PAR intelligent operation optimization system and method
By preprocessing and hierarchizing the real-time participating data in the PAR intelligent operating system, a standard data group is formed, and the problem of manual intervention in the system in automated control is solved, and the effect of reducing employment needs and improving the level of system intelligence is achieved.
Patent Information
- Application Number
- CN202411609251.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-12
- Publication Date
- 2025-06-27
AI Technical Summary
PAR intelligent operating system still requires manual intervention in the process of automated control, resulting in a large demand for enterprise employment and the inability to effectively achieve the goal of reducing costs and increasing efficiency.
By obtaining real-time participation data that requires manual participation by the operation object at the current moment, perform feature preprocessing and hierarchical processing, forming a standard data group, and prompting and operating according to different processing priorities.
It realizes that a person can handle a large number of operational processes that require manual supervision, reduces the employment needs of enterprises, and improves the production efficiency and the intelligence level of the system.
Smart Images
Figure QLYQS_1 
Figure QLYQS_8 
Figure QLYQS_12
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of PAR intelligent operation, and particularly relates to a PAR intelligent operation optimization system and method. Background Art
[0002] With the continuous development of technology, the PAR (Precision Automation and Robotics) intelligent operation system will be applied and promoted in more fields. The application of the PAR intelligent operation system is mainly reflected in industrial automation, intelligent manufacturing, and other fields that require high-precision, high-efficiency, and high-adaptability control. In addition to the fields of industrial automation and intelligent manufacturing, the PAR intelligent operation system can also be applied to other fields that require high-precision, high-efficiency, and high-adaptability control, such as aerospace, medical robots, intelligent transportation, etc. In these fields, the capabilities of the system such as precise control, adaptive adjustment, and predictive maintenance can also play an important role in improving the performance and efficiency of the overall system. In the future, the PAR intelligent operation system will be innovatively applied in more fields, such as smart homes, smart cities, etc. These innovative applications will further improve people's quality of life and the intelligent level of cities.
[0003] With the continuous maturity of technology and the continuous expansion of application fields, the standardization and normalization of the PAR intelligent operation system will become an inevitable trend. However, there are still some deficiencies in the continuous development and application of the PAR intelligent operation system. The main problem is that the PAR intelligent operation system still requires manual intervention during the process of automatic control, and the frequency of manual participation will increase in more complex precise control. This leads to the fact that the PAR intelligent operation system still requires a large amount of human supervision, and the demand for enterprise employment is still very high, which cannot effectively achieve the expected effect of cost reduction and efficiency improvement. Summary of the Invention
[0004] The purpose of the present invention is to provide a PAR intelligent operation optimization system and method, which can optimize and classify the manual participation data, enabling one person to handle a large number of operation processes that require human supervision and participation, thereby reducing the enterprise's employment demand, optimizing the PAR intelligent operation system, and improving its intelligent level.
[0005] The technical solution adopted by the present invention is specifically as follows: A PAR intelligent operation optimization method, comprising the following steps: Obtain the real-time participation data that requires manual participation of the operation object at the current moment, and perform feature preprocessing on it with the sample data set in the database to make its feature scales consistent to form a data group; Select the real-time participating data from the data group as the reference data center point, and determine the reference data clusters according to the reference data center point through different evaluation methods; Evenly distribute each remaining sample in the data group to the reference data clusters, and calculate its distance from the reference data center point in the reference data cluster; For each reference data cluster, calculate the mean value of the distances of all its samples to obtain a new reference data center point, calculate the coordinates of the new reference data center point, and determine a new reference data cluster according to the coordinates; Continuously repeat the calculation and update of the reference data center point until the change of the reference data center point of the reference data cluster is less than a preset threshold or the maximum number of iterations is reached, and continuously classify the real-time participating data into groups with the same similar characteristics to form a standard data group; As the data continuously converges, perform corresponding comparison processing on the standard data group generated by the real-time participating data and the preset data group in the database, and perform hierarchical processing on the real-time participating data.
[0006] As a preferred solution, it further includes hierarchical processing data feedback, feeding back the manual participation grading result and processing opinion to the database, and determining whether the grading result is accurate according to the feedback data; If the grading result is accurate, supplement the real-time participating data to the database as a sample; If the grading result does not meet the expectation, remove the current real-time participating data from the database, mark it as deviation data, and continuously improve the control strategy and adjust parameters based on the deviation data.
[0007] As a preferred solution, the acquisition of the real-time participating data specifically includes: Collect the real-time operation status information data of the operation object through a real-time monitoring system, and analyze and process the collected data using the algorithms built into the system; Determine the abnormalities and deviations in the operation process, and calculate the corresponding deviation coefficient and sensing signal coefficient; Determine whether the deviation coefficient and the sensing signal coefficient exceed a preset threshold; If it does not exceed the preset threshold, continue to operate the object through the operating system; If it exceeds the preset threshold, output the real-time participating data that requires manual participation and perform hierarchical processing on the real-time participating data.
[0008] As a preferred solution, the evaluation method can adopt one or more of the silhouette coefficient, variance ratio criterion, within-cluster sum of squared errors, between-cluster sum of squared errors, and gap statistic; During the evaluation process of determining the reference data center point as the reference data cluster, at least two of the evaluation methods are selected, and different evaluation algorithms are sequentially and cyclically selected according to the generation order of the real-time participating data to expand the capacity of the reference data cluster.
[0009] As a preferred solution, the distance between each sample and the reference data center point in its reference data cluster is determined by the following formula: ; Where is the reference data cluster, is a certain sample in the data group, is the reference data center point, then belongs to the reference data cluster of the reference data center point to which it is closest .
[0010] As a preferred solution, the coordinates of the reference data center point are continuously calculated and updated by the following formula: ; Where is the updated reference data center point, is the reference data cluster, is a certain sample in the data group.
[0011] As a preferred solution, the process of continuous convergence of the data is determined by the following formula: ; Where represents the sum of squares of each sample point to its reference data center point; If does not reach the minimum value, can be fixed update the reference data center point of each cluster After the reference data center point changes, fix the value of the reference data center point re-divide the cluster If reaches the minimum value, and also converge simultaneously.
[0012] As a preferred solution, the preset criteria for the hierarchical processing include an emergency processing data group, an important processing data group, a regular processing data group, and a negligible processing data group, and different prompt states are equipped according to different processing priorities to prompt the staff to perform manual participation operations according to the prompt states.
[0013] A PAR intelligent operation optimization system, which is applied to the above PAR intelligent operation optimization method, includes: A real-time participation data preprocessing module, which is used to obtain the real-time participation data that requires manual participation of the operation object at the current moment and perform preprocessing; A reference data center point initialization module, which is used to select a reference data center point and determine a reference data cluster according to the reference data center point through different evaluation methods; A data sample allocation module, which is used to allocate the remaining samples in the data group to the reference data cluster and calculate the distance between them and the reference data center point; A reference data center point update module, which is used to calculate the distance mean of all data samples, obtain a new reference data center point, and determine a new reference data cluster according to the new reference data center point; A repeated iteration module, which is used to continuously repeat the operations of the data sample allocation module and the reference data center point update module to form a standard data group; A hierarchical processing result output module, which is used to, after continuous repeated iteration, perform corresponding comparison processing on the standard data group and the preset data group in the database, and perform hierarchical processing on the real-time participation data.
[0014] A PAR intelligent operation optimization terminal, the optimization terminal includes a memory and a processor, the memory stores code, and the processor is configured to obtain the code and execute the above PAR intelligent operation optimization method.
[0015] The technical effects achieved by the present invention are: By collecting the real-time participation data that requires manual participation and performing hierarchical processing, the present invention can divide the real-time participation data into emergency processing data, important processing data, regular processing data, and ignorable processing data according to the priority of the data to be processed. Furthermore, when working, staff can process them one by one according to the specific hierarchical situation, perform manual participation operations, enabling one person to process a large number of operation processes that require manual supervision and participation, and perform operations in an orderly and reasonable manner, reducing the manpower requirement while reducing operation errors and improving production efficiency; thereby reducing the enterprise's labor cost and optimizing the PAR intelligent operation system, improving its intelligent level; secondly, after manual participation in the operation, the evaluation of the hierarchical processing result can be carried out and the evaluation can be fed back to the PAR intelligent operation system to continuously optimize the system, thereby ensuring the accuracy and stability of the hierarchical processing, and further ensuring the accuracy and stability of the entire operation system. Description of the Drawings
[0016] Figure 1 is a flowchart of the PAR intelligent operation optimization method according to an embodiment of the present invention; Figure 2 is the flowchart of the PAR intelligent operation optimization system according to the embodiments of the present invention; Figure 3 is the schematic structural diagram of the PAR intelligent operation optimization terminal according to the embodiments of the present invention. Specific embodiments
[0017] In order to make the objectives and advantages of the present invention clearer and more understandable, the present invention will be specifically described below in conjunction with embodiments. It should be understood that the following text is only used to describe one or several specific embodiments of the present invention, and does not strictly limit the scope of protection specifically claimed by the present invention.
[0018] As Figure 1 shown, a PAR intelligent operation optimization method includes the following steps: Step 1. Obtain the real-time participation data that requires manual participation of the operation object at the current moment, and perform feature preprocessing on it with the sample data set in the database to make its feature scales consistent to form a data group.
[0019] Among them, the data group can be expressed as , since this data group belongs to the intelligent unsupervised learning process, so this data group can only see x and has no class label y.
[0020] It should be noted that the acquisition of the real-time participation data specifically includes: Step 1.1. Collect the real-time operation state information data of the operation object through a real-time monitoring system, and analyze and process the collected data using the algorithms built in the system.
[0021] Step 1.2. Determine the abnormalities and deviations in the operation process, and calculate the corresponding deviation coefficient and sensing signal coefficient.
[0022] Step 1.3. Determine whether the deviation coefficient and the sensing signal coefficient exceed the preset threshold.
[0023] Step 1.4. If it does not exceed the preset threshold, continue to operate the object through the operating system.
[0024] Step 1.5. If it exceeds the preset threshold, output the real-time participation data that requires manual participation, and perform hierarchical processing on the real-time participation data.
[0025] In the above Step 1.1 - Step 1.5, in practical applications, the PAR intelligent operating system can automatically adjust control parameters to eliminate deviations by real - time monitoring the operating status of the robot and the overall status of the production line, and combining historical data and intelligent algorithms for deviation data judgment. However, in some cases, such as when the deviation is large or the system cannot be automatically adjusted, the system requires manual participation for further evaluation and decision - making. At this time, it is necessary to obtain real - time participation data that requires manual participation and perform the processing of Step 1.
[0026] Step 2. Select the real - time participation data from the data group as the benchmark data center point, and determine the benchmark data clusters according to the benchmark data center point through different evaluation methods.
[0027] It should be noted that if k clustering benchmark data center points are randomly selected, then there are k benchmark data clusters. .
[0028] In this embodiment, the evaluation methods include silhouette coefficient, variance ratio criterion, within - cluster sum of squared errors, between - cluster sum of squared errors, gap statistic, and are randomly and circularly assigned to different benchmark data center points to determine the benchmark data clusters.
[0029] Of course, in other embodiments, the evaluation method can adopt one or more of the silhouette coefficient, variance ratio criterion, within - cluster sum of squared errors, between - cluster sum of squared errors, gap statistic.
[0030] At the same time, in the evaluation process of determining the benchmark data clusters from the benchmark data center points, at least two evaluation methods are selected, and different evaluation algorithms are sequentially and circularly selected according to the generation order of the real - time participation data to expand the capacity of the benchmark data clusters, improve the data evaluation difference, thereby improving the data classification accuracy and ensuring the accuracy of the final classification processing results.
[0031] Step 3. Evenly distribute each remaining sample in the data group to the benchmark data clusters, and calculate its distance from the benchmark data center point in the benchmark data cluster.
[0032] In some embodiments, the distance between each sample and the benchmark data center point in its benchmark data cluster is determined by the following formula: ; where, is the benchmark data cluster, is a certain sample in the data group, is the benchmark data center point, then belongs to the benchmark data cluster of the benchmark data center point to which it is closest , form a preliminary reference data cluster, thereby realizing the initial classification processing of real-time participation data.
[0033] Step 4. For each reference data cluster, calculate the mean distance of all its samples to obtain a new reference data center point, calculate the coordinates of the new reference data center point, and determine the new reference data cluster according to the coordinates.
[0034] In some embodiments, the coordinates of the reference data center point are continuously calculated and updated as determined by the following formula: ; where is the updated reference data center point, is the reference data cluster, is a certain sample in the data group.
[0035] It should be noted that in the process of each repeated calculation and update, it is a reclassification of the real-time participation data, so as to accurately classify the real-time participation data relying on the existing database in the system as much as possible, ensure the accuracy of classification, and avoid misoperations caused by classification errors.
[0036] Step 5. Continuously repeat Step 3 and Step 4 until the change of the reference data center point of the reference data cluster reaches the maximum number of iterations, and continuously classify the real-time participation data into groups with the same similar characteristics to form a standard data group.
[0037] It should be noted that in this embodiment, the maximum number of iterations can be preset according to specific operation requirements to ensure the corresponding data classification accuracy in this operation environment; of course, in other embodiments, it can be set until the change of the reference data center point of the reference data cluster is less than the preset threshold or reaches the maximum number of iterations.
[0038] In some embodiments, the process of data continuous convergence is determined by the following formula: ; where represents the sum of squares of each sample point to its reference data center point; Step 5.1. If does not reach the minimum value, can be fixed, update the reference data center point of each cluster After the reference data center point changes, fix the value of the reference data center point The cluster Step 5.2. If reaches the minimum value, and also converge simultaneously, without reaching the maximum number of iterations.
[0039] In the above Step 5.1 - Step 5.2, by continuously converging the data until reaching the convergence minimum value or the maximum number of iterations, the real-time participating data is finally classified into the data group most similar to it, thereby forming a standard data group and realizing the classification of the real-time participating data.
[0040] Step 6. As the data continuously converges, the standard data group generated from the real-time participating data is compared with the preset data groups in the database, and the real-time participating data is classified.
[0041] Among them, the preset criteria for classification include an emergency processing data group, an important processing data group, a routine processing data group, and a negligible processing data group, and different prompt states are assigned according to different processing priorities to prompt the staff to perform manual participation operations according to the prompt states.
[0042] In the actual application process, depending on the terminal device, different prompt methods can be matched. For example, when using a computer as the terminal device, different priorities of classification can be prompted through a combination of pop-up windows, vibrations, and sounds, so as to ensure that the staff can obtain the processing order of different priorities according to different prompt methods, and then perform the processing of manual participation operations according to this order, ensuring the stable and efficient operation of the PAR intelligent operation system.
[0043] In some embodiments, it further includes Step 7. Feedback on classified data. The results and opinions of manual participation in classification are fed back to the database, and it is determined whether the classification results are accurate according to the feedback data; Step 7.1. If the classification results are accurate, the real-time participating data is supplemented to the database as a sample; Step 7.2. If the classification results do not meet expectations, the current real-time participating data is removed from the database and marked as deviation data, and the control strategy and parameters are continuously improved based on the deviation data.
[0044] It should be noted that after the staff evaluates the processing results as accurate, the classification results of this time will be supplemented into the corresponding data groups according to the types of emergency processing data, important processing data, routine processing data, and negligible processing data, so as to optimize the learning path in the operation system, learn from user behavior and feedback using machine learning and deep learning technologies, and continuously optimize and improve the system performance.
[0045] Secondly, after the staff evaluate the processing result as inaccurate, the current grading result will be removed from the database. At the same time, the staff can re - sort the grading result according to emergency processing data, important processing data, regular processing data, and negligible processing data, and supplement it to the corresponding data group, so as to optimize the learning path in the operating system, and use machine learning and deep learning technologies to learn from user behavior and feedback, continuously optimizing and improving system performance.
[0046] As Figure 2 shown, a PAR intelligent operation optimization system, which is applied to the above - mentioned PAR intelligent operation optimization method, includes: A real - time participation data pre - processing module, which is used to obtain the real - time participation data that requires manual participation of the operation object at the current moment and perform pre - processing; A reference data center point initialization module, which is used to select a reference data center point and determine a reference data cluster according to the reference data center point through different evaluation methods; A data sample allocation module, which is used to allocate the remaining samples in the data group to the reference data cluster and calculate the distance between them and the reference data center point; A reference data center point update module, which is used to calculate the distance mean of all data samples to obtain a new reference data center point and determine a new reference data cluster according to the new reference data center point; A repeated iteration module, which is used to continuously repeat the operations of the data sample allocation module and the reference data center point update module to form a standard data group; A grading result output module, which is used to, after continuous repeated iteration, perform corresponding comparison processing on the standard data group and the preset data group in the database, and perform grading processing on the real - time participation data.
[0047] In the actual application process, the real - time participation data pre - processing module is used to obtain the real - time participation data that requires manual participation of the operation object at the current moment, and perform feature pre - processing with the sample data set in the database to make their feature scales consistent to form a data group, so as to realize the pre - processing of the current real - time participation data, and mix the real - time participation data with the samples in the database to facilitate re - classification; Then, the reference data center point initialization module selects the real - time participation data from the data group as the reference data center point, and determines a reference data cluster according to the reference data center point through different evaluation methods to form a classification reference and realize the specific classification of each real - time participation data; After that, the data sample allocation module evenly allocates each remaining sample in the data group to the reference data cluster and calculates the distance between it and the reference data center point in the reference data cluster to realize the initial grading processing of the real - time participation data; Then, for each reference data cluster, the reference data center point update module calculates the mean distance of all its samples to obtain a new reference data center point, calculates the coordinates of the new reference data center point, and determines the new reference data cluster according to the coordinates, so as to achieve a re - classification of the real - time participating data once. And the repeated iteration module continuously calculates and updates the reference data center point until the change of the reference data center point of the reference data cluster is less than a preset threshold or the maximum number of iterations is reached. The real - time participating data is continuously classified into groups with similar characteristics to form a standard data group, realizing the classification of the real - time participating data. Subsequently, as the data continuously converges, the hierarchical processing result output module compares the standard data group generated from the real - time participating data with the preset data groups in the database, performs hierarchical processing on the real - time participating data, and assigns different prompt statuses according to different processing priorities. Then, when working, the staff can process them one by one according to the specific hierarchical situation, perform manual participation operations, enabling one person to handle a large number of operation processes that require manual supervision and participation, and operate in an orderly and reasonable manner, reducing the manpower requirement while reducing operation errors and improving production efficiency.
[0048] In some embodiments, it further includes a hierarchical processing data feedback module. By feeding back the results and opinions of manual participation in hierarchical processing to the database, it determines whether the hierarchical results are accurate according to the feedback data. When the hierarchical results are accurate, the real - time participating data is supplemented to the database as a sample. When the hierarchical results do not meet expectations, the current real - time participating data is removed from the database and marked as deviation data, and the control strategy and parameters are continuously improved based on the deviation data, so as to optimize the learning path in the operating system. Using machine learning and deep learning technologies, it learns from user behavior and feedback, and continuously optimizes and improves system performance.
[0049] Such as Figure 3 shown, a PAR intelligent operation optimization terminal, the optimization terminal includes a memory and a processor. The memory stores code, and the processor is configured to obtain the code and execute the above - mentioned PAR intelligent operation optimization method.
[0050] In some embodiments, the PAR intelligent operation optimization terminal includes at least one processor, a communication bus, a memory, and at least one communication interface.
[0051] Among them, the processor is a central processing unit (CPU), which is used to control the execution of the PAR intelligent operation optimization method.
[0052] The communication bus is used to connect the above - mentioned components and includes at least one path for transmitting information between the above - mentioned components.
[0053] The communication interface selects a device such as any transceiver for communicating with other devices or communication networks, such as Ethernet, radio access network, wireless local area network, etc.
[0054] The memory can be selected from semiconductor memories [such as random access memory (RAM) and read-only memory (ROM)], magnetic memories [magnetic disks (such as hard disks, floppy disks), magnetic tape memories], optical memories [optical disc memories (such as CD-ROM, DVD-ROM) and magneto-optical discs], etc., and is connected to the processor through a communication bus for storing codes and controlled by the processor for execution.
[0055] The above are only the preferred embodiments of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the principle of the present invention, several improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of the present invention. The structures, devices, and operation methods not specifically described and explained in the present invention, unless otherwise specified and limited, are implemented according to the conventional means in the art.
Claims
1. A PAR intelligent operation optimization system and method, characterized in that: The following steps are involved: Obtain the real-time participation data of the operation object that requires manual participation at the current moment, and perform feature preprocessing with the sample data set of the database to keep its feature scale consistent to form a data group; Selecting real-time participation data from the data group as a benchmark data center point, and determining a benchmark data cluster according to the benchmark data center point through different evaluation methods; Evenly distribute each remaining sample in the data group to the benchmark data cluster, and calculate the distance between the sample and the benchmark data center point in the benchmark data cluster; For each of the benchmark data clusters, the distance mean of all samples thereof is calculated to obtain a new benchmark data center point, and the coordinates of the new benchmark data center point are calculated, and a new benchmark data cluster is determined according to the coordinates; Continuously and repeatedly calculating and updating the benchmark data center point until the change of the benchmark data center point of the benchmark data cluster is less than a preset threshold or reaches a maximum number of iterations, and continuously classifying the real-time participation data into groups with the same or similar characteristics to form a standard data group; As the data continues to converge, the standard data group generated by the real-time participation data is subjected to corresponding comparison processing with the data group preset in the database, and the real-time participation data is subjected to hierarchical processing.
2. A PAR intelligent operation optimization method according to claim 1, characterized in that: It also includes feedback of graded processing data, feeding back the results and opinions of manual graded processing to the database, and judging whether the graded results are accurate based on the feedback data; If the classification result is accurate, the real-time participation data is added to the database as a sample; If the classification result does not meet expectations, the real-time participation data will be removed from the database and marked as deviation data, and the control strategy and adjustment parameters will be continuously improved based on the deviation data.
3. A PAR intelligent operation optimization method according to claim 1, characterized in that: The acquisition of the real-time participation data specifically includes: The real-time monitoring system is used to collect real-time operating status information data of the operating object, and the system's built-in algorithm is used to analyze and process the collected data; Determine anomalies and deviations during operation and calculate corresponding deviation coefficients and sensor signal coefficients; Determining whether the deviation coefficient and the sensor signal coefficient exceed a preset threshold; If the preset threshold is not exceeded, continue to operate the object through the operating system; If it exceeds the preset threshold, real-time participation data requiring human participation is output, and hierarchical processing of the real-time participation data is performed.
4. A PAR intelligent operation optimization method according to claim 1, characterized in that: The evaluation method may adopt one or more of the following: silhouette coefficient, variance ratio criterion, intra-cluster error sum of squares, inter-cluster error sum of squares, and gap statistics; In the evaluation process of determining the benchmark data center as a benchmark data cluster, at least two of the evaluation methods are selected, and different evaluation algorithms are cyclically selected in turn according to the real-time participating data generation order to expand the capacity of the benchmark data cluster.
5. A PAR intelligent operation optimization method according to claim 1, characterized in that: The distance between each sample and the benchmark data center point in its benchmark data cluster is calculated by the following formula: ； in, is the benchmark data cluster, is a sample in the data group, is the reference data center point, then Belongs to the reference data center point closest to it The benchmark data cluster .
6. A PAR intelligent operation optimization method according to claim 5, characterized in that: The coordinates of the reference center point are repeatedly calculated and updated according to the following formula: ； in, is the updated reference data center point, is the benchmark data cluster, is a sample in the data group.
7. A PAR intelligent operation optimization method according to claim 6, characterized in that: The process of continuous data convergence is determined by the following formula: ； in, Represents each sample point The sum of squares to its reference center point; like The minimum value has not been reached and can be fixed Update the benchmark data center point for each cluster , the value of the fixed benchmark data center point after the benchmark data center point changes Re-cluster It also changes and iterates; like When the minimum value is reached, and Also converges at the same time.
8. The PAR intelligent operation optimization method according to claim 1, characterized in that: The preset standards for hierarchical processing include an emergency processing data group, an important processing data group, a routine processing data group, and a negligible processing data group, and different prompt states are provided according to different processing priorities to prompt staff to perform manual participation operations according to the prompt states.
9. A PAR intelligent operation optimization system, applied to the PAR intelligent operation optimization method according to any one of claims 1 to 8, characterized in that: include: The real-time participation data preprocessing module is used to obtain the real-time participation data of the operation object that requires human participation at the current moment and perform preprocessing; A benchmark data center point initialization module is used to select a benchmark data center point and determine the benchmark data cluster according to the benchmark data center point through different evaluation methods; A data sample allocation module is used to allocate the remaining samples in the data group to the benchmark data cluster and calculate the distance between the data cluster and the central point of the benchmark data cluster; The benchmark data center point update module is used to calculate the distance mean of all data samples, obtain a new benchmark data center point, and determine a new benchmark data cluster based on the new benchmark data center point; A repeated iteration module, used for continuously repeating the operations of the data sample allocation module and the reference data center point update module to form a standard data group; The hierarchical processing result output module is used to perform corresponding comparison processing on the standard data group and the preset data group in the database after repeated iterations, and to perform hierarchical processing on the real-time participating data.
10. A PAR intelligent operation optimization terminal, characterized in that: The optimization terminal includes a memory and a processor, the memory stores codes, and the processor is configured to obtain the codes and execute the PAR intelligent operation optimization method according to any one of claims 1 to 8.