A 220-volt direct current system loss prediction method

Abstract

The application provides a 220-volt direct-current system loss prediction method, and relates to the technical field of electric power energy consumption, comprising the following steps: step 1: obtaining real-time operation parameters of each component in the system, simultaneously, obtaining electrical parameters of each component in the system and connecting lines, and constructing first data; step 2: preprocessing the first data and converting it into to-be-analyzed data in a preset format; step 3: performing feature extraction on the to-be-analyzed data, combining the obtained preset loss indicators, generating a loss feature set and a corresponding loss data set; step 4: using the selected preset model to perform prediction analysis on the loss feature set and the loss data set under the corresponding feature set, generating prediction data under each preset loss indicator, and outputting system loss prediction data. The application can perform advanced prediction on the loss data of each component in the direct-current system, timely discover potential risks in the system, and improve the safety and stability of the direct-current system.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power consumption, in particular to a 220-volt DC system loss prediction method. BACKGROUND

[0002] With the continuous progress of technology and the expansion of application fields, the application of DC system in various fields has been further developed. Compared with AC system, DC system has the advantages of high energy conversion efficiency, strong stability, high reliability, and wide application in various electronic devices. Among them, 220-volt DC system is widely used in industrial automation, telecommunication base station and other fields.

[0003] However, compared with AC system, the system structure of DC system is generally more complex, and the safety requirement is higher, so it is necessary to monitor the running state of each component in the system in real time, and find the abnormal situation in the system in time. Among them, judging the running state of the system through the loss data of each component is an effective monitoring strategy. However, the precision of the loss analysis method of 220-volt DC system is low, which is difficult to meet the increasing safety and precision requirements.

[0004] Therefore, the present application provides a 220-volt DC system loss prediction method. SUMMARY

[0005] The present application provides a 220-volt DC system loss prediction method to improve the loss analysis precision of DC system, and to predict the loss data of each component in the system in advance, so as to find the potential risk in the system in time, and to improve the safety and stability of DC system.

[0006] The present application provides a 220-volt DC system loss prediction method, comprising:

[0007] Step 1: obtaining the real-time operation parameters of each component in the system, at the same time, obtaining the electrical parameters of each component in the system and the connecting line, and constructing the first data;

[0008] Step 2: preprocessing the first data and converting it into the analysis data in the preset format;

[0009] Step 3: extracting the features of the analysis data, and combining the obtained preset loss index to generate the loss feature set and the corresponding loss data set;

[0010] Step 4: using the selected preset model to predict and analyze the loss feature set and the loss data set under the corresponding feature set, generating the prediction data under each preset loss index, and outputting the system loss prediction data.

[0011] Preferably, in step 1, it comprises:

[0012] obtaining voltage, current, temperature and load state data of each component in the system through a preset monitoring device, and outputting real-time operation parameters;

[0013] obtaining electrical parameters of each component and connection line in the system in a device database;

[0014] obtaining first data based on the real-time operation parameters and electrical parameters of each component.

[0015] Preferably, in step 2, it comprises:

[0016] performing classified analysis on the first data to obtain second data under each component in the system, and establishing a mapping relationship between the components based on the connection relationship between the components;

[0017] constructing third data based on the second data and the mapping relationship.

[0018] Preferably, in step 2, it further comprises:

[0019] performing data verification on the third data, and labeling abnormal data in the verification result;

[0020] performing data cleaning on the labeled third data, and converting the cleaned third data into analysis data according to a preset format selected in a format database.

[0021] Preferably, in step 3, it comprises:

[0022] extracting features from the analysis data through a preset feature extraction method, and selecting the extracted features based on a preset loss index to obtain a loss feature set;

[0023] extracting data corresponding to each feature from the analysis data based on the loss feature set to construct a loss data set.

[0024] Preferably, in step 3, it further comprises:

[0025] analyzing the extracted features to obtain category features of losses in the system;

[0026] The category features include resistance loss, switch loss, harmonic loss, magnetic coupling loss and other source loss.

[0027] Meanwhile, the numerical information corresponding to each category feature is obtained and identified through the preset feature extraction method, and a numerical feature is constructed.

[0028] Based on the category features and the numerical features corresponding to the category features, a first data set is generated.

[0029] inputting the first data set into a preset feature-factor matching model for matching, and outputting a matching factor;

[0030] Based on the matching factor, a corresponding preset loss indicator is selected from a preset loss indicator database, and a first loss indicator set is obtained by summarizing;

[0031] Each preset loss indicator in the first loss indicator set and the corresponding data in the first data set are subjected to data verification, and a second loss indicator set is constructed;

[0032] Obtain system selection instructions and manual selection instructions, and output an indicator selection instruction in combination with the execution priority of each selection instruction;

[0033] Based on the indicator selection instruction, an indicator is selected from the second loss indicator set, and a loss feature set is constructed according to the loss type characteristics of each indicator;

[0034] Based on the loss feature set, the numerical characteristics under each loss type characteristic are obtained from the first data set, and a first loss data set is constructed;

[0035] At the same time, based on the loss feature set, historical data corresponding to the characteristics are selected from the historical database, and a second loss data set is output;

[0036] Based on the first loss data set and the second loss data set, a loss data set is constructed.

[0037] Preferably, step 4 comprises:

[0038] Based on the loss feature set, a corresponding screening factor is obtained in combination with a preset feature-factor adaptation library;

[0039] Based on the screening factor, a first model with a matching degree greater than a preset degree is obtained from a preset model database;

[0040] Obtain the format conversion requirement corresponding to the first model, and perform format conversion on the loss data set to generate a to-be-divided data set;

[0041] In combination with a preset data division strategy, the to-be-divided data set is subjected to data division to obtain a loss training set, a loss test set, and a loss verification set, respectively;

[0042] The loss training set, the loss test set, and the loss verification set are subjected to data verification;

[0043] The loss training set that passes the verification is input into the first model for training, and the trained first model is output as a second model;

[0044] The second model is subjected to performance evaluation by the verified loss test set, and the second model is subjected to parameter adjustment by the verified loss verification set, and a model performance evaluation table is output according to the performance evaluation result and the parameter adjustment result;

[0045] Based on the preset loss index, corresponding evaluation standard index is selected in the standard database, and each evaluation result in the model performance evaluation table is compared and analyzed based on the evaluation standard index, and the second model satisfying the preset threshold condition is output as a prediction model;

[0046] The loss feature set in the preset period and the loss data set under the corresponding feature set are predicted and analyzed by the prediction model, and the prediction data under each preset loss index is generated;

[0047] The prediction data is classified and analyzed, and the sub-data of each component in the system under each preset loss index in the preset period is obtained, and the prediction curve of each component under the corresponding preset loss index is constructed;

[0048] Based on the prediction data and the prediction curve under each preset loss index, system loss prediction data is output.

[0049] Preferably, based on the prediction data and the prediction curve under each preset loss index, system loss prediction data is output, comprising:

[0050] Based on the data selection instruction, the loss feature set in the preset period and the loss data set under the corresponding feature set are obtained, and the to-be-predicted data is output;

[0051] Based on the loss data under the loss feature and the corresponding feature contained in the to-be-predicted data, historical loss data with a similarity greater than a preset degree is selected in the historical database;

[0052] The to-be-predicted data is input into the prediction model, the prediction data of each component in the system under the corresponding preset loss index is calculated, and the prediction data is compared and analyzed with the historical loss data, and the prediction offset coefficient Zi of the ith component under each preset loss index is calculated.

[0053]

[0054] Wherein, sj1 represents the predicted data value of the component under the jth preset loss index; sj2 represents the historical loss data value of the component under the jth preset loss index; bj represents the standard data value of the component under the jth preset index; mi1 represents the steady-state loss offset adjustment coefficient of the ith component; mi2 represents the transient loss offset adjustment coefficient of the ith component; m represents the total number of preset loss indexes corresponding to the ith component; and wj represents the offset coefficient corresponding to the jth preset loss index.

[0055] The predicted offset coefficients of each component in the system are counted, and a system offset data table is obtained by summarizing;

[0056] The system offset data table is compared and analyzed with a preset state judgment condition, the energy consumption states of each component in the system are judged based on the comparison result, the system loss prediction data are summarized based on the predicted curve, and a system energy consumption state table is output;

[0057] Meanwhile, the system offset data table and the predicted curve are input into an abnormality analysis module for abnormality analysis, and an abnormality analysis result is output;

[0058] Based on the system loss prediction data, the system energy consumption state table and the abnormality analysis result, corresponding energy consumption adjustment strategies and abnormality processing strategies are matched in a strategy database, and energy consumption optimization and abnormality disposal are performed on each component in the system.

[0059] The 220-volt DC system loss prediction method provided by the application can obtain first data with rich data types and high data sampling precision by obtaining real-time operation parameters and electrical parameters of each component in the DC system in real time, and then obtain to-be-analyzed data by preprocessing, improve the data quality of the first data, reduce the influence of invalid data on subsequent loss analysis, accurately identify the loss characteristics in the data through feature extraction, and predict the loss data of each component in the system corresponding to the loss characteristics based on a preset model. Not only is it convenient to optimize the loss of the system, but also potential risks in the system can be found in time, and the safety and stability of the DC system are improved. BRIEF DESCRIPTION OF DRAWINGS

[0060] In order to more clearly illustrate the technical solutions in the application or prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are some embodiments of the application, and other drawings can be obtained by those skilled in the art without creative labor.

[0061] Figure 1 is a flowchart of a 220-volt DC system loss prediction method provided by an embodiment of the application. DETAILED DESCRIPTION

[0062] In order to make the objects, technical solutions and advantages of the present application clearer, the technical solutions in the present application will be described clearly and completely below with reference to the drawings in the present application. Obviously, the described embodiments are only some but not all of the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall into the protection scope of the present application.

[0063] As shown in Figure 1 The 220-volt DC system loss prediction method provided by the embodiment of the present application comprises:

[0064] Step 1: Obtain real-time operation parameters of each component in the system, and at the same time, obtain electrical parameters of each component in the system and the connecting line, to obtain first data;

[0065] Step 2: Preprocess the first data and convert it into to-be-analyzed data in a preset format;

[0066] Step 3: Extract features from the to-be-analyzed data, and generate a loss feature set and a corresponding loss data set in combination with the obtained preset loss indicators;

[0067] Step 4: Use the selected preset model to predict and analyze the loss feature set and the loss data set under the corresponding feature set, generate prediction data under each preset loss indicator, and output system loss prediction data.

[0068] In this embodiment, the real-time operation parameters are operation data of each component in the DC system during operation, such as voltage, current, active power, reactive power, quality factor, etc.

[0069] In this embodiment, the electrical coefficient is an inherent electrical parameter of an element in each component in the DC system, such as resistance characteristic, capacitance characteristic, inductance characteristic, etc.

[0070] In this embodiment, the first data is data composed of real-time operation parameters and electrical parameters of each component in the DC system;

[0071] In this embodiment, preprocessing is a data processing method for improving data quality and facilitating subsequent data analysis, such as data cleaning, data conversion, etc.

[0072] In this embodiment, the preset format is preset format information to improve the compatibility of the system to data and facilitate the efficiency and accuracy of subsequent data processing and analysis.

[0073] In this embodiment, the data to be analyzed: that is, the first data after preprocessing and conversion to a preset format;

[0074] In this embodiment, the preset loss index: that is, the index for evaluating the loss of the DC system, is pre-set, for example, voltage drop, current loss, power loss under the steady-state loss index, overvoltage, overcurrent, instantaneous power under the transient loss index, etc.

[0075] In this embodiment, the loss feature set: that is, a set containing various loss features, for example, resistance loss, current loss, electromagnetic loss, harmonic loss, etc.

[0076] In this embodiment, the loss data set: that is, a set formed according to the loss data corresponding to the loss features, for example, resistance loss data, current loss data, electromagnetic loss data of each component, etc.

[0077] In this embodiment, the system loss prediction data: that is, the overall system prediction data obtained by aggregating the prediction data of each component in the system under each preset loss index.

[0078] The implementation principle and beneficial effects of the present embodiment are as follows: the present application obtains real-time operating parameters and electrical parameters of each component in the DC system, constructs first data with rich data types and high data sampling accuracy, and then obtains data to be analyzed through preprocessing, improves the data quality of the first data, reduces the influence of invalid data on subsequent loss analysis, accurately identifies the loss features in the data through feature extraction, and combines the preset model to predict the loss data of each component in the system corresponding to the loss features in advance, which not only facilitates loss optimization of the system, but also can timely discover potential risks in the system, and improves the safety and stability of the DC system.

[0079] The 220-volt DC system loss prediction method provided in the embodiment of the present application comprises the following steps:

[0080] The voltage, current, temperature, and load state data of each component in the system are obtained by a preset monitoring device, and real-time operating parameters are output.

[0081] The electrical parameters of each component and the connecting line in the device database are obtained.

[0082] Based on the real-time operating parameters and electrical parameters of each component, first data is obtained.

[0083] In this embodiment, the preset monitoring device: a device for monitoring the voltage, current, temperature, and load state data of each component in the system, is pre-set, for example, an ammeter, a voltmeter, an oscilloscope, a temperature sensor, a power analyzer, and other data acquisition devices, etc.

[0084] In this embodiment, the state data, i.e. the data containing the operating parameters such as voltage, current and temperature of each component in the system, can be used to judge the operating state of components such as power supply, cable, switch, inverter, transformer and load;

[0085] In this embodiment, the device database, i.e. the database storing the device information of each device in the DC system, for example, the input parameters, output parameters and control parameters of the inverter, the model, resistance and maximum limit current of each connecting cable, etc.

[0086] The implementation principle and beneficial effects of this embodiment are as follows: the preset monitoring device can accurately and comprehensively monitor the state data of each component and connecting line in the DC system in real time, and the electrical parameters obtained from the device library greatly improve the transparency of the operating data of each component in the DC system, which not only improves the accuracy and diversity of the data, but also provides convenience for subsequent loss analysis of each component and overall loss calculation of the system.

[0087] The 220-volt DC system loss prediction method provided by the embodiment of the present application comprises the following steps:

[0088] The first data is classified and analyzed to obtain second data under each component in the system, and meanwhile, the mapping relationship between the components is established in combination with the connection relationship between the components;

[0089] The third data is constructed based on the second data and the mapping relationship.

[0090] In this embodiment, the classification analysis is a method for improving the data analysis efficiency and helping relevant personnel better understand the relationship and difference between the data, and the present application can accurately divide the corresponding data into the corresponding system components, which is convenient for subsequent analysis and calculation of the loss of each component and the overall loss in combination with the connection relationship between the components;

[0091] In this embodiment, the second data is the data generated after the classification analysis of the first data;

[0092] In this embodiment, the mapping relationship is the mapping connection between the components constructed according to the connection relationship between the components;

[0093] In this embodiment, the third data is the data constructed according to the second data and the mapping relationship.

[0094] The implementation principle and beneficial effects of the embodiment are as follows: the application improves the readability of the loss data and the data structure by classifying and analyzing the first data and establishing the mapping relationship between the components, which not only facilitates the staff to deeply understand the relationship and rules between the components, but also optimizes the data structure and facilitates subsequent analysis and prediction of the loss data of the components in the direct current system.

[0095] The 220-volt direct current system loss prediction method provided by the embodiment of the application further includes the following steps in step 2.

[0096] The third data is subjected to data verification, and the abnormal data in the verification result is labeled.

[0097] The labeled third data is subjected to data cleaning, and the cleaned third data is converted according to the preset format selected in the format database to obtain the to-be-analyzed data.

[0098] In the embodiment, the data verification is a method for detecting and correcting the error information that may occur in the generation process of the third data, for example, the parity check, the cyclic redundancy check and the hash check.

[0099] In the embodiment, the abnormal data is the abnormal data selected from the third data after the verification, for example, the data loss, the data damage and the duplicate data.

[0100] In the embodiment, the data cleaning is a method for processing the abnormal data in the labeled third data, for example, the missing value filling, the abnormal value screening and the duplicate value deletion, which can improve the data quality of the third data.

[0101] In the embodiment, the format database is a database containing various data format adaptation information.

[0102] The implementation principle and beneficial effects of the embodiment are as follows: the application improves the data quality of the third data by verifying the third data and then cleaning the third data after the verification, reduces the situation of data error caused by data conversion and data transmission, further improves the accuracy of the data in the transmission and conversion process, and provides accurate data support for subsequent loss prediction.

[0103] The 220-volt direct current system loss prediction method provided by the embodiment of the application further includes the following steps in step 3.

[0104] The to-be-analyzed data is subjected to feature extraction by a preset feature extraction method, and the extracted features are selected based on the obtained preset loss index to obtain a loss feature set.

[0105] Based on the loss feature set, the data corresponding to each feature is extracted from the to-be-analyzed data to construct a loss data set.

[0106] In this embodiment, the preset feature extraction method is a method for identifying and extracting loss data features in the to-be-analyzed data, which is pre-set, and can reduce the complexity of data in the analysis process, for example, a hash feature extraction method.

[0107] The implementation principle and beneficial effects of the embodiment are as follows: The preset feature extraction method can accurately identify and extract loss data features in the to-be-analyzed data, and a loss feature set is constructed based on the preset loss indicators. Meanwhile, based on the loss feature set, the data under the corresponding feature is matched in the to-be-analyzed data to construct a loss data set. This not only greatly improves the accuracy of feature extraction, but also reduces the data calculation amount and complexity of subsequent loss analysis.

[0108] The 220-volt DC system loss prediction method provided in the embodiment of the application further includes the following steps in step 3.

[0109] The extracted features are analyzed to obtain the type features of the system internal loss;

[0110] The type features include resistance loss, switching loss, harmonic loss, magnetic coupling loss, and alien source loss.

[0111] Meanwhile, the numerical information corresponding to each type of feature is obtained and identified by the preset feature extraction method, and a numerical feature is constructed.

[0112] Based on the type features and the numerical features under the corresponding type features, a first data set is generated.

[0113] The first data set is input into a preset feature-factor matching model for matching, and a matching factor is output.

[0114] Based on the matching factor, a corresponding preset loss indicator is selected from a preset loss indicator database, and a first loss indicator set is obtained.

[0115] Each preset loss indicator in the first loss indicator set and the corresponding data in the first data set are verified, and a second loss indicator set is constructed.

[0116] The system self-selected instruction and the artificial selected instruction are obtained, and an indicator selection instruction is output in combination with the execution priority of each selected instruction.

[0117] Based on the indicator selection instruction, an indicator is selected in the second loss indicator set, and a loss feature set is constructed according to the loss type features corresponding to each indicator.

[0118] Based on the loss feature set, the numerical features under each loss category feature are obtained in the first data set, and the first loss data set is constructed;

[0119] At the same time, based on the loss feature set, the historical data corresponding to the features are selected in the historical database, and the second loss data set is output;

[0120] Based on the first loss data set and the second loss data set, the loss data set is constructed.

[0121] In this embodiment, the resistance loss: the loss caused by the resistance elements (such as cables, connectors, resistors, etc.) in the system;

[0122] In this embodiment, the switching loss: the energy loss generated by the switching devices (such as transistors, diodes, etc.) in the switching process;

[0123] In this embodiment, the harmonic loss: the energy loss caused by the harmonic current and voltage in the system;

[0124] In this embodiment, the magnetic coupling loss: the energy loss caused by the magnetic coupling in the transformer, inductor and other devices in the system;

[0125] In this embodiment, the other loss: such as iron loss, copper loss, contact resistance loss and heat dissipation loss;

[0126] In this embodiment, the numerical feature: the feature obtained by feature recognition and extraction of the numerical information corresponding to each category feature;

[0127] In this embodiment, the first data set: the data set generated according to the category feature and the numerical feature under the corresponding category feature;

[0128] In this embodiment, the preset feature-factor matching model: the matching model containing the mapping relationship between the category feature, the numerical feature and the matching factor, which is preset, used to obtain the corresponding matching factor according to the input category feature and numerical feature;

[0129] In this embodiment, the matching factor: the factor used to select the preset loss index in the preset loss index database;

[0130] In this embodiment, the preset loss index database: a database containing a large number of indexes for evaluating the loss of direct current system;

[0131] In this embodiment, the first loss index set: a set composed of each preset loss index selected in the preset loss index database according to the matching factor;

[0132] In this embodiment, the second loss index set: the first loss index set after data verification;

[0133] In this embodiment, the system self-selection instruction is an instruction that the system automatically selects;

[0134] In this embodiment, the manual selection instruction is a selection instruction input by a person;

[0135] In this embodiment, the execution priority is the priority of the system self-selection instruction and the manual selection instruction when executed, and generally, the execution priority of the manual selection instruction is higher than that of the system self-selection instruction;

[0136] In this embodiment, the index selection instruction is an instruction generated according to the system self-selection instruction and the manual selection instruction, and used for index selection in the second loss index set;

[0137] In this embodiment, the first loss data set is a data set composed of loss type characteristics and corresponding numerical characteristics of each loss type in the first data set according to the loss characteristic set;

[0138] In this embodiment, the second loss data set is a data set composed of historical data selected from the historical database according to the loss characteristic set.

[0139] The implementation principle and beneficial effects of the embodiment are as follows: the present application analyzes the extracted characteristics to obtain loss type characteristics of each component in the system and numerical characteristics under the corresponding type characteristics, and then selects corresponding loss indexes according to the obtained type characteristics and numerical characteristics, and subsequently constructs a loss data set according to the loss characteristic set, thereby laying a foundation for subsequent loss prediction analysis.

[0140] The 220-volt DC system loss prediction method provided by the embodiment of the present application comprises the following steps:

[0141] Based on the loss characteristic set, a corresponding screening factor is obtained by combining a preset characteristic-factor adaptation library;

[0142] Based on the screening factor, a first model with a matching degree greater than a preset degree is obtained from a preset model database;

[0143] The format conversion requirement corresponding to the first model is obtained, and the loss data set is format-converted to generate a to-be-divided data set;

[0144] The to-be-divided data set is divided according to a preset data division strategy, and loss training set, loss test set and loss verification set are obtained;

[0145] The loss training set, the loss test set and the loss verification set are verified;

[0146] The verified loss training set is input into the first model for training, and the trained first model is output as a second model;

[0147] The second model is evaluated based on the verified loss test set, and the second model is tuned based on the verified loss verification set. The model performance evaluation table is output based on the evaluation result and the tuning result;

[0148] Based on the preset loss index, the corresponding evaluation standard index is selected from the standard database, and each evaluation result in the model performance evaluation table is compared and analyzed based on the evaluation standard index. The second model that meets the preset threshold condition is output as a prediction model;

[0149] The loss feature set and the corresponding feature set within the preset period are predicted and analyzed by the prediction model to generate predicted data for each preset loss index;

[0150] The predicted data is classified and analyzed to obtain sub-data of each component in the system under each preset loss index within the preset period, and a prediction curve of each component under the corresponding preset loss index is constructed;

[0151] Based on the predicted data and the prediction curve under each preset loss index, system loss prediction data is output.

[0152] In this embodiment, the preset feature-factor adaptation library: contains the mapping relationship between the loss feature and the screening factor, and is pre-set, used to obtain the corresponding screening factor according to the input loss feature;

[0153] In this embodiment, the screening factor: is used to select the screening condition of the first model in the preset model database;

[0154] In this embodiment, the preset model database: contains a database of multiple data analysis models, and is pre-set, used to match the adapted training data model according to the input screening factor;

[0155] In this embodiment, the first model: is an initial model selected from the preset model database according to the screening factor;

[0156] In this embodiment, the format conversion requirement: is the data format conversion requirement corresponding to the first model. According to the format conversion requirement, the loss data to be trained can be format converted to improve the processing efficiency of the data;

[0157] In this embodiment, the to-be-divided data set: is the loss data set after format conversion;

[0158] In this embodiment, the preset data division strategy is that the data set division strategy for training and optimizing the first model is preset, for example, a K-Fold cross-validation division strategy;

[0159] In this embodiment, the loss training set is a data set for training the first model, and the mode and features of the loss data can be learned through the loss training set;

[0160] In this embodiment, the loss test set is a data set for evaluating the performance of the model on unobserved loss data;

[0161] In this embodiment, the loss validation set is a data set for adjusting the hyperparameters of the model;

[0162] In this embodiment, the second model is the first model after training;

[0163] In this embodiment, the model performance evaluation table is an evaluation table containing the performance evaluation results and the parameter adjustment results of the second model, which is used to represent the training and optimization effect of the second model and facilitate comparison and analysis with the subsequent preset threshold condition to determine whether the second model can be used as a prediction model of the system loss;

[0164] In this embodiment, the standard database contains evaluation standard indexes of each preset loss indicator for the rolling stock, and is used to select corresponding standard data as a reference according to the input preset loss indicator;

[0165] In this embodiment, the evaluation standard index is a standard index for comparison and analysis with each evaluation result in the model performance evaluation table;

[0166] In this embodiment, the preset threshold condition is a threshold condition for determining whether the second model can be used as a prediction model of the DC system loss, which is preset;

[0167] In this embodiment, the sub-data is data corresponding to each component in the system obtained after classification and analysis of the prediction data;

[0168] In this embodiment, the prediction curve is a data change curve generated according to the sub-data of each component according to time scale or other dimensions;

[0169] Principles and benefits of the embodiment: The application first obtains a screening factor through a loss data set, then screens an adaptive to-be-trained data model in a model database through the screening factor, subsequently performs format conversion on the loss data set in combination with a format conversion requirement corresponding to the model, trains and optimizes the model based on a training set, a test set and a verification set obtained based on a preset data division strategy, finally performs prediction analysis on the loss data set through the optimized model, and obtains system loss prediction data, which not only improves the generalization ability of the model, but also prevents overfitting of the model on the loss training set, thereby ensuring the precision of the prediction data of the system loss of the model.

[0170] The 220-volt DC system loss prediction method provided by the embodiment of the application outputs system loss prediction data based on prediction data and a prediction curve under each preset loss index, and comprises the following steps:

[0171] Based on a data selection instruction, a loss feature set in a preset period and a loss data set under the corresponding feature set are obtained, and to-be-predicted data are outputted and obtained;

[0172] Based on the loss features contained in the to-be-predicted data and the loss data under the corresponding features, historical loss data with a similarity greater than a preset degree are selected from a historical database;

[0173] The to-be-predicted data are inputted into a prediction model, prediction data of each component in the system under the corresponding preset loss index are calculated, and the prediction data are compared and analyzed with the historical loss data, and a prediction offset coefficient Zi of the ith component under each preset loss index is calculated;

[0174]

[0175] Wherein, sj1 represents a prediction data value of the component under the jth preset loss index; sj2 represents a historical loss data value of the component under the jth preset loss index; bj represents a standard data value of the component under the jth preset index; μi1 represents a steady-state loss offset adjustment coefficient of the ith component; μi2 represents a transient-state loss offset adjustment coefficient of the ith component; m represents a total number of preset loss indexes corresponding to the ith component; and ωj represents an offset coefficient corresponding to the jth preset loss index.

[0176] The prediction offset coefficients of each component in the system are counted, and a system offset data table is obtained by summarizing;

[0177] The system offset data table is compared and analyzed with a preset state determination condition, the energy consumption states of each component in the system are determined based on the comparison result, and the system loss prediction data are summarized in combination with the prediction curve, and a system energy consumption state table is outputted and obtained;

[0178] Meanwhile, the system offset data table and the prediction curve are input into the anomaly analysis module for anomaly analysis, and an anomaly analysis result is output.

[0179] Based on the system loss prediction data, the system energy consumption state table and the anomaly analysis result, corresponding energy consumption adjustment strategies and anomaly handling strategies are matched in the strategy database, and energy consumption optimization and anomaly handling are performed on each component in the system.

[0180] In this embodiment, the data selection instruction is an instruction for obtaining system loss characteristics and loss data in a preset period;

[0181] In this embodiment, the data to be predicted is system loss data to be predicted;

[0182] In this embodiment, the historical loss data is similar historical data matched in the historical database according to the loss characteristics and loss data in the preset period;

[0183] In this embodiment, the prediction offset coefficient is a coefficient for representing the size of the offset of the prediction data value of each component of the system under each preset loss indicator compared with the standard data value;

[0184] In this embodiment, the system offset data table is a data table containing the prediction offset coefficient of each component in the system;

[0185] In this embodiment, the preset state determination condition is a determination condition for determining the energy consumption state of each component according to the system offset data table, which is pre-set;

[0186] In this embodiment, the system energy consumption state table is a data table containing the energy consumption state of each component in the system and the system overall loss prediction data;

[0187] In this embodiment, the anomaly analysis module is a module for performing anomaly analysis on the system offset data table and the prediction curve, and is used to identify and analyze whether the running data and the running state of each component in the system are abnormal;

[0188] In this embodiment, the anomaly analysis result is an analysis result obtained by the anomaly analysis module;

[0189] In this embodiment, the strategy database is a database containing strategies for energy consumption adjustment and anomaly handling of each component in the system;

[0190] In this embodiment, the energy consumption adjustment strategy is a strategy for adjusting and optimizing the energy consumption state of each component in the system according to the system loss prediction data, the system energy consumption state table and the anomaly analysis result;

[0191] In this embodiment, the abnormality processing strategy is a strategy for processing abnormal data and abnormal states of each component in the system according to the abnormality analysis result.

[0192] The implementation principle and beneficial effects of the embodiment are as follows: the present application performs prediction analysis on the to-be-predicted data in a preset period, compares and analyzes the prediction data with the historical loss data selected from the historical database, further calculates the prediction offset coefficient of each component in the system under each preset loss index, determines the prediction accuracy of each component, and obtains the system offset data table by summarizing, then compares and analyzes the preset state determination condition, determines the energy consumption state of each component in the system, and can also accurately identify the abnormal data and state of each component, and combines the obtained strategy to perform energy consumption optimization adjustment and abnormality processing on the corresponding component, which not only improves the prediction accuracy of the data, but also can discover potential risks in the system in time, and ensures the stable operation of the direct current system.

[0193] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for predicting losses in a 220V DC system, characterized in that, include: Step 1: Obtain the real-time operating parameters of each component in the system, and at the same time, obtain the electrical parameters of each component and the connecting lines in the system to construct the first data; Step 2: Preprocess the first data and convert it into data to be analyzed in a preset format; Step 3: Extract features from the data to be analyzed, and combine them with the obtained preset loss indicators to generate a loss feature set and a corresponding loss dataset. Step 4: Use the selected preset model to perform predictive analysis on the loss feature set and the loss dataset under the corresponding feature set, generate predictive data under each preset loss index, and output the system loss prediction data. Step 4 includes: Based on the loss feature set, the corresponding screening factor is obtained by combining the preset feature-factor adaptation library. Based on the screening factors, a first model with a matching degree greater than the preset degree is obtained from the preset model database; Obtain the format conversion requirements corresponding to the first model, and convert the loss dataset to generate a dataset to be partitioned; The dataset to be divided is divided according to a preset data partitioning strategy to obtain a loss training set, a loss test set, and a loss validation set. Data validation is performed on the loss training set, loss test set, and loss validation set. The verified loss training set is input into the first model for training, and the trained first model is output as the second model; The second model will be evaluated using the verified loss test set, and the second model will be tuned using the verified loss verification set. A model performance evaluation table will be output based on the performance evaluation results and the tuning results. Based on the preset loss index, the corresponding evaluation standard index is selected from the standard database. Based on the evaluation standard index, each evaluation result in the model performance evaluation table is compared and analyzed. The second model whose comparison result meets the preset threshold condition is output as the prediction model. The prediction model is used to predict and analyze the loss feature set and the loss dataset under the corresponding feature set within a preset time period, and to generate prediction data for each preset loss index. The predicted data is categorized and analyzed to obtain sub-data for each component in the system under each preset loss index within a preset time period, and a prediction curve for each component under the corresponding preset loss index is constructed. Based on the predicted data and prediction curves under each preset loss index, the system loss prediction data is output.

2. The method for predicting losses in a 220V DC system according to claim 1, characterized in that, Step 1 includes: The system acquires voltage, current, temperature and load status data of each component by using pre-set monitoring equipment, and outputs real-time operating parameters. Obtain the electrical parameters of each component and connecting line within the system from the device database; The first data is obtained based on the real-time operating parameters and electrical parameters corresponding to each component.

3. The loss prediction method for a 220V DC system according to claim 1, characterized in that, Step 2 includes: The first data is categorized and analyzed to obtain the second data under each component in the system. At the same time, the mapping relationship between each component is established by combining the connection relationship between each component. Based on the second data and the mapping relationship, the third data is constructed.

4. The loss prediction method for a 220V DC system according to claim 3, characterized in that, Step 2 also includes: The third data is validated, and any abnormal data in the validation results is marked. The labeled third data is cleaned and then transformed according to the preset format selected in the format database to obtain the data to be analyzed.

5. The method for predicting losses in a 220V DC system according to claim 1, characterized in that, Step 3 includes: The data to be analyzed is subjected to feature extraction using a preset feature extraction method, and the extracted features are selected based on the obtained preset loss index to obtain a loss feature set. Based on the loss feature set, the data corresponding to each feature is extracted from the data to be analyzed, and a loss dataset is constructed.

6. The loss prediction method for a 220V DC system according to claim 5, characterized in that, Step 3 also includes: The extracted features are analyzed to obtain the types of losses within the system; The aforementioned characteristics include: resistive loss, switching loss, harmonic loss, magnetic coupling loss, and other source losses. Simultaneously, the numerical information corresponding to each type of feature is acquired and identified through the preset feature extraction method, and numerical features are constructed. Based on the aforementioned category features and the corresponding numerical features under those category features, a first dataset is generated; The first dataset is input into a preset feature-factor matching model for matching, and the matching factors are output. Based on the matching factor, the corresponding preset loss index is selected from the preset loss index database and summarized to obtain the first loss index set. For each preset loss index in the first loss index set, data verification is performed on the corresponding data in the first dataset, and a second loss index set is constructed. Obtain system-selected instructions and manual selection instructions, and combine the execution priority of each selection instruction to output the indicator selection instructions; Based on the indicator selection instructions, indicators are selected from the second set of loss indicators, and a loss feature set is constructed according to the loss type characteristics corresponding to each indicator. Based on the loss feature set, numerical features under each loss type feature are obtained in the first dataset, and the first loss dataset is constructed. Simultaneously, based on the loss feature set, historical data under the corresponding features are selected from the historical database, and a second loss dataset is output. A loss dataset is constructed based on the first loss dataset and the second loss dataset.

7. The method for predicting losses in a 220V DC system according to claim 6, characterized in that, Based on the predicted data and prediction curves under each preset loss index, the system loss prediction data is output, including: Based on the data selection instructions, the loss feature set and the corresponding loss dataset under the feature set are obtained within a preset time period, and the data to be predicted is output. Based on the loss features contained in the data to be predicted and the loss data under the corresponding features, historical loss data with a similarity greater than a preset degree are selected from the historical database. The data to be predicted is input into the prediction model to calculate the predicted data of each component in the system under the corresponding preset loss index. The predicted data is then compared and analyzed with historical loss data to calculate the prediction offset coefficient of the i-th component under each preset loss index. ; ； in, This represents the predicted data value of the component under the j-th preset loss index. bj represents the historical loss data value of the component under the j-th preset loss index; bj represents the standard data value of the component under the j-th preset index. This represents the steady-state loss offset adjustment coefficient for the i-th component; This represents the transient loss offset adjustment coefficient for the i-th component; m represents the total number of preset loss indicators corresponding to the i-th component. This represents the offset coefficient corresponding to the j-th preset loss index; The predicted offset coefficients of each component in the system are statistically analyzed and summarized to obtain a system offset data table; The system offset data table is compared and analyzed with the preset state determination conditions. Based on the comparison results, the energy consumption status of each component in the system is determined. The system loss prediction data is summarized in combination with the prediction curve and output as a system energy consumption status table. Simultaneously, the system offset data table and the prediction curve are input into the anomaly analysis module for anomaly analysis, and the anomaly analysis results are output. Based on the system loss prediction data, system energy consumption status table, and anomaly analysis results, corresponding energy consumption adjustment strategies and anomaly handling strategies are matched in the strategy database, and energy consumption optimization and anomaly handling are performed on each component in the system.

Images