Data prediction method and device, electronic equipment, storage medium and product
By using the global optimization process of search agents in the data prediction model and dynamically adjusting parameters, the problem of insufficient flexibility and accuracy of prediction models in the prior art is solved, and more efficient data prediction and decision support is achieved.
Patent Information
- Application Number
- CN202411151677.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-21
- Publication Date
- 2024-11-26
AI Technical Summary
In the prior art, the flexibility and accuracy of the prediction model are insufficient, it is difficult to adapt to data dimension changes, and depends on the experience of staff, resulting in large human resources consumption and inaccurate decision data.
The data prediction model trained by global optimization process based on search agents is adopted, and parameters are dynamically adjusted through genetic algorithms, simulation annealing algorithms, particle swarm optimization algorithms, etc. to reduce errors and improve the accuracy of prediction results.
It improves the flexibility and accuracy of the prediction model, reduces the dependence on human resources, and significantly improves the prediction accuracy and efficiency.
Smart Images

Figure CN119026748A_ABST
Abstract
Description
Technical Field
[0001] The present disclosure relates to the field of communication technology, and in particular to a data prediction method and device, electronic equipment, storage medium and product. Background Art
[0002] At present, in the existing technology, only a single parameter association method is often used to build a prediction model. However, the selection of a single parameter still depends on the experience of the staff to a certain extent. With the continuous development of technology and the increase in engineering complexity in reality, the requirements for prediction models are also increasing. How to make the prediction model flexibly adapt to various data dimension changes and obtain more accurate calculation results has become a technical problem that needs to be solved in this field. Summary of the invention
[0003] In view of the above problems, the present disclosure is proposed. The present disclosure provides a data prediction method and device, an electronic device, a storage medium and a product.
[0004] According to one aspect of the present disclosure, a data prediction method is provided, comprising: acquiring target project data; processing the target project data using a trained data prediction model to obtain target decision data output by the data prediction model; wherein the data prediction model is trained based on a global optimization process of at least one search agent.
[0005] According to another aspect of the present disclosure, a data prediction device is provided, comprising: a data acquisition module for acquiring target project data; a data processing module for processing the target project data using a trained data prediction model to obtain target decision data output by the data prediction model; wherein the data prediction model is trained based on a global optimization process of at least one search agent.
[0006] According to another aspect of the present disclosure, an electronic device is provided, including a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the method described in any one of the above embodiments.
[0007] According to another aspect of the present disclosure, a computer-readable storage medium is provided, on which a computer program / instruction is stored. When the computer program / instruction is executed by a processor, the method described in any of the above embodiments is implemented.
[0008] According to another aspect of the present disclosure, a computer program product is provided, including a computer program / instruction, which implements the method described in any of the above embodiments when the computer program / instruction is executed by a processor.
[0009] As will be described in detail below, according to a data prediction method and device, electronic device, storage medium and product of an embodiment of the present disclosure. After acquiring the target project data, the present disclosure will process the target project data by the trained data prediction model to output the target decision data. During the training process of the data prediction model, a search agent (such as a genetic algorithm, a simulated annealing algorithm, a particle swarm optimization algorithm, a pigeon flock algorithm) will be used to dynamically adjust the parameters in the data prediction model iteration process to reduce errors, so that the training prediction results output by the data prediction model are closer to the actual situation. During the training process, the present disclosure can flexibly construct a data prediction model directly according to the dimensions of the training data, without the need to select related data, thereby saving human resources.
[0010] It is to be understood that both the foregoing general description and the following detailed description are exemplary, and are intended to provide further explanation of the technology as claimed. BRIEF DESCRIPTION OF THE DRAWINGS
[0011] The above and other purposes, features and advantages of the present disclosure will become more apparent by describing the embodiments of the present disclosure in more detail in conjunction with the accompanying drawings. The accompanying drawings are used to provide a further understanding of the embodiments of the present disclosure and constitute a part of the specification. Together with the embodiments of the present disclosure, they are used to explain the present disclosure and do not constitute a limitation of the present disclosure. In the accompanying drawings, the same reference numerals generally represent the same components or steps.
[0012] Figure 1 It is a flowchart of a data prediction method provided by an embodiment of the present disclosure.
[0013] Figure 2 It is a test data graph of data prediction provided by an embodiment of the present disclosure.
[0014] Figure 3 This is another test data graph for data prediction provided by an embodiment of the present disclosure.
[0015] Figure 4 It is a flowchart of another data prediction method provided by an embodiment of the present disclosure.
[0016] Figure 5 It is a structural block diagram of a data prediction device provided by an embodiment of the present disclosure.
[0017] Figure 6 is a hardware block diagram illustrating an electronic device according to an embodiment of the present disclosure.
[0018] Figure 7 is a schematic diagram illustrating a computer program product according to an embodiment of the present disclosure. DETAILED DESCRIPTION
[0019] In order to make the purpose, technical solution and advantages of the present disclosure more obvious, the exemplary embodiments according to the present disclosure will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present disclosure, rather than all the embodiments of the present disclosure, and it should be understood that the present disclosure is not limited to the exemplary embodiments described here.
[0020] In the prior art, only a single parameter association method is often used to build a prediction model.
[0021] For example, by analyzing the existing workload quota, extracting key parameters, and establishing key parameter data source tables, workload transfer tables, and material transfer tables, these tables are interconnected through data processing software to form a budget model file. Use macro commands to process these tables, filter non-zero data entries, and automatically generate and export independent workload tables and material usage tables.
[0022] In the prior art, the key parameter extraction step often relies on the experience of the staff, and the interaction between various factors is captured by correctly setting the template, algorithm and device. Due to the different experience of the staff, the extraction of key factors may also be different, which affects the accuracy of the target decision data and consumes a lot of human resources.
[0023] Moreover, with the development of technology, the amount and complexity of data have also increased, and there are further requirements for the scalability of data prediction models.
[0024] In order to solve the above problems, the present disclosure provides a data prediction method, which optimizes parameters using a data prediction model trained by a global optimization process of at least one search agent, thereby obtaining more accurate target decision data. Figure 1 is a flow chart of a data prediction method provided by an embodiment of the present disclosure, such as Figure 1 As shown, the method specifically includes:
[0025] S101, obtaining target project data;
[0026] S102, using the trained data prediction model to process the target project data to obtain target decision data output by the data prediction model; wherein the data prediction model is trained based on a global optimization process of at least one search agent.
[0027] In this embodiment, the method can be applied to various fields, including but not limited to: communication engineering, machine learning, medical, and transportation. It can be used for cost estimation, sales strategy formulation, and inventory management.
[0028] The present disclosure can process the target project data through the trained data prediction model to obtain target decision data for auxiliary decision making.
[0029] The target project data may be a data set of various influencing factors obtained through screening, or may be a data set of influencing factors that have not been screened.
[0030] The target decision data is used to perform decision evaluation on the target project data; wherein the dimensions of the decision evaluation include at least one of the following: cost, feasibility, operational difficulty, and maintenance difficulty.
[0031] The type of data prediction model may include, but is not limited to, at least one of the following: a linear regression model, a BP neural network model, a logistic regression model, and a combined prediction model.
[0032] Before training, the data prediction model and the search agent need to be initialized, and the initialization content includes but is not limited to at least one of the following: the number of iterations (for example, 50 times), the boundary size, and the space dimension size.
[0033] The data prediction model will go through multiple iterations during the training process, thereby internally building a mapping relationship (i.e., parameters) between various influencing factors and training decision data. In the application process, the parameters obtained through training are used to obtain estimated target decision data based on the target project data.
[0034] In the process of training the data prediction model, the configuration of the initial parameters has a significant impact on the training process and final performance of the data prediction model, causing the network to converge to different local optimal solutions or global optimal solutions. A single search strategy may be difficult to find the global optimal mapping relationship. Therefore, during the training process, multiple groups of different parameter groups can be configured for the data prediction model for iterative search.
[0035] Furthermore, any parameter results obtained by iterating each parameter group based on any data prediction model can be optimized using a search agent, so as to quickly locate a better parameter area and find the globally optimal mapping relationship, thereby obtaining the global optimal solution of the data prediction model and improving the accuracy of the data prediction model.
[0036] The number of initially configured parameter groups is the same as the number of search agents.
[0037] The number of iterations N of the data prediction model and the number of iterations M of the search agent may be the same or different, wherein N is greater than or equal to M. For example, the number of iterations of the data prediction model is 50. After each iteration, if the obtained parameters need to be optimized once using the search agent, then M is also 50. Alternatively, after the data prediction model has been iterated 25 times, the search agent may be used for optimization once, and then M is 2.
[0038] The types of search agents include but are not limited to at least one of the following: genetic algorithm, particle swarm optimization, simulated annealing, ant colony algorithm, and pigeon colony algorithm.
[0039] In summary, after acquiring the target project data, the present disclosure will process the target project data through the trained data prediction model to output the target decision data. During the training process, the data prediction model will use a search agent (such as a genetic algorithm, a simulated annealing algorithm, a particle swarm optimization algorithm, and a pigeon flock algorithm) to dynamically adjust the parameters in the data prediction model iteration process to reduce errors, thereby making the training prediction results output by the data prediction model closer to the actual situation. During the training process, the present disclosure can flexibly construct a data prediction model directly according to the dimensions of the training data, without the need to select related data, thereby saving human resources.
[0040] In an illustrative embodiment, the iterative results of the data prediction model may be optimized using the pigeon colony algorithm.
[0041] The specific optimization steps may be: in any round of model training, based on the first model parameters of the first training model trained in the current round, the first model parameters are globally optimized using the pigeon swarm algorithm to obtain the second model parameters; wherein the first model parameters may be randomly generated or obtained by assigning values to other models. Next, the second model parameters are assigned to the first training model to obtain the second training model; when the second training model does not meet the first preset condition (for example, the maximum number of iterations of the data prediction model is reached, the maximum iteration time of the data prediction model is reached, and the error between the target decision data and is less than the preset threshold), the second training model is used as the initial data prediction model for the next round for model training. Among them, the training rounds can be set by yourself, and there is no specific restriction here.
[0042] In other words, the data prediction model can use the parameters optimized by the pigeon swarm algorithm (i.e., the first model parameters) to assign values to the data prediction model (i.e., the first training model) as a new data prediction model (i.e., the second training model), continue to train and verify the model, and repeat the above steps until the first preset condition is met and stop.
[0043] Among them, the parameter optimization step may include: S1, based on the first model parameters of the first training model and the fitness of each search agent, iteratively determine the global optimal fitness and the global optimal position; wherein the fitness is negatively correlated with the prediction error; any position is used to represent a set of model parameters.
[0044] Fitness can be understood as the quality of the current data prediction model, or the degree of similarity between the output obtained using the first training model and the actual result. Therefore, the prediction error used to measure the difference between the output and the actual result is negatively correlated with the fitness, that is, the higher the prediction error, the lower the fitness.
[0045] The prediction error may include but is not limited to at least one of the following: mean square error, root mean square error, mean absolute error, and cross entropy loss.
[0046] Exemplarily, the relationship between prediction error and fitness can be satisfied as follows:
[0047]
[0048] Where mse is the mean square error, TrainDataError is the first prediction error obtained by using the first training model to input the training data set, and TestDataError is the second prediction error obtained by using the first training model to input the test data set.
[0049] After completing an iteration and obtaining the fitness of each search agent, it is necessary to compare the highest fitness of the first training model with the current global optimal fitness. If the highest fitness is higher than the current global optimal fitness, the global optimal fitness needs to be updated, and its corresponding parameter configuration is the global optimal position.
[0050] S2, using the guide search agent to iteratively update the global optimal fitness and the global optimal position; wherein the guide search agent is a search agent with relatively high fitness; wherein the second model parameters are a set of model parameters corresponding to the global optimal position.
[0051] After obtaining the global optimal fitness and global optimal position, the algorithm will retain the parameter configuration that meets the second preset condition (such as the parameter configuration of the top 50% of the search agents in fitness) to calculate the central position. The central position can be used to provide the direction and position of movement for other search agents in subsequent iterations.
[0052] Exemplarily, the center position can be obtained by assigning a value to each search agent according to its fitness, thereby calculating a weighted average of the parameter configurations of each search agent that meets the second preset condition.
[0053] For example, the relationship can be:
[0054]
[0055] Among them, X c (t) is the center position of each search agent in the tth iteration; N p is the number of search agents that meet the second preset condition; i (t) is the position of the i-th agent in the t-th iteration; fitness(X i (t)) is the fitness of the i-th agent in the t-th iteration.
[0056] After obtaining the center position, the update direction of each search agent can be updated according to the center position. Specifically, each search agent will adjust its update direction according to the relative position between its current position and the center position, so that it is closer to the center position. This adjustment is usually achieved through certain rules. For example, the vector between the current position of the pigeon and the center position can be calculated, and the speed or position of the pigeon can be updated accordingly.
[0057] For example, the relationship may be satisfied as follows:
[0058] X′ i (t) = X i (t)+rand*(X c (t)-X i (t));
[0059] Among them, X′ i (t) is the position of the i-th agent in the t-th iteration after the update; rand is the random function.
[0060] In this way, the position update, that is, the parameter optimization, can be completed.
[0061] Before step S101, the collected data is raw data. After processing the raw data, the target project data can be obtained.
[0062] In an illustrative embodiment, the raw data may be preprocessed, wherein the preprocessing includes but is not limited to at least one of the following: data cleaning and normalization, so as to remove abnormal values and duplicate values and unify the value ranges of different features.
[0063] The methods used for data cleaning include but are not limited to at least one of the following: data screening, record number verification, value range judgment method, year-on-year and quarter-on-quarter methods.
[0064] The normalization method includes but is not limited to at least one of the following: Z-score, linear normalization, nonlinear normalization, and decimal normalization.
[0065] The following will explain in conjunction with specific embodiments.
[0066] For example, first, the historical data is evaluated, missing values are found through data screening, and the missing values are filled with the mean. Next, the Z-score method is used to identify outliers in the original data and delete them. For duplicate values, duplicate records in the data can be checked and deleted.
[0067] Next, the Z-score method can be used to normalize the original data after data cleaning:
[0068]
[0069] Among them, f(X) is the normalized value of the original data X after data cleaning; τ is the mean of the influencing factors to which X belongs; σ is the standard deviation of the influencing factors to which X belongs.
[0070] Next, the target project data can be obtained, and then the target project data can be input into the data prediction model for training.
[0071] Optionally, before preprocessing the original data, various influencing factors in the original data may be selected to obtain T relatively important influencing factors and establish a database, wherein the number of T may be set according to actual conditions.
[0072] Taking communication engineering as an example, we can infer 15 more important influencing factors through the historical data of wired communication engineering budget, which may include material data and engineering data.
[0073] The material data includes at least one of the following: bandwidth capacity, number of central computer rooms, size of central computer rooms, optical splitters, high-speed fiber switches, single-mode fiber materials, fiber length, cable length, number of signal amplifiers, number of uninterruptible power supplies, and number of temperature and humidity monitoring;
[0074] The engineering data includes at least one of the following: length of the excavation site, number of access points, maintenance time, and construction period.
[0075] Please refer to the following table to get the database:
[0076]
[0077] The following will explain the training process of the data prediction model in conjunction with specific embodiments.
[0078] For ease of understanding, the BP neural network is taken as an example below. The BP neural network may include: an input layer, a hidden layer and an output layer; wherein the hidden layer is used to learn the relationship between input data and output data; the hidden layer includes: a weight function, a threshold function, and an activation function.
[0079] Among them, the number of arrays P input in the input layer is the same as the number of influencing factors in the database, the number of hidden layers can be determined according to actual conditions, and the number of nodes in the output layer is 1.
[0080] Among them, the weight function, threshold function, and activation function can be set according to actual conditions.
[0081] Among them, the weight matrix is used to adjust the impact of input data on the hidden layer neuron nodes.
[0082] The maximum number of trainings is set to A (e.g., 1000 times), the learning rate is set to B (e.g., 0.1), the target minimum error of training is set to C (e.g., 0.001), the momentum factor is set to D (e.g., 0.01), the minimum performance gradient is set to E (1e-6), and the maximum number of consecutive failures is set to F (e.g., 7 times). A, B, C, D, E, and F can all be determined according to actual conditions, and the present disclosure does not make specific restrictions here.
[0083] For the input data column F(X i (t)), where X i (t)>0, i=1, 2, 3, ..., n; t=1, 2, ..., 15. In the neural network architecture, the input memory of each hidden layer neuron node is V i,j , i = 1, 2, ..., n; j = 1, 2, ..., d, d is the number of hidden layer nodes. Specifically, the weight of the jth hidden layer neuron node corresponding to the tth element of the i-th group of input data is ω t,j , when each input data set corresponds to the jth hidden layer neuron node, there is also a specific output threshold. The output threshold of the jth hidden layer neuron node corresponding to the i-th group of input data is θ i,j , the relationship between the input data of the input layer and the input of the hidden neuron node can be satisfied as follows:
[0084]
[0085] When the input of the jth hidden layer neuron node is greater than 0, it will continue to output U i,j , the relationship can be expressed as:
[0086]
[0087] in, is the activation function.
[0088] According to U i, And the activation function can get the error function.
[0089] The gradient of the weights of each layer can be obtained according to the error function, so as to adjust the weights (i.e., parameters). The weights obtained in the output layer can be optimized using the above-mentioned search agent, which will not be elaborated here.
[0090] To more intuitively demonstrate the optimization effect of this disclosure, please refer to Figure 2 and Figure 3 . Figure 2 is a test data graph of data prediction provided by an embodiment of the present disclosure, Figure 3 is another test data diagram for data prediction provided by the embodiment of the present disclosure. Figure 2 and Figure 3 As shown in the table above, after entering the database, the average error of the model in predicting the budget results is 5%, which is significantly better than the 80% of the traditional budget method. Compared with the traditional method, the model has significantly improved in both prediction accuracy and efficiency. Specifically, the prediction accuracy of the model has increased from 80% before the search agent optimization to 95% after optimization.
[0091] In summary, this paper verifies the efficiency and accuracy of the model through actual cases. Compared with traditional budgeting methods, it has achieved remarkable results in improving accuracy while avoiding the problem of high dependence on professionals in traditional methods.
[0092] The above is a data prediction method provided by the present disclosure. The above embodiments can be combined according to actual conditions and are not limited here.
[0093] In an illustrative embodiment, Figure 4 is a flow chart of another data prediction method provided by an embodiment of the present disclosure, such as Figure 4 As shown, the process of using the data prediction method can be: obtaining the historical budget (i.e., original data), preprocessing the historical budget to obtain the target project data, inputting the target project data into the input layer of the BP neural network, processed by the hidden layer, and the output layer outputs the static budget result. Next, the pigeon swarm algorithm can be used to optimize the parameters of the BP neural network to obtain GBestF (i.e., global optimal fitness) and GBestX (i.e., global optimal position), so as to optimize each parameter and finally obtain the trained data prediction model.
[0094] The present disclosure also provides a data prediction device. Figure 5 A structural block diagram of a data prediction device provided by an embodiment of the present disclosure, such as Figure 5 As shown, the data prediction device 500 includes: a data acquisition module 501, which is used to acquire target project data.
[0095] The data processing module 502 is used to process the target project data using the trained data prediction model to obtain the target decision data output by the data prediction model; wherein the data prediction model is trained based on the global optimization process of at least one search agent.
[0096] In an exemplary embodiment, the data processing module 502 is also configured to, during any round of model training, use the first model parameters of the first training model as the basis and use the pigeon swarm algorithm to perform global optimization processing to obtain second model parameters; the first training model is the initial data prediction model for training in the current round; the second model parameters are assigned to the first training model to obtain a second training model; when the second training model does not meet the first preset condition, the second training model is used as the initial data prediction model for the next round for model training.
[0097] In an exemplary embodiment, the data processing module 502 is also configured to iteratively determine the global optimal fitness and the global optimal position based on the first model parameters of the first training model and the fitness of each search agent; wherein the fitness is negatively correlated with the prediction error; any one position is used to represent a set of model parameters; using the guide search agent, iteratively update the global optimal fitness and the global optimal position; wherein the guide search agent is a search agent with a relatively high fitness; wherein the second model parameters are a set of model parameters corresponding to the global optimal position.
[0098] In an exemplary embodiment, the data processing module 502 is also configured to use the first training model to obtain a first prediction error of a training data set; use the first training model to obtain a second prediction error of a test data set; obtain the inverse of a positive correlation parameter between the first prediction error and the second prediction error to obtain the fitness.
[0099] In an exemplary embodiment, the data processing module 502 is also configured so that the data prediction model is a neural network model; the neural network model includes: an input layer, a hidden layer and an output layer; wherein the hidden layer is used to learn the relationship between input data and output data; the hidden layer includes: a weight function, a threshold function, and an activation function.
[0100] In an exemplary embodiment, the data processing module 502 is also configured to use the target decision data to perform decision evaluation on the target project data; wherein the dimensions of the decision evaluation include at least one of the following: cost, feasibility, ease of operation, and ease of maintenance.
[0101] In an exemplary embodiment, the data processing module 502 is also configured so that the input data of the data prediction model includes at least one of the following: material data and engineering data; wherein the material data includes at least one of the following: bandwidth capacity, number of central computer rooms, scale of central computer rooms, splitter, high-speed fiber optic switch, single-mode fiber optic material, fiber length, cable length, number of signal amplifiers, number of uninterruptible power supplies, number of temperature and humidity monitoring; wherein the engineering data includes at least one of the following: length of the excavation site, number of access points, maintenance time, and construction period.
[0102] In an exemplary embodiment, the data processing module 502 is further configured to pre-process the target project data; wherein the pre-processing includes at least one of the following: data cleaning and normalization.
[0103] Figure 6 The electronic device 600 according to the embodiment of the present disclosure includes at least a memory, a processor and a computer program stored in the memory, and the processor executes the computer program to implement the data prediction method described in any of the above embodiments.
[0104] Figure 6 The electronic device 600 shown specifically includes: a central processing unit (CPU) 601, a graphics processing unit (GPU) 602 and a memory 603. These units are interconnected via a bus 604. The central processing unit (CPU) 601 and / or the graphics processing unit (GPU) 602 can be used as the above-mentioned processor, and the memory 603 can be used as the above-mentioned memory for storing computer-readable instructions. In addition, the electronic device 600 may also include a communication unit 605, a storage unit 606, an output unit 607, an input unit 608 and an external device 609, which are also connected to the bus 604.
[0105] Figure 7 A schematic diagram of a computer-readable storage medium provided in an embodiment of the present disclosure. Figure 7 As shown, a computer-readable storage medium 700 according to an embodiment of the present disclosure has a computer program / instruction 701 stored thereon. When the computer program / instruction 701 is executed by a processor, the signal transmission method described in any of the foregoing embodiments of the present disclosure is implemented. The computer-readable storage medium includes, but is not limited to, for example, volatile memory and / or non-volatile memory. Volatile memory may include, for example, random access memory (RAM) and / or cache memory (cache), etc. Non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, optical disk, magnetic disk, etc.
[0106] The present disclosure further provides a computer program product, including a computer program / instruction, which, when executed by a processor, implements the signal transmission method described in any of the foregoing embodiments of the present disclosure.
[0107] The basic principles of the present disclosure are described above in conjunction with specific embodiments. However, it should be noted that the advantages, strengths, effects, etc. mentioned in the present disclosure are only examples and not limitations, and it cannot be considered that these advantages, strengths, effects, etc. are required by each embodiment of the present disclosure. In addition, the specific details disclosed above are only for the purpose of illustration and ease of understanding, and are not limitations. The above details do not limit the present disclosure to the necessity of adopting the above specific details to be implemented.
[0108] The block diagrams of the devices, apparatuses, equipment, and systems involved in this disclosure are only illustrative examples and are not intended to require or imply that they must be connected, arranged, and configured in the manner shown in the block diagrams. As will be appreciated by those skilled in the art, these devices, apparatuses, equipment, and systems can be connected, arranged, and configured in any manner. Words such as "including," "comprising," "having," and the like are open words, referring to "including but not limited to," and can be used interchangeably therewith. The words "or" and "and" used herein refer to the words "and / or," and can be used interchangeably therewith, unless the context clearly indicates otherwise. The word "such as" used herein refers to the phrase "such as but not limited to," and can be used interchangeably therewith.
[0109] Additionally, as used herein, "or" used in a list of items beginning with "at least one" indicates a separate list, so that, for example, a list of "at least one of A, B, or C" means A or B or C, or AB or AC or BC, or ABC (i.e., A and B and C). Furthermore, the word "exemplary" does not mean that the example described is preferred or better than other examples.
[0110] It should also be noted that in the system and method of the present disclosure, each component or each step can be decomposed and / or recombined. Such decomposition and / or recombination should be regarded as equivalent solutions of the present disclosure.
[0111] Various changes, substitutions, and modifications of the techniques described herein may be made without departing from the teachings defined by the appended claims. Furthermore, the scope of the claims of the present disclosure is not limited to the specific aspects of the processes, machines, manufactures, compositions of events, means, methods, and actions described above. Currently existing or later to be developed processes, machines, manufactures, compositions of events, means, methods, or actions that perform substantially the same functions or achieve substantially the same results as the corresponding aspects described herein may be utilized. Thus, the appended claims include such processes, machines, manufactures, compositions of events, means, methods, or actions within their scope.
[0112] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use the present disclosure. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein may be applied to other aspects without departing from the scope of the present disclosure. Therefore, the present disclosure is not intended to be limited to the aspects shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.
[0113] The above description has been given for the purpose of illustration and description. In addition, this description is not intended to limit the embodiments of the present disclosure to the forms disclosed herein. Although multiple example aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, changes, additions and sub-combinations thereof.
Claims
1. A data prediction method, characterized in that: The method comprises: Get target project data; The target project data is processed using the trained data prediction model to obtain target decision data output by the data prediction model; wherein the data prediction model is trained based on a global optimization process of at least one search agent.
2. The method according to claim 1, characterized in that The method further comprises: In any round of model training, based on the first model parameters of the first training model, the pigeon swarm algorithm is used to perform global optimization processing to obtain the second model parameters; the first training model is the initial data prediction model trained in the current round; Assigning the second model parameters to the first training model to obtain a second training model; When the second training model does not meet the first preset condition, the second training model is used as the initial data prediction model for the next round of model training.
3. The method according to claim 2, characterized in that The method of performing global optimization processing based on the first model parameters of the first training model and using the pigeon colony algorithm to obtain the second model parameters includes: Iteratively determine the global optimal fitness and the global optimal position based on the first model parameter of the first training model and the fitness of each search agent; wherein the fitness is negatively correlated with the prediction error; and any one position is used to represent a set of model parameters; The global optimal fitness and the global optimal position are iteratively updated by using a guide search agent; wherein the guide search agent is a search agent with relatively high fitness; wherein the second model parameters are a set of model parameters corresponding to the global optimal position.
4. The method according to claim 3, characterized in that The method further comprises: Using the first training model, obtaining a first prediction error of a training data set; Using the first training model, obtaining a second prediction error of the test data set; The inverse of a positive correlation parameter between the first prediction error and the second prediction error is obtained to obtain the fitness.
5. The method according to any one of claims 1 to 4, characterized in that: The data prediction model is a neural network model; the neural network model includes: an input layer, a hidden layer and an output layer; Wherein, the hidden layer is used to learn the relationship between input data and output data; The hidden layer includes: a weight function, a threshold function, and an activation function.
6. The method according to any one of claims 1 to 5, characterized in that: The target decision data is used to make a decision evaluation on the target project data; The decision evaluation dimensions include at least one of the following: cost, feasibility, operational difficulty, and maintenance difficulty.
7. The method according to any one of claims 1 to 4, characterized in that: The input data of the data prediction model includes at least one of the following: material data and engineering data; The material data includes at least one of the following: bandwidth capacity, number of central computer rooms, size of central computer rooms, optical splitters, high-speed fiber switches, single-mode fiber materials, fiber length, cable length, number of signal amplifiers, number of uninterruptible power supplies, and number of temperature and humidity monitoring; The engineering data includes at least one of the following: length of the excavation site, number of access points, maintenance time, and construction period.
8. The method according to claim 7, characterized in that Before processing the target project data using the trained data prediction model, the method further includes: Preprocessing the target project data; Wherein, the preprocessing includes at least one of the following: data cleaning and normalization.
9. A data prediction device, characterized in that: The device comprises: Data acquisition module, used to acquire target project data; A data processing module is used to process the target project data using a trained data prediction model to obtain target decision data output by the data prediction model; wherein the data prediction model is trained based on a global optimization process of at least one search agent.
10. A computer device / equipment / system comprising a memory, a processor and a computer program stored in the memory, characterized in that: The processor executes the computer program to implement the method according to any one of claims 1 to 8.
11. A computer-readable storage medium having a computer program / instruction stored thereon, characterized in that: When the computer program / instructions are executed by a processor, the method according to any one of claims 1 to 8 is implemented.
12. A computer program product comprising a computer program / instructions, characterized in that When the computer program / instructions are executed by a processor, the method according to any one of claims 1 to 8 is implemented.
