Platform data optimization method, system and equipment for mathematical modeling training

Abstract

The invention discloses a platform data optimization method, system and equipment for mathematical modeling training, and relates to the technical field of teaching management. The platform data optimization method for mathematical modeling training comprises the following steps: judging a response state; processing data; and obtaining a platform stability operation score. The response state of the mathematical thinking platform in the preset simulation time period is monitored in real time to judge whether mathematical modeling training is carried out or not, if yes, the model learning parameters to be regulated and controlled are screened, and meanwhile the platform stability operation score in the model learning parameter regulation and control process is obtained; the mathematical thinking platform parameters are automatically regulated and controlled in combination with the model learning parameter adjustment range, so that the effect of improving the synchronization performance when the mathematical thinking training platform processes teaching data in modeling training is achieved; the problem that in the prior art, when a mathematical thinking training platform processes teaching data in modeling training, the synchronization performance is not high is solved.

Description

Technical Field

[0001] The present invention relates to the field of teaching management technology, and in particular to a platform data optimization method, system and device for mathematical modeling training. Background Art

[0002] With the rapid development of computer technology, the continuous expansion of mathematical application fields, and the rise of data science and artificial intelligence, significant progress has been made. These technological advances provide more powerful computing support, richer data sources, and more efficient algorithmic tools for mathematical modeling, thereby promoting the widespread application and development of mathematical modeling in various fields. On the mathematical modeling training platform, accurate and efficient data processing is not only a prerequisite for data optimization, but also the core of improving model accuracy and efficiency. In the process of data optimization, the platform needs to use a variety of optimization algorithms and techniques, such as gradient descent, stochastic gradient descent, Newton's method, etc., to find the optimal solution for model parameters.

[0003] It is worth mentioning that with the continuous maturity of artificial intelligence technology, the mathematical modeling training platform has begun to have intelligent teaching functions. The platform can use machine learning algorithms to analyze students' learning behaviors and habits based on their learning data and feedback, automatically adjust teaching content and difficulty, and provide students with personalized learning experiences. This intelligent teaching method can not only stimulate students' interest in learning, but also help them better understand and master the knowledge and skills of mathematical modeling.

[0004] Existing technology tracks students' learning progress in real time through a learning management system, collects students' feedback on mathematical thinking training courses, then conducts principal component analysis to extract data features and perform data dimensionality reduction and data compression to form a complete data set. Finally, by analyzing students' learning progress and test scores, the effectiveness of students' mathematical thinking training is evaluated, thereby improving the efficiency and accuracy of mathematical modeling training.

[0005] For example, the learning engine open platform management system announced in the invention patent with announcement number: CN112017084B includes: pre-processing and parsing the multimodal information of the target user in terms of vision, voice and touch during the teaching interaction process, so as to determine the command interaction between the target user and the open platform during the teaching interaction process, and then optimize the learning engine model according to the command interaction situation, so as to play images and / or sounds to interact with the target user for knowledge education.

[0006] For example, the patent application with publication number: CN108648123A discloses a network teaching platform based on big data and a method for managing the network teaching process using the same, including: a teaching system, an academic affairs management system and a big data analysis system; the teaching system and the academic affairs management system are bidirectionally connected via the HTTP protocol, which is used for bidirectional synchronization of teaching and academic affairs related business data such as student and course basic data and course selection information; the teaching system is connected to the big data analysis system, which is used for collecting structured, semi-structured and unstructured data in the teaching system and storing them in the big data analysis system; the big data analysis system is connected to the teaching system and the academic affairs management system respectively, and feeds back learning behavior characteristics and teaching quality analysis results to the teaching system and the academic affairs management system.

[0007] However, in the process of implementing the technical solution of the invention in the embodiments of the present application, it is found that the above-mentioned technology has at least the following technical problems: in the prior art, before processing the teaching data, the synchronization delay problem of the online teaching thinking training platform when simultaneously processing a large amount of real-time data is not fully considered; secondly, the design of the existing learning effectiveness standards often relies on the database, resulting in the model in the modeling course process being unable to timely and comprehensively reflect the students' actual training situation, thereby affecting the teaching effect of mathematical modeling and the effectiveness of parameter control measures; there is a problem of low synchronization performance when the mathematical thinking training platform processes the teaching data in the modeling training. Summary of the invention

[0008] The embodiments of the present application solve the problem of low synchronization performance of the mathematical thinking training platform in the prior art when processing the teaching data in modeling training by providing a platform data optimization method, system and equipment for mathematical modeling training, thereby achieving improved synchronization performance when the mathematical thinking training platform processes the teaching data in modeling training.

[0009] An embodiment of the present application provides a platform data optimization method for mathematical modeling training, comprising the following steps: step one, real-time monitoring of the response state of a mathematical thinking platform within a preset simulation period to determine whether to perform mathematical modeling training; step two, if mathematical modeling training is performed, comparing the acquired platform operation data of the mathematical thinking platform with the initial platform operation data to screen the model learning parameters to be regulated; step three, obtaining the platform stability operation score during the model learning parameter regulation process, and automatically regulating the mathematical thinking platform parameters based on the acquired platform stability operation score and the model learning parameter adjustment range, wherein the platform stability operation score represents quantitative data of the degree of influence of the first platform stability operation score, the second platform stability operation score and the third platform stability operation score on the operation stability of the mathematical thinking platform.

[0010] Furthermore, the real-time monitoring of the response status of the mathematical thinking platform within a preset simulation time period comprises the following specific steps: obtaining a first synchronous response target value, a second synchronous response target value, a third synchronous response target value and a fourth synchronous response target value of the mathematical thinking platform within the preset simulation time period, wherein: the first synchronous response target value is obtained by weighting the degree of difference between the obtained number of synchronous simulation training students and the target number of synchronous simulation training students through the synchronous simulation training student number weight factor; the second synchronous response target value is obtained by weighting the degree of difference between the obtained number of synchronous online users and the target number of synchronous online users through the synchronous online user number weight factor; the second synchronous response target value is obtained by weighting the degree of difference between the obtained number of synchronous online users and the target number of synchronous online users through the client synchronous response time weight factor. The degree of difference between the obtained client synchronous response time and the target client synchronous response time is weighted to obtain a third synchronous response target value; the degree of difference between the obtained client synchronous interaction time and the target client synchronous interaction time is weighted by the client synchronous interaction time weight factor to obtain a fourth synchronous response target value; the obtained first synchronous response target value, second synchronous response target value, third synchronous response target value and fourth synchronous response target value are averaged to obtain the synchronous response target value; the synchronous response target value represents quantitative data on the degree of influence of the first synchronous response target value, the second synchronous response target value, the third synchronous response target value and the fourth synchronous response target value on the synchronization performance of the mathematical thinking platform.

[0011] Furthermore, the specific process of determining whether to perform mathematical modeling training is as follows: determining whether the acquired synchronous response target value is greater than the synchronous response target value preset in the database; if so, sending a mathematical modeling training instruction to the mathematical thinking platform; otherwise, sending a warning information analysis instruction to the PLC control unit in the mathematical thinking platform, and performing initial data adjustment on the initial platform operation data of the mathematical thinking platform; the warning information analysis instruction is used to distinguish the type of warning information; the warning information includes a first warning information and a second warning information; the first warning information indicates that the acquired synchronous response target value is equal to the warning level corresponding to the synchronous response target value preset in the database; the second warning information indicates that the acquired synchronous response target value is less than the warning level corresponding to the synchronous response target value preset in the database; the initial platform operation data includes a first initial platform operation data and a second initial platform operation data; the first initial platform operation data includes an initial training switching duration and an initial network bandwidth; the second initial platform operation data includes an initial number of parallel requests and an initial learning step; the initial data adjustment includes a first initial data adjustment and a second initial data adjustment.

[0012] Furthermore, the specific process of adjusting the first initial data includes: reading the current initial training switching duration and initial network bandwidth from the first parameter storage area of ​​the mathematical thinking platform, and obtaining the current average platform data acquisition duration; when the obtained average platform data acquisition duration is greater than the average platform data acquisition duration preset in the database, prompting the PLC control unit to reduce the initial training switching duration by a preset amount and increase the initial network bandwidth by a preset amount according to the platform data acquisition average duration deviation; the specific process of adjusting the second initial data includes: reading the current initial number of parallel requests and initial learning step size from the second parameter storage area of ​​the mathematical thinking platform, and obtaining the current average platform data acquisition duration; when the obtained average platform data acquisition duration is greater than the average platform data acquisition duration preset in the database When the average duration of platform data acquisition is long, the PLC control unit is prompted to reduce the initial number of parallel requests of the preset simulation training period according to the first platform data acquisition average duration deviation; when the acquired platform data acquisition average duration is less than the platform data acquisition average duration preset in the database, the PLC control unit is prompted to increase the initial learning step of the preset simulation training period according to the second platform data acquisition average duration deviation; the platform data acquisition average duration deviation includes the first platform data acquisition average duration deviation and the second platform data acquisition average duration deviation; the first platform data acquisition average duration deviation represents the difference between the platform data acquisition average duration and the platform data acquisition average duration preset in the database; the second platform data acquisition average duration deviation represents the difference between the platform data acquisition average duration preset in the database and the platform data acquisition average duration.

[0013] Furthermore, the specific steps of screening the model learning parameters to be regulated are as follows: obtaining a first operating state compliance coefficient, and performing weighted summation and post-processing based on the platform operation data obtained within a preset simulation training period, the operating state deviation weight factor in the database, and the initial platform operation data to obtain a second operating state compliance coefficient; summing the obtained first operating state compliance coefficient and the second operating state compliance coefficient to obtain the operating state compliance coefficient; the platform operation data include training switching duration, network bandwidth, number of parallel requests and learning step size; the first operating state compliance coefficient is obtained by processing the synchronous response target value and the synchronous response target value weight factor; the operating state deviation weight factor includes training switching duration weight factor, network bandwidth weight factor, number of parallel requests weight factor and learning step size weight factor; the operating state compliance coefficient represents quantitative data of the degree of influence of the first operating state compliance coefficient and the second operating state compliance coefficient on the degree of compliance of the mathematical thinking platform data.

[0014] Furthermore, the specific process of screening the model learning parameters to be regulated is: determine whether the obtained operating state compliance coefficient is less than the operating state compliance coefficient preset in the database, if so, mark the model learning parameters of the model constructed by the corresponding student as qualified training parameters, otherwise mark the model learning parameters of the model constructed by the corresponding student as parameters to be regulated, and adjust the model learning parameters at the same time; the model learning parameters include weight parameters and bias parameters; the specific steps of adjusting the model learning parameters are: obtain the weight parameter deviation and the bias parameter deviation and send them to the PLC control unit in the mathematical thinking platform, and increase the weight parameter and the bias parameter respectively according to the obtained weight parameter deviation and bias parameter deviation; when the weight parameter deviation and the bias parameter deviation are both 0, record the corresponding weight parameter regulation times and bias parameter regulation times at the same time and record them as the optimal weight parameter regulation times and the optimal bias parameter regulation times; the weight parameter deviation represents the difference between the weight parameter when the weight parameter regulation is completed and the weight parameter before the weight parameter regulation; the bias parameter deviation represents the difference between the bias parameter when the bias parameter regulation is completed and the bias parameter before the bias parameter regulation.

[0015] Furthermore, the platform stability operation score is obtained by the following method: obtaining the weight parameter update frequency and bias parameter update frequency of the mathematical thinking platform during the model learning parameter adjustment period, and obtaining the first platform stability operation score in combination with the reference platform convergence parameters in the database; obtaining the second platform stability operation score and the third platform stability operation score and summing them up and analyzing the obtained first platform stability operation score to obtain the platform stability operation score; the reference platform convergence parameters include the mutual influence weight of the model learning parameter update frequency, the weight parameter update frequency weight factor, the bias parameter update frequency weight factor, the reference weight parameter update frequency and the reference bias parameter update frequency; the second platform stability operation score is obtained by processing the operating state compliance coefficient weight factor and the operating state compliance coefficient; the third platform stability operation score is obtained by processing the synchronous response batch weight factor, the synchronous response batch and the reference synchronous response batch.

[0016] Furthermore, the specific process of automatically controlling the parameters of the mathematical thinking platform based on the obtained platform stability operation score and the model learning parameter adjustment range is as follows: determine whether the obtained platform stability operation score is greater than the platform stability operation score preset in the database, and if so, send a convergence completion instruction to the PLC control unit in the mathematical thinking platform, otherwise send an automatic control instruction to the PLC control unit in the mathematical thinking platform; increase the mathematical thinking platform parameters according to the gain deviation and output saturation deviation obtained for a preset number of times until the gain deviation and output saturation deviation are equal to 0; the mathematical thinking platform parameters include gain and output saturation; the gain deviation represents the difference between the gain preset in the database and the initial gain of the mathematical thinking platform; the output saturation deviation represents the difference between the output saturation preset in the database and the initial output saturation of the mathematical thinking platform.

[0017] An embodiment of the present application provides a platform data optimization system for mathematical modeling training, including: a response state judgment module, a data processing module and a platform stability operation score acquisition module; wherein the response state judgment module is used to monitor the response state of the mathematical thinking platform in real time within a preset simulation period to determine whether to perform mathematical modeling training; the data processing module is used to compare the acquired platform operation data of the mathematical thinking platform with the initial platform operation data if mathematical modeling training is performed, and screen the model learning parameters to be regulated; the platform stability operation score acquisition module is used to obtain the platform stability operation score during the model learning parameter regulation process, and at the same time, automatically regulate the mathematical thinking platform parameters based on the acquired platform stability operation score and the model learning parameter adjustment range, and the platform stability operation score represents quantitative data of the degree of influence of the first platform stability operation score, the second platform stability operation score and the third platform stability operation score on the operation stability of the mathematical thinking platform.

[0018] An embodiment of the present application provides a device, which includes a memory for storing computer program instructions and a processor for executing the program instructions, wherein when the computer program instructions are executed by the processor, the device is triggered to execute the platform data optimization method for mathematical modeling training.

[0019] One or more technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages: 1. Real-time monitoring of the response status of the mathematical thinking platform within a preset simulation period is used to determine whether mathematical modeling training is to be performed. If so, the model learning parameters to be adjusted are screened, and the platform stability operation score during the model learning parameter adjustment process is obtained. The mathematical thinking platform parameters are automatically adjusted in combination with the model learning parameter adjustment range, thereby achieving a more accurate assessment of the operating stability of the mathematical thinking platform during the platform operation data processing process, and then achieving an improvement in the synchronization performance of the mathematical thinking training platform when processing the teaching data in the modeling training, effectively solving the problem of low synchronization performance of the mathematical thinking training platform when processing the teaching data in the modeling training in the prior art.

[0020] 2. By obtaining the first operating state compliance coefficient, and performing weighted summation based on the platform operating data obtained during the preset simulation training period, the operating state deviation weight factor in the database, and the initial platform operating data, the second operating state compliance coefficient is obtained. The first operating state compliance coefficient obtained is summed with the second operating state compliance coefficient to obtain the operating state compliance coefficient, thereby improving the accuracy and reliability of obtaining the operating state compliance coefficient, and further achieving a more accurate assessment of the operating state of the mathematical thinking platform.

[0021] 3. By obtaining the weight parameter update frequency and bias parameter update frequency of the mathematical thinking platform during the model learning parameter adjustment period, and then combining the reference platform convergence parameters in the database to obtain the first platform stability operation score, finally obtaining the second platform stability operation score and the third platform stability operation score and combining them with the obtained first platform stability operation score for summing and analysis to obtain the platform stability operation score, thereby improving the accuracy and reliability of the platform stability operation score, thereby achieving a more accurate evaluation of the stable operation status of the mathematical thinking platform. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 A flowchart of a platform data optimization method for mathematical modeling training provided in an embodiment of the present application; Figure 2 A schematic diagram of the structure of a platform data optimization system for mathematical modeling training provided in an embodiment of the present application; Figure 3 A data analysis flow chart of the platform operation data provided in an embodiment of the present application. DETAILED DESCRIPTION

[0023] The embodiments of the present application solve the problem of low synchronization performance of the mathematical thinking training platform in the prior art when processing the teaching data in modeling training by providing a platform data optimization method, system and equipment for mathematical modeling training. The response state of the mathematical thinking platform in a preset simulation period is monitored in real time to obtain a synchronization response target value, and then whether to perform mathematical modeling training is determined based on the obtained synchronization response target value. If mathematical modeling training is performed, the platform operation data of the acquired mathematical thinking platform is compared with the initial platform operation data during the data processing period, and the operation state compliance coefficient is obtained and the model learning parameters to be adjusted are screened. Then, the platform stability operation score in the process of model learning parameter adjustment is obtained. Finally, the mathematical thinking platform parameters are automatically adjusted based on the obtained platform stability operation score and the model learning parameter adjustment range, thereby achieving an improvement in the synchronization performance of the mathematical thinking training platform when processing the teaching data in modeling training.

[0024] The technical solution in the embodiment of the present application is to solve the problem that the above-mentioned mathematical thinking training platform has low synchronization performance when processing teaching data in modeling training. The overall idea is as follows: By real-time monitoring of the response status of the mathematical thinking platform within the preset simulation period to determine whether mathematical modeling training is to be carried out, if so, the model learning parameters to be adjusted are screened, and the platform stability operation score during the model learning parameter adjustment process is obtained. The mathematical thinking platform parameters are automatically adjusted in combination with the model learning parameter adjustment range, thereby achieving the effect of improving the synchronization performance of the mathematical thinking training platform when processing the teaching data in modeling training.

[0025] In order to better understand the above technical solution, the above technical solution will be described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0026] like Figure 1 As shown, it is a flow chart of a platform data optimization method for mathematical modeling training provided in an embodiment of the present application, and the method includes the following steps: step one, real-time monitoring of the response state of the mathematical thinking platform within a preset simulation time period to determine whether to perform mathematical modeling training; step two, if mathematical modeling training is performed, the platform operation data of the mathematical thinking platform obtained during the data processing period is compared with the initial platform operation data to screen the model learning parameters to be regulated; step three, obtaining the platform stability operation score during the model learning parameter regulation process, and automatically regulating the mathematical thinking platform parameters based on the obtained platform stability operation score and the model learning parameter adjustment range, the platform stability operation score represents quantitative data of the degree of influence of the first platform stability operation score, the second platform stability operation score and the third platform stability operation score on the operation stability of the mathematical thinking platform.

[0027] In this embodiment, in the preliminary round of the geometric mathematical modeling competition, participating students usually conduct mathematical modeling training through the mathematical thinking platform, which usually includes three processes: model preparation and hypothesis, model establishment, and model solution and verification. The following is a specific scenario description, a geometric model example, and the reason why the mathematical thinking platform considers synchronization during data processing: Participating students log in to the mathematical thinking platform and select the geometric mathematical modeling training module. The mathematical thinking platform provides a series of geometric modeling questions, and students perform modeling according to the requirements of the questions. For example, the question requires the establishment of a surface area model of a three-dimensional geometric body.

[0028] Students first prepare the model and make assumptions about the shape, size and other parameters of the geometric body. Then, they use the tools provided by the platform to build a mathematical model of the geometric body. Then, they solve the model to obtain the surface area of ​​the geometric body. Finally, they test the model to ensure the accuracy of the results.

[0029] During the modeling training process, students need to adjust the model parameters in real time and observe the model operation results fed back by the platform. In order to ensure the accuracy of the model results, the platform needs to update the model parameters in real time and calculate the model results synchronously. Therefore, during the data processing process, it is necessary to consider the synchronization of the platform operation to ensure the real-time synchronization of the model parameters and model results.

[0030] The data processing method provided in the present application scheme can ensure the stability and accuracy of the platform during the modeling training process by real-time monitoring of the platform response status, comparing the platform operation data, screening and automatically adjusting the model learning parameters, and realizing parameter optimization and stable operation of the mathematical thinking platform during the mathematical modeling training process. This process-based management method helps to improve the training efficiency and stability of the platform, thereby realizing the improvement of the synchronization performance of the mathematical thinking training platform when processing the teaching data in the modeling training.

[0031] Furthermore, the response status of the mathematical thinking platform within a preset simulation period is monitored in real time, and the specific steps are as follows: obtain the first synchronous response target value, the second synchronous response target value, the third synchronous response target value and the fourth synchronous response target value of the mathematical thinking platform within the preset simulation period, wherein: the first synchronous response target value is obtained by weighting the degree of difference between the obtained number of synchronous simulation training students and the target number of synchronous simulation training students through the synchronous simulation training student number weight factor; the second synchronous response target value is obtained by weighting the degree of difference between the obtained number of synchronous online users and the target number of synchronous online users through the synchronous online user number weight factor; the third synchronous response target value is obtained by weighting the degree of difference between the obtained client synchronous response time and the target client synchronous response time through the client synchronous response time weight factor; the fourth synchronous response target value is obtained by weighting the degree of difference between the obtained client synchronous interaction time and the target client synchronous interaction time through the client synchronous interaction time weight factor; the first synchronous response target value (i.e. ), the second synchronous response target value (i.e. ), the third synchronous response target value (i.e. ) and the fourth synchronous response target value (i.e. ) are summed and averaged to obtain the synchronous response target value; the synchronous response target value represents the quantitative data of the degree of influence of the first synchronous response target value, the second synchronous response target value, the third synchronous response target value and the fourth synchronous response target value on the synchronization performance of the mathematical thinking platform.

[0032] Among them, the specific restriction expression of the synchronous response target value is: ; Where m is the number of the preset simulation period, , M is the total number of preset simulation periods, e is a natural constant, It indicates the synchronous response target value of the mathematical thinking platform within the preset simulation period. represents the weight factor of the number of students in synchronous simulation training, It indicates the number of students who have received synchronous simulation training on the Mathematical Thinking Platform in the mth preset simulation period. Indicates the number of students in the target synchronous simulation training, Represents the weight factor of the number of simultaneous online users, represents the number of simultaneous online users of the Mathematical Thinking Platform in the mth preset simulation period, Indicates the number of target simultaneous online users. Indicates the weight factor of the client's synchronous response time. It represents the client synchronous response time of the mathematical thinking platform in the mth preset simulation period. Indicates the target client's synchronous response time. Indicates the weight factor of the client's synchronous interaction duration. It represents the client synchronous interaction duration of the mathematical thinking platform in the mth preset simulation period. Indicates the target client synchronous interaction duration.

[0033] In this embodiment, the units of the number of synchronous simulation training students and the target number of synchronous simulation training students are the same, both are individuals; the units of the number of synchronous online users and the target number of synchronous online users are the same, both are individuals; the unit of the client synchronous response time and the target client synchronous response time are the same, both are milliseconds (ms); the unit of the client synchronous interaction time and the target client synchronous interaction time are the same, both are seconds (s).

[0034] Specifically, the number of synchronous online users represents the sum of the number of participating students and teachers who are simultaneously online during the preset simulation period; the background management system of the mathematical thinking platform has its own performance monitoring, data recording and statistical functions, which can directly obtain the number of synchronous simulation training students, the number of synchronous online users, the client synchronous response time and the client synchronous interaction time; the target number of synchronous simulation training students, the target number of synchronous online users, the target client synchronous response time and the target client synchronous interaction time are set by preset personnel.

[0035] When evaluating the synchronization performance of the mathematical thinking platform, four key weight factors are used: the weight factor of the number of students in synchronous simulation training, the weight factor of the number of synchronous online users, the weight factor of the client synchronous response time, and the weight factor of the client synchronous interaction time. These weight factors are pre-set from the database and correspond to the number of students in synchronous simulation training, the number of synchronous online users, the client synchronous response time, and the client synchronous interaction time, respectively. There is a pre-set mapping relationship between these preset weight factors and the indicators they represent. This relationship can be a direct one-to-one correspondence or a more complex many-to-one relationship. For example, in actual operation, the real-time number of synchronous simulation training students, the number of synchronous online users, the client synchronous response time, and the client synchronous interaction time are input into this mapping relationship to quickly obtain the corresponding weight factors.

[0036] What is particularly important is that the values ​​of these four weight factors are limited to between 0 and 1, and their sum is 1. This setting ensures that all relevant factors can be comprehensively considered when calculating the fault risk assessment value, and guarantees the accuracy and reliability of the assessment results.

[0037] Before designing the platform data optimization method for mathematical modeling training, a special database was built to store a variety of preset data to support subsequent analysis and evaluation work. The content of this database includes but is not limited to the preset synchronous response target value, the preset average duration of platform data acquisition, the preset operation status compliance coefficient, the preset platform stability operation score, and the preset simulation period and data processing period. These values ​​are not set arbitrarily, but are set by professional and technical personnel based on the actual application scenarios and needs of the mathematical thinking platform. For example, the preset synchronous response target value is represented by the sum and average of the historical synchronous response target values ​​of the mathematical thinking platform in each historical simulation period in the database.

[0038] Of course, these values ​​in the database are not static. Technical personnel can set and fine-tune these values ​​according to actual debugging conditions and needs to ensure that they can truly and accurately reflect the actual performance and stability requirements of the mathematical thinking platform.

[0039] To simplify the analysis, we define , , , , the simplified calculation formula of the synchronous response target value is: , where represents the first synchronous response target value of the mathematical thinking platform in the mth preset simulation period, represents the second synchronous response target value of the mathematical thinking platform in the mth preset simulation period, represents the third synchronous response target value of the mathematical thinking platform in the mth preset simulation period, Represents the fourth synchronous response target value of the mathematical thinking platform in the mth preset simulation period.

[0040] It needs to be understood that the synchronous response target value increases with the increase in the number of synchronous simulation training students and the number of synchronous online users, and decreases with the increase in the client synchronous response time and the client synchronous interaction time. Among them, the increase in the number of synchronous simulation training students and the number of synchronous online users may increase the load of the platform, thereby affecting the client synchronous response time. At the same time, the extension of the client synchronous response time may reduce user satisfaction and participation, thereby affecting the client synchronous interaction time and the overall synchronization performance of the platform.

[0041] Therefore, when optimizing the synchronization performance of the mathematical thinking platform, it is necessary to comprehensively consider these parameters and their mutual influence, and take comprehensive measures to improve the platform's synchronous response capability, thereby achieving the improvement of the synchronization performance of the mathematical thinking training platform when processing the teaching data in modeling training, and effectively solving the problem of low synchronization performance of the mathematical thinking training platform when processing the teaching data in modeling training in the prior art.

[0042] Furthermore, the specific process of determining whether to conduct mathematical modeling training is as follows: determining whether the acquired synchronous response target value is greater than the synchronous response target value preset in the database, if so, sending a mathematical modeling training instruction to the mathematical thinking platform, otherwise sending a warning information parsing instruction to the PLC control unit in the mathematical thinking platform, and performing initial data adjustment on the initial platform operation data of the mathematical thinking platform; the warning information parsing instruction is used to distinguish the types of warning information; the warning information includes the first warning information and the second warning information; the first warning information indicates that the acquired synchronous response target value is equal to the warning level corresponding to the synchronous response target value preset in the database; the second warning information indicates that the acquired synchronous response target value is less than the warning level corresponding to the synchronous response target value preset in the database; the initial platform operation data includes the first initial platform operation data and the second initial platform operation data; the first initial platform operation data includes the initial training switching time and the initial network bandwidth; the second initial platform operation data includes the initial number of parallel requests and the initial learning step size; the initial data adjustment includes the first initial data adjustment (i.e., adjusting the first initial platform operation data) and the second initial data adjustment (i.e., adjusting the second initial platform operation data).

[0043] Among them, the specific process of adjusting the first initial data includes: reading the current initial training switching duration and initial network bandwidth from the first parameter storage area of ​​the mathematical thinking platform, and obtaining the current average platform data acquisition duration; when the obtained average platform data acquisition duration is greater than the average platform data acquisition duration preset in the database, the PLC (Programmable Logic Controller) control unit is prompted to reduce the initial training switching duration by a preset amount and increase the initial network bandwidth by a preset amount according to the deviation of the average platform data acquisition duration.

[0044] The specific process of adjusting the second initial data includes: reading the current initial number of parallel requests and the initial learning step from the second parameter storage area of ​​the mathematical thinking platform, and obtaining the current average duration of platform data acquisition; when the obtained average duration of platform data acquisition is greater than the average duration of platform data acquisition preset in the database, the PLC control unit is prompted to reduce the initial number of parallel requests for the preset simulation training period according to the first platform data acquisition average duration deviation; when the obtained average duration of platform data acquisition is less than the average duration of platform data acquisition preset in the database, the PLC control unit is prompted to increase the initial learning step for the preset simulation training period according to the second platform data acquisition average duration deviation; the platform data acquisition average duration deviation includes the first platform data acquisition average duration deviation and the second platform data acquisition average duration deviation; the first platform data acquisition average duration deviation represents the difference between the platform data acquisition average duration and the platform data acquisition average duration preset in the database; the second platform data acquisition average duration deviation represents the difference between the platform data acquisition average duration preset in the database and the platform data acquisition average duration.

[0045] In this embodiment, the preset amplitude is set by a preset personnel. Assuming that the average platform data acquisition time deviation is equal to 1.2 seconds, and the initial training switching time is equal to 5 minutes, the preset amplitude can be set to 10 seconds, that is, the initial training switching time is reduced from 5 minutes to 4 minutes and 50 seconds. It should be noted that each time the initial training switching time is reduced, it is necessary to obtain the average platform data acquisition time deviation once. When the platform data acquisition average time deviation is equal to 0, stop reducing the initial training switching time. Similarly, the adjustment of the initial network bandwidth, the initial number of parallel requests and the initial learning step is as shown above.

[0046] Specifically, the average duration of platform data acquisition is measured by a timer, which represents the average duration of response corresponding to the feedback from participating students obtained by the mathematical thinking platform within the preset simulation period; the platform operation data and the initial platform operation data are directly read through the background records of the background management system in the mathematical thinking platform, and the preset average duration of platform data acquisition is represented by the sum and average of the historical average duration of platform data acquisition of the mathematical thinking platform in the historical simulation period in the database.

[0047] This example achieves real-time adjustment of the initial platform operation data by dividing the initial platform operation data into regions (i.e., the first parameter storage area and the second parameter storage area) and comparing the average duration of platform data acquisition with the average duration of platform data acquisition preset in the database. This helps the mathematical thinking platform to better adapt to different operating environments, thereby improving the synchronous response performance of the mathematical thinking platform.

[0048] Further, the model learning parameters to be regulated are screened, and the specific steps are as follows: Obtain the first operating state compliance coefficient (i.e. ), and then the second operation state compliance coefficient (i.e. ); the first running state compliance coefficient and the second running state compliance coefficient are summed to obtain the running state compliance coefficient; the platform operation data includes training switching time, network bandwidth, number of parallel requests and learning step size; the first running state compliance coefficient is obtained by processing the synchronous response target value and the synchronous response target value weight factor; the running state deviation weight factor includes the training switching time weight factor, the network bandwidth weight factor, the parallel request number weight factor and the learning step size weight factor; the running state compliance coefficient represents the quantitative data of the influence of the first running state compliance coefficient and the second running state compliance coefficient on the degree of compliance of the mathematical thinking platform data.

[0049] Among them, when the acquired synchronous response target value is greater than the synchronous response target value preset in the database, the specific restriction expression of the operating state compliance coefficient is: ; Where t is the number of the data processing period, , T is the total number of data processing periods, e is a natural constant, It indicates the synchronous response target value of the mathematical thinking platform within the preset simulation period. It indicates that the running status of the mathematical thinking platform during the data processing period meets the coefficient, represents the synchronous response target value weight factor, represents the training switching duration weight factor, represents the training switching duration of the mathematical thinking platform in the tth data processing period, Indicates the initial training switching duration, represents the network bandwidth weight factor, represents the network bandwidth of the mathematical thinking platform during the tth data processing period, represents the initial network bandwidth, Represents the weight factor of the number of parallel requests, represents the number of parallel requests of the Mathematical Thinking Platform in the tth data processing period, Indicates the initial number of parallel requests, represents the learning step weight factor, represents the learning step length of the mathematical thinking platform in the tth data processing period, represents the initial learning step size.

[0050] In this embodiment, the unit of the training switching duration is the same as that of the initial training switching duration, which is milliseconds (ms); the unit of the network bandwidth is the same as that of the initial network bandwidth, which is megabits per second (Mbps); the unit of the number of parallel requests is the same as that of the initial number of parallel requests, which is pieces.

[0051] The synchronous response target value weight factor, the training switching time weight factor, the network bandwidth weight factor, the parallel request number weight factor and the learning step weight factor are respectively the influence of the synchronous response target value, the training switching time, the network bandwidth, the parallel request number and the learning step number preset in the database on the process of obtaining the running state compliance coefficient. Specifically, the database stores the preset weight factors corresponding to the synchronous response target value, the training switching time, the network bandwidth, the parallel request number and the learning step number. There is a pre-set mapping relationship between these weight factors and the synchronous response target value, the training switching time, the network bandwidth, the parallel request number and the learning step number. This mapping relationship can be one-to-one or many-to-one. For example, in actual applications, the real-time synchronous response target value, the training switching time, the network bandwidth, the parallel request number and the learning step number can be input into this mapping relationship, so as to quickly obtain the corresponding weight factor, and then more accurately calculate the running state compliance coefficient.

[0052] In this example, the value ranges of the synchronous response target value weight factor, the training switching time weight factor, the network bandwidth weight factor, the parallel request number weight factor, and the learning step weight factor are all limited to between 0 and 1, and the sum of the five is 1.

[0053] It should be understood that the running state compliance coefficient increases with the increase of the synchronous response target value and increases with the training switching time deviation (i.e. ), network bandwidth deviation (i.e. ), the number of parallel requests deviation (i.e. ) and the learning step size deviation (i.e. ) increases and decreases. When the training switching duration deviation increases, it indicates that the training switching efficiency of the mathematical thinking platform is reduced, which in turn affects the synchronous response performance of the mathematical thinking platform, that is, the synchronous response target value is correspondingly reduced.

[0054] When the training switching duration deviation increases, the mathematical thinking platform may need a longer time to adapt to the new training task or environment. In this case, the convergence speed of the generalization ability of the mathematical thinking platform during the adaptation process may decrease, that is, the running stability performance will decrease, thereby increasing the learning step of the mathematical thinking platform, aggravating the instability of the running performance of the mathematical thinking platform, and thus reducing the running state compliance coefficient.

[0055] To sum up, the operating status compliance coefficient is an indicator that comprehensively reflects the operating status and performance of the system. It is affected by multiple factors, including the synchronous response target value, training switching duration deviation, network bandwidth deviation, parallel request number deviation and learning step deviation. In order to optimize the system's operating status compliance coefficient, it is necessary to comprehensively consider the mutual influence relationship between these factors, thereby achieving the improvement of the synchronization performance of the mathematical thinking training platform when processing the teaching data in modeling training, and effectively solving the problem of low synchronization performance of the mathematical thinking training platform when processing the teaching data in modeling training in the prior art.

[0056] Furthermore, the specific process of screening the model learning parameters to be regulated is as follows: determine whether the obtained operating state compliance coefficient is less than the preset operating state compliance coefficient in the database; if so, mark the model learning parameters of the model constructed by the corresponding student as qualified training parameters, otherwise mark the model learning parameters of the model constructed by the corresponding student as parameters to be regulated, and adjust the model learning parameters at the same time; the model learning parameters include weight parameters and bias parameters; the specific steps of adjusting the model learning parameters are as follows: obtain the weight parameter deviation and the bias parameter deviation and send them to the PLC control unit in the mathematical thinking platform, and increase the weight parameter and the bias parameter respectively according to the obtained weight parameter deviation and the bias parameter deviation; when the weight parameter deviation and the bias parameter deviation are both 0, record the corresponding weight parameter regulation times and bias parameter regulation times at the same time and record them as the optimal weight parameter regulation times and the optimal bias parameter regulation times; the weight parameter deviation represents the difference between the weight parameter when the weight parameter regulation is completed and the weight parameter before the weight parameter regulation; the bias parameter deviation represents the difference between the bias parameter when the bias parameter regulation is completed and the bias parameter before the bias parameter regulation.

[0057] In this embodiment, it is assumed that before the weight parameter adjustment, the weight parameter of the model constructed by student A is 0.6, and after the weight parameter adjustment, the weight parameter of the model constructed by student A is 0.8. At this time, the weight parameter deviation is 0.2. According to the adjustment rules of the PLC control unit, the weight parameter of the mathematical thinking platform is gradually increased by 0.05 each time until the weight parameter is equal to 0.8, that is, the weight parameter deviation is 0, and the weight parameter adjustment is completed. Similarly, the adjustment of the bias parameter is the same as the adjustment steps of the weight parameter.

[0058] The preset operating state compliance coefficient is represented by the sum and average of the historical operating state compliance coefficients of the mathematical thinking platform in each historical data processing period in the database. The weight parameters and bias parameters are automatically learned by the gradient descent algorithm during the modeling training phase. This example realizes the automation and intelligence of the process by automatically judging the operating state compliance coefficient and adjusting the model learning parameters accordingly, reducing the need for manual intervention. Secondly, precise adjustment based on the weight parameter deviation and the bias parameter deviation helps to quickly converge to the optimal parameter combination, thereby improving the efficiency and accuracy of model learning parameter regulation, and further improving the stability of the mathematical thinking platform during data processing.

[0059] Furthermore, the model learning parameter adjustment range includes a weight parameter adjustment range and a bias parameter adjustment range; the weight parameter adjustment range indicates the range corresponding to the weight parameter when the weight parameter adjustment is completed and the weight parameter before the weight parameter adjustment; the bias parameter adjustment range indicates the range corresponding to the bias parameter when the bias parameter adjustment is completed and the bias parameter before the bias parameter adjustment; the platform stability operation score is obtained by the following method: obtaining the weight parameter update frequency and the bias parameter update frequency of the mathematical thinking platform during the model learning parameter adjustment period, and combining the reference platform convergence parameters in the database to obtain the first platform stability operation score (i.e. ); Get the second platform stability running score (i.e. ) and the third platform stability running score (i.e. ) and combined with the obtained first platform stability operation score, the platform stability operation score is obtained by summing up and analyzing; the reference platform convergence parameters include the mutual influence weight of the model learning parameter update frequency, the weight parameter update frequency weight factor, the bias parameter update frequency weight factor, the reference weight parameter update frequency and the reference bias parameter update frequency; the second platform stability operation score is obtained by the operation state compliance coefficient weight factor and the operation state compliance coefficient; the third platform stability operation score is obtained by the synchronous response batch weight factor, the synchronous response batch and the reference synchronous response batch.

[0060] Among them, when the obtained operation status compliance coefficient is not less than the operation status compliance coefficient preset in the database, the specific restriction expression of the platform stability operation score is: ; In the formula, e is a natural constant, It represents the platform stability operation score of the Mathematical Thinking Platform during the model learning parameter adjustment period. Indicates that the operating status complies with the coefficient weight factor, It indicates that the running status of the mathematical thinking platform during the data processing period meets the coefficient, represents the synchronous response batch weight factor, It represents the synchronous response batch of the mathematical thinking platform during the model learning parameter adjustment period. Indicates the reference synchronous response batch, Indicates that the update frequency of model learning parameters affects each other's weights, Represents the weight factor of the weight parameter update frequency, Indicates the weight parameter update frequency of the mathematical thinking platform during the model learning parameter adjustment period. represents the reference weight parameter update frequency, represents the bias parameter update frequency weight factor, Indicates the frequency of updating the bias parameters of the mathematical thinking platform during the model learning parameter adjustment period. Indicates the reference bias parameter update frequency.

[0061] In this embodiment, the units of the synchronization response batch and the reference synchronization response batch are the same, both are pieces; the units of the weight parameter update frequency and the reference weight parameter update frequency are the same, both are times / iteration; the units of the bias parameter update frequency and the reference bias parameter update frequency are the same, both are times / iteration.

[0062] The synchronous response batch, weight parameter update frequency and bias parameter update frequency are directly read through the background records of the background management system in the mathematical thinking platform; the reference synchronous response batch is represented by the sum and average of the historical synchronous response batches of the mathematical thinking platform in each historical model learning parameter adjustment period in the database, and the reference weight parameter update frequency and reference bias parameter update frequency are represented by the sum and average of the historical weight parameter update frequency and historical bias parameter update frequency of the mathematical thinking platform in each historical model learning parameter adjustment period.

[0063] The weight factors of the weight factor of the running state compliance coefficient, the weight factor of the synchronous response batch and the model learning parameter update frequency respectively represent the influence of the running state compliance coefficient, the synchronous response batch and the model learning parameter update frequency preset in the database on the platform stability running score acquisition process. Specifically, the database stores preset weight factors corresponding to the running state compliance coefficient, the synchronous response batch and the model learning parameter update frequency. There is a pre-set mapping relationship between these weight factors and the running state compliance coefficient, the synchronous response batch and the model learning parameter update frequency. This mapping relationship can be one-to-one or many-to-one. For example, in actual applications, the real-time running state compliance coefficient, the synchronous response batch and the model learning parameter update frequency can be input into this mapping relationship, so as to quickly obtain the corresponding weight factors, and then more accurately calculate the platform stability running score.

[0064] In this example, the operating state conformity coefficient weight factor, the synchronous response batch weight factor, and the model learning parameter update frequency, which influence each other, are all limited to a value range between 0 and 1, and the sum of the three is 1.

[0065] The database stores preset weight factors that are closely related to the platform stability operation score. A pre-defined mapping relationship is established between these weight factors and the corresponding weight parameter update frequency weight factors and bias parameter update frequency weight factors. It is worth noting that this mapping is not set arbitrarily. It can be one-to-one or many-to-one. In practical applications, when it is necessary to evaluate the stability of the mathematical thinking platform, the real-time weight parameter update frequency and bias parameter update frequency can be directly input into this preset mapping relationship, so that the weight parameter update frequency weight factors and bias parameter update frequency weight factors that match the platform stability operation score can be obtained quickly and accurately.

[0066] It is particularly important that in order to ensure the consistency and comparability of the evaluation, the value ranges of the weight parameter update frequency weight factor and the bias parameter update frequency weight factor in this example are strictly limited to between 0 and 1, and the sum of the two is 1.

[0067] Furthermore, the specific process of automatically adjusting the parameters of the mathematical thinking platform based on the obtained platform stability operation score and the adjustment range of the model learning parameters is as follows: determine whether the obtained platform stability operation score is greater than the platform stability operation score preset in the database, and if so, send a convergence completion instruction to the PLC control unit in the mathematical thinking platform, otherwise send an automatic control instruction to the PLC control unit in the mathematical thinking platform; increase the mathematical thinking platform parameters according to the gain deviation and output saturation deviation obtained for a preset number of times until the gain deviation and output saturation deviation are equal to 0; the mathematical thinking platform parameters include gain and output saturation; the gain deviation represents the difference between the gain preset in the database and the initial gain of the mathematical thinking platform; the output saturation deviation represents the difference between the output saturation preset in the database and the initial output saturation of the mathematical thinking platform.

[0068] In this embodiment, the preset platform stability operation score is represented by the sum and average of the historical platform stability operation scores of the mathematical thinking platform in each historical model learning parameter adjustment period in the database; this example uses a PLC (Programmable Logic Controller) control unit to implement an automated control process, which improves the efficiency and accuracy of the mathematical thinking platform parameter control. At the same time, the intelligent judgment and adjustment mechanism can make corresponding adjustment measures according to the real-time status of the mathematical thinking platform, ensuring the stable operation of the mathematical thinking platform and avoiding the blindness and uncertainty of traditional control methods.

[0069] The initial gain and initial output saturation of the mathematical thinking platform are usually the initial setting parameters of the mathematical thinking platform, where the gain is used to reflect the response degree of the mathematical thinking platform, and the output saturation is used to reflect the output range of the mathematical thinking platform. In practical applications, it is assumed that the initial gain of the mathematical thinking platform is 1.0, and the preset gain in the database is 1.2. At this time, the gain deviation is 0.2. According to the control rules of the PLC control unit, the gain of the mathematical thinking platform is gradually increased, such as increasing the gain by 0.05 at a time until the gain reaches 1.2. At this time, the corresponding gain deviation is 0, and the gain control is stopped. Similarly, the output saturation control is the same as the gain control steps.

[0070] It should be noted that during the adjustment of the mathematical thinking platform parameters, it is necessary to monitor the update time and update amplitude of the mathematical thinking platform parameters in real time, and the number of gain adjustments should not be greater than the number of optimal weight parameter adjustments, and the number of output saturation adjustments should not be greater than the number of optimal bias parameter adjustments, thereby improving the accuracy and reliability of the automatic control of the mathematical thinking platform parameters.

[0071] like Figure 2 As shown, it is a structural schematic diagram of a platform data optimization system for mathematical modeling training provided in an embodiment of the present application. The platform data optimization system for mathematical modeling training provided in an embodiment of the present application includes: a response state judgment module, a data processing module and a platform stability operation score acquisition module; wherein, the response state judgment module is used to monitor the response state of the mathematical thinking platform in a preset simulation time period in real time to determine whether to perform mathematical modeling training; the data processing module is used to compare the platform operation data of the mathematical thinking platform obtained with the initial platform operation data within the data processing period if mathematical modeling training is performed, and screen the model learning parameters to be regulated; the platform stability operation score acquisition module is used to obtain the platform stability operation score in the process of model learning parameter regulation, and at the same time, automatically regulate the mathematical thinking platform parameters based on the obtained platform stability operation score and the model learning parameter adjustment range, and the platform stability operation score represents the quantitative data of the degree of influence of the first platform stability operation score, the second platform stability operation score and the third platform stability operation score on the operation stability of the mathematical thinking platform.

[0072] In this embodiment, if Figure 3As shown, it is a data analysis flow chart of the platform operation data provided in the embodiment of the present application. Through the collaborative work among the response state judgment module, the data processing module and the platform stability operation score acquisition module, efficient and automated management and optimization of the mathematical thinking platform are realized. This system not only improves the stability and operation efficiency of the mathematical thinking platform, but also provides more reliable and efficient support for mathematical thinking training, thereby realizing the improvement of the synchronization performance of the mathematical thinking training platform when processing the teaching data in modeling training.

[0073] An embodiment of the present application provides a device, which includes a memory for storing computer program instructions and a processor for executing the program instructions, wherein when the computer program instructions are executed by the processor, the device is triggered to execute a platform data optimization method for mathematical modeling training.

[0074] To summarize, the embodiment of the present application monitors the response state of the mathematical thinking platform within a preset simulation period in real time to determine whether to conduct mathematical modeling training. If so, the model learning parameters to be adjusted are screened, and the platform stability operation score during the model learning parameter adjustment process is obtained. The mathematical thinking platform parameters are automatically adjusted in combination with the model learning parameter adjustment range, thereby achieving a more accurate assessment of the operating stability of the mathematical thinking platform during the platform operation data processing process, and further achieving an improvement in the synchronization performance of the mathematical thinking training platform when processing the teaching data in the modeling training, effectively solving the problem of low synchronization performance of the mathematical thinking training platform when processing the teaching data in the modeling training in the prior art.

[0075] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented 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.

[0076] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0077] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0078] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0079] Although the preferred embodiments of the present invention have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0080] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A platform data optimization method for mathematical modeling training, characterized in that: The following steps are involved: Step 1: real-time monitoring of the response status of the mathematical thinking platform within a preset simulation period to determine whether to conduct mathematical modeling training; Step 2: If mathematical modeling training is performed, the platform operation data of the acquired mathematical thinking platform is compared with the initial platform operation data to select the model learning parameters to be regulated; Step three, obtain the platform stability operation score during the model learning parameter adjustment process, and automatically adjust the mathematical thinking platform parameters based on the obtained platform stability operation score and the model learning parameter adjustment range. The platform stability operation score represents the quantitative data of the degree of influence of the first platform stability operation score, the second platform stability operation score and the third platform stability operation score on the mathematical thinking platform operation stability.

2. The platform data optimization method for mathematical modeling training according to claim 1, characterized in that: The real-time monitoring of the response status of the mathematical thinking platform within a preset simulation period comprises the following specific steps: The first synchronous response target value, the second synchronous response target value, the third synchronous response target value and the fourth synchronous response target value of the mathematical thinking platform within a preset simulation period are obtained, wherein: The first synchronous response target value is obtained by weighting the difference between the obtained number of synchronous simulation training students and the target number of synchronous simulation training students through the synchronous simulation training student number weight factor; The second synchronous response target value is obtained by weighting the difference between the acquired number of synchronous online users and the target number of synchronous online users through the weight factor of the number of synchronous online users; The third synchronization response target value is obtained by weighting the difference between the acquired client synchronization response time and the target client synchronization response time through the client synchronization response time weight factor; The fourth synchronous response target value is obtained by weighting the difference between the acquired client synchronous interaction time and the target client synchronous interaction time through the client synchronous interaction time weight factor; Calculating the average value of the first synchronous response target value, the second synchronous response target value, the third synchronous response target value, and the fourth synchronous response target value to obtain a synchronous response target value; The synchronous response target value represents quantitative data of the degree of influence of the first synchronous response target value, the second synchronous response target value, the third synchronous response target value and the fourth synchronous response target value on the synchronization performance of the mathematical thinking platform.

3. The platform data optimization method for mathematical modeling training according to claim 2, characterized in that: The specific process of determining whether to perform mathematical modeling training is as follows: Determine whether the acquired synchronous response target value is greater than the synchronous response target value preset in the database. If so, send a mathematical modeling training instruction to the mathematical thinking platform, otherwise send a warning information analysis instruction to the PLC control unit in the mathematical thinking platform, and at the same time make initial data adjustments to the initial platform operation data of the mathematical thinking platform; The warning information parsing instruction is used to distinguish the types of warning information; The warning information includes first warning information and second warning information; The first warning information indicates that the acquired synchronization response target value is equal to the warning level corresponding to the synchronization response target value preset in the database; The second warning information indicates that the acquired synchronization response target value is less than the warning level corresponding to the synchronization response target value preset in the database; The initial platform operation data includes first initial platform operation data and second initial platform operation data; The first initial platform operation data includes initial training switching duration and initial network bandwidth; The second initial platform operation data includes the initial number of parallel requests and the initial learning step length; The initial data adjustment includes a first initial data adjustment and a second initial data adjustment.

4. The platform data optimization method for mathematical modeling training according to claim 3, characterized in that: The specific process of the first initial data adjustment includes: Read the current initial training switching duration and initial network bandwidth from the first parameter storage area of ​​the mathematical thinking platform, and obtain the current average duration of platform data acquisition; When the average duration of platform data acquisition obtained is greater than the average duration of platform data acquisition preset in the database, the PLC control unit is prompted to reduce the initial training switching duration by a preset amount and increase the initial network bandwidth by a preset amount according to the deviation of the average duration of platform data acquisition; The specific process of the second initial data adjustment includes: Read the current initial number of parallel requests and initial learning step length from the second parameter storage area of ​​the mathematical thinking platform, and obtain the current average duration of platform data acquisition; When the average duration of platform data acquisition obtained is greater than the average duration of platform data acquisition preset in the database, the PLC control unit is prompted to reduce the number of initial parallel requests for the preset simulation training period according to the deviation of the first platform data acquisition average duration; When the average duration of platform data acquisition obtained is less than the average duration of platform data acquisition preset in the database, the PLC control unit is prompted to increase the initial learning step length of the preset simulation training period according to the deviation of the average duration of second platform data acquisition; The platform data acquisition average time deviation includes a first platform data acquisition average time deviation and a second platform data acquisition average time deviation; The first platform data acquisition average duration deviation represents the difference between the platform data acquisition average duration and the platform data acquisition average duration preset in the database; The second platform data acquisition average time deviation represents the difference between the platform data acquisition average time preset in the database and the platform data acquisition average time.

5. The platform data optimization method for mathematical modeling training according to claim 2, characterized in that: The specific steps of screening the model learning parameters to be regulated are as follows: Obtaining a first operating state compliance coefficient, and performing weighted summation and post-processing based on the obtained platform operating data within a preset simulation training period, the operating state deviation weight factor in the database, and the initial platform operating data to obtain a second operating state compliance coefficient; The first operating state compliance coefficient and the second operating state compliance coefficient are summed to obtain the operating state compliance coefficient; The platform operation data includes training switching duration, network bandwidth, number of parallel requests and learning step length; The first operation state compliance coefficient is obtained by processing the synchronous response target value and the synchronous response target value weight factor; The running state deviation weight factors include a training switching duration weight factor, a network bandwidth weight factor, a parallel request quantity weight factor and a learning step length weight factor; The operating state compliance coefficient represents quantitative data of the degree of influence of the first operating state compliance coefficient and the second operating state compliance coefficient on the degree of compliance of the mathematical thinking platform data.

6. The platform data optimization method for mathematical modeling training according to claim 5, characterized in that: The specific process of screening the model learning parameters to be regulated is: Determine whether the obtained running state compliance coefficient is less than the running state compliance coefficient preset in the database. If so, mark the model learning parameters of the corresponding student-built model as qualified training parameters. Otherwise, mark the model learning parameters of the corresponding student-built model as proposed adjustment parameters, and adjust the model learning parameters at the same time. The model learning parameters include weight parameters and bias parameters; The specific steps of adjusting the model learning parameters are: Obtain weight parameter deviation and bias parameter deviation and send them to the PLC control unit in the mathematical thinking platform, and increase the weight parameter and bias parameter respectively according to the obtained weight parameter deviation and bias parameter deviation; When the weight parameter deviation and the bias parameter deviation are both 0, the corresponding weight parameter adjustment times and bias parameter adjustment times are recorded at the same time and recorded as the optimal weight parameter adjustment times and the optimal bias parameter adjustment times; The weight parameter deviation represents the difference between the weight parameter after the weight parameter adjustment is completed and the weight parameter before the weight parameter adjustment; The bias parameter deviation represents the difference between the bias parameter when the bias parameter adjustment is completed and the bias parameter before the bias parameter adjustment.

7. The platform data optimization method for mathematical modeling training according to claim 5, characterized in that: The model learning parameter adjustment range includes a weight parameter adjustment range and a bias parameter adjustment range; The weight parameter adjustment range represents the range corresponding to the weight parameter after the weight parameter adjustment is completed and the weight parameter before the weight parameter adjustment; The bias parameter adjustment range represents the range corresponding to the bias parameter when the bias parameter adjustment is completed and the bias parameter before the bias parameter adjustment; The platform stability operation score is obtained by the following method: Obtain the weight parameter update frequency and bias parameter update frequency of the mathematical thinking platform during the model learning parameter adjustment period, and obtain the stability operation score of the first platform in combination with the reference platform convergence parameters in the database; Obtaining a second platform stability operation score and a third platform stability operation score and summing them with the obtained first platform stability operation score to obtain a platform stability operation score after analysis; The reference platform convergence parameters include the model learning parameter update frequency mutual influence weight, the weight parameter update frequency weight factor, the bias parameter update frequency weight factor, the reference weight parameter update frequency and the reference bias parameter update frequency; The second platform stability operation score is obtained by processing the operation state compliance coefficient weight factor and the operation state compliance coefficient; The third platform stability operation score is obtained by processing the synchronous response batch weight factor, the synchronous response batch and the reference synchronous response batch.

8. The platform data optimization method for mathematical modeling training according to claim 1, characterized in that: The specific process of automatically regulating the parameters of the mathematical thinking platform based on the obtained platform stability operation score and the model learning parameter adjustment range is as follows: Determine whether the obtained platform stability operation score is greater than the platform stability operation score preset in the database. If so, send a convergence completion instruction to the PLC control unit in the mathematical thinking platform; otherwise, send an automatic control instruction to the PLC control unit in the mathematical thinking platform; According to the gain deviation and output saturation deviation obtained for a preset number of times, the mathematical thinking platform parameters are increased until the gain deviation and the output saturation deviation are equal to 0; The mathematical thinking platform parameters include gain and output saturation; The gain deviation represents the difference between the gain preset in the database and the initial gain of the mathematical thinking platform; The output saturation deviation represents the difference between the output saturation preset in the database and the initial output saturation of the mathematical thinking platform.

9. A platform data optimization system for mathematical modeling training, characterized in that: include: Response status judgment module, data processing module and platform stability operation score acquisition module; The response state judgment module is used to monitor the response state of the mathematical thinking platform in real time within a preset simulation period to determine whether to perform mathematical modeling training; The data processing module is used to compare the acquired platform operation data of the mathematical thinking platform with the initial platform operation data to screen the model learning parameters to be regulated if mathematical modeling training is performed; The platform stability operation score acquisition module is used to obtain the platform stability operation score during the model learning parameter adjustment process, and at the same time automatically adjust the mathematical thinking platform parameters based on the acquired platform stability operation score and the model learning parameter adjustment range. The platform stability operation score represents quantitative data on the degree of influence of the first platform stability operation score, the second platform stability operation score and the third platform stability operation score on the operation stability of the mathematical thinking platform.

10. A device, characterized in that: The device includes a memory for storing computer program instructions and a processor for executing the program instructions, wherein, when the computer program instructions are executed by the processor, the device is triggered to execute the platform data optimization method for mathematical modeling training as described in any one of claims 1-8.

Images