Electromagnetic flow testing method and system applied to robot

Abstract

This invention relates to the field of electromagnetic flow analysis technology, providing an electromagnetic flow testing method and system for robots. The method involves obtaining electromagnetic flow state description variables from the target electromagnetic flow test data by mining flow state indicators based on the target electromagnetic flow test data. Then, based on these variables, task discrimination is performed to obtain past injection task discrimination results and amplified injection task discrimination results. Based on these results, the target robot is controlled to perform automated injection processing. After accurately and comprehensively identifying the past and amplified injection task discrimination results of the target electromagnetic flow test data, the target robot can be controlled using these results to flexibly and adaptively achieve automated injection processing.

Description

Technical Field

[0001] This invention relates to the field of electromagnetic flow analysis technology, and more specifically, to an electromagnetic flow testing method and system for robots. Background Technology

[0002] Electromagnetic flow measurement technology is a method for measuring fluid flow rate using Faraday's law of electromagnetic induction. This technology is commonly used in the water treatment, chemical, pharmaceutical, food, and beverage industries for non-invasive measurement of the flow rate of conductive liquids.

[0003] With the continuous development of artificial intelligence, intelligent robots are being used more and more widely in automated production lines. Automated liquid injection systems, as a comprehensive intelligent production line, are closely related to intelligent robots and electromagnetic flow testing. However, in the actual operation of automated liquid injection systems, traditional technologies still struggle to effectively achieve flexible and adaptive automated liquid injection processing. Summary of the Invention

[0004] To address the aforementioned problems, this invention provides an electromagnetic flow testing method and system for robots.

[0005] A first aspect of this invention provides an electromagnetic flow testing method for robots, applied to an electromagnetic flow testing and analysis system, the method comprising:

[0006] Acquire target electromagnetic flow test data; wherein, the target electromagnetic flow test data is obtained after performing electromagnetic flow test on the target automated liquid injection system, and the electromagnetic flow test of the target automated liquid injection system is implemented based on a pre-set electromagnetic flowmeter;

[0007] The flow state mining indication of the target electromagnetic flow test data is obtained by processing at least one of the following: a past electromagnetic flow test data example carrying a first global electromagnetic flow monitoring annotation and an amplified electromagnetic flow test data example carrying a second global electromagnetic flow monitoring annotation. The first global electromagnetic flow monitoring annotation is obtained by interacting with the amplified injection task prediction view of the past electromagnetic flow test data example and the past injection task prior view of the past electromagnetic flow test data example. The second global electromagnetic flow monitoring annotation is obtained by interacting with the past injection task prediction view of the amplified electromagnetic flow test data example and the amplified injection task prior view of the amplified electromagnetic flow test data example.

[0008] Based on the flow state mining instructions, electromagnetic flow state description mining is performed on the target electromagnetic flow test data to obtain the electromagnetic flow state description variables of the target electromagnetic flow test data.

[0009] Based on the electromagnetic flow state description variables, injection task discrimination is performed to obtain the past injection task discrimination results and amplified injection task discrimination results of the target electromagnetic flow test data;

[0010] Based on the judgment results of the previous injection tasks and the judgment results of the amplified injection tasks, the target robot is controlled to perform automated injection processing.

[0011] In some alternative schemes, the determination step of the traffic state mining indication includes:

[0012] Examples of past electromagnetic flow test data and prior viewpoints on past injection tasks based on those past electromagnetic flow test data examples;

[0013] Based on the completed and debugged amplification injection task identification branch, the past electromagnetic flow test data examples are identified to obtain the amplification injection task prediction view of the past electromagnetic flow test data examples.

[0014] By merging the prior viewpoints of the past injection missions and the predictive viewpoints of the amplified injection missions, a first global electromagnetic flow monitoring annotation is obtained from the past electromagnetic flow test data examples.

[0015] Based on the first global electromagnetic flow monitoring annotation and the past electromagnetic flow test data example, the injection task identification network to be debugged is debugged to obtain the first injection task identification network for identifying amplified injection task items and past injection task items.

[0016] The flow state mining indication is determined based on the neural network configuration weights of the state description mining subnet of the first injection task identification network.

[0017] In some alternative schemes, the debugged amplification injection task identification branch is obtained through the following steps:

[0018] A priori view of the amplification injection task for obtaining examples of amplified electromagnetic flow test data and the examples of amplified electromagnetic flow test data.

[0019] Based on the example of amplification electromagnetic flow test data carrying the prior viewpoint of the amplification injection task, the basic amplification injection task identification branch is debugged to obtain the debugged amplification injection task identification branch.

[0020] In some alternative schemes, the determination step of the traffic state mining indication includes:

[0021] A priori view of the amplification injection task for obtaining examples of amplified electromagnetic flow test data and the examples of amplified electromagnetic flow test data.

[0022] Based on the identified branch of the past injection task that has been debugged, the example of the amplified electromagnetic flow test data is identified to obtain the past injection task prediction view of the example of the amplified electromagnetic flow test data.

[0023] By merging the prior view of the amplification injection task and the prediction view of the past injection task, a second global electromagnetic flow monitoring annotation is obtained for the example of the amplification electromagnetic flow test data.

[0024] Based on the second global electromagnetic flow monitoring annotation and the example of the amplified electromagnetic flow test data, the injection task identification network to be debugged is debugged to obtain a second injection task identification network for identifying amplified injection tasks and past injection tasks.

[0025] The flow state mining indication is determined based on the neural network configuration weights of the state description mining subnet of the second injection task identification network.

[0026] In some alternative solutions, the previously identified injection task branch that has been debugged is obtained through the following steps:

[0027] Examples of past electromagnetic flow test data and prior viewpoints on past injection tasks based on those past electromagnetic flow test data examples;

[0028] Based on the past electromagnetic flow test data examples that carry the prior viewpoints of the past injection tasks, the basic past injection task identification branch is debugged to obtain the debugged past injection task identification branch.

[0029] In some alternative schemes, the steps for obtaining the prediction insights for the amplification injection task are as follows:

[0030] Based on the identification branches in the completed and debugged amplification injection task identification branches, the past electromagnetic flow test data examples are identified to obtain the task identification information sequence of the past electromagnetic flow test data examples. The task identification information sequence includes the first task text mask sequence of the past electromagnetic flow test data examples. The first task text mask sequence is the sequence of task text masks predicted by each identification branch for the past electromagnetic flow test data examples.

[0031] Based on a pre-set first mask correlation threshold, redundancy is eliminated from the first task text mask sequence of the past electromagnetic flow test data example to obtain the current task text mask of the past electromagnetic flow test data example.

[0032] The prediction view of the amplification injection task is determined based on the current task text mask of the past electromagnetic flow test data example.

[0033] In some optional schemes, the task identification information sequence further includes the identification confidence of each task text mask in the first task text mask sequence, and the redundancy elimination of the first task text mask sequence of the past electromagnetic flow test data example to obtain the current task text mask of the past electromagnetic flow test data example includes:

[0034] Redundancy elimination is performed on the first task text mask sequence of the past electromagnetic flow test data example to obtain the first deduplicated task text mask of the past electromagnetic flow test data example.

[0035] From the first task text mask sequence, obtain each candidate task text mask of the past electromagnetic flow test data example to obtain a second task text mask sequence, wherein the second task text mask sequence is a sequence of the candidate task text masks, and the candidate task text mask is a task text mask whose mask correlation with the first deduplicated task text mask is greater than the second mask correlation threshold.

[0036] Based on the recognition confidence of each task text mask in the second task text mask sequence, the first deduplicated task text mask is adjusted to obtain the current task text mask of the past electromagnetic flow test data example.

[0037] A second aspect of the present invention provides an electromagnetic flow testing and analysis system, comprising:

[0038] The data acquisition module is used to acquire target electromagnetic flow test data; wherein, the target electromagnetic flow test data is obtained after performing electromagnetic flow test on the target automated liquid injection system, and the electromagnetic flow test of the target automated liquid injection system is implemented based on a pre-set electromagnetic flowmeter;

[0039] The instruction acquisition module is used to acquire the flow status mining instruction of the target electromagnetic flow test data. The flow status mining instruction is obtained by processing at least one of the following: a past electromagnetic flow test data example carrying a first global electromagnetic flow monitoring annotation and an amplified electromagnetic flow test data example carrying a second global electromagnetic flow monitoring annotation. The first global electromagnetic flow monitoring annotation is obtained by interacting with the amplified injection task prediction view of the past electromagnetic flow test data example and the past injection task prior view of the past electromagnetic flow test data example. The second global electromagnetic flow monitoring annotation is obtained by interacting with the past injection task prediction view of the amplified electromagnetic flow test data example and the amplified injection task prior view of the amplified electromagnetic flow test data example.

[0040] The state mining module is used to perform electromagnetic flow state description mining on the target electromagnetic flow test data according to the flow state mining instruction, and obtain the electromagnetic flow state description variables of the target electromagnetic flow test data.

[0041] The task discrimination module is used to discriminate the injection task based on the electromagnetic flow state description variable, and to obtain the past injection task discrimination results and the amplified injection task discrimination results of the target electromagnetic flow test data;

[0042] The liquid injection control module is used to control the target robot to perform automated liquid injection based on the judgment results of the previous liquid injection tasks and the judgment results of the amplified liquid injection tasks.

[0043] A third aspect of the present invention provides an electromagnetic flow testing and analysis system, comprising: a processor and a memory and a bus connected to the processor; the processor and the memory communicate with each other through the bus; the processor is used to call a computer program in the memory to execute the above-described electromagnetic flow testing method applied to a robot.

[0044] In a fourth aspect, the present invention provides a computer-readable storage medium having a program stored thereon that, when executed by a processor, implements the above-described electromagnetic flow testing method for robots.

[0045] The electromagnetic flow testing method and system for robots provided in this invention obtain electromagnetic flow state description variables for the target electromagnetic flow test data by performing electromagnetic flow state description mining based on the flow state mining indication of the target electromagnetic flow test data, and obtain past injection task discrimination results and amplified injection task discrimination results based on the electromagnetic flow state description variables of the target electromagnetic flow test data. Finally, the target robot is controlled to perform automated injection processing based on the past injection task discrimination results and amplified injection task discrimination results.

[0046] Considering that the flow status mining indication is obtained by processing at least one of the past electromagnetic flow test data examples carrying the first global electromagnetic flow monitoring annotation and the amplified electromagnetic flow test data examples carrying the second global electromagnetic flow monitoring annotation, this allows for accurate and comprehensive identification of the different injection tasks corresponding to the past injection task sequences and the amplified injection task sequences from the perspective of electromagnetic flow test data.

[0047] Furthermore, given that the identified amplification injection task prediction viewpoints are used as examples of past electromagnetic flow test data for amplification injection task prior knowledge, and the identified past injection task prediction viewpoints are used as examples of amplification electromagnetic flow test data for past injection task prior knowledge, there is no need to reconfigure prior knowledge, thereby improving the timeliness of the overall solution.

[0048] Furthermore, since the first global electromagnetic flow monitoring annotation is obtained by interacting the prediction view of amplification injection tasks based on past electromagnetic flow test data examples with the prior view of past injection tasks based on past electromagnetic flow test data examples, the prior knowledge of amplification injection tasks in past electromagnetic flow test data examples is stable, improving the anti-interference capability of neural network debugging; even if the flow state mining indication is obtained in the case of small sample, the task discrimination can still accurately and comprehensively distinguish different injection tasks in past injection task sequences and amplification injection task sequences by using the electromagnetic flow state description variables of the target electromagnetic flow test data mined based on the flow state mining indication.

[0049] Finally, after accurately and comprehensively identifying the past injection task discrimination results and the amplified injection task discrimination results of the target electromagnetic flow test data, the target robot can be controlled based on the past injection task discrimination results and the amplified injection task discrimination results, thereby flexibly and adaptively realizing automated injection processing. Attached Figure Description

[0050] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0051] Figure 1 This is a flowchart illustrating an electromagnetic flow testing method for robots, provided as an embodiment of the present invention.

[0052] Figure 2 This is a functional block diagram of an electromagnetic flow testing and analysis system provided in an embodiment of the present invention.

[0053] Figure 3 This is a schematic diagram of a product module of an electromagnetic flow testing and analysis system provided in an embodiment of the present invention. Detailed Implementation

[0054] Exemplary embodiments of the present invention will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

[0055] To better understand the above technical solutions, the technical solutions of the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the embodiments of the present invention and the specific features in the embodiments are detailed descriptions of the technical solutions of the present invention, rather than limitations on the technical solutions of the present invention. In the absence of conflict, the embodiments of the present invention and the technical features in the embodiments can be combined with each other.

[0056] Please see Figure 1 The flowchart below illustrates an electromagnetic flow testing method for robots, provided by an embodiment of the present invention. This method is applied to an electromagnetic flow testing and analysis system, and the specific content of the method includes steps S110-S150.

[0057] S110: Obtain target electromagnetic flow test data.

[0058] In S110, the target electromagnetic flow test data is obtained after performing an electromagnetic flow test on the target automated liquid injection system. The electromagnetic flow test of the target automated liquid injection system is based on a pre-set electromagnetic flowmeter.

[0059] First, the target electromagnetic flow rate test data is the electromagnetic flow rate test data to be processed. This data is typically collected in real time and represents the flow rate information of the current operating status of the automated liquid injection system.

[0060] Secondly, the target automated fluid injection system is a system that integrates an electromagnetic flow meter to automatically control the fluid injection process. It can include pumps, valves, sensors, and the software and hardware used to control these components. The target automated fluid injection system is the object whose flow data is monitored and analyzed.

[0061] The next step, performing electromagnetic flow testing on the target automated injection system, involves using an electromagnetic flowmeter to measure the flow rate of fluid through a portion of the injection system. This test provides data on flow velocity, total flow rate, and other parameters, and is a crucial step in ensuring the system's accurate operation.

[0062] Furthermore, to ensure the accuracy of test results, electromagnetic flowmeters need to be properly configured before testing. This may include calibration, configuration parameters (such as flow range and velocity range), and installation location. The term "pre-configuration" emphasizes the importance of pre-test preparation for obtaining reliable data.

[0063] In the S110 process, electromagnetic flow test data is first acquired from the target automated liquid injection system. This data is based on pre-set readings of the electromagnetic flowmeter. This data is recorded and used for subsequent data analysis and system control procedures.

[0064] Understandably, the S110 focuses on acquiring electromagnetic flow rate test data from a specific automated liquid filling system. To illustrate, imagine a typical industrial scenario: a beverage production company has an automated production line for bottling beverages. This line is equipped with an automated liquid filling system that precisely controls the amount of beverage injected into each bottle. A key component of the system is an electromagnetic flow meter, installed on the pipeline conveying the liquid, monitoring the flow rate of the beverage through the pipeline in real time.

[0065] Before production begins, the electromagnetic flow meter is set up and the equipment is calibrated to ensure accurate readings. The flow range is set to match the expected flow rate, and the software parameters are ensured to be suitable for the current production batch (such as a specific type of beverage).

[0066] Once the production process begins, the electromagnetic flow meter continuously measures the flow rate of the beverage passing through the pipe. Because the electromagnetic flow meter does not rely on fluid contact with any sensor components, it is ideally suited for the hygiene requirements of the food and beverage industry.

[0067] In this example, the "target" data refers to the flow rate reading during each filling cycle in the production process. This data is crucial to ensuring that each bottle is filled correctly. If the flow rate reading is lower or higher than expected, it may indicate a leak, blockage, or other mechanical problem.

[0068] The target automated filling system includes all the equipment and control algorithms used for the automated filling process. In addition to electromagnetic flow meters, it may also include servo controllers for opening and closing valves, sensors for detecting bottle position, and program logic for executing filling commands.

[0069] In S110, the operator or control system collects data from the electromagnetic flowmeter and uses it as the basis for subsequent analysis and decision-making. For example, if the data shows that the flow rate deviates from the normal range, the system can automatically adjust the injection rate or trigger an alarm to prompt the operator to check.

[0070] In other words, S110 is the data collection stage in the entire automated liquid injection process, which ensures that subsequent steps can optimize and adjust the production process based on accurate and reliable information.

[0071] S120: Obtain the flow status mining indication of the target electromagnetic flow test data.

[0072] In S120, the flow status mining indication is obtained by processing at least one of the following: a past electromagnetic flow test data example carrying a first global electromagnetic flow monitoring annotation and an amplified electromagnetic flow test data example carrying a second global electromagnetic flow monitoring annotation. The first global electromagnetic flow monitoring annotation is obtained by interacting with the amplified injection task prediction view of the past electromagnetic flow test data example and the past injection task prior view of the past electromagnetic flow test data example. The second global electromagnetic flow monitoring annotation is obtained by interacting with the past injection task prediction view of the amplified electromagnetic flow test data example and the amplified injection task prior view of the amplified electromagnetic flow test data example.

[0073] The following is an explanation of the relevant technical terms used in S120 in this embodiment of the invention.

[0074] Flow state mining indicators are a set of analytical parameters or algorithmic instructions used to guide the analysis of electromagnetic flow test data during data processing. They determine how to identify and extract characteristics of electromagnetic flow state and can include applied statistical methods, machine learning models, or other data analysis techniques.

[0075] The first global electromagnetic flow monitoring annotations are tags or metadata added based on historical flow data, providing comprehensive information about flow status and injection task performance. "Global" means that these annotations take into account data from a variety of different scenarios and are applicable to the evaluation of the entire system.

[0076] The past electromagnetic flow test data examples are real flow data points recorded in the past, reflecting the results of historical fluid flow conditions and injection tasks.

[0077] The second global electromagnetic flow monitoring annotation is similar to the first annotation; it is also based on flow data, but it is for an augmented dataset. The augmented dataset can include data that has been processed to simulate possible future scenarios.

[0078] The example of amplified electromagnetic flow test data is simulated flow data generated through data amplification technology, used to predict future flow conditions and the performance of injection tasks.

[0079] The prediction perspective for amplified injection tasks can be understood as predictive labels generated based on amplified datasets, representing the system's prediction of future injection task performance based on historical and simulated data.

[0080] The prior view of past injection missions is based on real labels of real historical data, representing the actual performance and results of past injection missions.

[0081] Interaction refers to the comparative analysis between predictive and prior viewpoints to evaluate the accuracy of predictions and improve the model.

[0082] The predictive view of past injection tasks is a predictive label based on historical data of past injection task performance, which can be obtained through retrospective analysis.

[0083] The prior view of the amplification injection task is the true label of the injection task performance in the amplification dataset. Although the data is simulated, these labels represent the expected or considered correct task performance.

[0084] The "Flow Status Mining Indicators" in S120 are generated through interactive analysis using historical real data and augmented simulated data, combined with their predicted and actual labels. These indicators will be used to analyze target electromagnetic flow test data, helping to better understand the current flow status and supporting subsequent data processing and automated injection task identification.

[0085] At this stage, the goal is to gain further insights into the flow status by analyzing the collected target electromagnetic flow test data. To achieve this, information needs to be extracted from historical and augmented data to generate indications.

[0086] Further explanations of terms within the context of a beverage company are as follows.

[0087] Examples of past electromagnetic flow test data are previously collected flow test data, including flow readings during the injection process under various conditions (such as different types of beverages, different production batches, and different environmental conditions). This data is stored in the factory's database and can be used to understand flow behavior under normal operating conditions.

[0088] First Global Electromagnetic Flow Monitoring Annotations: Based on past electromagnetic flow test data, the analysis system can create a global view, highlighting key patterns, outliers, or other important characteristics in the data. These annotations can be automatically generated by the algorithm or manually added by experienced engineers based on their understanding of system performance.

[0089] Example of augmented electromagnetic flow test data: To predict future flow behavior or simulate hypothetical operating conditions, analysis systems can use techniques such as machine learning or data synthesis to create additional electromagnetic flow test data. This data can help predict potential future scenarios, such as the impact of new beverage types or faster production rates on flow.

[0090] Second, global electromagnetic flow monitoring annotations: Similar to processing historical data, augmented data also requires annotation. These annotations reflect the data analysis software's or engineers' understanding of this augmented data, such as its similarity to known operating conditions and potential trend changes.

[0091] Flow Status Mining Instructions: Finally, combining the first and second global electromagnetic flow monitoring annotations, the analysis system generates instructions or parameters that will guide how to conduct in-depth analysis of the target electromagnetic flow test data. For example, if historical data shows that flow typically increases within a certain time period, the flow status mining instructions might include parameters to check whether the current data for that time period conforms to this pattern.

[0092] Through the above process, S120 lays the foundation for the subsequent flow state description mining (S130). Flow state mining instructions help determine which data features should be focused on, how to interpret these features, and what possible measures need to be taken to adjust the injection task to ensure product quality and production efficiency.

[0093] S130: Based on the flow state mining instruction, perform electromagnetic flow state description mining on the target electromagnetic flow test data to obtain the electromagnetic flow state description variables of the target electromagnetic flow test data.

[0094] In S130, electromagnetic flow state description mining is a data analysis process that utilizes advanced algorithms to analyze target electromagnetic flow test data and extract information about the fluid flow state. This process involves data mining techniques and pattern recognition methods, with the aim of identifying key features and regularities in the flow data.

[0095] Electromagnetic flow state descriptor variables are specific quantitative indicators or variables obtained through the electromagnetic flow state description mining process. They characterize the physical state behind the data collected by electromagnetic flowmeters, such as flow velocity, flow rate, and flow stability. Electromagnetic flow state descriptor variables are a mathematical expression of flow conditions and can be used for further analysis or as input features for machine learning models.

[0096] In S130, the system uses the flow state mining indication obtained from S120 to process the target electromagnetic flow test data collected in S110. The flow state mining indication may include the application of specific statistical methods, machine learning algorithms, or other analytical tools to gain a deeper understanding of the current flow state and convert this information into a format that can be used for subsequent processing steps.

[0097] For example, in an automated fluid injection system, electromagnetic flow state description mining might uncover certain periodic fluctuation patterns related to the operation of specific stages on the production line. The system then translates these findings into electromagnetic flow state description variables, such as fluctuation frequency and amplitude. This information is invaluable for diagnosing problems and optimizing processes.

[0098] As can be seen, the electromagnetic flow state description mining in S130 is an analytical process that extracts key information from the target electromagnetic flow test data and generates electromagnetic flow state description variables. These variables are precise mathematical representations of the flow conditions and are used for further flow analysis and control decisions.

[0099] In stage S130, the goal is to analyze the target electromagnetic flow test data in order to extract a feature vector reflecting the flow state. The feature vector typically contains multiple feature values, each describing an aspect of the flow state.

[0100] Suppose we are monitoring an automated filling line that uses electromagnetic flow meters to measure the flow rate of beverage entering the bottles. For example, we can consider constructing a feature vector based on the following flow state characteristics:

[0101] Average flow rate (L / min): The average flow rate of beverage through the flow meter within a specific time window;

[0102] Peak flow rate (L / min): The maximum flow rate recorded during the measurement period;

[0103] Flow velocity stability (L / min^2): The degree of fluctuation in flow velocity over time, which can be expressed by the variance or standard deviation of the flow velocity;

[0104] Total flow rate (L): The total volume of fluid passing through the flow meter during the measurement cycle;

[0105] Acceleration (L / min^2): The rate at which the flow velocity changes, which can be understood as the time derivative of the flow velocity.

[0106] Based on this, we can assume that the electromagnetic flowmeter collects the following data during a certain measurement cycle:

[0107] Average flow rate: 2 L / min;

[0108] Peak flow rate: 3 L / min;

[0109] Flow rate stability: 0.1 L / min^2 (indicating relatively stable flow rate);

[0110] Total flow rate: 120L;

[0111] Acceleration: 0.05 L / min^2 (indicating that the increase and decrease of flow velocity are relatively gradual).

[0112] Based on this data, an electromagnetic flow state feature vector can be constructed as follows:

[0113] [\text{eigenvector}=[2, 3, 0.1, 120, 0.05]].

[0114] This feature vector will be further used in pattern recognition, anomaly detection, or prediction models to help assess whether the current traffic status is normal or whether adjustments to the production process are needed.

[0115] In practice, even more complex features may be calculated and added to the feature vector, such as time-frequency domain features and statistical features, to more comprehensively describe the traffic status. This vector may also be used in conjunction with the traffic status mining indicators previously obtained from S120 to perform more accurate traffic status analysis and subsequent processing steps.

[0116] S140: Based on the electromagnetic flow state description variables, perform injection task discrimination to obtain the past injection task discrimination results and amplified injection task discrimination results of the target electromagnetic flow test data.

[0117] In S140, the injection task determination is an evaluation process in which the electromagnetic flow state description variables obtained from S130 are used to evaluate whether the injection task was performed as expected. This determination process may include comparing current flow data with established quality standards, historical performance data, or model predictions.

[0118] The past injection task judgment results are based on historical data, illustrating how the target electromagnetic flow rate test data compares to the performance of past actual injection tasks. If the current data shows characteristics similar to historical data, the system may determine that the injection task is operating normally; if there are significant differences, it may indicate that adjustments or intervention are needed.

[0119] The amplified injection task discrimination results are based on amplified (simulated) data and are used to evaluate how the target electromagnetic flow rate test data compares to the injection task performance under predicted future conditions. By analyzing this data, the system can identify potential problems in advance and take measures to optimize the injection process to cope with possible future changes.

[0120] For example, suppose a factory recently introduced a new type of bottle with different shapes and capacities. In S140, the system would use electromagnetic flow state description variables to analyze the filling process of these new bottles and compare it with historical data (results from past filling tasks) and predictive models (results from expanded filling tasks). Such analysis can help determine whether filling parameters need to be adjusted or production line settings modified to accommodate the requirements of the new bottles.

[0121] As can be seen, the injection task determination in step S140 is a crucial evaluation process. It combines electromagnetic flow state description variables, historical data, and predictive models to ensure the quality and efficiency of the injection task. Past injection task determination results and amplified injection task determination results provide operators or automated systems with the necessary information to optimize the injection process.

[0122] S150: Based on the judgment results of the previous injection tasks and the judgment results of the amplified injection tasks, control the target robot to perform automated injection processing.

[0123] In S150, the target robot refers to the machine device responsible for actually performing the liquid filling task in an automated liquid filling system. The target robot is typically an industrial robot or automated device equipped with appropriate mechanical devices, such as pumps, valves, and flow meters, to control the precise filling of liquids (e.g., beverages, pharmaceuticals, chemicals). These robots can be programmed to perform complex sequences of actions and can adapt to containers of different shapes and sizes. In this automated liquid filling system, the target robot is tightly integrated with an electromagnetic flow meter. The electromagnetic flow meter provides real-time data for monitoring and regulating the flow rate of liquid through the pipeline. The target robot is directed by a central control system or local control unit. This system receives test data from the electromagnetic flow meter and issues instructions based on the analysis results, such as adjusting the filling speed, starting or stopping the flow, etc.

[0124] After completing S140, the target robot needs to adjust its operation based on the judgment results of past injection tasks and the judgment results of expanded injection tasks. For example, if the judgment result indicates that the current flow rate may lead to insufficient or excessive injection, the robot will adjust its filling parameters to ensure that each container receives the correct amount of liquid.

[0125] In S150, the target robot uses the discrimination results to perform actual physical operations, which may include starting or stopping the injection, changing the filling speed, or taking corrective measures when an anomaly is detected. The robot's movements are controlled by advanced software algorithms to ensure high efficiency and accuracy.

[0126] In the S150, the target robot is a key component in the automated injection system, performing the actual operations. It uses data from previous steps to drive decision-making, optimizing the injection task to ensure product quality, improve production efficiency, and minimize waste. Through an intelligent control system, the target robot can respond to complex data inputs and automatically execute precise operations.

[0127] As can be seen, by applying S110-S150, this embodiment of the invention obtains electromagnetic flow state description variables of the target electromagnetic flow test data by performing electromagnetic flow state description mining based on the flow state mining indication of the target electromagnetic flow test data, and obtains past injection task discrimination results and amplified injection task discrimination results of the target electromagnetic flow test data by performing task discrimination based on the electromagnetic flow state description variables of the target electromagnetic flow test data, and finally controls the target robot to perform automated injection processing based on the past injection task discrimination results and amplified injection task discrimination results.

[0128] Considering that the flow status mining indication is obtained by processing at least one of the past electromagnetic flow test data examples carrying the first global electromagnetic flow monitoring annotation and the amplified electromagnetic flow test data examples carrying the second global electromagnetic flow monitoring annotation, this allows for accurate and comprehensive identification of the different injection tasks corresponding to the past injection task sequences and the amplified injection task sequences from the perspective of electromagnetic flow test data.

[0129] Furthermore, given that the identified amplification injection task prediction viewpoints are used as examples of past electromagnetic flow test data for amplification injection task prior knowledge, and the identified past injection task prediction viewpoints are used as examples of amplification electromagnetic flow test data for past injection task prior knowledge, there is no need to reconfigure prior knowledge, thereby improving the timeliness of the overall solution.

[0130] Furthermore, since the first global electromagnetic flow monitoring annotation is obtained by interacting the prediction view of amplification injection tasks based on past electromagnetic flow test data examples with the prior view of past injection tasks based on past electromagnetic flow test data examples, the prior knowledge of amplification injection tasks in past electromagnetic flow test data examples is stable, improving the anti-interference capability of neural network debugging; even if the flow state mining indication is obtained in the case of small sample, the task discrimination can still accurately and comprehensively distinguish different injection tasks in past injection task sequences and amplification injection task sequences by using the electromagnetic flow state description variables of the target electromagnetic flow test data mined based on the flow state mining indication.

[0131] Finally, after accurately and comprehensively identifying the past injection task discrimination results and the amplified injection task discrimination results of the target electromagnetic flow test data, the target robot can be controlled based on the past injection task discrimination results and the amplified injection task discrimination results, thereby flexibly and adaptively realizing automated injection processing.

[0132] In some optional embodiments, the step of determining the traffic status mining indication includes S210-S250.

[0133] S210: Obtain past electromagnetic flow test data examples and the prior view of past injection tasks based on the past electromagnetic flow test data examples.

[0134] S220: Based on the completed and debugged amplification injection task identification branch, the past electromagnetic flow test data examples are identified to obtain the amplification injection task prediction view of the past electromagnetic flow test data examples.

[0135] S230: Combine the prior view of the past injection mission and the predictive view of the amplified injection mission to obtain the first global electromagnetic flow monitoring annotation of the past electromagnetic flow test data example.

[0136] S240: Based on the first global electromagnetic flow monitoring annotation and the past electromagnetic flow test data example, debug the injection task identification network to be debugged to obtain a first injection task identification network for identifying amplified injection task items and past injection task items.

[0137] S250: Determine the flow state mining indication based on the neural network configuration weights of the state description mining subnet of the first injection task identification network.

[0138] The S210-S250 series is illustrated using the example of a beverage production company seeking to improve the efficiency and quality of its automated dispensing system, in order to better understand and optimize electromagnetic flow monitoring.

[0139] First, in S210, flow test data recorded by historically used electromagnetic flowmeters and corresponding actual injection task results (i.e., prior viewpoints) are collected. Next, in S220, the calibrated amplified injection task identification branch is used to analyze this historical data, generating predictive labels or evaluations (i.e., predictive viewpoints) to estimate possible future injection performance under similar conditions. Subsequently, in S230, the prior viewpoints and predictive viewpoints are merged to generate a first global electromagnetic flow monitoring annotation that comprehensively reflects past flow and injection performance. In S240, these annotations are used to calibrate the injection task identification network, resulting in an accurate model capable of simultaneously identifying new and historical injection task events; this model can be considered the first injection task identification network. Finally, in S250, flow state mining indicators are determined based on the neural network configuration weights of the state description mining subnet in the calibrated first injection task identification network. These indicators are algorithm parameters or rules used to guide how to process and analyze real-time flow test data.

[0140] Overall, this series of steps yields a powerful tool that integrates historical experience and predictive insights. It not only helps monitor current flow conditions but also anticipates and adapts to upcoming changes. This approach allows for more precise liquid injection tasks, reduces waste, increases production efficiency, and ensures consistent product quality. In summary, implementing this technology leads to significant improvements in production efficiency and cost savings, while enhancing the ability to predict future production challenges.

[0141] In this embodiment of the invention, a deep convolutional network can be designed to identify and classify various patterns and anomalies in electromagnetic flow test data to ensure the quality and consistency of the injection process. That is, the injection task identification network can be a deep convolutional network. The following is an introduction to deep convolutional networks.

[0142] Data preprocessing: Before inputting electromagnetic flow test data into a convolutional network, several preprocessing steps are typically required. These can include normalization, denoising, and data augmentation to improve the model's robustness and performance.

[0143] Convolutional Layers: Convolutional layers are the core component of CNNs. They perform convolution operations on the input data using filters (or convolutional kernels) to extract local features. In the fluid injection task recognition network, convolutional layers can be used to detect local patterns in the flow data, such as periodic fluctuations and abrupt changes.

[0144] Activation functions: Activation functions introduce non-linearity, enabling networks to capture complex relationships and patterns. ReLU (Rectified Linear Unit) is one such activation function.

[0145] Pooling Layer: A pooling layer is used to reduce the spatial dimensionality of data while retaining important information. In time series analysis, pooling can help the network abstract more stable traffic state characteristics and reduce computational load.

[0146] Fully connected layers: After multiple convolutional and pooling layers, the network typically contains one or more fully connected layers, which can integrate local features and perform high-level inference.

[0147] Output Layer: The final output layer is responsible for generating the network's final prediction. In the injection task identification, the output layer may output a classification result, indicating whether the current flow rate is normal, insufficient, or excessive.

[0148] Training and optimization: Training a convolutional network involves tuning the network parameters using a large amount of labeled training data. Optimization algorithms such as stochastic gradient descent (SGD) and its variants (e.g., Adam) are used to minimize the loss function, which is the difference between the network's predictions and the true labels.

[0149] Evaluation and Validation: During training, the network's performance needs to be evaluated on a separate validation set to monitor for overfitting and tune hyperparameters. Finally, the network should undergo a final evaluation on a test set to determine its ability to generalize to unseen data.

[0150] In automated liquid injection applications, the liquid injection task identification network can analyze the data stream from the electromagnetic flowmeter in real time and identify any deviations from the normal flow pattern. If an anomaly is detected, it can trigger an alarm, adjust the injection speed or other relevant parameters, or even stop the injection process for inspection or maintenance.

[0151] As an example of a deep convolutional network, the liquid injection task recognition network can efficiently process and analyze complex electromagnetic flow test data, providing intelligent decision support for automated liquid injection systems, ensuring product quality and improving production efficiency.

[0152] In some other possible design approaches, the completed amplification injection task identification branch is obtained through the following S310-S320 debugging.

[0153] S310: Prior view of the amplification injection task for obtaining examples of amplified electromagnetic flow test data and the examples of amplified electromagnetic flow test data.

[0154] S320: Based on the example of amplification electromagnetic flow test data carrying the prior viewpoint of the amplification injection task, the basic amplification injection task identification branch is debugged to obtain the debugged amplification injection task identification branch.

[0155] In an exemplary beverage production scenario, consider finding ways to improve the accuracy and adaptability of its dispensing system. Specifically, this involves tuning the amplified dispensing task identification branch to improve system performance.

[0156] First, simulated electromagnetic flow test data are collected in the S310. This data represents various possible future scenarios, such as the use of new containers or changes in raw material flow characteristics. This data, along with corresponding prior knowledge of the amplification and injection mission, provides the basis for the upcoming commissioning work. These prior knowledge may be derived from expert experience, theoretical calculations, or actual results from similar past scenarios.

[0157] Next, in S320, these data examples with prior knowledge of the amplification injection task are used to fine-tune and optimize the underlying amplification injection task identification branch. This tuning process may involve training a machine learning model, through which the model learns how to accurately predict injection results based on the input flow data examples.

[0158] This design allows for better adaptation to new situations and reduces the need for manual adjustments to constantly changing production conditions. By introducing extensive simulation data and prior knowledge during the commissioning phase, a more intelligent and adaptive identification branch is created, which not only improves the accuracy of injection tasks but also enhances the system's robustness in the face of new challenges.

[0159] In yet another exemplary embodiment, the step of determining the traffic status mining indication includes S410-S450.

[0160] S410: Prior view of the amplification injection task for obtaining examples of amplified electromagnetic flow test data and the examples of amplified electromagnetic flow test data.

[0161] S420: Based on the identified branch of the past injection task that has been debugged, the amplified electromagnetic flow test data example is identified to obtain the past injection task prediction view of the amplified electromagnetic flow test data example.

[0162] S430: Combine the prior view of the amplification injection task and the prediction view of the past injection task to obtain the second global electromagnetic flow monitoring annotation of the amplification electromagnetic flow test data example.

[0163] S440: Based on the second global electromagnetic flow monitoring annotation and the example of amplified electromagnetic flow test data, debug the injection task identification network to be debugged to obtain a second injection task identification network for identifying amplified injection tasks and past injection tasks.

[0164] S450: Determine the flow state mining indication based on the neural network configuration weights of the state description mining subnet of the second injection task identification network.

[0165] In the above exemplary scheme, the determination of flow status mining indicators is to improve the predictive capability and adaptability of the automated injection system. S410 to S450 provide a process for optimizing flow monitoring and injection task identification by combining historical and simulated data.

[0166] S410 begins by collecting simulated (amplified) electromagnetic flow test data and corresponding preset evaluations or expected results (prior insights). This data can be generated by changing injection conditions or parameters to simulate different production scenarios or potential problems. Subsequently, in S420, this amplified data is processed using a previously debugged injection task identification branch to generate predictive labels (predictive insights) based on historical performance. Next, S430 merges these two insights to form a comprehensive second global electromagnetic flow monitoring annotation, which integrates practical experience and predictive insights. Based on these annotations, in S440, the injection task identification network to be debugged is fine-tuned to form a second injection task identification network capable of handling both new and existing injection tasks. Finally, in S450, flow state mining indicators are customized according to the neural network configuration weights of the state description mining subnet in the debugged second injection task identification network. These indicators will serve as a basis for decision-making, guiding how to analyze and process real-time flow test data.

[0167] Overall, this series of steps provides more robust decision support for automated liquid filling systems. Combined with expanded data models, the system can not only react based on past experience but also adapt to various future scenarios. This means production lines can adjust more quickly to new container types, formulation changes, or other production parameter variations, thereby reducing downtime, improving product quality, and ensuring the stability and reliability of the entire liquid filling process. Furthermore, through precise adjustment of the identification network, the production system can achieve more efficient operation, thereby saving resources, reducing waste, and improving overall production efficiency.

[0168] Furthermore, the previously identified injection task branch that has completed debugging is obtained through the following S510-S520 debugging process.

[0169] S510: Obtain past electromagnetic flow test data examples and the prior view of past injection tasks for the past electromagnetic flow test data examples.

[0170] S520: Based on the example of past electromagnetic flow test data carrying the prior viewpoint of the past injection task, the basic past injection task identification branch is debugged to obtain the debugged past injection task identification branch.

[0171] In an example of an automated liquid injection system, suppose a chemical plant wants to improve the accuracy and efficiency of its liquid injection process. To achieve this goal, the method described above was used to debug the identification branch of its past liquid injection tasks.

[0172] First, historical electromagnetic flow test data was collected using the S510, which records the flow rates observed during past production processes. Simultaneously, prior knowledge of past injection tasks was also gathered; these knowledge points may be expectations or assumptions based on past production experience, expert knowledge, or specific problems encountered previously.

[0173] Then, in S520, historical data with prior insights from previous injection missions are used to refine the underlying past injection mission identification branch. This refinement may involve retraining the machine learning model, adjusting parameters, or employing a new algorithm to enable the identification branch to more accurately predict injection mission outcomes based on electromagnetic flow test data.

[0174] By applying this method, automated injection systems become more intelligent, capable of accurately reviewing and learning lessons from historical data. Such systems not only improve the accuracy of injection tasks but also identify potential problems and prevent their recurrence. The optimized identification of past injection task branches gives the system better predictive capabilities, leading to fewer product defects, higher production efficiency, and lower operating costs.

[0175] In some examples, the steps for obtaining the prediction insights from the amplification injection task include S610-S630.

[0176] S610: Based on each identification branch in the completed amplification injection task identification branch, the past electromagnetic flow test data example is identified to obtain the task identification information sequence of the past electromagnetic flow test data example. The task identification information sequence includes the first task text mask sequence of the past electromagnetic flow test data example. The first task text mask sequence is the sequence of task text masks predicted by each identification branch for the past electromagnetic flow test data example.

[0177] S620: Based on a pre-set first mask correlation threshold, redundancy elimination is performed on the first task text mask sequence of the past electromagnetic flow test data example to obtain the current task text mask of the past electromagnetic flow test data example.

[0178] S630: Determine the prediction view of the amplification injection task based on the current task text mask of the past electromagnetic flow test data example.

[0179] Obtaining insights into the prediction of injection tasks is a key aspect of the intelligent upgrade of automated injection systems. S610 to S630 describe a process of extracting accurate predictions through in-depth analysis of historical flow data and removal of information redundancy.

[0180] First, in the S610, the pre-tuned amplification injection task identification branches are used to process historical electromagnetic flow test data. These identification branches analyze the data individually and generate a series of task text masks based on each branch's understanding and predictions of the data. This sequence can be viewed as a collection of multiple expert opinions, with each mask representing an identification result for a specific aspect.

[0181] Next, in S620, to optimize information utilization and eliminate redundancy, a mask correlation threshold is set. Only when the correlation between identified information exceeds this threshold will the corresponding information be retained. This step effectively simplifies the information sequence, leaving only the most representative and important identification results.

[0182] Finally, in S630, the predicted viewpoint for the amplification injection task is determined using the filtered current task text mask. This predicted viewpoint integrates the analysis results of various identification branches and eliminates unnecessary duplicate information, making the final prediction more accurate and reliable.

[0183] Overall, this series of steps enables the automated liquid injection system to handle complex production scenarios more intelligently and precisely. By integrating predictions from different identification branches and eliminating redundant information, the system receives clearer and more accurate guidance, thereby improving the effectiveness of liquid injection tasks. This approach not only improves production efficiency and product quality but also helps reduce errors and waste, further lowering costs. Furthermore, due to the system's deep understanding and analysis of data, it can adapt to new production changes.

[0184] In some examples, the task identification information sequence also includes the identification confidence level of each task text mask in the first task text mask sequence. Based on this, redundancy elimination is performed on the first task text mask sequence of the past electromagnetic flow test data example in S620 to obtain the current task text mask of the past electromagnetic flow test data example, including S621-S623.

[0185] S621: Redundancy elimination is performed on the first task text mask sequence of the past electromagnetic flow test data example to obtain the first deduplicated task text mask of the past electromagnetic flow test data example.

[0186] S622: Obtain each candidate task text mask of the past electromagnetic flow test data example from the first task text mask sequence to obtain a second task text mask sequence, wherein the second task text mask sequence is a sequence of the candidate task text masks, and the candidate task text mask is a task text mask whose mask correlation with the first deduplicated task text mask is greater than a second mask correlation threshold.

[0187] S623: Based on the recognition confidence of each task text mask in the second task text mask sequence, adjust the first deduplicated task text mask to obtain the current task text mask of the past electromagnetic flow test data example.

[0188] In the mentioned example, a specific application scenario is a complex automated electromagnetic flow monitoring system deployed on a production line in a smart manufacturing plant. This system is responsible for collecting data during the injection process and analyzing the flow status through a series of complex algorithms to ensure that each step of the operation meets predetermined standards. Now, the system introduces the concept of a task identification information sequence, which includes a first task text mask sequence and the identification confidence level of each task text mask. In S620, the system optimizes the task text mask sequence of past electromagnetic flow test data through a series of sub-steps (S621 to S623), eliminating redundancy, thereby obtaining a more accurate and efficient current task text mask.

[0189] First, in S621, the system processes the task text mask sequence of historical data, removing duplicate or unnecessary information to generate a deduplicated task text mask. Next, in S622, the system selects candidate task text masks with high correlation to the deduplicated task text masks, forming a second task text mask sequence. Finally, in S623, the system further adjusts the deduplicated task text masks based on the recognition confidence of these candidate task text masks to determine the current task text mask.

[0190] This design enables the electromagnetic flow monitoring system to perform injection tasks with higher accuracy and reliability. Through refined information processing, the system eliminates data redundancy, reduces noise, and improves its ability to reflect real-time conditions. This ensures a more stable and controllable injection process, reduces production deviations and errors, and enhances the efficiency and product quality of the entire production process.

[0191] In addition, in some other possible embodiments, the control of the target robot to perform automated injection processing based on the discrimination results of the previous injection task and the discrimination results of the amplification injection task described in S150 includes S151-S156.

[0192] S151: Obtain the control text of the automated liquid injection production line to be adjusted.

[0193] S152: Perform past injection task matching and expanded injection task matching on multiple robot injection events in the automated injection production line control text to obtain past injection task matching result set and expanded injection task matching result set.

[0194] S153: By setting a first update strategy, the set of past liquid injection task matching results is updated to obtain a first automated liquid injection production line control log that includes past liquid injection task control events.

[0195] S154: By setting a second update strategy, the matching result set of the amplification injection task is updated to obtain a second automated injection production line control log that includes amplification injection task control events.

[0196] S155: Based on the integration of the first automated liquid injection production line control log and the second automated liquid injection production line control log, an automated liquid injection production line control report matching the target control item in the automated liquid injection production line control text is obtained; the target control item includes at least one of past liquid injection task control events and expanded liquid injection task control events.

[0197] S156: Adjust the control scheme of the automated liquid injection production line control text based on the automated liquid injection production line control report.

[0198] In an improved solution for an automated liquid injection system, the liquid injection process for the target robot was refined and optimized. This process involves a series of steps, from acquiring the current control text to adjusting the control strategy based on historical and predictive data.

[0199] First, through S151, the control text currently used to guide the operation of multiple robots on the automated liquid injection production line was obtained. This text details the liquid injection events of the robots, i.e., the specific operating parameters and procedures.

[0200] Next, in S152, each robot injection event described in the control text is analyzed, and its association with known past injection tasks and preset amplification injection tasks is matched to generate two result sets: past injection task matching result set and amplification injection task matching result set.

[0201] Subsequently, the two result sets are processed in S153 and S154 using two different update strategies. The first update strategy integrates past experience, while the second update strategy incorporates predictions of possible future scenarios. These two strategies generate first and second automated liquid injection line control logs containing relevant control events, respectively.

[0202] In S155, these two control logs are integrated into a single control report, which includes information matching the target control items, providing comprehensive guidance for robot operation. Finally, in S156, based on this integrated control report, the control text for the automated liquid injection production line is precisely adjusted to ensure more efficient and accurate liquid injection operations.

[0203] This design allows the automated liquid injection system to combine historical successes with insights into potential future changes, resulting in a more flexible and intelligent control strategy. Such system upgrades not only improve production efficiency and product quality but also reduce error rates and material waste.

[0204] Based on the above, please refer to the following: Figure 2 The above is a block diagram of an electromagnetic flow testing and analysis system 100 provided in an embodiment of the present invention. The electromagnetic flow testing and analysis system 100 may include:

[0205] The data acquisition module 101 is used to acquire target electromagnetic flow test data; wherein, the target electromagnetic flow test data is obtained after performing electromagnetic flow test on the target automated liquid injection system, and the electromagnetic flow test of the target automated liquid injection system is implemented based on a pre-set electromagnetic flowmeter;

[0206] The instruction acquisition module 102 is used to acquire the flow status mining instruction of the target electromagnetic flow test data. The flow status mining instruction is obtained by processing at least one of the following: a past electromagnetic flow test data example carrying a first global electromagnetic flow monitoring annotation and an amplified electromagnetic flow test data example carrying a second global electromagnetic flow monitoring annotation. The first global electromagnetic flow monitoring annotation is obtained by interacting with the amplified injection task prediction view of the past electromagnetic flow test data example and the past injection task prior view of the past electromagnetic flow test data example. The second global electromagnetic flow monitoring annotation is obtained by interacting with the past injection task prediction view of the amplified electromagnetic flow test data example and the amplified injection task prior view of the amplified electromagnetic flow test data example.

[0207] The state mining module 103 is used to perform electromagnetic flow state description mining on the target electromagnetic flow test data according to the flow state mining instruction, and obtain the electromagnetic flow state description variables of the target electromagnetic flow test data.

[0208] Task discrimination module 104 is used to discriminate injection tasks based on the electromagnetic flow state description variables, and obtain past injection task discrimination results and amplified injection task discrimination results of the target electromagnetic flow test data;

[0209] The liquid injection control module 105 is used to control the target robot to perform automated liquid injection based on the judgment results of the past liquid injection tasks and the judgment results of the amplified liquid injection tasks.

[0210] For a description of the above functional modules, please refer to the corresponding method descriptions above.

[0211] Please refer to the following: Figure 3This invention also provides an electromagnetic flow testing and analysis system 100, including a processor 111, a memory 112, and a bus 113 connected to the processor 111. The processor 111 and the memory 112 communicate with each other via the bus 113. The processor 111 is used to call program instructions in the memory 112 to execute the aforementioned electromagnetic flow testing method applied to robots.

[0212] In summary, this embodiment of the invention obtains electromagnetic flow state description variables of the target electromagnetic flow test data by performing electromagnetic flow state description mining based on the flow state mining indication of the target electromagnetic flow test data, and obtains past injection task discrimination results and amplified injection task discrimination results of the target electromagnetic flow test data by performing task discrimination based on the electromagnetic flow state description variables of the target electromagnetic flow test data, and finally controls the target robot to perform automated injection processing based on the past injection task discrimination results and amplified injection task discrimination results.

[0213] Considering that the flow status mining indication is obtained by processing at least one of the past electromagnetic flow test data examples carrying the first global electromagnetic flow monitoring annotation and the amplified electromagnetic flow test data examples carrying the second global electromagnetic flow monitoring annotation, this allows for accurate and comprehensive identification of the different injection tasks corresponding to the past injection task sequences and the amplified injection task sequences from the perspective of electromagnetic flow test data.

[0214] Furthermore, given that the identified amplification injection task prediction viewpoints are used as examples of past electromagnetic flow test data for amplification injection task prior knowledge, and the identified past injection task prediction viewpoints are used as examples of amplification electromagnetic flow test data for past injection task prior knowledge, there is no need to reconfigure prior knowledge, thereby improving the timeliness of the overall solution.

[0215] Furthermore, since the first global electromagnetic flow monitoring annotation is obtained by interacting the prediction view of amplification injection tasks based on past electromagnetic flow test data examples with the prior view of past injection tasks based on past electromagnetic flow test data examples, the prior knowledge of amplification injection tasks in past electromagnetic flow test data examples is stable, improving the anti-interference capability of neural network debugging; even if the flow state mining indication is obtained in the case of small sample, the task discrimination can still accurately and comprehensively distinguish different injection tasks in past injection task sequences and amplification injection task sequences by using the electromagnetic flow state description variables of the target electromagnetic flow test data mined based on the flow state mining indication.

[0216] Finally, after accurately and comprehensively identifying the past injection task discrimination results and the amplified injection task discrimination results of the target electromagnetic flow test data, the target robot can be controlled based on the past injection task discrimination results and the amplified injection task discrimination results, thereby flexibly and adaptively realizing automated injection processing.

[0217] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, product, or cloud server that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, product, or cloud server. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, product, or cloud server that includes that element.

[0218] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0219] The above are merely embodiments of the present invention and are not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principle of the present invention should be included within the scope of the claims of the present invention.

Claims

1. A method for testing electromagnetic flow in robots, characterized in that, The method, applied to an electromagnetic flow testing and analysis system, includes: Acquire target electromagnetic flow test data; wherein, the target electromagnetic flow test data is obtained after performing electromagnetic flow test on the target automated liquid injection system, and the electromagnetic flow test of the target automated liquid injection system is implemented based on a pre-set electromagnetic flowmeter; The flow state mining indication of the target electromagnetic flow test data is obtained by processing at least one of the following: a past electromagnetic flow test data example carrying a first global electromagnetic flow monitoring annotation and an amplified electromagnetic flow test data example carrying a second global electromagnetic flow monitoring annotation. The first global electromagnetic flow monitoring annotation is obtained by interacting with the amplified injection task prediction view of the past electromagnetic flow test data example and the past injection task prior view of the past electromagnetic flow test data example. The second global electromagnetic flow monitoring annotation is obtained by interacting with the past injection task prediction view of the amplified electromagnetic flow test data example and the amplified injection task prior view of the amplified electromagnetic flow test data example. Based on the flow state mining instructions, electromagnetic flow state description mining is performed on the target electromagnetic flow test data to obtain the electromagnetic flow state description variables of the target electromagnetic flow test data. Based on the electromagnetic flow state description variables, injection task discrimination is performed to obtain the past injection task discrimination results and amplified injection task discrimination results of the target electromagnetic flow test data; Based on the judgment results of the previous injection tasks and the judgment results of the amplified injection tasks, the target robot is controlled to perform automated injection processing.

2. The electromagnetic flow testing method for robots as described in claim 1, characterized in that, The steps for determining the flow status mining indication include: Examples of past electromagnetic flow test data and prior viewpoints on past injection tasks based on those past electromagnetic flow test data examples; Based on the completed and debugged amplification injection task identification branch, the past electromagnetic flow test data examples are identified to obtain the amplification injection task prediction view of the past electromagnetic flow test data examples. By merging the prior viewpoints of the past injection missions and the predictive viewpoints of the amplified injection missions, a first global electromagnetic flow monitoring annotation is obtained from the past electromagnetic flow test data examples. Based on the first global electromagnetic flow monitoring annotation and the past electromagnetic flow test data example, the injection task identification network to be debugged is debugged to obtain the first injection task identification network for identifying amplified injection task items and past injection task items. The flow state mining indication is determined based on the neural network configuration weights of the state description mining subnet of the first injection task identification network.

3. The electromagnetic flow testing method for robots as described in claim 2, characterized in that, The debugged amplification injection task identification branch was obtained through the following steps: A priori view of the amplification injection task for obtaining examples of amplified electromagnetic flow test data and the examples of amplified electromagnetic flow test data. Based on the example of amplification electromagnetic flow test data carrying the prior viewpoint of the amplification injection task, the basic amplification injection task identification branch is debugged to obtain the debugged amplification injection task identification branch.

4. The electromagnetic flow testing method for robots as described in claim 1, characterized in that, The steps for determining the flow status mining indication include: A priori view of the amplification injection task for obtaining examples of amplified electromagnetic flow test data and the examples of amplified electromagnetic flow test data. Based on the identified branch of the past injection task that has been debugged, the example of the amplified electromagnetic flow test data is identified to obtain the past injection task prediction view of the example of the amplified electromagnetic flow test data. By merging the prior view of the amplification injection task and the prediction view of the past injection task, a second global electromagnetic flow monitoring annotation is obtained for the example of the amplification electromagnetic flow test data. Based on the second global electromagnetic flow monitoring annotation and the example of the amplified electromagnetic flow test data, the injection task identification network to be debugged is debugged to obtain a second injection task identification network for identifying amplified injection tasks and past injection tasks. The flow state mining indication is determined based on the neural network configuration weights of the state description mining subnet of the second injection task identification network.

5. The electromagnetic flow testing method for robots as described in claim 4, characterized in that, The previously identified injection task branch that has been successfully debugged was obtained through the following steps: Examples of past electromagnetic flow test data and prior viewpoints on past injection tasks based on those past electromagnetic flow test data examples; Based on the past electromagnetic flow test data examples that carry the prior viewpoints of the past injection tasks, the basic past injection task identification branch is debugged to obtain the debugged past injection task identification branch.

6. The electromagnetic flow testing method for robots as described in claim 1, characterized in that, The steps for obtaining the prediction insights for the amplification injection task are as follows: Based on the identification branches in the completed and debugged amplification injection task identification branches, the past electromagnetic flow test data examples are identified to obtain the task identification information sequence of the past electromagnetic flow test data examples. The task identification information sequence includes the first task text mask sequence of the past electromagnetic flow test data examples. The first task text mask sequence is the sequence of task text masks predicted by each identification branch for the past electromagnetic flow test data examples. Based on a pre-set first mask correlation threshold, redundancy is eliminated from the first task text mask sequence of the past electromagnetic flow test data example to obtain the current task text mask of the past electromagnetic flow test data example. The prediction view of the amplification injection task is determined based on the current task text mask of the past electromagnetic flow test data example.

7. The electromagnetic flow testing method for robots as described in claim 6, characterized in that, The task identification information sequence further includes the identification confidence level of each task text mask in the first task text mask sequence. The redundancy elimination of the first task text mask sequence of the past electromagnetic flow test data example to obtain the current task text mask of the past electromagnetic flow test data example includes: Redundancy elimination is performed on the first task text mask sequence of the past electromagnetic flow test data example to obtain the first deduplicated task text mask of the past electromagnetic flow test data example. From the first task text mask sequence, obtain each candidate task text mask of the past electromagnetic flow test data example to obtain a second task text mask sequence, wherein the second task text mask sequence is a sequence of the candidate task text masks, and the candidate task text mask is a task text mask whose mask correlation with the first deduplicated task text mask is greater than the second mask correlation threshold. Based on the recognition confidence of each task text mask in the second task text mask sequence, the first deduplicated task text mask is adjusted to obtain the current task text mask of the past electromagnetic flow test data example.

8. An electromagnetic flow testing and analysis system, characterized in that, include: The data acquisition module is used to acquire target electromagnetic flow test data; wherein, the target electromagnetic flow test data is obtained after performing electromagnetic flow test on the target automated liquid injection system, and the electromagnetic flow test of the target automated liquid injection system is implemented based on a pre-set electromagnetic flowmeter; The instruction acquisition module is used to acquire the flow status mining instruction of the target electromagnetic flow test data. The flow status mining instruction is obtained by processing at least one of the following: a past electromagnetic flow test data example carrying a first global electromagnetic flow monitoring annotation and an amplified electromagnetic flow test data example carrying a second global electromagnetic flow monitoring annotation. The first global electromagnetic flow monitoring annotation is obtained by interacting with the amplified injection task prediction view of the past electromagnetic flow test data example and the past injection task prior view of the past electromagnetic flow test data example. The second global electromagnetic flow monitoring annotation is obtained by interacting with the past injection task prediction view of the amplified electromagnetic flow test data example and the amplified injection task prior view of the amplified electromagnetic flow test data example. The state mining module is used to perform electromagnetic flow state description mining on the target electromagnetic flow test data according to the flow state mining instruction, and obtain the electromagnetic flow state description variables of the target electromagnetic flow test data. The task discrimination module is used to discriminate the injection task based on the electromagnetic flow state description variable, and to obtain the past injection task discrimination results and the amplified injection task discrimination results of the target electromagnetic flow test data; The liquid injection control module is used to control the target robot to perform automated liquid injection based on the judgment results of the previous liquid injection tasks and the judgment results of the amplified liquid injection tasks.

9. An electromagnetic flow testing and analysis system, characterized in that, It includes a processor, a memory, and a bus connected to the processor; the processor and the memory communicate with each other via the bus. The processor is used to call the computer program in the memory to execute the electromagnetic flow testing method for robots according to any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, It stores a program that, when executed by a processor, implements the electromagnetic flow testing method for robots as described in any one of claims 1-7.

Images