A Big Data-Based System for Predicting and Evaluating Scientific and Technological Achievements

Abstract

This invention discloses a big data-based system for predicting and evaluating scientific and technological achievements, relating to the field of scientific and technological achievement prediction and evaluation technology. Specifically, it includes: a research and development process data collection module, an information technology achievement prediction module, and an information technology achievement comprehensive evaluation module. The research and development process data collection module collects historical research data, feature data, and network transmission performance data of scientific research projects in real time during the research and development process. The information technology achievement prediction module establishes a prediction model for the feature data of the research process of scientific and technological achievements, sets research process status indicators based on the feature data, and selects network transmission performance data within a stable time range for further training and optimization of the LSTM prediction model. The information technology achievement comprehensive evaluation module comprehensively evaluates the stability of the research process of information technology achievements by predicting real-time network transmission performance data and setting information research fluctuation indicators to predict the stability of the research process.

Description

Technical Field

[0001] This invention belongs to the field of technology for predicting and evaluating scientific and technological achievements, and specifically relates to a big data-based system for predicting and evaluating scientific and technological achievements. Background Technology

[0002] Technology foresight is a systematic and structured approach that aims to identify emerging technologies or combinations of technologies that may have a significant impact on the economy, society, and environment by predicting and analyzing future trends in technological development. It emphasizes a systematic, long-term, and strategic approach, involving comprehensive analysis across multiple fields such as economics, society, and the environment, and aims to select strategic research areas and general new technologies that are likely to generate the greatest economic and social benefits.

[0003] Previous systems primarily focused on static data such as the application and approval of research projects, neglecting dynamic data during the research and development process, such as laboratory equipment operation data, equipment status data, and research and development progress data. This data is crucial for a comprehensive evaluation of scientific and technological achievements. Traditional prediction methods are often based on simple statistical models or expert experience, lacking precise mathematical models and big data support, resulting in inaccurate predictions and failing to provide strong support for scientific research decisions. Existing scientific and technological achievement evaluation systems often only focus on the final research results, neglecting indicators such as stability and data integrity during the research process. These indicators are equally important for ensuring the smooth progress of research work.

[0004] Therefore, there is an urgent need for a big data-based predictive evaluation system for scientific and technological achievements, which can predict and evaluate the research process of scientific and technological achievements by integrating laboratory equipment operation data, equipment status data, and R&D progress data. Summary of the Invention

[0005] The purpose of this invention is to provide a big data-based system for predicting and evaluating scientific and technological achievements, which addresses the technical problem in existing technologies that focus only on the final research results while neglecting the stability and data integrity during the research process.

[0006] To achieve the above objectives, the present invention adopts the following technical solution:

[0007] A big data-based system for predicting and evaluating scientific and technological achievements includes:

[0008] The system includes a data collection module for the research and development process, a prediction module for information technology achievements, and a comprehensive evaluation module for information technology achievements. The data collection module for the research and development process is used to collect historical research data, characteristic data, and network transmission performance data of scientific research projects in real time during the research and development process, and to determine the characteristic data set corresponding to different research objectives.

[0009] Among them, historical research data includes research objectives and technical routes. The time period from preparation to completion of each research objective is recorded as a research stage, and the characteristic data and network transmission performance data of each research stage in the R&D laboratory are monitored in real time.

[0010] Feature data includes equipment operation data, equipment status data, and R&D progress data;

[0011] Network transmission performance data includes the number of error bits, total number of bits transmitted, and number of packet losses;

[0012] The information technology achievement prediction module sets research process status indicators based on feature data, selects network transmission performance data within a stable time range, establishes a prediction model for the feature data of the scientific and technological achievement research process based on LSTM network structure, and trains and optimizes the LSTM prediction model using network transmission performance data within a stable time range.

[0013] The comprehensive evaluation module for information technology achievements is used to comprehensively evaluate the stability of the research process of information technology achievements. By predicting real-time network transmission performance data, it sets information research fluctuation indicators to predict the stability of the research process.

[0014] Furthermore, the characteristic data sets corresponding to different research objectives are determined using the following methods:

[0015] The research objectives of the scientific and technological achievements are sorted according to the research sequence, and the resulting set of research objectives in the research process is denoted as... Where Y(u) represents the set of research objectives for scientific and technological achievement u, i represents the research objective numbered i, and n represents the total number of research objectives that scientific and technological achievement u needs to complete. The set of characteristic data of the research stage where research objective i is located is denoted as , where yx(i) represents the set of operational data of the R&D laboratory equipment in the research stage represented by the i-th research objective, xh(i) represents the set of material consumption data of the R&D laboratory in the research stage represented by the i-th research objective, and zt(i) represents the set of status data of the R&D laboratory equipment in the research stage represented by the i-th research objective.

[0016] The equipment operation data includes the actual operating time of the equipment within a specific time period and the energy consumption of the equipment during operation;

[0017] Equipment status data includes temperature changes of different equipment during the research process;

[0018] The R&D progress data represents the amount of data processed during the R&D process.

[0019] Furthermore, based on the feature data, the research process status indicators are set, specifically using the following method:

[0020] Based on the characteristic data set of the research phase in which research objective i is located, the actual operating time of the equipment during the research process is determined. Each consecutive segment of actual operating time is recorded as a time period of the research phase. The characteristic data of each time period are then combined using the formula... The following are the research process status indicators, where m represents the m-th time period, t represents time, A(m,t) represents the research process status indicator changing with time t in the m-th time period, sj(m,t) represents the data processing volume of the research process changing with time t in the m-th time period, obtained directly from the data processing tasks (such as computation and storage) recorded in the equipment log or data processing system within a specific time period, wd(m,t) represents the local temperature change of the equipment changing with time t in the m-th time period, obtained by real-time acquisition of temperature data during equipment operation by a temperature sensor and processing by the data collection module, and ny(m,t) represents the energy consumption value changing with time t in the m-th time period, measured in real-time using an electricity meter or energy monitoring equipment. Weighting coefficients representing the amount of data processed. Weighting coefficients representing temperature change values. The weighting coefficient represents the energy consumption value.

[0021] Furthermore, network transmission performance data within a stable time range is selected using the following method:

[0022] Based on historical data, a stable range of research process status indicators is set, denoted as (a, b). The research process status indicators for each research objective in each time period are calculated. By comparing the stable range of research process status indicators, the time intervals in which the research process status indicators fall within the stable range of research process status indicators for each time period are selected and denoted as the stable time range. The network transmission performance data of each research objective in the process of researching information technology achievements within the stable time range of each time period are retained and the retained network transmission performance data is denoted as stable transmission performance data.

[0023] Furthermore, a predictive model for the characteristic data of the scientific and technological achievement research process is established. The specific method is as follows:

[0024] A prediction model is constructed based on an LSTM network. The LSTM network structure includes input units, output units, forget units, state units, and recurrent units. New input information is determined by the input unit and then stored in the state unit. The state units are connected to the recurrent units, allowing the information within the state units to undergo linear self-circulation through the recurrent units. By pre-setting the weight values ​​of the forget unit, the output unit controls whether to output. The output value function of each unit is determined based on the sigmoid function, the input vector, the hidden layer vector, and different weight coefficients. The weight coefficients include: different bias weight coefficients for each unit, input weight coefficients, and recurrent weight coefficients for the input unit, forget unit, and output unit. The update method of the internal state of the LSTM model is determined based on the output value functions of the input unit, forget unit, and output unit.

[0025] Furthermore, the update method of the internal state of the LSTM model is determined based on the output value function of the input unit, forget unit, and output unit. The specific method is as follows:

[0026] The output values ​​of the input unit, forget unit, and output unit are calculated. The forget unit controls the degree to which information from the previous time step is retained in the state unit. The input unit controls the degree to which new information flows into the state unit, updating the value of the state unit. The output unit controls the output of information from the state unit to the hidden state. This is achieved by using the formula... This indicates the update method of the state unit, where t represents time. The state unit value represents time t. This represents the output of the forgetting unit at time t. This represents the output of the input unit at time t. tanh represents element-wise multiplication, and tanh represents the hyperbolic tangent function. The input value represents time t. This represents the hidden state at time t. Enter the weighting coefficients. Represents the hidden state weight coefficient. These represent the bias weighting coefficients of the state unit; these weighting coefficients can be determined based on specific circumstances and requirements. This represents the state unit value at the moment preceding time t.

[0027] Furthermore, the hidden state specifically includes:

[0028] Using formula This represents the hidden state, where t represents time. This represents the hidden state at time t. The state unit value represents time t. The input value represents time t. This represents the sigmoid function. tanh represents element-wise multiplication, and tanh represents the hyperbolic tangent function. The input weight coefficients represent the state unit. Enter the weighting coefficients. Represents the hidden state weight coefficient. These represent the bias weighting coefficients of the output unit; these weighting coefficients can be determined based on specific circumstances and requirements. This represents the hidden state at the moment preceding time t.

[0029] Furthermore, the LSTM prediction model is trained and optimized using network transmission performance data within a stationary time range. The specific method is as follows:

[0030] Extract a portion of the stable transmission performance data as training data, and use the remaining stable transmission performance data as control data. Use the training data as input data to train the LSTM prediction model. By comparing the output value of the LSTM prediction model with the control data, determine the actual prediction error value. Set an error threshold R. When the average error value between the output value of the LSTM prediction model and the control data within the time range j is less than R, it is determined that the LSTM prediction model training is complete.

[0031] Furthermore, an information research fluctuation index is set to predict the stationarity of the research process. The specific method is as follows:

[0032] Using formula Let p represent the information research fluctuation index, j represent the prediction duration j, t represent time, g(t) represent the number of error bits at time t, obtained by counting the number of failed verification bits during transmission using network monitoring tools (such as Wireshark) or hardware devices (such as routers), f(t) represent the number of data packets lost at time t, obtained by comparing the sequence numbers of data packets at the sending and receiving ends to count the number of unacknowledged or lost data packets, and v(t) represent the total number of bits transmitted at time t, obtained by using a network traffic counter (such as SNMP protocol) to count the total number of successfully transmitted bits in real time. Weighting coefficients representing the number of error bits. The weighting coefficient represents the number of lost packets. Weighting coefficients representing the total number of bits transmitted;

[0033] A threshold P is set for the information research fluctuation index. When the predicted information research fluctuation index p is less than the threshold P, it means that the research process of current information technology achievements is proceeding stably. When the predicted information research fluctuation index p is greater than or equal to the threshold P, it indicates that the future network transmission performance may have large fluctuations or instability.

[0034] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are:

[0035] 1. This invention establishes a predictive model for characteristic data of the scientific and technological achievement research process. It extracts some stable transmission performance data as training data and uses the remaining data as control data to test the prediction accuracy of the prediction model and ensure the reliability of the prediction results. By setting error thresholds and time ranges, it judges the training completion status of the LSTM prediction model, avoids the problems of overtraining or undertraining, improves the efficiency and effectiveness of model training, and realizes effective prediction of characteristic data of the scientific and technological achievement research process.

[0036] 2. Based on the characteristic data set of the research stage of the research target, this invention determines the actual running time of the equipment and sets the research process status indicators, which helps to understand the status of the research process more intuitively. By comprehensively using historical data to set the stable range of the research process status indicators and filtering out the stable time range, key network transmission performance data can be retained, providing a basis for subsequent decision-making.

[0037] 3. This invention, by predicting network transmission performance data over a future period and combining the predicted values ​​of network transmission performance data, obtains a predicted value for the information research fluctuation index. This can quantitatively assess the stability of the research process of information technology achievements, helping researchers to determine whether to continue promoting research work. Based on the prediction results, it can help to more accurately assess the future demand of scientific and technological achievements for research process resources, thereby optimizing resource allocation in advance, improving the stability and efficiency of the research process, and promoting the transformation of information technology achievements. Attached Figure Description

[0038] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0039] Figure 1 A flowchart of a big data-based scientific and technological achievement prediction and evaluation system is shown.

[0040] Figure 2 A flowchart illustrating the steps of a big data-based method for predicting and evaluating scientific and technological achievements is shown.

[0041] Figure 3 A flowchart illustrating the steps of the stationary time range screening method is shown. Detailed Implementation

[0042] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0043] like Figure 1 , Figure 2 , Figure 3 The system shown is a big data-based prediction and evaluation system for scientific and technological achievements, which specifically includes the following: a data collection module for the R&D process, a prediction module for information technology achievements, and a comprehensive evaluation module for information technology achievements.

[0044] The R&D process data collection module establishes a data interface with the scientific research project management system to achieve automatic synchronization and updating of information technology achievement data. It determines the historical research data of information technology achievements by using historical scientific research project application data, scientific research project approval data, and research process data. The historical research data includes the key elements of the big data scientific research achievement research plan: research objectives and technical routes. It determines the research objectives that need to be achieved to realize the information technology achievements, determines the technical routes designed to achieve different research objectives, and records the time period from preparation to completion of each research objective as a research stage. Using Internet of Things technology, it monitors the characteristic data and network transmission performance data of each research stage in the R&D laboratory in real time.

[0045] The characteristic data for each research phase includes equipment operation data, equipment status data, and research and development progress data. Among them, equipment operation data includes the actual operating time of the equipment within a specific time period and the energy consumption of the equipment during operation. In this embodiment, the energy consumption only considers the consumption of electrical energy. Equipment status data includes the temperature changes of different equipment under different operating conditions during the research process. Research and development progress data represents the amount of data processed during the research and development process.

[0046] The research objective set is sorted according to the research sequence of scientific and technological achievements, and the resulting research objective set in the scientific and technological achievement research process is denoted as... Where Y(u) represents the set of research objectives for scientific and technological achievement u, i represents the research objective numbered i, and n represents the total number of research objectives that scientific and technological achievement u needs to complete. The set of characteristic data of the research stage where research objective i is located is denoted as , where yx(i) represents the set of operational data of the R&D laboratory equipment in the research stage represented by the i-th research objective, xh(i) represents the set of material consumption data of the R&D laboratory in the research stage represented by the i-th research objective, and zt(i) represents the set of status data of the R&D laboratory equipment in the research stage represented by the i-th research objective.

[0047] In the process of researching information technology achievements, network transmission performance data that occurs during transmission is recorded through network monitoring tools or equipment logs. The network transmission performance data includes the number of error bits, the total number of transmitted bits, and the number of lost data packets.

[0048] The information technology achievement prediction module is used to establish a prediction model for the characteristic data of the research process of scientific and technological achievements. It uses historical characteristic data collected from different research stages as the input data set of the prediction model, and uses the prediction model to predict the characteristic data of the research process of scientific and technological achievements. It uses real-time characteristic data collected from different research stages as a control data set to verify the prediction accuracy of the prediction model.

[0049] Based on the characteristic data set of the research stage where research objective i is located, the actual operating time of the equipment during the research process is determined. Each consecutive segment of actual operating time is recorded as a time period of the research stage. The research process status index is set by comprehensively considering the characteristic data of each time period. The specific formulas for the research process status index of research objective i in each time period are as follows:

[0050] ;

[0051] Where m represents the m-th time period, t represents time, A(m,t) represents the research process status index changing with time t in the m-th time period, sj(m,t) represents the data processing volume of the research process changing with time t in the m-th time period, which is directly extracted from the data processing tasks (such as computation and storage) completed within a specific time period recorded by the equipment log or data processing system, wd(m,t) represents the local temperature change value of the equipment changing with time t in the m-th time period, which is obtained by collecting temperature data of the equipment in real time through temperature sensors and processing it after data collection module, and ny(m,t) represents the energy consumption value changing with time t in the m-th time period, which is measured in real time by using an electricity meter or energy monitoring equipment. Weighting coefficients representing the amount of data processed. Weighting coefficients representing temperature change values. The weighting coefficient represents the energy consumption value.

[0052] Based on historical data, a stable range of research process status indicators is set, denoted as (a, b). The research process status indicators for each research objective in each time period are calculated. By comparing the stable range of research process status indicators, the time intervals in which the research process status indicators fall within the stable range of research process status indicators for each time period are selected and denoted as the stable time range. The network transmission performance data of each research objective in the process of researching information technology achievements within the stable time range of each time period are retained and the retained network transmission performance data is denoted as stable transmission performance data.

[0053] The prediction model is constructed based on the LSTM network. The LSTM network structure includes input units, output units, forget units, state units, and recurrent units. New input information is determined by the input unit and then stored in the state unit. The state unit is connected to the recurrent unit, so that the information in the state unit is linearly self-circulated through the recurrent unit. The output unit controls whether to output by pre-setting the weight value of the forget unit. The output value function of each unit is determined based on the sigmoid function, the input vector, the hidden layer vector, and different weight coefficients. The weight coefficients include: different bias weight coefficients for each unit, input weight coefficients, and recurrent weight coefficients of the input unit, forget unit, and output unit.

[0054] The update method of the internal state of the LSTM model is determined based on the output value functions of the input unit, forget unit, and output unit. Specifically, this method involves calculating the output values ​​of the input unit, forget unit, and output unit; using the forget unit to control the degree of information retention of the state units from the previous time step; using the input unit to control the degree of new information flowing into the state units; updating the values ​​of the state units; and using the output unit to control the output of information from the state units to the hidden state. The update method of the state units is expressed by the following formula:

[0055] ;

[0056] Where t represents time. The state unit value represents time t. This represents the output of the forgetting unit at time t. This represents the output of the input unit at time t. tanh represents element-wise multiplication, and tanh represents the hyperbolic tangent function. The input value represents time t. This represents the hidden state at time t. Enter the weighting coefficients. Represents the hidden state weight coefficient. These represent the bias weighting coefficients of the state unit; these weighting coefficients can be determined based on specific circumstances and requirements. This represents the state unit value at the moment before time t. In this embodiment, each moment is defined as 0.4 seconds.

[0057] Furthermore, the hidden state is obtained from the output value of the output unit, and the specific calculation formula is as follows:

[0058] ;

[0059] Where t represents time. This represents the hidden state at time t. The state unit value represents time t. The input value represents time t. This represents the sigmoid function. tanh represents element-wise multiplication, and tanh represents the hyperbolic tangent function. The input weight coefficients represent the state unit. Enter the weighting coefficients. Represents the hidden state weight coefficient. These represent the bias weighting coefficients of the output unit; these weighting coefficients can be determined based on specific circumstances and requirements. This represents the hidden state at the moment preceding time t.

[0060] A portion of the stable transmission performance data is extracted as training data, and the remaining stable transmission performance data is used as control data. The training data is used as input data to train the LSTM prediction model. The actual prediction error value is determined by comparing the output value of the LSTM prediction model with the control data. A preset error threshold R is set. When the average error value between the output value of the LSTM prediction model and the control data within the time range j is less than R, the LSTM prediction model is considered to have completed training. In this embodiment, the length of the time range j is set to 60 seconds.

[0061] The comprehensive evaluation module for information technology achievements uses a trained LSTM prediction model to predict the collected real-time network transmission performance data. Taking the current time point as the end point, it predicts the network transmission performance data for the next j-hour period from the network transmission performance data included within a time period of y. This yields the predicted network transmission performance data value for j-hour period. The predicted network transmission performance data value is then used to set an information research fluctuation index. The specific formula for the information research fluctuation index is as follows:

[0062] ;

[0063] Where p represents the information research fluctuation index, j represents the prediction duration j, t represents time, g(t) represents the number of error bits at time t, obtained by counting the number of failed verification bits during transmission using network monitoring tools (such as Wireshark) or hardware devices (such as routers), f(t) represents the number of data packets lost at time t, obtained by comparing the data packet sequence numbers at the sending and receiving ends to count the number of unacknowledged or lost data packets, and v(t) represents the total number of bits transmitted at time t, obtained by using a network traffic counter (such as the SNMP protocol) to count the total number of successfully transmitted bits in real time. Weighting coefficients representing the number of error bits. The weighting coefficient represents the number of lost packets. The weighting coefficient represents the total number of bits transmitted.

[0064] A threshold P is set for the information research fluctuation index. When the predicted information research fluctuation index p is less than the threshold P, it indicates that the current research process of information technology achievements is proceeding stably. When the predicted information research fluctuation index p is greater than or equal to the threshold P, it indicates that the future network transmission performance may fluctuate significantly or become unstable. This may adversely affect the data integrity, accuracy, real-time performance, and overall research progress of information technology achievements. At this time, the system issues an early warning signal, prompting corresponding measures to be taken, such as optimizing network configuration, increasing network bandwidth, and using more reliable data transmission protocols, to ensure the smooth progress of the research work.

[0065] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

[0066] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to specific implementations. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A big data-based system for predicting and evaluating scientific and technological achievements, characterized in that, include: The system includes a data collection module for the research and development process, a prediction module for information technology achievements, and a comprehensive evaluation module for information technology achievements. The data collection module for the research and development process is used to collect historical research data, characteristic data, and network transmission performance data of scientific research projects in real time during the research and development process, and to determine the characteristic data set corresponding to different research objectives. Among them, historical research data includes research objectives and technical routes. The time period from preparation to completion of each research objective is recorded as a research stage, and the characteristic data and network transmission performance data of each research stage in the R&D laboratory are monitored in real time. Feature data includes equipment operation data, equipment status data, and R&D progress data; Network transmission performance data includes the number of error bits, total number of bits transmitted, and number of packet losses; The information technology achievement prediction module sets research process status indicators based on feature data, selects network transmission performance data within a stable time range, establishes a prediction model for the feature data of the scientific and technological achievement research process based on the LSTM network structure, and trains and optimizes the LSTM prediction model using network transmission performance data within a stable time range. Based on the characteristic data set of the research phase in which research objective i is located, the actual operating time of the equipment during the research process is determined. Each consecutive segment of actual operating time is recorded as a time period of the research phase. The characteristic data of each time period are then combined using the formula... Let A(m,t) represent the state indicators of the research process, where m represents the m-th time period, t represents time, A(m,t) represents the state indicator of the research process as a function of time t in the m-th time period, sj(m,t) represents the amount of data processed in the research process as a function of time t in the m-th time period, wd(m,t) represents the change in local temperature of the equipment as a function of time t in the m-th time period, and ny(m,t) represents the energy consumption as a function of time t in the m-th time period. Weighting coefficients representing the amount of data processed. Weighting coefficients representing temperature change values. Weighting coefficients representing energy consumption values; The comprehensive evaluation module for information technology achievements is used to comprehensively evaluate the stability of the research process of information technology achievements. By predicting real-time network transmission performance data, it sets information research fluctuation indicators to predict the stability of the research process. Using formula Let p represent the information research volatility index, j represent the prediction duration j, t represent time, g(t) represent the number of erroneous bits at time t, f(t) represent the number of data packets lost at time t, and v(t) represent the total number of transmitted bits at time t. Weighting coefficients representing the number of error bits. The weighting coefficient represents the number of lost packets. Weighting coefficients representing the total number of bits transmitted; A threshold P is set for the information research fluctuation index. When the predicted information research fluctuation index p is less than the threshold P, it means that the research process of current information technology achievements is proceeding stably. When the predicted information research fluctuation index p is greater than or equal to the threshold P, it indicates that the future network transmission performance will fluctuate significantly or be unstable.

2. The big data-based scientific and technological achievement prediction and evaluation system according to claim 1, characterized in that, The specific method for determining the characteristic data sets corresponding to different research objectives is as follows: The research objectives of the scientific and technological achievements are sorted according to the research sequence, and the resulting set of research objectives in the research process is denoted as... Where Y(u) represents the set of research objectives for scientific and technological achievement u, i represents the research objective numbered i, and n represents the total number of research objectives that scientific and technological achievement u needs to complete. The set of characteristic data of the research stage where research objective i is located is denoted as , where yx(i) represents the set of operational data of the R&D laboratory equipment in the research stage represented by the i-th research objective, xh(i) represents the set of material consumption data of the R&D laboratory in the research stage represented by the i-th research objective, and zt(i) represents the set of status data of the R&D laboratory equipment in the research stage represented by the i-th research objective. The equipment operation data includes the actual operating time of the equipment within a specific time period and the energy consumption of the equipment during operation; Equipment status data includes temperature changes of different equipment during the research process; The R&D progress data represents the amount of data processed during the R&D process.

3. The big data-based scientific and technological achievement prediction and evaluation system according to claim 1, characterized in that, The specific method for filtering network transmission performance data within a stable time range is as follows: Based on historical data, a stable range of research process status indicators is set, denoted as (a, b). The research process status indicators for each research objective in each time period are calculated. By comparing the stable range of research process status indicators, the time intervals in which the research process status indicators fall within the stable range of research process status indicators for each time period are selected and denoted as the stable time range. The network transmission performance data of each research objective in the process of researching information technology achievements within the stable time range of each time period are retained and the retained network transmission performance data is denoted as stable transmission performance data.

4. The big data-based scientific and technological achievement prediction and evaluation system according to claim 1, characterized in that, The specific method for establishing a predictive model for characteristic data of scientific and technological achievements research process is as follows: A prediction model is constructed based on an LSTM network. The LSTM network structure includes input units, output units, forget units, state units, and recurrent units. New input information is determined by the input unit and then stored in the state unit. The state units are connected to the recurrent units, allowing the information within the state units to undergo linear self-circulation through the recurrent units. By pre-setting the weight values ​​of the forget unit, the output unit controls whether to output. The output value function of each unit is determined based on the sigmoid function, the input vector, the hidden layer vector, and different weight coefficients. The weight coefficients include: different bias weight coefficients for each unit, input weight coefficients, and recurrent weight coefficients for the input unit, forget unit, and output unit. The update method of the internal state of the LSTM model is determined based on the output value functions of the input unit, forget unit, and output unit.

5. The big data-based scientific and technological achievement prediction and evaluation system according to claim 4, characterized in that, The update method for the internal state of the LSTM model is determined based on the output value functions of the input unit, forget unit, and output unit. The specific method is as follows: The output values ​​of the input unit, forget unit, and output unit are calculated. The forget unit controls the degree to which information from the previous time step is retained in the state unit. The input unit controls the degree to which new information flows into the state unit, updating the value of the state unit. The output unit controls the output of information from the state unit to the hidden state. This is achieved by using the formula... Let represent the update method of the state unit, where t represents time, c(t) represents the state unit value at time t, y(t) represents the output of the forget unit at time t, and i(t) represents the output of the input unit at time t. Let represent element-wise multiplication, tanh represent the hyperbolic tangent function, and x(t) represent the input value at time t. Let h(t) represent the input weight coefficients, and h(t) represent the hidden state at time t. Represents the hidden state weight coefficient. The bias weight coefficient represents the state unit, and c(t-1) represents the state unit value at the time t before time t.

6. The big data-based scientific and technological achievement prediction and evaluation system according to claim 5, characterized in that, Hidden state, specifically including: Using formula Let represent the hidden state, where t represents time, h(t) represents the hidden state at time t, c(t) represents the state cell value at time t, x(t) represents the input value at time t, and σ represents the sigmoid function. tanh represents element-wise multiplication, and tanh represents the hyperbolic tangent function. The input weight coefficients represent the state units. Enter the weighting coefficients. Represents the hidden state weight coefficient. The bias weight coefficient represents the output unit, and h(t-1) represents the hidden state at the time t before time t.

7. The big data-based scientific and technological achievement prediction and evaluation system according to claim 1, characterized in that, The LSTM prediction model is trained and optimized using network transmission performance data within a stationary time range. The specific method is as follows: Extract a portion of the stable transmission performance data as training data, and use the remaining stable transmission performance data as control data. Use the training data as input data to train the LSTM prediction model. By comparing the output value of the LSTM prediction model with the control data, determine the actual prediction error value. Set an error threshold R. When the average error value between the output value of the LSTM prediction model and the control data within the time range j is less than R, it is determined that the LSTM prediction model training is complete.

Images