Product production process scheduling method and system based on intelligent manufacturing

Abstract

The invention relates to the technical field of intelligent manufacturing, and discloses a product production process scheduling method and system based on intelligent manufacturing, and the method comprises the steps: 1, data collection and analysis; 2, establishing a production process model; 3, applying an intelligent scheduling algorithm; 4, a real-time monitoring and feedback mechanism; and 5, continuously optimizing and learning. According to the invention, in a real-time monitoring and feedback mechanism, the scheme combines historical data to dynamically correct real-time production data, and performs anomaly detection. By comparing the trend of the current production data with the trend of the historical data in real time, the system can identify and correct data abnormity in time, and it is ensured that each link in the production process is smoothly carried out according to a predetermined plan. The mechanism has the advantages that the accuracy of production data can be ensured by quickly detecting and correcting the deviation, the production delay caused by data errors or lagging is avoided, and the production efficiency and the scheduling precision are improved.

Description

Technical Field

[0001] The present invention relates to the field of intelligent manufacturing technology, and in particular to a product production process scheduling method and system based on intelligent manufacturing. Background Art

[0002] With the transformation and upgrading of the manufacturing industry, intelligent manufacturing has gradually become a trend in the development of the industry. Intelligent manufacturing systems usually involve multiple production processes and a large number of production resources, including machinery and equipment, manual operations, raw materials, transportation tools, etc. In traditional production scheduling methods, schedulers usually rely on manual experience or rule-based systems to allocate and schedule resources. This method is inefficient and cannot fully utilize the potential of equipment and resources, which can easily lead to problems such as unreasonable production plans, waste of resources, and long production cycles.

[0003] At present, the methods and systems for production scheduling in intelligent manufacturing generally achieve optimal scheduling of production processes through the application of intelligent algorithms and real-time data analysis, thereby improving production efficiency, reducing production costs, and optimizing resource utilization.

[0004] However, there is still the problem of incomplete or inaccurate data collection. The scheduling effect of the system is highly dependent on the accuracy and completeness of production data. If there are omissions or errors in data collection (such as sensor failure, network delay, etc.), it may cause deviations in the results of production scheduling and affect production efficiency.

[0005] To this end, we propose a product production process scheduling method and system based on intelligent manufacturing. Summary of the invention

[0006] The present invention mainly solves the technical problems existing in the above-mentioned prior art and provides a product production process scheduling method and system based on intelligent manufacturing.

[0007] In order to achieve the above object, the present invention adopts the following technical solution, a product production process scheduling method based on intelligent manufacturing, comprising the following steps: Step 1: Data collection and analysis: multiple data collections are performed through different types of sensors. After data collection, edge computing and preprocessing are performed. The data quality monitoring module is used to check the integrity, accuracy and timeliness of the data collected by all sensors and devices in real time to form real-time production data. Step 2: Establish a production process model, complete the processing of noise in historical data based on the Kalman filter algorithm, and generate a new data set through data enhancement technology based on relevant data of other production lines, industry standard data and simulation data, and establish a production process model based on the historical data after noise filtering and the new data set; Step 3: Application of intelligent scheduling algorithm. A data quality assessment module is used in the scheduling algorithm. Each batch of input data is scored based on a scoring mechanism. The specific scoring indicators are the accuracy, completeness and timeliness of the data. A fault-tolerant compensation module is introduced. When data inconsistency occurs in the fault-tolerant compensation module, the scheduling algorithm makes compensation adjustments based on the remaining reliable data. Step 4: Real-time monitoring and feedback mechanism, build a dynamic data correction module, and compare real-time production data with historical data to find anomalies and make corrections. Based on the anomaly detection algorithm, the real-time production data is evaluated; Step 5: Deep learning, introducing the incremental learning module. The incremental learning module compares the real-time production data collected by different sensors with the historical data to update and improve the existing scheduling strategy.

[0008] Preferably, the edge computing and preprocessing in step 1 include the following specific steps: First, the raw data collected by the sensor is processed by a low-pass filter to filter out the error data generated by the sensor. Then, the filtered data is subjected to abnormal data detection, and the data that exceeds the preset normal range is marked and discarded.

[0009] Preferably, the scoring mechanism in step 3 is calculated according to the following formula:

[0010] Among them, Q i represents the quality score of the i-th batch of data, A i Score the accuracy of the data, C i is the data integrity score, T i It is the timeliness score of the data. w1, w2, and w3 are weighted coefficients, representing the importance of each indicator in the total score.

[0011] Preferably, the dynamic data correction module in step 4 includes the following correction steps: When the system detects abnormal equipment status or delayed production progress, the dynamic data correction module will make predictions and corrections based on historical trends to fill in missing or inaccurate data.

[0012] Preferably, the core formula of the incremental learning module in step 5 is as follows:

[0013] Among them, θ t represents the model parameters after the tth iteration, represents the model parameters of the previous round, α is the learning rate, represents the gradient of the loss function, x tis the input data at the current moment, y t is the target value at the current moment.

[0014] The present invention also provides a product production process scheduling system based on intelligent manufacturing, comprising: type sensor, each of the type sensors is provided with at least one backup sensor, Edge computing and preprocessing modules for noise filtering and simple abnormal data detection of real-time production data; Data quality monitoring module, used to check the integrity, accuracy and timeliness of data collected by all sensors and devices in real time; The production process model building module builds the production process model based on the processed historical data, trains and optimizes the production process model through machine learning algorithms, and uses supervised learning algorithms to optimize the production process model according to the multi-dimensional production requirements of the product; The data quality assessment module scores the quality of each real-time production data generated. The specific scoring indicators include the accuracy, completeness and timeliness of the data. The fault-tolerant compensation module is used to compensate or adjust data based on the remaining reliable data when data inconsistency occurs.

[0015] The present invention provides a method and system for scheduling product production processes based on intelligent manufacturing. It has the following beneficial effects: 1. Incremental learning and dynamic optimization. This solution adopts an incremental learning module, which enables the system to dynamically optimize according to the production data collected in real time. The advantage of this method is that the system can quickly adapt to changes in the production environment and adjust the scheduling strategy in real time without the need for full retraining. Through incremental learning, the system can continuously update and improve the scheduling model to keep it highly consistent with production needs and the environment. With the continuous accumulation of production data, the scheduling effect of the system will gradually improve, which can more effectively solve the production bottleneck problem and improve the overall production efficiency.

[0016] 2. Adaptive feedback and error correction mechanism. This solution uses the adaptive feedback and error correction mechanism of real-time data to enable the system to quickly respond to sudden anomalies in production and avoid the impact of data problems or external factors on production progress. When the system detects abnormal data or production delays, it can automatically adjust the production plan and provide feedback to the dispatcher. The advantage of this mechanism is that the system can flexibly respond to various unforeseen challenges in the production process, such as equipment failure or resource shortages, to ensure the efficiency and continuity of scheduling in the production process.

[0017] 3. Fault tolerance of intelligent scheduling algorithms. The system adopts intelligent scheduling algorithms based on deep learning, genetic algorithms, and particle swarm optimization, and introduces fault-tolerant mechanisms. These algorithms can handle partial data loss or errors to ensure smooth production scheduling. Even when data is abnormal or inconsistent, the system can still schedule reasonably, avoiding production bottlenecks or equipment idleness caused by data problems. This fault tolerance enables the system to ensure high scheduling efficiency and production continuity when faced with imperfect data.

[0018] 4. Real-time data correction and anomaly detection. In the real-time monitoring and feedback mechanism, this solution combines historical data to dynamically correct real-time production data and perform anomaly detection. By comparing the trends of current production data and historical data in real time, the system can promptly identify and correct data anomalies to ensure that each link in the production process proceeds smoothly according to the predetermined plan. The advantage of this mechanism is that it can ensure the accuracy of production data by quickly detecting and correcting deviations, avoid production delays caused by data errors or lags, and improve production efficiency and scheduling accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the implementation of the present invention or the technical solution in the prior art, the following is a brief introduction to the drawings required for the implementation or the prior art description. Obviously, the drawings described below are only exemplary, and for ordinary technicians in this field, other implementation drawings can be derived from the provided drawings without creative work.

[0020] Figure 1 It is a flow chart of the data collection steps of the present invention; Figure 2 Establish a production process model flow chart for the present invention; Figure 3 This is a flow chart of the intelligent scheduling algorithm of the present invention; Figure 4 This is a flow chart of the real-time monitoring and feedback mechanism of the present invention; Figure 5 This is a system architecture diagram of the present invention. DETAILED DESCRIPTION

[0021] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0022] The present invention provides a product production process scheduling method based on intelligent manufacturing, comprising the following steps: Step 1: Data collection and analysis In order to ensure that the data collected during the production process is accurate and uninterrupted, this method adopts a multi-level data collection and redundancy design. Specifically, key data is collected multiple times through different types of sensors. For example, key parameters such as equipment status and temperature will be collected by sensors of different brands and types, and summarized into the system through IoT devices. If a sensor fails or data is lost, the backup sensor will immediately start and take over the data collection task. This redundant design can effectively prevent data loss caused by a single sensor failure, thereby ensuring data continuity and reliability. To this end, the system sets up a redundant judgment mechanism, which uses a voting algorithm to automatically judge the validity and reliability of the data source when collecting sensor data.

[0023] Specifically, assuming that multiple sensors measure the same parameter at the same time, and set the validity threshold of data collection, the system will select the data source that meets the threshold as the priority reference. For example, if the temperature data output by a sensor deviates from the value of other sensors by more than the set threshold, the data will be considered abnormal and replaced by the backup sensor data. The core purpose of the redundancy mechanism is to ensure that sensor failure does not affect production scheduling decisions.

[0024] In order to reduce network bandwidth pressure and transmission delay, the system performs edge computing and preprocessing after data collection. This process is first performed on the local device to perform noise filtering and abnormal data detection. Specifically, the raw data collected by the sensor is processed by a low-pass filter to eliminate short-term burst noise. The low-pass filter uses the following formula to filter the sensor data:

[0025] Among them, x filtered (t) is the filtered data at the current time t, x(t) is the raw data, and β is the filter coefficient, which is usually between 0 and 1. The filtering process can smooth the raw data and reduce errors caused by sensor jitter or instantaneous interference. Through this algorithm, the system can effectively remove high-frequency noise in the data and provide a smoother input for subsequent data processing.

[0026] Preprocessing also includes simple anomaly detection, such as using threshold detection or Z-score method to check whether the data has significant deviations. If the data exceeds the preset normal range, the system will mark the data and exclude it, or fill in the missing values ​​through interpolation algorithm. The purpose of this preprocessing process is to ensure that only clean and expected data is uploaded to the cloud platform, thereby reducing the pressure of data transmission and ensuring the accuracy of subsequent analysis.

[0027] In order to further improve the accuracy and completeness of data, this system introduces a data quality monitoring module, which checks the completeness, accuracy and timeliness of data collected by all sensors and equipment in real time. During the data transmission process, the system will continuously monitor the status of each data item, especially focusing on the operating status of the equipment and the real-time nature of data collection. That is, the system will perform timestamp verification on each collected data packet to ensure that the collected data matches the real-time status of the production process.

[0028] When data is abnormal, such as sensor failure, data loss, or exceeding the preset threshold, the system will immediately trigger an alarm and perform automatic verification. The specific automatic verification process can be achieved through the following methods: Assuming that the reading data of a certain sensor is too different from the data of other related sensors, the system will automatically query historical data through a backtracking algorithm, find other reliable data sources, perform a rollback operation, and restore the data to a reasonable range. If multiple sensors fail, the system will make inferences based on historical trends and fill in the missing data through linear regression or other interpolation algorithms.

[0029] The data quality monitoring module is used to monitor the data status in real time. Once an abnormality is found, such as errors or loss of data from a certain sensor, the system will automatically perform corrections to ensure the accuracy of scheduling decisions. For example, in a production scheduling, a temperature sensor on a certain production line had a minor failure, resulting in a significant deviation in the sensor output data. The system immediately enabled the backup sensor through a redundant mechanism and eliminated the error through edge computing and filtering algorithms. During the data transmission process, the system also automatically verified to ensure the integrity and timeliness of all data. The final scheduling result was not affected in any way, ensuring the continuity and efficiency of the production process.

[0030] Step 2: Build a production process model In the process of building a production process model, we first need to define various aspects of the production process based on the production requirements of the product and comprehensively consider multiple factors, including production procedures, resource requirements, equipment capabilities, production cycles and other multi-dimensional information. The core goal of this model is to provide accurate time, space and resource requirements for each production link, so that the production process can be scheduled and executed under the most optimized conditions.

[0031] In order to optimize the production process, machine learning algorithms are used to train and optimize the production model. Specifically, the system performs regression analysis and pattern recognition on various parameters in the production process based on historical data to extract key factors that affect production efficiency and quality. Using supervised learning algorithms, support vector machines or decision tree models, resource requirements, time allocation and potential bottlenecks in each production link are predicted based on historical data. In this way, the system can automatically generate the optimal production process model when facing different production needs.

[0032] In the process of building a production process model, the quality of data directly affects the accuracy and effectiveness of the model. To this end, the system introduces a data cleaning and enhancement mechanism to ensure that the data used is clean, accurate and representative.

[0033] First, the system uses the Kalman filter algorithm to process the noise in the data. Kalman filtering is a recursive algorithm used to extract the best estimate from noisy data. After setting the initial value and error covariance matrix, the system updates the data through the filtering algorithm every time new data is collected. The Kalman filter formula is as follows:

[0034] in, is the predicted value at the current moment, y k is the current measured value, K k is the Kalman gain, and Hk is the measurement matrix. The role of the Kalman gain is to determine the weight of the current data update based on the current error covariance and the uncertainty of the measurement value. The goal of the Kalman filter is to minimize the estimation error, make the output data as close to the true value as possible, and reduce the impact of noise on the construction of the production process model.

[0035] In order to improve the generalization ability of the production process model, the system not only relies on historical data from a single source, but also combines relevant data from other production lines, industry standard data, and simulation data for enhancement. Through data enhancement technology, the system can generate new data sets to make up for possible omissions and deviations in historical data, thereby enhancing the robustness and adaptability of the model. Specifically, it involves randomly perturbing, rotating, or scaling existing data to simulate data changes in different production environments. These enhanced data help the model maintain a high level of prediction accuracy when facing different situations that may arise in the actual production process.

[0036] In order to continuously optimize the production process model, the system improves the accuracy of the model through continuous self-learning and updating. After the initial model is established, the system will continuously collect new production data and use machine learning algorithms to iteratively optimize the production process model. Through a feedback mechanism based on historical data, the system can adjust key parameters in the model and maximize production efficiency and resource utilization each time it is optimized.

[0037] Through data cleaning and enhancement, the production process model can effectively deal with noise and missing data, thereby improving the accuracy and robustness of the overall model. Taking Kalman filtering as an example, assuming that during the equipment status acquisition process, some data has large noise due to sensor failure. After using Kalman filtering, more accurate equipment status information can be extracted from the noisy sensor data, thereby improving the accuracy of the production process model and avoiding the impact of misleading data on production scheduling.

[0038] In the process of data enhancement, if the data of a production line is insufficient, a new production data set can be generated by combining simulation data with data from other production lines for enhancement, so that the model can maintain good performance in different production scenarios. For example, in a certain production task, the system may establish a preliminary production process model based on historical data, and use data enhancement technology to simulate the resource requirements and production cycles of multiple production links to ensure that the model can quickly adapt and make optimal scheduling decisions under new production requirements.

[0039] Step 3: Application of intelligent scheduling algorithm In order to achieve efficient production scheduling, this system combines intelligent algorithms such as deep learning, genetic algorithms, and particle swarm optimization to make accurate scheduling decisions in complex production environments. These algorithms are based on real-time production data and production process models, and can dynamically adjust the allocation of production resources to ensure smooth processes and minimize equipment idleness and production bottlenecks.

[0040] In production scheduling, the quality of data directly affects the accuracy and effectiveness of scheduling. In order to deal with data quality issues, this system introduces a data quality assessment module in the scheduling algorithm. Each batch of input data will be scored based on its quality. The specific scoring indicators include data accuracy, completeness and timeliness. The scoring mechanism is calculated according to the following formula:

[0041] Among them, Q i represents the quality score of the i-th batch of data, A i Score the accuracy of the data, C i is the data integrity score, T i Score the timeliness of the data. w1, w2, and w3 are weighted coefficients, representing the importance of each indicator in the total score. Through this formula, the system can automatically adjust the weights of different data according to their quality. The impact of low-quality data in scheduling decisions will be reduced, thereby avoiding the impact of erroneous data on scheduling results.

[0042] If a batch of sensor data has a large deviation in accuracy but good timeliness and completeness, the quality score of the batch of data may be low, and the system will reduce the weight of the data in the scheduling decision to ensure the accuracy and rationality of the scheduling decision.

[0043] In the actual production process, some data may be lost or erroneous, which will affect the effect of production scheduling. In order to meet this challenge, the system introduces a robust fault-tolerant mechanism in the intelligent scheduling algorithm, which can ensure the rationality and accuracy of the scheduling results even in the case of data loss or error.

[0044] The core idea of ​​the fault-tolerance mechanism is that when data inconsistency occurs, the scheduling algorithm can compensate or adjust based on the remaining reliable data. Through the intelligent compensation algorithm, the system can infer the reasonable value of the lost data based on historical data, or gradually adjust the scheduling plan through iterative optimization to minimize the impact of missing data or errors.

[0045] For example, if the real-time data of a device is not updated in time due to network delay, the system can compensate through the trend of historical data, estimate the current status of the device, and adjust the production scheduling plan accordingly. In addition, the algorithm will automatically select reliable data sources for scheduling decisions based on data quality to avoid relying on low-quality data for optimization.

[0046] Through these optimization methods, the production scheduling system can more accurately deal with data quality issues in the production process and make reasonable scheduling decisions when there are missing or erroneous data. For example, if the system encounters sensor data delays for some equipment in a certain production link, traditional scheduling methods may cause the production line to stagnate due to incomplete data. After the introduction of intelligent algorithms, the system can compensate based on existing data and historical trends, quickly adjust the scheduling plan, and ensure that production tasks can continue to be executed.

[0047] In terms of data quality scoring, if the data received by the system contains accuracy deviations of some sensors, the system will reduce the weight of these low-quality data according to the above quality scoring formula, thereby avoiding the adverse impact of erroneous data on the scheduling results. In this way, the system can always maintain a high scheduling accuracy even when the data quality is not completely ideal.

[0048] Through these optimization methods, the intelligence level of the scheduling system has been significantly improved. It can not only cope with complex and changing production environments, but also maintain efficient and accurate production scheduling capabilities in the face of incomplete or inaccurate data.

[0049] Step 4: Real-time monitoring and feedback mechanism During the production process, the core of real-time monitoring is to continuously collect and analyze production data to ensure that the production process proceeds according to the predetermined plan. Once an abnormality or deviation from the plan is found, the system can make timely adjustments to prevent production stagnation or bottlenecks.

[0050] In order to deal with inaccurate production data or transmission delays, this step establishes a dynamic data correction mechanism. By comparing real-time production data with historical data, the system can automatically detect anomalies and make corrections. When the system detects problems such as abnormal equipment status and lagging production progress, the data correction process will be initiated. Specifically, the system will make forecast corrections based on historical trends to fill in missing or inaccurate data, thereby improving the accuracy of scheduling decisions.

[0051] If there is a deviation in the status data of a certain device in a certain production link, the system will use the trend line in the historical data to make corrections, infer the actual status of the device, and adjust the scheduling of subsequent production tasks accordingly. For example, if the status data of the device is abnormal, the system can infer the repair cycle of the device based on the past equipment operation mode and adjust the production plan based on this.

[0052] In the real-time monitoring process, this step also combines anomaly detection algorithms to quickly evaluate production data. By setting rules or thresholds, the system can detect any deviations in the production process in real time and respond quickly. For example, if the resource consumption of a production link exceeds the predetermined range, the system will immediately trigger an alarm and start the exception handling procedure. Anomaly detection usually uses a threshold-based method, that is, if the temperature or vibration value of a certain device exceeds the set standard, the system will determine it as an abnormality and perform corresponding processing.

[0053] In the production process, the feedback mechanism is crucial, especially when facing production data errors or changes in external factors. The system needs to be able to automatically adjust the feedback strategy according to changes in real-time data to ensure the flexibility and accuracy of the production plan.

[0054] When the system finds errors in the collected data or problems with the sensors, it will automatically adjust the feedback mechanism. For example, in the case of errors in data collection, the system will prioritize prompting the operator to check the relevant equipment or sensors and provide more precise adjustment suggestions. Specifically, the system will adjust the feedback response strategy based on changes in real-time data quality. If the data quality deteriorates, the feedback mechanism will guide the operator to calibrate the equipment or repair the sensor; if the data quality is good, the system will automatically perform regular production scheduling without human intervention.

[0055] In addition to relying on preset rules, the system also introduces an intelligent reasoning mechanism to make feedback adjustments more intelligent. When anomalies occur in real-time data, the system not only provides feedback based on fixed rules, but also performs intelligent reasoning through machine learning algorithms to predict potential problems and make advance adjustments. For example, if a production process lags behind, the system can infer the cause based on historical data (such as equipment failure, raw material shortage, etc.) and provide corresponding adjustment plans.

[0056] The core algorithm of the intelligent reasoning mechanism can be expressed by the following formula:

[0057] Among them, f(x t ) represents the current production status, x t is the data value at the current moment, is the adjustment factor, is the data value at the previous moment. The function of this formula is to adjust the coefficient according to the changing trend of real-time data. To predict the production status at the next moment and adjust the production plan based on this prediction to avoid delayed production.

[0058] Through real-time data correction and anomaly detection mechanisms, the system can effectively deal with data anomalies that occur during the production process. If a temperature sensor of a device fails, the system will first compare historical temperature data to determine whether there is an anomaly and use historical data to make predictions and corrections. If the sensor data is completely lost, the system will use the device's historical operation data to fill in the gaps to ensure that production scheduling is not affected.

[0059] In terms of adaptive feedback adjustment, suppose that in a certain production link, the sensor of the equipment has an error, resulting in deviations in some data. The system will correct the error data through the inference algorithm based on the current feedback mechanism and automatically adjust the subsequent production task allocation. If the system detects that the production progress of a certain equipment is lagging behind and the data quality is low, the feedback mechanism will remind the operator to check the equipment and adjust the production plan according to the historical trend of the equipment to ensure smooth production.

[0060] Step 5: Deep Learning In the continuous optimization process, this step introduces an incremental learning module so that each data collection and analysis can dynamically optimize the production scheduling model. Incremental learning allows the system to continuously update and improve the existing scheduling strategy as the data stream continues to arrive, thereby improving the efficiency and accuracy of scheduling. Through this mechanism, the system can not only learn new production patterns, but also self-check and repair the differences between new data and historical data, ensuring that scheduling decisions are based on accurate and high-quality data.

[0061] Whenever a new batch of data is collected, the system compares it with historical data through the incremental learning module to optimize the existing production scheduling model. Unlike traditional batch learning methods, incremental learning can avoid retraining the entire model, save computing resources, and adapt to changes in the production environment in real time. The algorithm gradually adapts the model parameters to the new data. The core formula of incremental learning is as follows:

[0062] Among them, θ t represents the model parameters after the tth iteration, represents the model parameters of the previous round, α is the learning rate, represents the gradient of the loss function, x t is the input data at the current moment, y t is the target value at the current moment.

[0063] The purpose of this formula is to enable the model to update its parameters each time data is input to better predict and optimize the scheduling plan. The benefit of incremental learning is that it can handle the ever-changing production environment and avoid system performance degradation due to frequent data updates through gradual adjustments.

[0064] The present invention also provides a product production process scheduling system based on intelligent manufacturing. The system includes multiple types of sensors, each type of sensor is provided with at least one backup sensor, an edge computing and preprocessing module, a data quality monitoring module, a production process model building module, a data quality assessment module, and a fault-tolerant compensation module.

[0065] Among them, the edge computing processing module is used to perform noise filtering and simple abnormal data detection on real-time production data.

[0066] The data quality monitoring module is used to check the integrity, accuracy and timeliness of data collected by all sensors and devices in real time.

[0067] The production process model building module establishes a production process model based on processed historical data, trains and optimizes the production process model through machine learning algorithms, and uses supervised learning algorithms to optimize the production process model based on the multi-dimensional production requirements of the product.

[0068] The data quality assessment module scores the real-time production data generated each time based on its quality. Specific scoring indicators include data accuracy, completeness and timeliness.

[0069] When data inconsistency occurs in the fault-tolerant compensation module, the scheduling algorithm can compensate or adjust according to the remaining reliable data.

[0070] The above shows and describes the basic principles and main features of the present invention and the advantages of the present invention. It should be understood by those skilled in the art that the present invention is not limited to the above embodiments. The above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention to be protected. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. A product production process scheduling method based on intelligent manufacturing, characterized in that: The steps include: Step 1: Data collection and analysis: multiple data collections are performed through different types of sensors. After data collection, edge computing and preprocessing are performed. The data quality monitoring module is used to check the integrity, accuracy and timeliness of the data collected by all sensors and devices in real time to form real-time production data. Step 2: Establish a production process model, complete the processing of noise in historical data based on the Kalman filter algorithm, and generate a new data set through data enhancement technology based on relevant data of other production lines, industry standard data and simulation data, and establish a production process model based on the historical data after noise filtering and the new data set; Step 3: Application of intelligent scheduling algorithm. A data quality assessment module is used in the scheduling algorithm. Each batch of input data is scored based on a scoring mechanism. The specific scoring indicators are the accuracy, completeness and timeliness of the data. A fault-tolerant compensation module is introduced. When data inconsistency occurs in the fault-tolerant compensation module, the scheduling algorithm makes compensation adjustments based on the remaining reliable data. Step 4: Real-time monitoring and feedback mechanism, build a dynamic data correction module, and compare real-time production data with historical data to find anomalies and make corrections. Based on the anomaly detection algorithm, the real-time production data is evaluated; Step 5: Deep learning, introducing the incremental learning module. The incremental learning module compares the real-time production data collected by different sensors with the historical data to update and improve the existing scheduling strategy.

2. According to the method for scheduling product production process based on intelligent manufacturing according to claim 1, it is characterized in that: The edge computing and preprocessing in step 1 include the following specific steps: First, the raw data collected by the sensor is processed by a low-pass filter to filter out the error data generated by the sensor. Then, the filtered data is subjected to abnormal data detection, and the data that exceeds the preset normal range is marked and discarded.

3. The method for scheduling product production process based on intelligent manufacturing according to claim 1, characterized in that: The scoring mechanism in step 3 is calculated according to the following formula: Among them, Q i represents the quality score of the i-th batch of data, A i Score the accuracy of the data, C i is the data integrity score, T i is the timeliness score of the data, w1, w2, and w3 are weighting coefficients, representing the importance of each indicator in the total score.

4. The method for scheduling product production process based on intelligent manufacturing according to claim 1, characterized in that: The dynamic data correction module in step 4 includes the following correction steps: When the system detects abnormal equipment status or delayed production progress, the dynamic data correction module will make predictions and corrections based on historical trends to fill in missing or inaccurate data.

5. The method for scheduling a product production process based on intelligent manufacturing according to claim 1, characterized in that: The core formula of the incremental learning module in step 5 is as follows: Among them, θ t represents the model parameters after the tth iteration, represents the model parameters of the previous round, α is the learning rate, represents the gradient of the loss function, x t is the input data at the current moment, y t is the target value at the current moment.

6. A product production process scheduling system based on intelligent manufacturing, applied to the method according to any one of claims 1 to 5, characterized in that: include: type sensor, each of the type sensors is provided with at least one backup sensor, Edge computing and preprocessing modules for noise filtering and simple abnormal data detection of real-time production data; Data quality monitoring module, used to check the integrity, accuracy and timeliness of data collected by all sensors and devices in real time; The production process model building module builds the production process model based on the processed historical data, trains and optimizes the production process model through machine learning algorithms, and uses supervised learning algorithms to optimize the production process model according to the multi-dimensional production requirements of the product; The data quality assessment module scores the quality of each real-time production data generated. The specific scoring indicators include the accuracy, completeness and timeliness of the data. The fault-tolerant compensation module is used to compensate or adjust data based on the remaining reliable data when data inconsistency occurs.

Images