Intelligent prediction method and system for operation platform operation and maintenance data

Abstract

The invention discloses an intelligent prediction method and system for operation platform operation and maintenance data, and the method comprises the steps: firstly, widely collecting all kinds of operation platform operation and maintenance data closely related to a prediction object, and carrying out the cleaning and preprocessing of the collected data; secondly, for the processed data, automatically extracting features with key value for a prediction task from the original data through a machine learning algorithm; then, on the basis of the extracted data features and a specific prediction target, a prediction model is constructed on the basis of a machine learning algorithm; training the prediction model by using historical data to obtain a trained prediction model; inputting new operation and maintenance data into the trained prediction model for prediction and evaluation; and finally, performing real-time response processing on the generated prediction result. According to the scheme, the intelligent prediction model for the operation and maintenance data of the operation platform is formed based on fusion of automatic data processing and a machine learning model, and efficient and accurate prediction for the operation and maintenance data generated by the operation platform can be realized.

Description

Technical Field

[0001] This invention relates to the field of data operation technology, and more specifically to a smart prediction solution for operation and maintenance data generated by an operation platform. Background Technology

[0002] With the rapid development of information technology, the scale of data faced by the operation and maintenance field is expanding at an unprecedented rate. This has brought abundant information resources to operation and maintenance management, but it has also posed a severe challenge to its processing and analysis capabilities. Against this background, how to efficiently mine and process this massive amount of data, and accurately predict the operation and maintenance status from it, so as to ensure the stable operation of the system and optimize resource allocation, has become a key issue that urgently needs to be addressed in current operation and maintenance management.

[0003] Traditional operation and maintenance data processing methods mostly rely on rules and thresholds preset by the personal experience of operation and maintenance personnel. This method may be effective when dealing with small-scale, simple-structured datasets, but its limitations become apparent when facing today's large-scale, high-dimensional, complex and ever-changing operation and maintenance data. Manual rules are difficult to fully cover all possible abnormal scenarios, and the reaction speed is slow, often failing to respond to rapidly changing operation and maintenance needs in a timely manner, resulting in decreased prediction accuracy and low operation and maintenance efficiency.

[0004] Therefore, intelligent prediction of the operation and maintenance data generated by the operation platform to ensure the stability and reliability of the operation platform is an urgent problem to be solved in this field. Summary of the Invention

[0005] To address the problems existing in current operational platform data processing solutions, the present invention aims to provide an intelligent prediction solution for operational platform data. This solution is based on the integration of automated data processing and machine learning models to form an intelligent prediction model for operational platform data, which can achieve efficient and accurate prediction of operational platform data.

[0006] To achieve the above objectives, the present invention provides a smart prediction method for operation and maintenance data of an operation platform, the smart prediction method comprising: Step 1: Collect a wide range of operational platform maintenance data closely related to the target of prediction, and clean and preprocess the collected data; Step 2: For the data processed in Step 1, machine learning algorithms are used to automatically extract features that are of key value to the prediction task from the original data. The feature extraction process is as follows: F′=FeatureExtract(F); Wherein, FeatureExtract represents the feature extraction function, including feature selection, feature scaling, and feature transformation operations; the original data feature set is F={f1, f2, ..., fk}; and the extracted new feature set is F′={f1′, f2′, ..., fl′} (l≥k); Step 3: Based on the data features extracted in Step 2 and the specific prediction target, construct a prediction model using machine learning algorithms; Step 4: Train the prediction model using historical data, and obtain the completed prediction model by continuously adjusting the model parameters to minimize the prediction error; Step 5: Input the new operation and maintenance data into the trained prediction model, obtain the prediction results, and use evaluation metrics to evaluate the prediction results; Step 6: Based on the preset early warning and processing mechanism, perform real-time response processing on the prediction results generated in Step 5.

[0007] Furthermore, in step (4), an adaptive update mechanism is introduced for the completed training prediction model. Based on the prediction model's real-time prediction results based on the received real-time data stream, a new dataset is formed, and the prediction model is automatically updated and optimized through an online learning algorithm.

[0008] Furthermore, the method also performs multimodal data fusion processing on the received data, fusing heterogeneous data from different sources and of different types.

[0009] Furthermore, the method also includes a prediction result interpretation step, in which the prediction process and results of the prediction model are analyzed using visualization methods and interpretation algorithms for the prediction results obtained in step (5).

[0010] To achieve the above objectives, the present invention also provides an intelligent prediction system for operation and maintenance data of an operation platform, the intelligent prediction system comprising: The data collection and processing module is used to collect various operational platform maintenance data closely related to the prediction object, and to clean and preprocess the collected data. An automated feature engineering module is configured to automatically extract features that are of key value to the prediction task from the data collected and processed by the data collection and processing module using machine learning algorithms. The feature extraction process is as follows: F′=FeatureExtract(F); Wherein, FeatureExtract represents the feature extraction function, including feature selection, feature scaling, and feature transformation operations; the original data feature set is F={f1, f2, ..., fk}; and the extracted new feature set is F′={f1′, f2′, ..., fl′} (l≥k); A prediction model building module is configured to build a prediction model based on the data features extracted by the automated feature engineering module and the specific prediction target, and then based on machine learning algorithms. The model training and optimization module uses historical data to train the prediction model constructed by the prediction model construction module, and continuously adjusts the model parameters to minimize the prediction error to obtain a completed prediction model. The prediction and evaluation module is configured to input new operation and maintenance data into the trained prediction model, obtain prediction results, and evaluate the prediction results using evaluation metrics. The real-time response and feedback module is configured to respond to the prediction results generated by the prediction and evaluation module in real time based on a preset early warning and processing mechanism.

[0011] Furthermore, the model training and optimization module also introduces a loss function to optimize the prediction model: ; Among them, Model opt For the optimized prediction model, the loss function is L(Model, H), and R is the regularization term.

[0012] Furthermore, the intelligent prediction system also includes a real-time data stream processing module, which is configured to receive and process data streams in real time and feed the processed real-time data streams back to the prediction and evaluation module for real-time prediction and response.

[0013] Furthermore, the intelligent prediction system also includes an adaptive model update module, which is configured to generate a new dataset based on the real-time prediction results of the prediction model based on the received real-time data stream, and automatically update and optimize the prediction model accordingly through an online learning algorithm.

[0014] Furthermore, the intelligent prediction system also includes a multimodal data fusion module, which can fuse heterogeneous data from different sources and of different types.

[0015] Furthermore, the intelligent prediction system also includes a prediction result interpretation module, which is configured to analyze the prediction process and results of the prediction model through visualization means and interpretation algorithms for the prediction results obtained by the prediction and evaluation module.

[0016] The intelligent prediction solution for operation and maintenance data of the operation platform provided by this invention forms an intelligent prediction model for operation and maintenance data of the operation platform by integrating automated data processing and machine learning models, which can achieve efficient and accurate prediction of the operation and maintenance data generated by the operation platform.

[0017] Furthermore, the present invention achieves automated processing of operation and maintenance data and construction of prediction models through steps such as data collection and processing module, automated feature engineering module and prediction model construction module, thereby improving the accuracy and efficiency of operation and maintenance prediction.

[0018] Furthermore, the present invention achieves automatic optimization and updating of the prediction model through a model training and optimization module and an adaptive model update module, enabling it to adapt to the ever-changing data environment.

[0019] Furthermore, the present invention achieves real-time prediction and response through a real-time data stream processing module and a real-time response and feedback module, which can promptly detect and handle potential operation and maintenance problems, thereby improving the stability and reliability of the system.

[0020] Furthermore, the present invention utilizes a multimodal data fusion module and a prediction result interpretation module to fully leverage information from different types of data, thereby improving the accuracy and robustness of predictions. It also helps users understand the prediction process and results through visualization and interpretation algorithms. Attached Figure Description

[0021] The present invention will be further described below with reference to the accompanying drawings and specific embodiments.

[0022] Figure 1 This is a schematic diagram of the intelligent prediction system for operation and maintenance data of the operation platform in this invention. Figure 2 This is a flowchart illustrating the intelligent prediction process for operational platform maintenance data in this invention. Detailed Implementation

[0023] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below with reference to specific illustrations.

[0024] This invention provides a smart prediction solution for the large-scale, high-dimensional, complex and ever-changing operation and maintenance management data of operation platforms, thereby achieving efficient and highly accurate smart prediction and management of operation and maintenance management data.

[0025] Specifically, the intelligent prediction scheme for operation and maintenance data of the operation platform provided by this invention mainly includes three parts: prediction model construction and training, prediction and evaluation of operation and maintenance data, and adaptive updating of the prediction model.

[0026] (i) In the prediction model building and training phase, complete the construction and training of a large model for the prediction management needs of operation and maintenance data of the operation platform, and form a prediction model that can make efficient and highly accurate predictions of operation and maintenance data of the operation platform.

[0027] Specifically, the first step is to collect operational and maintenance data from various operational platforms related to the target of prediction, including historical data and real-time data extracted from historical databases; and after data collection, data cleaning and preprocessing are performed to ensure the accuracy and completeness of the data.

[0028] Next, machine learning algorithms are used to automatically extract features that are of key value to the prediction task from the preprocessed data, reducing human intervention and improving the efficiency and accuracy of feature selection.

[0029] Next, based on the extracted data features and specific prediction targets, a prediction model is constructed using machine learning algorithms; Next, the prediction model is trained using historical data, and the prediction error is minimized by continuously adjusting the model parameters to obtain a fully trained prediction model.

[0030] As further explanation, this solution includes the following processing steps when cleaning and preprocessing the received data: Deduplication: Use the formula n_new=n_old-|D_repeated|, where n_new is the number of duplicate samples after deduplication, n_old is the original number of samples, and |D_repeated| is the size of the duplicate sample set; Missing value handling: Imputation using the mean x_filled=\overline{x}, where x_filled is the imputed value and \overline{x} is the mean of feature x; Outlier detection and handling: The Z-score method is used, Z=(x-\overline{x}) / \sigma, where \sigma is the standard deviation of feature x. Data with a Z-score between plus and minus 3 are considered normal.

[0031] This process ensures that all input data is highly accurate and complete, laying a solid foundation for subsequent analysis.

[0032] As further explanation, this scheme extracts features that are of key value to the prediction task in the following way: F′=FeatureExtract(F); Wherein, FeatureExtract represents the feature extraction function, including feature selection, feature scaling, and feature transformation operations; the original data feature set is F={f1, f2, ..., fk}; and the extracted new feature set is F′={f1′, f2′, ..., fl′} (l≥k).

[0033] Furthermore, the FeatureExtract function here can employ various machine learning algorithms for feature selection and transformation, such as Principal Component Analysis (PCA) and Linear Discriminant Analysis (LDA), thereby improving the efficiency and accuracy of feature selection.

[0034] As further explanation, this solution can use machine learning algorithms (Algorithms) to construct the prediction model.

[0035] The machine learning algorithm here can be selected from algorithms such as Support Vector Machine (SVM), Random Forest (RF), and Neural Network (NN), depending on the characteristics of the data and the prediction target.

[0036] As further explanation, when using historical data to fully train the prediction model, the specific model training process is as follows: Model = Algorithm(H); Where Model represents the trained prediction model; Algorithm represents the selected machine learning algorithm; and H is a historical dataset containing multiple training samples, H={(x1, y1), (x2, y2), ..., (xp, yp)}.

[0037] Furthermore, this scheme introduces a prediction model optimization mechanism during the model training process to further optimize the prediction model and improve the accuracy of the prediction results.

[0038] The prediction model optimization mechanism here is based on a loss function, as detailed below: ; Among them, Model opt The optimized prediction model is defined as Model, which represents the trained prediction model. The loss function is L(Model, H), and R is the regularization term.

[0039] (ii) Based on the above scheme, after the prediction model is built and trained, this scheme will carry out the prediction and evaluation stage of operation and maintenance data.

[0040] In the prediction and evaluation phase of operation and maintenance data, the operation and maintenance data of the operation platform are predicted, analyzed and processed based on the completed prediction model.

[0041] Specifically, in this stage, the new operation and maintenance data is first input into the trained prediction model to obtain the prediction results, and the prediction results are evaluated using evaluation metrics. Next, the generated prediction results are processed in real time based on the preset early warning and processing mechanism.

[0042] As a further explanation, when inputting new operation and maintenance data into the trained prediction model, this solution can achieve real-time prediction and response by introducing a real-time data stream processing mechanism to receive and process the real-time operation and maintenance data stream generated by the operation platform and input it into the trained prediction model.

[0043] Specifically, the real-time prediction process here can be represented as: yt=Model opt (st); Where yt represents the prediction result for the real-time data stream st; the real-time data stream is S={s1, s2, ..., st}.

[0044] Based on this, and considering that the real-time operation and maintenance data stream generated by the operating platform includes heterogeneous data from different sources and of different types, this solution further introduces a multimodal data fusion mechanism to fuse the acquired heterogeneous data from different sources and of different types, making full use of the information from different types of data to improve the accuracy and robustness of prediction.

[0045] Specifically, for the two heterogeneous datasets D1 and D2, the fusion process in this scheme can be represented as follows: D fused =Fuse(D1, D2); Among them, D fused This represents the merged dataset, and Fuse represents the data fusion function.

[0046] As a further explanation, when evaluating the prediction results of the prediction model using evaluation metrics, if the evaluation results do not meet the requirements, the model is readjusted or features are reselected, and the model is retrained, thereby ensuring the accuracy of the final prediction results.

[0047] Specifically, based on the evaluation index M, the evaluation process in this scheme can be expressed as follows: Score = M (Model) new (NewData); Here, "Score" represents the evaluation score, and "NewData" represents the new operation and maintenance dataset.

[0048] As a further explanation, this solution introduces a prediction result interpretation mechanism in addition to the prediction and evaluation results of the prediction model. Through visualization methods and interpretation algorithms, it helps users understand the prediction process and results of the model, increasing the transparency and credibility of the prediction. Specifically, the interpretation process of the prediction results in this scheme can be expressed as follows: Explanation = Explain(Model) New ,y); Here, Explanation represents the interpretation of the prediction result, Explain represents the explanation function, and y represents the prediction result.

[0049] As a further explanation, in the specific implementation of the interpretation of prediction results, this scheme selects a target variable, makes predictions on nearby samples, and uses a linear model to fit these prediction results to interpret individual prediction decisions.

[0050] As a further explanation, when implementing real-time response processing of the generated prediction results, this solution constructs an early warning and processing mechanism by setting a pre-set early warning threshold θ. By judging the relative relationship between the prediction result and the set early warning threshold θ, it determines whether a fault situation is predicted and automatically triggers the corresponding early warning and processing mechanism to take timely measures to prevent the fault from occurring.

[0051] Specifically, the real-time response process here is based on the following computational model: ; Where Response represents the response action, TriggerWarning represents the triggering of the warning action, NoAction represents no action taken, and θ represents the warning threshold.

[0052] (iii) The adaptive update phase of the prediction model in this scheme is synchronized with the prediction model construction and training phase and the prediction and evaluation phase of operation and maintenance data. It can automatically update and optimize the prediction model based on new data during the prediction model training or operation process.

[0053] Specifically, the present invention introduces an adaptive model update mechanism for the prediction model, which can form a new dataset based on the prediction results and / or evaluation results of the prediction model based on the received real-time data stream, and automatically update and optimize the prediction model accordingly through an online learning algorithm.

[0054] The adaptive model update mechanism here can work in conjunction with the optimization mechanism during the training phase of the prediction model to further improve the accuracy of the prediction results.

[0055] The adaptive update of the prediction model in this scheme can be achieved through the following computational model: Model new =Update(Model opt (N); Among them, Model new This represents the updated prediction model, where Update represents the model update function, and the new dataset is N = {(x...} p+1 y p+1 ), ..., (x q y q )}.

[0056] Furthermore, the model update function Update here can use online learning algorithms to automatically update and optimize the model, such as stochastic gradient descent (SGD) and the Adam algorithm.

[0057] The intelligent prediction scheme for operation and maintenance data of the operation platform provided in this invention can be configured into a corresponding software program to form an intelligent prediction system for operation and maintenance data of the operation platform. When running, this software program will execute the aforementioned intelligent prediction method for operation and maintenance data of the operation platform and store it in a corresponding storage medium for the processor to retrieve and execute.

[0058] See Figure 1 The diagram shown is a system block diagram of the intelligent prediction system for the operation and maintenance data of the formed operation platform.

[0059] Based on the diagram, the intelligent prediction system 100 for operation and maintenance data of this operation platform mainly includes the following functional modules: data collection and organization module 101, automated feature engineering module 102, prediction model construction module 103, model training and optimization module 104, prediction and evaluation module 105, prediction result interpretation module 106, real-time response and feedback module 107, and adaptive model update module 108.

[0060] The data collection and organization module 101 is used to collect various operational platform maintenance data closely related to the prediction object, and to clean and preprocess the collected data.

[0061] The automated feature engineering module 102 is configured to automatically extract features that are of key value to the prediction task from the data collected and processed by the data collection and processing module 101 using machine learning algorithms. The feature extraction process is as follows: F′=FeatureExtract(F); Wherein, FeatureExtract represents the feature extraction function, including feature selection, feature scaling, and feature transformation operations; the original data feature set is F={f1, f2, ..., fk}; and the extracted new feature set is F′={f1′, f2′, ..., fl′} (l≥k).

[0062] The prediction model building module 103 is configured to build a prediction model based on the data features extracted by the automated feature engineering module 102 and the specific prediction target, and then based on machine learning algorithms.

[0063] The model training and optimization module 104 is configured to train the prediction model constructed by the prediction model construction module 103 based on historical data, and to obtain a completed prediction model by continuously adjusting the model parameters to minimize the prediction error.

[0064] The prediction and evaluation module 105 is configured to input new operation and maintenance data into the trained prediction model, obtain prediction results, and evaluate the prediction results using evaluation metrics.

[0065] The prediction result interpretation module 106 is configured to analyze the prediction process and results of the prediction model using visualization methods and interpretation algorithms based on the prediction results obtained by the prediction and evaluation module 105.

[0066] The real-time response and feedback module 107 is configured to respond to the prediction results generated by the prediction and evaluation module 105 in real time based on a preset early warning and processing mechanism.

[0067] The adaptive model update module 108 is configured to form a new dataset based on the real-time prediction results and / or evaluation results of the prediction and evaluation module 105 based on the received real-time data stream, and thereby automatically update and optimize the prediction model trained and optimized by the model training and optimization module 104 through an online learning algorithm.

[0068] Regarding this system solution, the following details the specific composition of each functional module in the system and the technical means that may be involved.

[0069] The data collection and processing module 101 in this system collects various data related to the prediction object, including historical data and real-time data, during the data collection and processing process. After data collection, it performs data cleaning and preprocessing to ensure the accuracy and integrity of the data.

[0070] The data cleaning and preprocessing here specifically includes: Deduplication: Use the formula n_new=n_old-|D_repeated|, where n_new is the number of duplicate samples after deduplication, n_old is the original number of samples, and |D_repeated| is the size of the duplicate sample set; Missing value handling: Imputation using the mean x_filled=\overline{x}, where x_filled is the imputed value and \overline{x} is the mean of feature x; Outlier detection and handling: The Z-score method is used, Z=(x-\overline{x}) / \sigma, where \sigma is the standard deviation of feature x. Data with a Z-score between plus and minus 3 are considered normal.

[0071] The automated feature engineering module 102 in this system automatically extracts useful features from the raw data through machine learning algorithms, reducing manual intervention and improving the efficiency and accuracy of feature selection.

[0072] The FeatureExtract function used in this module can employ various machine learning algorithms for feature selection and transformation, such as Principal Component Analysis (PCA) and Linear Discriminant Analysis (LDA), thereby improving the efficiency and accuracy of feature selection.

[0073] The prediction model building module 103 in this system specifically uses a machine learning algorithm to build the prediction model. This machine learning algorithm can be selected from algorithms such as Support Vector Machine (SVM), Random Forest (RF), and Neural Network (NN), depending on the characteristics of the data and the prediction objective.

[0074] When the model training and optimization module 104 in this system performs comprehensive training on the prediction model using historical data, the specific model training process is as follows: Model = Algorithm(H); Where Model represents the trained prediction model; Algorithm represents the selected machine learning algorithm; and H is a historical dataset containing multiple training samples, H={(x1, y1), (x2, y2), ..., (xp, yp)}.

[0075] Furthermore, the model training and optimization module 104 introduces a prediction model optimization mechanism to further optimize the prediction model and improve the accuracy of the prediction results.

[0076] The prediction model optimization mechanism here is based on a loss function, as detailed below: ; Among them, Model opt The optimized prediction model is defined as Model, which represents the trained prediction model. The loss function is L(Model, H), and R is the regularization term.

[0077] The prediction and evaluation module 105 in this system introduces and runs the prediction model trained by the model training and optimization module 104, and can interact with the operating platform to obtain the operation and maintenance data generated by the operating platform, and then inputs it into the prediction model to obtain the prediction results.

[0078] Based on this, the prediction and evaluation module 105 introduces a real-time data stream processing module 109. This data stream processing module 109 can receive and process real-time operation and maintenance data streams generated by the operating platform in real time and input them into the trained prediction model. This allows it to work with the real-time response and feedback module 107 to achieve real-time prediction and response.

[0079] Specifically, the real-time data stream processing module 109 here can be represented as: yt=Model opt (st); Where yt represents the prediction result for the real-time data stream st; the real-time data stream is S={s1, s2, ..., st}.

[0080] Based on this, considering that the real-time operation and maintenance data stream generated by the operating platform includes heterogeneous data from different sources and of different types, this system further introduces a multimodal data fusion module 110 in the prediction and evaluation module 105. This multimodal data fusion module 110 is configured to interact with the real-time data stream processing module 109, the data collection and processing module 101, etc., and can fuse the heterogeneous data from different sources and of different types, making full use of the information of different types of data to improve the accuracy and robustness of prediction.

[0081] Specifically, for the two heterogeneous datasets D1 and D2, the fusion process in this scheme can be represented as follows: D fused =Fuse(D1, D2); Among them, D fused This represents the merged dataset, and Fuse represents the data fusion function.

[0082] The prediction and evaluation module 105 in this system evaluates the prediction results by constructing corresponding evaluation indicators M.

[0083] Specifically, when using evaluation metrics for evaluation, if the evaluation results do not meet the requirements, the model is readjusted or features are reselected, and the model is retrained to ensure the accuracy of the final prediction results.

[0084] Furthermore, this prediction and evaluation module 105 can evaluate the prediction results through the evaluation module: Score = M (Model) new (NewData); Where Score represents the evaluation score, NewData represents the new operation and maintenance dataset, and M represents the evaluation metric.

[0085] The prediction result interpretation module 106 in this system uses visualization methods and interpretation algorithms to help users understand the prediction process and results of the model, thereby increasing the transparency and credibility of the prediction.

[0086] Specifically, the prediction result interpretation module 106 interprets the prediction results using the following calculation model: Explanation = Explain(Model) New ,y); Here, Explanation represents the interpretation of the prediction result, Explain represents the explanation function, and y represents the prediction result.

[0087] As a further explanation, this prediction result interpretation module 106 specifically interprets individual prediction decisions by selecting a target variable, making predictions for nearby samples, and fitting these prediction results with a linear model.

[0088] In this system, the real-time response and feedback module 107, when performing real-time response processing on the generated prediction results, specifically constructs an early warning and processing mechanism by setting a pre-set early warning threshold θ. By judging the relative relationship between the prediction result and the pre-set early warning threshold θ, it determines whether a fault situation is predicted and automatically triggers the corresponding early warning and processing mechanism to take timely measures to prevent the fault from occurring.

[0089] Specifically, this real-time response and feedback module 107 can realize the real-time response processing through the following calculation model: ; Where Response represents the response action, TriggerWarning represents the triggering of the warning action, NoAction represents no action taken, and θ represents the warning threshold.

[0090] The adaptive model update module 108 in this system introduces an adaptive model update mechanism for the prediction model. It can form a new dataset based on the prediction results and / or evaluation results of the prediction model based on the received real-time data stream, and automatically update and optimize the prediction model through an online learning algorithm. This enables the prediction model to be automatically updated and optimized when new data arrives.

[0091] The adaptive model update mechanism here can work in conjunction with the optimization mechanism during the training phase of the prediction model to further improve the accuracy of the prediction results.

[0092] Specifically, the adaptive model update module 108 can achieve adaptive updates to the prediction model through the following model update: Model new =Update(Model opt (N); Among them, Model new This represents the updated prediction model, where Update represents the model update function, and the new dataset is N = {(x...} p+1 y p+1 ), ..., (x q y q )}.

[0093] Furthermore, the model update function Update here can use online learning algorithms to automatically update and optimize the model, such as stochastic gradient descent (SGD) and the Adam algorithm.

[0094] The following example illustrates the implementation process of the intelligent prediction system for operation and maintenance data of the operation and maintenance platform based on the present invention.

[0095] See Figure 2 The process of using the intelligent prediction system based on the operation and maintenance data of this operation platform to make intelligent predictions on the operation and maintenance data includes the following steps: Step 1: Utilize the data collection and organization module to extensively collect various types of data closely related to the prediction object. This data includes historical data extracted from historical databases as well as the latest data generated in real time. After the data collection is completed, the data cleaning and preprocessing stage begins. This stage ensures that all input data has a high degree of accuracy and completeness through a series of meticulous operations, such as deduplication, missing value imputation, and outlier detection and processing, laying a solid foundation for subsequent analysis.

[0096] Step 2: The automated feature engineering module uses advanced machine learning algorithms to automatically mine and extract features that are of key value to the prediction task from the raw data, reducing the intervention of manual operations, thereby improving the efficiency of feature selection and ensuring the accuracy of feature selection.

[0097] For example, given the original data feature set F, after processing by the automated feature engineering module, a new feature set F′ will be generated, where the number of features in F′ may be greater than or equal to F. The specific feature extraction process can be represented as F′=FeatureExtract(F). FeatureExtract, as a feature extraction function, covers key operations such as feature selection, feature scaling, and feature transformation.

[0098] Step 3: The prediction model building module carefully selects suitable machine learning algorithms to build a prediction model based on the characteristics of the collected data and the specific prediction objectives. After the model is built, it will be comprehensively trained using rich historical data to improve the model's prediction performance.

[0099] The modeling process can be represented as Model = Algorithm(H), where Algorithm represents the selected machine learning algorithm and H is a historical dataset containing multiple training samples.

[0100] Step 4: Model Training and Optimization Module. In the model training phase, historical data will be used to train the prediction model, and the prediction error will be minimized by continuously adjusting the model parameters; at the same time, a loss function will be introduced to optimize the prediction model.

[0101] Step 5: The real-time data stream processing module receives and processes the data stream in real time, inputs the data into the prediction model to achieve real-time prediction and response, thereby ensuring that the model has excellent predictive capabilities.

[0102] In this real-time prediction and response process, the adaptive model update module can automatically update and optimize the prediction model when new data arrives; the model training and optimization module introduces the adaptive model update function, and uses online learning algorithms (such as SGD, Adam, etc.) to realize the automatic updating and optimization of the model, ensuring that the model can continuously adapt to the ever-changing data environment.

[0103] In this real-time prediction and response process, a multimodal data fusion module can also be used to fuse heterogeneous data from different sources and of different types, thereby improving the accuracy and robustness of the prediction.

[0104] In this real-time prediction and response process, the prediction and evaluation module is responsible for inputting new operation and maintenance data into the trained model, obtaining prediction results, and evaluating the prediction results using evaluation metrics.

[0105] Step Six: As needed, the prediction result interpretation module can use visualization methods and interpretation algorithms to analyze the prediction process and results of the prediction model based on the prediction results obtained by the prediction and evaluation module.

[0106] Step 7: When the real-time response and feedback module performs real-time response processing on the generated prediction results, it specifically constructs an early warning and processing mechanism by setting a pre-set early warning threshold θ. By judging the relative relationship between the prediction result and the set early warning threshold θ, it determines whether a fault situation is predicted and automatically triggers the corresponding early warning and processing mechanism to take timely measures to prevent the fault from occurring.

[0107] This invention also provides a processor for running a program, wherein the program executes the steps of the above-described intelligent prediction method for operation and maintenance data of the operation platform.

[0108] This invention also provides a terminal device, which includes a processor, a memory, and a program stored in the memory and executable on the processor. The program code is loaded and executed by the processor to implement the steps of the above-described intelligent prediction method for operation and maintenance data of the operation platform.

[0109] The present invention also provides a computer program product, which, when executed on a data processing device, is suitable for performing the steps of the above-described intelligent prediction method for operation and maintenance data of the operation platform.

[0110] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0111] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and modules described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0112] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0113] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0114] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0115] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0116] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0117] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, like read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0118] Computer-readable media, including both permanent and non-permanent, removable and non-removable media, can store information using any method or technology. Information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transfer medium that can be used to store information accessible by a computing device.

[0119] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0120] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0121] The method, specific system unit, or part thereof of the present invention described above is a pure software architecture. It can be deployed via program code on physical media, such as hard disks, optical discs, or any electronic device (such as smartphones or computer-readable storage media). When a machine loads and executes the program code (e.g., a smartphone loads and executes it), the machine becomes a device for implementing the present invention. The method and device of the present invention can also be transmitted in program code form via transmission media, such as cables, optical fibers, or any other transmission method. When the program code is received, loaded, and executed by a machine (e.g., a smartphone), the machine becomes a device for implementing the present invention.

[0122] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of this invention is defined by the appended claims and their equivalents.

Claims

1. A method for intelligent prediction of operation and maintenance data of an operation platform, characterized in that, The intelligent prediction method includes: Step 1: Collect a wide range of operational platform maintenance data closely related to the target of prediction, and clean and preprocess the collected data; Step 2: For the data processed in Step 1, machine learning algorithms are used to automatically extract features that are of key value to the prediction task from the original data. The feature extraction process is as follows: F′=FeatureExtract(F); Wherein, FeatureExtract represents the feature extraction function, including feature selection, feature scaling, and feature transformation operations; the original data feature set is F={f1, f2, ..., fk}; and the extracted new feature set is F′={f1′, f2′, ..., fl′} (l≥k); Step 3: Based on the data features extracted in Step 2 and the specific prediction target, construct a prediction model using machine learning algorithms; Step 4: Train the prediction model using historical data, and obtain the completed prediction model by continuously adjusting the model parameters to minimize the prediction error; Step 5: Input the new operation and maintenance data into the trained prediction model, obtain the prediction results, and use evaluation metrics to evaluate the prediction results; Step 6: Based on the preset early warning and processing mechanism, perform real-time response processing on the prediction results generated in Step 5.

2. The intelligent prediction method for operation and maintenance data of the operation platform according to claim 1, characterized in that, In step (4), an adaptive update mechanism is introduced for the completed training prediction model. Based on the prediction model's real-time prediction results based on the received real-time data stream, a new dataset is formed, and the prediction model is automatically updated and optimized through an online learning algorithm.

3. The intelligent prediction method for operation and maintenance data of the operation platform according to claim 1, characterized in that, The method also performs multimodal data fusion processing on the data received, fusing heterogeneous data from different sources and of different types.

4. The intelligent prediction method for operation and maintenance data of the operation platform according to claim 1, characterized in that, The method also includes a prediction result interpretation step, in which the prediction process and results of the prediction model are analyzed using visualization methods and interpretation algorithms for the prediction results obtained in step (5).

5. A smart prediction system for operation and maintenance data of an operation platform, characterized in that, The intelligent prediction system includes: The data collection and processing module is used to collect various operational platform maintenance data closely related to the prediction object, and to clean and preprocess the collected data. An automated feature engineering module is configured to automatically extract features that are of key value to the prediction task from the data collected and processed by the data collection and processing module using machine learning algorithms. The feature extraction process is as follows: F′=FeatureExtract(F); Wherein, FeatureExtract represents the feature extraction function, including feature selection, feature scaling, and feature transformation operations; the original data feature set is F={f1, f2, ..., fk}; and the extracted new feature set is F′={f1′, f2′, ..., fl′} (l≥k); A prediction model building module is configured to build a prediction model based on the data features extracted by the automated feature engineering module and the specific prediction target, and then based on machine learning algorithms. The model training and optimization module uses historical data to train the prediction model constructed by the prediction model construction module, and continuously adjusts the model parameters to minimize the prediction error to obtain a completed prediction model. The prediction and evaluation module is configured to input new operation and maintenance data into the trained prediction model, obtain prediction results, and evaluate the prediction results using evaluation metrics. The real-time response and feedback module is configured to respond to the prediction results generated by the prediction and evaluation module in real time based on a preset early warning and processing mechanism.

6. The intelligent prediction system for operation and maintenance data of the operation platform according to claim 5, characterized in that, The model training and optimization module also introduces a loss function to optimize the prediction model: ； Among them, Model opt For the optimized prediction model, the loss function is L(Model, H), and R is the regularization term.

7. The intelligent prediction system for operation and maintenance data of the operation platform according to claim 5, characterized in that, The intelligent prediction system also includes a real-time data stream processing module, which is configured to receive and process data streams in real time and feed the processed real-time data streams back to the prediction and evaluation module for real-time prediction and response.

8. The intelligent prediction system for operation and maintenance data of the operation platform according to claim 5, characterized in that, The intelligent prediction system also includes an adaptive model update module, which is configured to generate a new dataset based on the prediction model's real-time prediction results from the received real-time data stream, and then automatically update and optimize the prediction model using an online learning algorithm.

9. The intelligent prediction system for operation and maintenance data of the operation platform according to claim 5, characterized in that, The intelligent prediction system also includes a multimodal data fusion module, which can fuse heterogeneous data from different sources and of different types.

10. The intelligent prediction system for operation and maintenance data of the operation platform according to claim 5, characterized in that, The intelligent prediction system also includes a prediction result interpretation module, which is configured to analyze the prediction process and results of the prediction model through visualization methods and interpretation algorithms based on the prediction results obtained by the prediction and evaluation module.

Images