processed data form a set of silver powder characteristic parameters, providing a data basis for subsequent quality monitoring.

[0062] A real-time dual-threshold feedback control model for impurity content is constructed based on a set of silver powder characteristic parameters. This model employs two monitoring thresholds: a primary threshold for total impurity content and a secondary threshold for a single element. The primary threshold controls the overall impurity level in the silver powder and is typically set at 100 ppm. The secondary threshold sets stricter limits for specific harmful elements (such as copper and lead), typically at 10 ppm. When the impurity level approaches or exceeds the threshold, the system generates warning signals of varying levels based on the degree of deviation. These warning signals are categorized as prompt, warning, and emergency, corresponding to different response strategies, enabling graded monitoring and control of silver powder purity. Based on the silver powder quality warning signals, the system automatically generates a silver powder test report detailing the silver powder's particle size distribution curve, particle size uniformity index (calculated as (D90-D10) / D50), purity level (total impurity content and major element content), and morphological characteristics derived from electron microscopy sampling. These data are then compared with product standards to automatically assess the silver powder's quality grade. The test report also includes records of production process parameters, establishing a correspondence between process parameters and product quality, forming a complete silver powder production batch quality file, and achieving product quality traceability. For example, during the production of silver powder for conductive paste, the IoT system monitored in real time that the silver powder particle size in the reactor gradually increased from an initial 0.2μm to 1.5μm. After eliminating sampling noise through Butterworth filtering, the calculated values ​​were D10 = 1.2μm, D50 = 1.5μm, and D90 = 1.9μm, with a particle size uniformity index of 0.47. Simultaneously, the conductivity data showed that the total impurity content was approximately 85ppm, including 8ppm of copper, which was below the set primary and secondary thresholds. The system generated a green warning signal, indicating good product quality. The resulting quality file showed that this batch of silver powder met the requirements for use in high-end conductive pastes.

[0063] In a specific embodiment, the process of executing step S101 may specifically include the following steps:

[0064] A dual-wavelength laser scattering sensor is installed inside the chemical reactor, with visible light and near-infrared light wavelength ranges set to collect raw data on the particle size distribution of silver powder and obtain real-time particle size information;

[0065] A four-electrode conductivity sensor was installed on each of the separation device and the drying system. The maximum measurement range and measurement accuracy parameters were set to collect the conductivity characteristics of the silver powder and obtain the raw conductivity data.

[0066] The real-time information of particle size and raw conductivity data are transmitted to the edge computing unit, and a data communication network is established using a star topology to obtain a data transmission channel;

[0067] The data in the edge computing unit is timestamped and synchronized, and the network time protocol is used to control the synchronization error to obtain synchronized silver powder particle size and conductivity data.

[0068] Specifically, a dual-wavelength laser scattering sensor was installed inside the chemical reactor, utilizing both visible light at a wavelength of 650 nm and near-infrared light at a wavelength of 980 nm for measurement. The dual-wavelength design offers the advantage of visible light being suitable for detecting micron-sized silver powder particles, while near-infrared light is more effective for detecting submicron-sized particles. The combination of these two wavelengths enables full particle size detection capabilities across the 0.1-10 μm range. The sensor measures scattered light intensity at different angles and calculates the particle size distribution using Mie scattering theory. It collects 100 data points per second, generating a raw data stream of particle size distribution. After preliminary processing, this data is converted into a particle size frequency distribution histogram and a cumulative distribution curve, providing real-time information on the silver powder's particle size. Four-electrode conductivity sensors, comprising two excitation electrodes and two measuring electrodes, are installed in both the separation unit and the drying system. This four-electrode design effectively avoids the polarization effects and electrode contamination issues associated with conventional two-electrode designs. The sensors have a maximum measurement range of 0-2000 μS / cm, adequately covering the conductivity variations experienced during the silver powder production process. The measurement accuracy is set to ±0.5% to ensure data reliability. Sensors in the separation unit primarily monitor the ion concentration in the washing liquid, reflecting the effectiveness of impurity removal. Sensors in the drying system monitor ion migration that may occur during the drying process. The sensors collect data every 10 seconds, recording conductivity and its temporal trends to generate raw conductivity data.

[0069] Real-time particle size information and raw conductivity data are transmitted to the edge computing unit, using a star topology to establish a data communication network. In this star topology, each sensor node is directly connected to the central edge computing unit, forming a point-to-point connection, eliminating the latency and data loss risks associated with multi-hop transmission. Data transmission utilizes the Industrial Ethernet protocol, with a transmission rate of 100 Mbps, sufficient to meet the requirements of real-time data streaming. The edge computing unit is located at the production site, close to the data source, reducing the physical distance and latency of data transmission, ensuring the real-time availability of critical data. This structural design and protocol selection establish a stable and reliable data transmission channel. Data in the edge computing unit is timestamped, using the Network Time Protocol (NTP) to control synchronization errors. Timestamp synchronization is a key step in ensuring the comparability of data from multiple sources. Using the NTP protocol, all sensor nodes regularly calibrate their time with a timing server, keeping time synchronization errors within 10 milliseconds. The synchronization process first determines the clock offset of each sensor, then corrects the data timestamps. Finally, the data from different sources is chronologically reordered to form a time-aligned data stream. The processed data is consistent in time, ensuring an accurate correspondence between particle size data and conductivity data, and providing a foundation for subsequent data analysis and correlation studies. For example, during the production of a batch of silver powder, the particle size distribution data collected by the dual-wavelength laser scattering sensor showed that the average particle size of the silver powder increased from 0.8μm to 1.6μm over time. At the same time, the conductivity sensor monitored that the conductivity of the washing liquid in the separation device gradually decreased from an initial 180μS / cm to 15μS / cm. Through timestamp synchronization, the correspondence between the particle size growth process and the impurity removal process was clarified, and it was found that the washing effect was best when the particle size reached 1.2μm, and the conductivity decreased at the fastest rate.

[0070] In a specific embodiment, the process of executing step S102 may specifically include the following steps:

[0071] Based on the Butterworth bandpass filter, low-frequency cutoff value and high-frequency cutoff value are set for the real-time silver powder particle size data to filter out high-frequency interference and low-frequency drift, and obtain purified particle size data;

[0072] Perform particle cumulative distribution statistics on the purified particle size data, calculate the cumulative distribution function using the standard particle size analysis algorithm, and obtain the particle size distribution curve;

[0073] Based on the particle size distribution curve, the particle size value D10 at which the cumulative distribution percentage is 10%, the particle size value D50 at which the cumulative distribution percentage is 50%, and the particle size value D90 at which the cumulative distribution percentage is 90% are extracted to obtain a particle size characteristic vector;

[0074] The conductivity data is subjected to a four-layer decomposition process using the db4 wavelet transform method to eliminate electromagnetic interference signals in the production environment and obtain denoised conductivity data.

[0075] Conductivity feature extraction algorithm is executed based on the denoised conductivity data to calculate the slope, peak point and steady-state value of the conductivity curve to obtain conductivity characteristic parameters;

[0076] The particle size characteristic vector and the conductivity characteristic parameter are combined into a data structure according to a preset format to obtain a normalized silver powder characteristic parameter set.

[0077] Specifically, real-time silver powder particle size data was processed using a Butterworth bandpass filter with a low-frequency cutoff of 0.01 Hz and a high-frequency cutoff of 10 Hz, creating a filter within a specific frequency band. The low-frequency cutoff is used to filter out slow-changing signals such as temperature drift and baseline offset generated by long-term equipment operation, while the high-frequency cutoff is used to filter out fast-changing interference signals such as electrical noise and mechanical vibration. The advantage of the Butterworth filter is that its frequency response curve within the passband is extremely flat, preserving the original signal characteristics while effectively removing out-of-band noise. The resulting purified particle size data retains true information about the silver powder particle size variations and eliminates interference from various non-target signals. The purified particle size data was statistically analyzed for cumulative distribution, and the cumulative distribution function (CDF) was calculated using a standard particle size analysis algorithm. The particle size data was first sorted by size and divided into several intervals. The number of particles in each interval was counted, and the percentage of particles in each interval relative to the total number of particles was calculated. The percentages for each interval were then summed to obtain a cumulative percentage value. A curve was plotted with particle size as the abscissa and cumulative percentage as the ordinate to form a particle size distribution curve. This curve intuitively shows the overall distribution of silver powder particle size and is the basis for subsequent particle size feature extraction.

[0078] Based on the particle size distribution curve, three key particle size characteristics are extracted: D10, D50, and D90. D10 is the particle size value corresponding to the 10% cumulative distribution percentage, representing the characteristic size of the smaller particles in the distribution; D50 is the particle size value at the 50% cumulative distribution percentage, i.e., the median particle size, representing the overall particle size level; and D90 is the particle size value at the 90% cumulative distribution percentage, representing the characteristic size of the larger particles in the distribution. The specific method for extracting these three characteristic points from the curve is to find the 10%, 50%, and 90% points on the cumulative percentage axis and then read the corresponding particle size values. The particle size characteristic vector [D10, D50, D90] composed of these three values ​​comprehensively reflects the particle size distribution characteristics of the silver powder.

[0079] The conductivity data was processed using the db4 wavelet transform, a wavelet basis function with excellent time-frequency localization. Four-level decomposition decomposes the signal into approximate and detail coefficients at four scales. The approximate coefficients represent the low-frequency portion of the signal, while the detail coefficients represent the high-frequency portion. By setting thresholds for the detail coefficients at different levels, meaningful signal components are retained while random noise components are removed, achieving denoising of the conductivity signal. This method is particularly suitable for processing conductivity signals with multi-scale features and effectively eliminates electromagnetic interference in the silver powder production environment.

[0080] A conductivity feature extraction algorithm is executed on the de-noised conductivity data to calculate three key characteristic parameters: the conductivity curve slope, peak point, and steady-state value. The slope of the curve is a quantitative representation of the rate of change of conductivity, obtained by calculating the difference between the conductivity values ​​at adjacent time points; the peak point is the point where the conductivity reaches its maximum value, determined by finding the global maximum; and the steady-state value is the final stable value of the conductivity, obtained by calculating the average value over a period of time at the end of the signal. These three parameters form a comprehensive set of conductivity characteristic parameters that fully describes the dynamic changes in the conductive properties of the silver powder.

[0081] The particle size characteristic vector and the conductivity characteristic parameter are combined into a unified data structure to form a set of silver powder characteristic parameters. During this combination process, each parameter is normalized to allow for comparison and comprehensive analysis of parameters of different dimensions. Normalization is performed by subtracting the minimum value from each parameter and dividing it by the difference between the maximum and minimum values, so that all parameters are normalized to a range between 0 and 1.

[0082] In a specific embodiment, the process of executing step S103 may specifically include the following steps:

[0083] The silver powder characteristic parameter set is input into a multi-layer convolutional neural network structure, the number of convolution kernels is set to increase layer by layer, the silver powder impurity feature correlation vector is extracted, and the impurity content feature model is obtained;

[0084] Based on the impurity content characteristic model, the main threshold of total impurity content is set as the maximum allowable impurity content of silver powder, and the secondary thresholds of copper, iron, and lead are set as the maximum allowable content of a single element, thus obtaining the dual-threshold judgment condition.

[0085] Based on the dual-threshold judgment conditions, a sliding window mechanism is established for the real-time detected silver powder impurity data. The window width is set to the number of continuous sampling points, and the sliding step is set to the fixed sampling point interval to obtain a dynamic monitoring sequence for impurity content.

[0086] A threshold comparison operation is performed on the dynamic monitoring sequence of impurity content. When the detection value exceeds the secondary threshold of a single element but does not exceed the main threshold of the total impurity content, the first-level warning state mark is triggered. When the detection value exceeds the main threshold of the total impurity content, the second-level warning state mark is triggered, and different levels of silver powder quality warning signals are obtained;

[0087] Based on the early warning signals of silver powder quality at different levels, an adaptive process drift estimation algorithm is executed to adjust the noise compensation parameters to obtain accurate correction values ​​for impurity content detection;

[0088] The precise correction value is combined with different levels of silver powder quality warning signals to generate a warning data structure containing the warning level, impurity type, content value and timestamp to obtain the final silver powder quality warning signal.

[0089] Specifically, the silver powder characteristic parameter set is input into a multi-layer convolutional neural network for deep learning analysis. The network structure contains three convolutional layers, and the number of convolution kernels in each layer is set to 16, 32 and 64 respectively, showing an increasing trend layer by layer. The first layer of convolution kernels mainly extracts the basic characteristics of silver powder particle size and conductivity, the second layer of convolution kernels captures the relationship between these features, and the third layer of convolution kernels integrates to form complex high-level features. Each convolution layer is followed by a ReLU activation function and a maximum pooling layer to compress the feature dimensions while retaining key information. Through this deep network structure, an impurity feature association vector is extracted from the silver powder characteristic parameter set. This vector contains the mapping relationship between the content of various impurities in the silver powder and physical properties such as particle size and conductivity, forming an impurity content characteristic model. Based on the established impurity content characteristic model, a dual-threshold judgment mechanism is set, including a main threshold for total impurity content and a secondary threshold for a single element. The primary threshold for total impurity content is set at 100ppm, representing the maximum allowable value for the sum of all impurity elements in silver powder. Sub-thresholds for individual elements set stricter limits for specific harmful elements, including 20ppm for copper, 15ppm for iron, and 5ppm for lead. These thresholds are determined based on the purity requirements of silver powder for high-end applications such as electronic pastes and conductive inks, forming a dual-threshold judgment criteria for silver powder quality monitoring.

[0090] A sliding window mechanism is established for real-time silver powder impurity data to achieve dynamic monitoring. The sliding window width is set to 20 consecutive sampling points, representing 10 minutes of detection time, and the sliding step is set to 5 sampling points, meaning that the monitoring results are updated every 2.5 minutes. This mechanism can smooth short-term fluctuations while maintaining sensitivity to trend changes. The data within the sliding window are calculated through weighted averaging to obtain the window representative value. The weights are assigned using an exponential decay method, which makes the recent data have a greater impact on the monitoring results. By processing the impurity data through a continuously moving window, a dynamic monitoring sequence of impurity content is obtained.

[0091] A threshold comparison operation is performed on the dynamic impurity content monitoring sequence to determine the silver powder quality status. When the content of a single element exceeds its secondary threshold but the total impurity content does not exceed the primary threshold, a level one warning status flag is triggered, requiring close monitoring but not immediate intervention. When the total impurity content exceeds the primary threshold or the content of any single element exceeds twice its secondary threshold, a level two warning status flag is triggered, requiring immediate adjustment of process parameters or suspension of production. The warning status flag contains information such as the trigger time, the type of element exceeding the standard, and the extent of the excess, forming different levels of silver powder quality warning signals.

[0092] Based on the generated silver powder quality warning signal, an adaptive process drift estimation algorithm is executed. This algorithm analyzes impurity content trends within the same production batch to identify systematic errors caused by process parameter drift, raw material changes, or equipment status changes. The algorithm calculates the deviation pattern between the measured value and historical data and dynamically adjusts noise compensation parameters, including zero offset correction and gain adjustment factors, based on the deviation characteristics. This eliminates systematic errors in the silver powder detection process and produces more accurate impurity content correction values. The precise correction value is combined with the warning signal to form structured warning data, which contains four key fields: warning level (level 1 / level 2), impurity type (copper / iron / lead / total impurities), content value (corrected ppm value), and timestamp (time of occurrence accurate to the second). This structured warning data accurately conveys information about silver powder quality anomalies, enabling production managers to quickly identify problems and take appropriate measures.

[0093] In a specific embodiment, the process of executing the step of setting the total impurity content main threshold as the maximum allowable impurity content of the silver powder may specifically include the following steps:

[0094] The historical silver powder purity data is input into the impurity content characteristic model, and the threshold prediction function is trained using the supervised learning method to obtain the threshold reference curve;

[0095] According to the threshold reference curve and the application field standards of silver powder products, the silver powder quality is divided into three grades: electronic paste grade, conductive glue grade and ordinary grade, and a grading standard table is obtained;

[0096] Based on the classification standard table, a primary threshold value is set for the total impurity content, and secondary threshold values ​​are set for copper, iron, and lead, respectively, to obtain a threshold parameter matrix;

[0097] Detection environment compensation calculation is performed on the threshold parameter matrix, and the actual execution value of the threshold is adjusted to obtain the dual-threshold judgment condition.

[0098] Specifically, the impurity content feature model was trained using historical silver powder purity data, and a supervised learning approach was used to establish a threshold prediction function. The training set consisted of impurity content data from different batches of silver powder produced over the past year and the corresponding product application results, with impurity content as the input feature and the product application results (pass / fail) as the labels. The model was trained using the Support Vector Machine (SVM) algorithm, and the optimal decision boundary was found by adjusting the penalty coefficient C and kernel function parameters. After training, a threshold prediction function was generated that predicts the acceptable impurity content threshold based on the silver powder's particle size and conductivity characteristics. This function was then plotted as a threshold benchmark curve, visually displaying the corresponding impurity content limits for silver powders with different characteristics. Silver powder products were then quality-graded based on the threshold benchmark curve and industry standards. The three grades were categorized: electronic paste grade, which has the highest requirements and is suitable for high-end electronic packaging and microelectronic interconnects; conductive adhesive grade, which has the next highest requirements and is used for general electronic assembly and conductive adhesives; and general grade, which has the most relaxed restrictions and is suitable for non-electronic applications such as catalysis and antimicrobial applications. The grading standard comprehensively considers particle size uniformity, conductivity, and impurity content to generate a grading standard table. The table is presented in tabular form, with the horizontal axis representing the quality level and the vertical axis representing the various indicators. Each cell contains the corresponding limit requirements, providing a basis for subsequent threshold setting.

[0099] Based on the grading standard table, specific threshold parameters are set for each quality level. A primary threshold is set for total impurity content: 100ppm for electronic paste, 200ppm for conductive adhesive, and 500ppm for general grade. At the same time, stricter secondary threshold limits are set for key harmful elements, such as 20ppm, 50ppm, and 100ppm for copper; 15ppm, 40ppm, and 80ppm for iron; and 5ppm, 10ppm, and 30ppm for lead. These values ​​form a 4-row, 3-column threshold parameter matrix, with each row representing an impurity type and each column representing a quality level. The matrix element values ​​correspond to the corresponding threshold values.

[0100] Compensation calculations for the detection environment are performed on the threshold parameter matrix, taking into account the impact of actual production environment factors on the threshold. Compensation calculations include three aspects: temperature compensation, humidity compensation, and electromagnetic interference compensation. Temperature compensation uses a formula to calculate the impact of temperature on the measured value based on the temperature coefficient of the detection instrument. Humidity compensation considers the impact of ambient humidity on the conductivity sensor. Electromagnetic interference compensation adjusts the measurement tolerance based on the electromagnetic noise level in the production environment. The various compensation values ​​are comprehensively calculated to form a compensation coefficient matrix, which is multiplied by the original threshold parameter matrix to obtain the actual threshold values ​​to be implemented, forming the dual-threshold judgment conditions. For example, on a certain electronic paste-grade silver powder production line, a threshold prediction function trained through analysis of historical data shows that when the silver powder D50 value is 1.5μm and the steady-state conductivity value is 10μS / cm, the total impurity content threshold should be set to 100ppm. Taking into account the high detection environment temperature of the production line (35°C) and strong electromagnetic interference, the calculated comprehensive compensation coefficient is 1.05. The adjusted main threshold of the actual total impurity content is 105ppm, and the secondary threshold of a single element such as copper is adjusted to 21ppm accordingly, forming a dual-threshold judgment condition suitable for this production environment.

[0101] In a specific embodiment, the process of executing the step of performing the detection environment compensation calculation on the threshold parameter matrix may specifically include the following steps:

[0102] The temperature, humidity, and pressure parameters of the silver powder production environment are collected in real time. The environmental parameter sensor network is used to obtain the production site environmental data and obtain the environmental parameter vector.

[0103] A calculation model for the factors affecting the test results was constructed based on the environmental parameter vectors. The influence weights of each environmental parameter on the test results were determined through multivariate linear regression analysis to obtain the environmental impact coefficient matrix.

[0104] Perform matrix operations on the threshold parameter matrix and the environmental impact coefficient matrix, calculate the compensation coefficient under environmental conditions, and obtain the threshold dynamic adjustment formula;

[0105] According to the threshold dynamic adjustment formula, the main threshold and the auxiliary threshold are corrected in real time, and the corrected values ​​are applied to the dual-threshold feedback control model to obtain the dual-threshold judgment conditions.

[0106] Specifically, high-precision temperature, humidity, and pressure sensors are installed in the reaction, separation, and drying areas of the production workshop, forming an environmental parameter sensor network. The temperature sensor uses a PT100 platinum resistance thermometer with an accuracy of ±0.1°C and a measurement range of 0-100°C; the humidity sensor uses a capacitive design with an accuracy of ±2% RH and a measurement range of 10-95% RH; and the pressure sensor uses a piezoresistive design with an accuracy of ±0.1 kPa and a measurement range of 85-110 kPa. These sensors collect data once per minute and transmit it to a central processor via the Industrial Internet of Things (IIoT) protocol, generating an environmental parameter vector consisting of three components: temperature, humidity, and pressure.

[0107] Based on the collected environmental parameter vectors, a calculation model for factors influencing test results is constructed. This model uses a multivariate linear regression analysis method, taking the test result deviations in historical test data as the dependent variable and the environmental parameters at the corresponding time as the independent variables. The regression coefficients are calculated using the least squares method. Specifically, the environmental parameters and corresponding test deviation data from the past three months are collected to form a training set matrix. The regression coefficients are then solved using normal equations. The regression analysis results form an environmental impact coefficient matrix, with rows representing different test items (such as total impurities, copper, iron, and lead), and columns representing different environmental parameters (temperature, humidity, and pressure). The matrix elements are the corresponding impact coefficients, quantifying the degree of influence of environmental factors on the test results. Matrix operations are performed on the threshold parameter matrix and the environmental impact coefficient matrix to calculate the environmental compensation coefficient. The specific calculation process is to subtract the standard environmental parameter vector (typically 25°C, 50% RH, and 101.3 kPa) from the current environmental parameter vector to obtain the environmental parameter deviation vector. The environmental impact coefficient matrix is ​​then multiplied by the environmental parameter deviation vector to obtain the deviation correction value for each test item. Finally, the deviation correction value is added to the baseline threshold to form the threshold dynamic adjustment formula. This formula expresses the quantitative relationship between changes in environmental conditions and the threshold adjustment amount, ensuring that the accuracy of the detection results can be maintained under different environmental conditions.

[0108] Based on the dynamic threshold adjustment formula, the corrected threshold values ​​for the current environmental conditions are calculated in real time. This process involves substituting the most recently acquired environmental parameter vector into the dynamic adjustment formula to calculate the corrected values ​​for the primary threshold for total impurity content and the secondary thresholds for each individual element. These corrected thresholds are directly applied to the dual-threshold feedback control model to determine whether the silver powder meets quality requirements. This dynamic adjustment mechanism ensures the environmental adaptability of threshold determination and eliminates environmental interference with test results.

[0109] It should be noted that the threshold dynamic adjustment formula is a quantitative calculation model based on the impact of environmental parameters on the test results. The calculation process first extracts the influence coefficient of each environmental factor on a specific test item from the environmental impact coefficient matrix, such as the influence coefficient of temperature on the detection of total impurity content, the influence coefficient of humidity on the detection of copper content, etc. Then measure the difference between the current environmental parameters (temperature, humidity, pressure) and the standard calibration environmental parameters (usually 25°C, 50% relative humidity, 101.3kPa) to obtain the environmental parameter deviation value. Then multiply the deviation value of each environmental parameter by the corresponding influence coefficient to obtain the detection deviation caused by each environmental factor. Add up the deviations caused by all environmental factors to obtain the total environmental deviation correction value. Finally, add or subtract the baseline threshold (the threshold set under a standard environment) from the environmental deviation correction value (depending on the direction of the deviation) to calculate the correction threshold applicable to the current environmental conditions. For example, if the total impurity content primary threshold under standard conditions is 100ppm, and the current temperature is 10°C higher than the standard environment, with a temperature influence coefficient of -0.3ppm / °C, the temperature-induced deviation correction is -3ppm, and the corrected primary threshold should be 97ppm. This formula ensures accurate threshold determination under varying environmental conditions and eliminates interference from environmental changes on test results.

[0110] In a specific embodiment, the process of executing step S104 may specifically include the following steps:

[0111] Perform time series integration on silver powder quality warning signals, group and integrate warning data according to production batch numbers, form batch-level quality monitoring records, and obtain batch quality data sets;

[0112] Generate particle size distribution statistics based on batch quality data sets, including particle size distribution histogram, D value curve and particle size uniformity index, and obtain silver powder particle size distribution report;

[0113] Classify and summarize the impurity content data in the batch quality data set, generate a statistical table of the content of each impurity element and the purity grade determination results, and obtain a silver powder purity level report;

[0114] Based on the morphological feature data in the batch quality data set, generate the silver powder shape classification results and morphological feature parameter table, and obtain the silver powder morphological feature report;

[0115] Combine the silver powder particle size distribution report, silver powder purity level report and silver powder morphology feature report into a test report document, add production time, batch number and quality grade information to obtain the silver powder test report;

[0116] The silver powder test report is stored in a distributed database system, a unique identification code is generated and associated with the corresponding batch of silver powder, and a quality file of the silver powder production batch is obtained.

[0117] Specifically, a time series integration process is performed on silver powder quality warning signals. By extracting the batch number field from the warning data and grouping and integrating it, all warning signals generated during the same production batch are chronologically arranged and merged. This grouping method shifts the data structure from sampling point to production batch, which better meets the requirements of product quality management. The integration process also includes time window smoothing and outlier removal of the warning data, using a moving median filter to eliminate the influence of short-term fluctuations. The processed data forms a batch-level quality monitoring record, containing the quality trend of the silver powder batch throughout the production process, forming a batch quality dataset. Particle size distribution statistics are generated based on the batch quality dataset. The raw particle size data is first partitioned into intervals, typically dividing the 0.1-10μm range into 20 equal intervals. The percentage of particles within each interval is calculated and plotted as a particle size distribution histogram. A cumulative percentage curve is then calculated, and the three characteristic values ​​(D10, D50, and D90) are extracted and plotted as a D value curve, visually demonstrating the concentration trend and dispersion of the particle size distribution. The particle size uniformity index is also calculated. The index is defined as (D90-D10) / D50. The smaller the value, the more concentrated and uniform the particle size distribution. These charts and indicators together constitute the silver powder particle size distribution report, which fully reflects the particle size characteristics of the silver powder.

[0118] The impurity content data from the batch quality data set is categorized and summarized. First, the content data for each impurity element (copper, iron, lead, etc.) is extracted, and statistical indicators such as the mean, standard deviation, maximum, and minimum values ​​are calculated to generate a statistical table of impurity element content. The total impurity content and the content of each individual element are then compared with preset thresholds to determine the purity grade. Purity grade determination is based on the "barrel principle," where the lowest-level indicator determines the final grade. For example, if the total impurity content meets electronic paste grade but the copper content only reaches conductive adhesive grade, the final grade is determined to be conductive adhesive grade. The determination results, together with the statistical table, form a report on the silver powder's purity level.

[0119] A morphological analysis report is generated based on the morphological feature data from the batch quality dataset. First, digital image processing is performed on the silver powder morphology images acquired by a scanning electron microscope, including image segmentation, edge detection, and feature extraction. Morphological parameters such as circularity, aspect ratio, and surface roughness are calculated to classify the silver powder shape as spherical, flake, dendritic, or irregular. The shape classification results, combined with the morphological feature parameters, form a silver powder morphological characteristic report, which intuitively demonstrates the morphological properties of the silver powder.

[0120] Combine the silver powder particle size distribution report, purity level report, and morphology report into a complete test report document. Include basic information such as the production date, batch number, production line number, and final quality grade on the front page of the report. Arrange the three sections in order within the body of the report, and add conclusions and recommendations at the end. The report format uses a standardized template for quick reading and comparison. This combined document, the silver powder test report, comprehensively documents the quality characteristics of the batch of silver powder.

[0121] Silver powder test reports are stored in a distributed database system, using blockchain technology to ensure data immutability and traceability. During the storage process, a unique identification code is generated for each report. This code, composed of a batch number, timestamp, and hash value, ensures the uniqueness of each report. RFID tags with the identification code are also affixed to the physical silver powder packaging, establishing a virtual-physical connection. These processed test reports constitute a quality archive for the silver powder production batch, supporting quality traceability throughout the product lifecycle. For example, during the production of silver powder for conductive paste, the IoT detection system monitored the particle size distribution of this batch (Batch No. SP20250519A) in real time, revealing a normal distribution: D10 = 1.2μm, D50 = 1.5μm, D90 = 1.9μm, and a particle size uniformity index of 0.47. Impurity analysis revealed a total impurity level of 85ppm, including 15ppm copper, 12ppm iron, and 3ppm lead, meeting electronic paste grade standards. Morphological analysis revealed that 94% of the particles were spherical, with a circularity of 0.92, meeting the requirements for high-end conductive paste applications. The system integrates this data into a comprehensive test report, stores it in a distributed database, and generates an identification code, "SP20250519A-20250519143022-7A3B9C," which is linked to the actual product, forming a batch quality profile that allows downstream customers to verify product quality information.

[0122] The above describes the silver powder production detection method based on the Internet of Things in the embodiment of the present application. The following describes the silver powder production detection system based on the Internet of Things in the embodiment of the present application. Figure 2 In the embodiment of the present application, an embodiment of the silver powder production detection system based on the Internet of Things includes:

[0123] Deployment module 201 is used to deploy laser scattering sensors and conductivity sensors on the chemical reactor, separation device, drying system, and packaging unit in the silver powder production line to form a silver powder production detection Internet of Things and obtain real-time silver powder particle size and conductivity data;

[0124] Extraction module 202, for performing Butterworth filtering and feature extraction on the real-time silver powder particle size and conductivity data, calculating three particle size characteristic points D10, D50, and D90 and conductivity characteristic parameters, and obtaining a silver powder characteristic parameter set;

[0125] A construction module 203 is used to construct a real-time impurity content dual-threshold feedback control model based on the silver powder characteristic parameter set, set a main threshold for total impurity content and a secondary threshold for a single element, implement graded monitoring of silver powder purity, and obtain a silver powder quality warning signal;

[0126] The recording module 204 is used to generate a silver powder detection report based on the silver powder quality warning signal, record the silver powder particle size distribution, purity level and morphological characteristics, and obtain a silver powder production batch quality file.

[0127] The present invention also provides a computer-readable storage medium, which can be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. The computer-readable storage medium stores instructions. When the instructions are run on a computer, the computer executes the steps of the silver powder production detection method based on the Internet of Things.

[0128] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, systems and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0129] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling an IoT-based silver powder production and testing device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes various media that can store program code, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0130] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A silver powder production detection method based on the Internet of Things, characterized in that: The method comprises: Laser scattering sensors and conductivity sensors are deployed in the chemical reactors, separation devices, drying systems, and packaging units in the silver powder production line to form a silver powder production detection Internet of Things, obtaining real-time silver powder particle size and conductivity data; Performing Butterworth filtering and feature extraction on the real-time silver powder particle size and conductivity data, calculating three particle size characteristic points D10, D50, and D90 and conductive characteristic parameters, and obtaining a silver powder characteristic parameter set; Based on the silver powder characteristic parameter set, a real-time impurity content dual-threshold feedback control model is constructed, and a main threshold of total impurity content and a secondary threshold of a single element are set to realize graded monitoring of silver powder purity and obtain a silver powder quality early warning signal; A silver powder test report is generated based on the silver powder quality warning signal, recording the silver powder particle size distribution, purity level and morphological characteristics, and obtaining a silver powder production batch quality file.

2. The method for silver powder production and detection based on the Internet of Things according to claim 1, characterized in that: The chemical reaction kettle, separation device, drying system and packaging unit in the silver powder production line are deployed with laser scattering sensors and conductivity sensors to form a silver powder production detection Internet of Things, which can obtain real-time silver powder particle size and conductivity data, including: A dual-wavelength laser scattering sensor is installed inside the chemical reactor, with visible light and near-infrared light wavelength ranges set to collect raw data on the particle size distribution of the silver powder to obtain real-time particle size information; A four-electrode conductivity sensor is installed on each of the separation device and the drying system, and the maximum measurement range and measurement accuracy parameters are set to collect conductivity characteristic information of the silver powder to obtain raw conductivity data; The real-time information of the particle size and the raw data of the conductivity are transmitted to the edge computing unit, and a data communication network is established using a star topology to obtain a data transmission channel; The data in the edge computing unit is timestamped and synchronized, and the synchronization error is controlled using the network time protocol to obtain synchronized silver powder particle size and conductivity data.

3. The method for silver powder production and detection based on Internet of Things according to claim 1, characterized in that: The real-time silver powder particle size and conductivity data are subjected to Butterworth filtering and feature extraction, and three particle size characteristic points D10, D50, and D90 and conductive characteristic parameters are calculated to obtain a silver powder characteristic parameter set, including: Based on a Butterworth bandpass filter, a low-frequency cutoff value and a high-frequency cutoff value are set for the real-time silver powder particle size data to filter out high-frequency interference and low-frequency drift to obtain purified particle size data; Performing particle cumulative distribution statistics on the purified particle size data, calculating the cumulative distribution function using a standard particle size analysis algorithm, and obtaining a particle size distribution curve; Extracting a particle size value D10 at a cumulative distribution percentage of 10%, a particle size value D50 at a cumulative distribution percentage of 50%, and a particle size value D90 at a cumulative distribution percentage of 90% based on the particle size distribution curve to obtain a particle size characteristic vector; The conductivity data is subjected to a four-layer decomposition process using the db4 wavelet transform method to eliminate electromagnetic interference signals in the production environment and obtain denoised conductivity data; Executing a conductivity feature extraction algorithm based on the de-noised conductivity data to calculate the slope, peak point, and steady-state value of the conductivity curve to obtain conductivity characteristic parameters; The particle size characteristic vector and the conductivity characteristic parameter are combined into a data structure according to a preset format to obtain a normalized silver powder characteristic parameter set.

4. The method for silver powder production and detection based on the Internet of Things according to claim 1, wherein: The method of constructing a real-time impurity content dual-threshold feedback control model based on the silver powder characteristic parameter set, setting a main threshold of total impurity content and a secondary threshold of a single element, realizing graded monitoring of silver powder purity, and obtaining a silver powder quality early warning signal includes: Inputting the silver powder characteristic parameter set into a multi-layer convolutional neural network structure, setting the number of convolution kernels to increase layer by layer, extracting the silver powder impurity feature correlation vector, and obtaining an impurity content feature model; Based on the impurity content characteristic model, the main threshold of the total impurity content is set as the maximum allowable impurity content of the silver powder, and the secondary thresholds of the copper element, the iron element, and the lead element are set as the maximum allowable content of a single element, thereby obtaining a dual-threshold judgment condition; According to the dual-threshold judgment condition, a sliding window mechanism is established for the real-time detected silver powder impurity data, the window width is set to the number of continuous sampling points, and the sliding step is set to the fixed sampling point interval, to obtain a dynamic monitoring sequence of impurity content; Performing a threshold comparison operation on the impurity content dynamic monitoring sequence, when the detection value exceeds the single element secondary threshold but does not exceed the total impurity content primary threshold, triggering a first-level warning state mark; when the detection value exceeds the total impurity content primary threshold, triggering a second-level warning state mark, obtaining different levels of silver powder quality warning signals; Based on the silver powder quality warning signals of different levels, an adaptive process drift estimation algorithm is executed to adjust the noise compensation parameters to obtain an accurate correction value for impurity content detection; The precise correction value is combined with the silver powder quality warning signals of different levels to generate a warning data structure including the warning level, impurity type, content value and time stamp, and obtain a final silver powder quality warning signal.

5. The method for silver powder production and detection based on Internet of Things according to claim 4, characterized in that: Based on the impurity content characteristic model, the main threshold of the total impurity content is set as the maximum allowable impurity content of silver powder, and the secondary thresholds of copper element, iron element and lead element are set as the maximum allowable content of a single element, to obtain the dual-threshold judgment condition, including: Inputting historical silver powder purity data into the impurity content characteristic model, using a supervised learning method to train a threshold prediction function to obtain a threshold reference curve; According to the threshold reference curve and the application field standards of silver powder products, three silver powder quality grades are divided into electronic paste grade, conductive glue grade and ordinary grade, and a grading standard table is obtained; Based on the grading standard table, a main threshold value is set for the total impurity content, and secondary threshold values ​​are set for copper, iron, and lead, respectively, to obtain a threshold parameter matrix; A detection environment compensation calculation is performed on the threshold parameter matrix, and the actual execution value of the threshold is adjusted to obtain a dual-threshold determination condition.

6. The method for silver powder production and detection based on the Internet of Things according to claim 5, characterized in that: The performing of detection environment compensation calculation on the threshold parameter matrix, adjusting the actual execution value of the threshold, and obtaining the dual-threshold determination condition includes: The temperature, humidity, and pressure parameters of the silver powder production environment are collected in real time. The environmental parameter sensor network is used to obtain the production site environmental data and obtain the environmental parameter vector. A test result influencing factor calculation model is constructed based on the environmental parameter vector, and the influence weight of each environmental parameter on the test result is determined by multivariate linear regression analysis to obtain an environmental influence coefficient matrix; Performing matrix operations on the threshold parameter matrix and the environmental impact coefficient matrix to calculate the compensation coefficient under environmental conditions to obtain a threshold dynamic adjustment formula; According to the threshold dynamic adjustment formula, the main threshold and the auxiliary threshold are corrected in real time, and the corrected values ​​are applied to the dual-threshold feedback control model to obtain the dual-threshold judgment condition.

7. The method for silver powder production and detection based on Internet of Things according to claim 1, characterized in that: The silver powder test report is generated according to the silver powder quality warning signal, recording the silver powder particle size distribution, purity level and morphological characteristics, and obtaining a silver powder production batch quality file, including: Performing time series integration on the silver powder quality warning signals, grouping and integrating the warning data according to the production batch number, forming batch-level quality monitoring records, and obtaining a batch quality data set; Generate particle size distribution statistical charts based on the batch quality data set, including a particle size distribution histogram, a D value curve graph, and a particle size uniformity index, to obtain a silver powder particle size distribution report; Classify and summarize the impurity content data in the batch quality data set, generate a statistical table of the content of each impurity element and a purity grade determination result, and obtain a silver powder purity level report; Based on the morphological feature data in the batch quality data set, generating a silver powder shape classification result and a morphological feature parameter table to obtain a silver powder morphological feature report; Merging the silver powder particle size distribution report, the silver powder purity level report, and the silver powder morphology characteristic report into a test report document, adding the production time, batch number, and quality grade information to obtain a silver powder test report; The silver powder test report is stored in a distributed database system, a unique identification code is generated and associated with the corresponding batch of silver powder, and a silver powder production batch quality file is obtained.

8. A silver powder production detection system based on the Internet of Things, characterized in that: For implementing the silver powder production detection method based on the Internet of Things according to any one of claims 1 to 7, the silver powder production detection system based on the Internet of Things comprises: A deployment module is used to deploy laser scattering sensors and conductivity sensors on the chemical reactor, separation device, drying system, and packaging unit in the silver powder production line, forming a silver powder production detection Internet of Things and obtaining real-time silver powder particle size and conductivity data; An extraction module is used to perform Butterworth filtering and feature extraction on the real-time silver powder particle size and conductivity data, calculate the three particle size characteristic points D10, D50, and D90 and the conductivity characteristic parameters, and obtain a silver powder characteristic parameter set; A construction module is used to construct a real-time impurity content dual-threshold feedback control model based on the silver powder characteristic parameter set, set a main threshold of total impurity content and a secondary threshold of a single element, realize graded monitoring of silver powder purity, and obtain a silver powder quality early warning signal; The recording module is used to generate a silver powder detection report based on the silver powder quality warning signal, record the silver powder particle size distribution, purity level and morphological characteristics, and obtain a silver powder production batch quality file.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the processor executes the silver powder production detection method based on the Internet of Things according to any one of claims 1 to 7.

Images