Electromechanical equipment operation state evaluation method and device, terminal equipment and storage medium
By preprocessing data with state labels and balancing sample sizes, the method addresses the challenge of inaccurate state predictions in mechanical and electrical equipment, enhancing model learning and fault detection accuracy.
Patent Information
- Application Number
- CN202510488178.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-04-18
- Publication Date
- 2025-07-15
AI Technical Summary
When traditional methods process historical operation data of electromechanical equipment, the data volume is huge and messy, resulting in weak generalization capabilities of decision trees and random forest models, making it difficult to accurately evaluate the operating status of the equipment.
By assigning the run state label to the initial historical running dataset for classification, balancing the sample count, generating training sets and test sets using rewinding sampling, training decision trees, and integrating random forest models for evaluation.
It improves the accuracy of the model's evaluation of the operating state of small samples, enhances the generalization ability and robustness of the model, improves the reliability of the evaluation results, and supports the maintenance of electromechanical equipment and fault prediction.
Smart Images
Figure CN120316634A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of electromechanical equipment monitoring, and in particular to a method, device, terminal device and storage medium for evaluating the operating state of electromechanical equipment. Background Art
[0002] When dealing with the historical operation data of electromechanical equipment by traditional methods, there are many intractable problems. The initial historical operation data generated by electromechanical equipment is extremely large and chaotic, and the historical operation data of different electromechanical equipment is mixed with each other without effective classification and sorting. This situation directly leads to weak generalization ability of the decision tree and random forest models trained based on these data, making it difficult to accurately evaluate the actual operating state of electromechanical equipment, and further resulting in inaccurate prediction of the operating state of electromechanical equipment in the prior art. Summary of the Invention
[0003] The present invention provides a method, device, terminal device and storage medium for evaluating the operating state of electromechanical equipment, and the method can solve the problem of inaccurate prediction of the operating state of electromechanical equipment existing in the prior art.
[0004] An embodiment of the present invention provides a method for evaluating the operating state of electromechanical equipment, including:
[0005] Obtaining an initial historical operation data set; wherein, the initial historical operation data set includes the historical operation data of a plurality of electromechanical equipment;
[0006] Assigning corresponding operating state labels to each historical operation data in the initial historical operation data set to obtain a first historical operation data set;
[0007] Classifying each historical operation data in the first historical operation data set based on the operating state label to obtain a second historical operation data set; wherein, the second historical operation data set includes a plurality of sample subsets for representing different operating states;
[0008] Balancing the number of samples in each sample subset in the second historical operation data set to obtain a third historical operation data set;
[0009] Performing sampling with replacement on the third historical operation data set to obtain a plurality of training sets and test sets;
[0010] Training a preset decision tree according to the training set and the test set to obtain a trained decision tree, and integrating the trained decision trees to obtain a random forest model;
[0011] Evaluating the operating state of the electromechanical equipment according to the random forest model.
[0012] Further, the categories of the operation status tags include normal, abnormal, and potential failure;
[0013] Classifying each historical operation data in the first historical operation dataset based on the operation status tags, to obtain a second historical operation dataset, including:
[0014] Classifying each historical operation data in the first historical operation dataset according to the operation status tags of the historical operation data, to obtain a number of sample subsets; wherein, the sample subsets include a normal sample subset, an abnormal sample subset, and a potential failure sample subset.
[0015] Further, balancing the number of samples in each sample subset in the second historical operation dataset, to obtain a third historical operation dataset, including:
[0016] Identifying the number of samples in the normal sample subset, the abnormal sample subset, and the potential failure sample subset respectively, to obtain a sample subset identified as a minority class, a sample subset identified as a majority class, and a sample subset identified as a class that does not need to be adjusted;
[0017] For the sample subset identified as the minority class, randomly generating a number of synthetic samples according to the distances between the samples in the sample subset, to obtain an expanded sample subset;
[0018] For the sample subset identified as the majority class, randomly deleting a number of samples in the sample subset, to obtain a reduced sample subset;
[0019] Integrating the expanded sample subset, the reduced sample subset, and the sample subset identified as the class that does not need to be adjusted into the third historical operation dataset.
[0020] Further, training a preset decision tree according to the training set and the test set, to obtain a trained decision tree, including:
[0021] For each training set, inputting the training set into a preset untrained decision tree, so that the untrained decision tree performs iterative training according to the historical operation data and the corresponding operation status tags in the training set, until reaching a preset convergence condition, stopping the iterative training, to obtain a preliminarily trained decision tree;
[0022] For each preliminarily trained decision tree, using the test set as the input of the decision tree, and determining the evaluation accuracy of the decision tree according to the evaluation result output by the decision tree;
[0023] According to the evaluation accuracy of each decision tree completed by the preliminary training, assign corresponding weights to each decision tree completed by the preliminary training to obtain each trained decision tree.
[0024] Further, the evaluation of the operating state of the electromechanical equipment according to the random forest model includes:
[0025] Obtain the real-time operating data of the electromechanical equipment;
[0026] Input the real-time operating data into the random forest model, so that each trained decision tree in the random forest model evaluates the real-time operating data respectively, obtain the evaluation results output by each trained decision tree, use the weights of each trained decision tree to perform weighted analysis on the evaluation results output by each trained decision tree, obtain the weighted analysis result, and determine the operating state of the real-time operating data according to the weighted analysis result.
[0027] Further, the electromechanical equipment includes a compressor and a DAC device; the initial historical operating data set includes the historical vibration signal of the compressor, the historical temperature signal of the compressor, the historical pressure signal of the compressor, and the historical voltage signal of the DAC device.
[0028] Further, the method for evaluating the operating state of the electromechanical equipment further includes: triggering an alarm when it is determined that the operating state of the real-time operating data is abnormal or potentially faulty.
[0029] An embodiment of the present invention further provides an apparatus for evaluating the operating state of an electromechanical equipment, including: an initial historical operating data set acquisition module, a first historical operating data set acquisition module, a second historical operating data set acquisition module, a third historical operating data set acquisition module, a training set and test set acquisition module, a model training module, and an operating state evaluation module;
[0030] The initial historical operating data set acquisition module is used to acquire an initial historical operating data set; wherein, the initial historical operating data set includes the historical operating data of several electromechanical equipments;
[0031] The first historical operating data set acquisition module is used to assign corresponding operating state labels to each historical operating data in the initial historical operating data set to obtain a first historical operating data set;
[0032] The second historical operating data set acquisition module is used to classify each historical operating data in the first historical operating data set based on the operating state label to obtain a second historical operating data set; wherein, the second historical operating data set includes several sample subsets used to represent different operating states;
[0033] The third historical operation dataset acquisition module is configured to balance the number of samples in each sample subset in the second historical operation dataset to obtain a third historical operation dataset;
[0034] The training set and test set acquisition module is configured to perform sampling with replacement on the third historical operation dataset to obtain a number of training sets and test sets;
[0035] The model training module is configured to train a preset decision tree according to the training set and the test set to obtain a trained decision tree, and integrate the trained decision trees to obtain a random forest model;
[0036] The operation state evaluation module is configured to evaluate the operation state of the electromechanical device according to the random forest model.
[0037] This application also provides a terminal device, including:
[0038] One or more processors;
[0039] A memory coupled to the processor for storing one or more programs;
[0040] When the one or more programs are executed by the one or more processors, the one or more processors implement the electromechanical device operation state evaluation method as described in the above-mentioned invention embodiment.
[0041] This application also provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the electromechanical device operation state evaluation method as described in the above-mentioned invention embodiment is implemented.
[0042] By implementing the present invention, the following beneficial effects are achieved:
[0043] The present invention provides a method, device, terminal device and storage medium for evaluating the operating state of electromechanical equipment. The method assigns corresponding operating state labels to each historical operating data in the initial historical operating dataset, and then classifies them based on the operating state labels to obtain a precise second historical operating dataset containing different operating state sample subsets, effectively separating the data features. The number of samples in each sample subset in the second historical operating dataset is balanced to obtain a third historical operating dataset. This operation effectively avoids the problem of class bias in the model training process, enables the model to fully learn the data features of various operating states, improves the evaluation accuracy of the model for small-sample operating states, enhances the generalization ability and robustness of the model, and then performs sampling with replacement from the third historical operating dataset to generate several training sets and test sets, providing a scientific sample combination for the training of decision trees. The decision tree and random forest model trained based on this can more accurately reflect the true operating condition of the equipment when evaluating the operating state of electromechanical equipment, greatly improving the reliability of the evaluation results, and providing support for the maintenance, management and fault prediction of electromechanical equipment. BRIEF DESCRIPTION OF THE DRAWINGS
[0044] In order to more clearly illustrate the technical solutions of the present application, the accompanying drawings required for the embodiments will be briefly introduced below. Obviously, the accompanying drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.
[0045] Figure 1 is a schematic flowchart of a method for evaluating the operating state of electromechanical equipment provided by an embodiment of the present application;
[0046] Figure 2 is a schematic diagram of the evaluation accuracy of a traditional random forest provided by an embodiment of the present application;
[0047] Figure 3 is a schematic diagram of the evaluation accuracy of a random forest after integrating decision trees provided by an embodiment of the present application;
[0048] Figure 4 is a flowchart of the weighted analysis of a random forest model based on each decision tree provided by an embodiment of the present application;
[0049] Figure 5 is a schematic flowchart of the analysis process for evaluating the operating state of electromechanical equipment based on a random forest provided by an embodiment of the present application;
[0050] Figure 6 is a schematic structural diagram of an apparatus for evaluating the operating state of electromechanical equipment provided by an embodiment of the present application;
[0051] Figure 7 It is a schematic structural diagram of a terminal device provided by an embodiment of the present application. Detailed implementation manners
[0052] To make the objectives, technical solutions and advantages of the present application clearer, the technical solutions in the present application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present application. Apparently, the described embodiments are some but not all of the embodiments of the present application. All other embodiments obtained by those of ordinary skill in the art without making creative efforts based on the embodiments in the present application belong to the scope of protection of the present application.
[0053] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which this application belongs; the terms used herein are only for the purpose of describing specific embodiments and are not intended to limit this application; the terms "including" and "having" and any variations thereof in the specification and claims of this application and the above drawings are intended to cover non-exclusive inclusion.
[0054] In the description of the embodiments of the present application, technical terms such as "first" and "second" are only used to distinguish different objects and cannot be understood as indicating or implying relative importance or implicitly indicating the quantity, specific order or primary-secondary relationship of the indicated technical features. In the description of the embodiments of the present application, "a plurality of" means two or more, unless otherwise specifically defined.
[0055] Referring to "embodiments" herein means that specific features, structures or characteristics described in connection with the embodiments can be included in at least one embodiment of the present application. The phrase appears in various places in the specification and does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment mutually exclusive with other embodiments. Those skilled in the art will explicitly and implicitly understand that the embodiments described herein can be combined with other embodiments.
[0056] In the description of the embodiments of the present application, the term "and / or" is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist simultaneously, and B exists alone. In addition, the character " / " in this article generally represents an "or" relationship between the associated objects before and after.
[0057] In the description of the embodiments of the present application, the term "a plurality of" refers to two or more (including two). Similarly, "a plurality of groups" refers to two or more groups (including two groups), and "a plurality of pieces" refers to two or more pieces (including two pieces).
[0058] In the description of the embodiments of the present application, unless otherwise clearly specified and limited, technical terms such as "installation", "connection", "connection", "fixation", etc. should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or integrated; it can also be a mechanical connection or an electrical connection; it can be directly connected, or indirectly connected through an intermediate medium, and can be the communication inside two components or the interaction relationship between two components. For those of ordinary skill in the art, the specific meanings of the above terms in the embodiments of the present application can be understood according to specific situations.
[0059] See Figure 1 , to solve the problem of inaccurate prediction of the operating state of electromechanical equipment in the prior art, an electromechanical equipment operating state evaluation method provided by an embodiment of the present invention includes:
[0060] S1. Obtain an initial historical operation data set; wherein, the initial historical operation data set includes historical operation data of several electromechanical devices;
[0061] In a preferred embodiment, the electromechanical devices include a compressor and a DAC device; the initial historical operation data set includes the historical vibration signal of the compressor, the historical temperature signal of the compressor, the historical pressure signal of the compressor, and the historical voltage signal of the DAC device;
[0062] Specifically, sensors in the quantum metrology device are used to collect the operation data of the electromechanical devices to ensure that it can continuously and real-time monitor the operation data; these collected operation data cover the vibration signal of the compressor, the temperature signal of the compressor, the pressure signal of the compressor, and the voltage signal of the DAC device, and each operation data presents one of the three different operation states of normal operation, abnormal operation, and potential failure;
[0063] Specifically, the collected historical operation data is denoised. By means of filters, smoothing processing, etc., the data noise caused by sensor errors or environmental interference is reduced, and at the same time, missing values are checked and filled by methods such as mean filling and interpolation filling, and then these historical operation data are integrated into an initial historical operation data set.
[0064] S2. Assign corresponding operation state labels to each historical operation data in the initial historical operation data set to obtain a first historical operation data set;
[0065] Schematically, the operation states of each historical operation data in the initial historical operation data set are marked;
[0066] Specifically, for the historical vibration signals of the compressor, judgment is made based on indicators such as its vibration frequency, amplitude, and stability of the vibration waveform. For example, if the vibration frequency is within the frequency range set for the normal operation of the equipment, the amplitude is within the allowable fluctuation range, and the vibration waveform shows a regular and stable state, then the corresponding historical operation data is marked with a "normal" operation status label; once the vibration frequency exceeds the normal range, or the amplitude shows a large abnormal fluctuation, or the vibration waveform is disordered, combined with the actual operating conditions of the equipment, if this abnormal situation has affected the normal operation of the equipment, it is marked with an "abnormal" operation status label, and if it has not yet had an obvious impact on the operation of the equipment but may cause a failure in the long-term development, it is marked with a "potential failure" operation status label;
[0067] For the historical temperature signals of the compressor, referring to the designed temperature range of the equipment and the temperature fluctuation range during previous normal operations, if the temperature signal is within the normal fluctuation range, it is marked as "normal". When the temperature value is higher than the normal upper limit and lasts for a period of time, based on the exceeding amplitude and the duration, it is judged whether the equipment has an abnormal working condition. If so, it is marked as "abnormal", and if not, it is marked as "potential failure"; similarly, for the historical pressure signals of the compressor, by comparing with the standard pressure value and the reasonable rate of pressure change, the determination of the operation status label is carried out; for the historical voltage signals of the DAC equipment, according to parameters such as the rated voltage value of the equipment and the allowable voltage deviation range, the corresponding operation status labels are assigned to each historical operation data;
[0068] It should be noted that the specific judgment criteria can be adjusted accordingly according to the actual situation;
[0069] Specifically, after completing the operation status marking, each historical operation data in the initial historical operation data set is checked to ensure that each historical operation data corresponds to an operation status label, so as to integrate the historical operation data with the assigned operation status labels to form the first historical operation data set.
[0070] S3. Classify each historical operation data in the first historical operation data set based on the operation status label to obtain the second historical operation data set; among them, the second historical operation data set includes several sample subsets used to represent different operation states;
[0071] In a preferred embodiment, the categories of the operation status labels include normal, abnormal, and potential failure;
[0072] The classifying each historical operation data in the first historical operation data set based on the operation status label to obtain the second historical operation data set includes:
[0073] Classify each piece of historical operation data in the first historical operation data set according to the operation status label of each piece of historical operation data, to obtain several sample subsets; wherein, the sample subsets include a normal sample subset, an abnormal sample subset, and a potential fault sample subset;
[0074] Specifically, after obtaining the first historical operation data set, taking the operation status label as the classification basis, classify the historical operation data with the operation status label of "normal" in the first historical operation data set into the normal sample subset, and classify the historical operation data with the operation status label of "abnormal" in the first historical operation data set into the abnormal sample subset, and classify the historical operation data with the operation status label of "potential fault" in the first historical operation data set into the potential fault sample subset.
[0075] S4. Balance the number of samples in each sample subset in the second historical operation data set to obtain a third historical operation data set;
[0076] In a preferred embodiment, the balancing the number of samples in each sample subset in the second historical operation data set to obtain a third historical operation data set includes:
[0077] Identify the number of samples in the normal sample subset, the abnormal sample subset, and the potential fault sample subset respectively, to obtain a sample subset identified as a minority class, a sample subset identified as a majority class, and a sample subset identified as a class that does not need to be adjusted;
[0078] For the sample subset identified as the minority class, randomly generate several synthetic samples according to the distance between the samples in the sample subset to obtain an expanded sample subset;
[0079] For the sample subset identified as the majority class, randomly delete several samples in the sample subset to obtain a reduced sample subset;
[0080] Integrate the expanded sample subset, the reduced sample subset, and the sample subset identified as the class that does not need to be adjusted into the third historical operation data set;
[0081] Schematically, use a synthetic data algorithm to expand the samples in the sample subset identified as the minority class;
[0082] It should be noted that after obtaining the sample numbers of the normal sample subset, the abnormal sample subset, and the potential fault sample subset, it is necessary to first sort the sample numbers of these three sample subsets. The sample subset with the sample number at the median is used as the benchmark subset, the sample subset with the sample number higher than the benchmark subset is used as the majority class, and the sample subset identified as the majority class is recognized as such. The sample subset with the sample number lower than the benchmark subset is used as the minority class, and the sample subset identified as the minority class is recognized as such. When there are two subsets with the same sample number, if the sample numbers of these two subsets with the same number are higher than the remaining sample subsets, then the remaining sample subset is identified as the minority class, and both of these two sample subsets with the same number are identified as the majority class. Conversely, these two sample subsets with the same number are identified as the minority class, and the remaining sample subset is identified as the majority class;
[0083] Specifically, taking the SMOTE method as an example for the synthetic data algorithm, assuming that the abnormal sample subset is identified as the minority class, synthetic samples are generated by the SMOTE method to expand the samples in the abnormal sample subset. That is, one or more neighbors are selected for the samples in the abnormal sample subset, and new synthetic samples are generated by the interpolation method, thereby effectively increasing the number of samples in the abnormal sample subset and avoiding the impact on the model accuracy caused by the unbalanced sample number. The specific calculation expression is as follows:
[0084] new_data = x + rand * (y i - x) (i = 1, 2,..., K);
[0085] In the formula: rand is a random number between (0, 1); new_data is the new sample, x is the sample in the minority class sample subset, y i is one of the neighboring samples, and K is the total number of neighboring samples;
[0086] Specifically, for the sample subset identified as the majority class, a method of randomly deleting some samples is adopted to reduce the sample number of the sample subset identified as the majority class, so as to balance it with the sample number of the sample subset identified as the minority class. During the deletion process, it is ensured that the deleted samples will not affect the representativeness of the data set, and the diversity of the samples is retained as much as possible;
[0087] After adjusting the sample subset identified as the minority class and the sample subset identified as the majority class, it is evaluated whether the sample numbers between the adjusted sample subsets reach balance. The sample subsets that reach balance and the sample subsets that do not need to be adjusted are integrated to form the third historical operation data set;
[0088] It should be noted that if the difference between the number of samples in the adjusted minority class sample subset and the number of samples in the sample subset of the class that does not need to be adjusted is within the pre-set reasonable error range, and the difference between the number of samples in the adjusted majority class sample subset and the number of samples in the sample subset of the class that does not need to be adjusted is also within the pre-set reasonable error range, then it is considered that the number of samples in each sample subset has reached balance.
[0089] S5. Perform sampling with replacement on the third historical operation data set to obtain a number of training sets and test sets;
[0090] Specifically, perform sampling with replacement (Bootstrap) on the third historical operation data set to obtain a number of training sets and test sets; it should be noted that both the training sets and the test sets are mixed with "fault" historical operation data, "normal" historical operation data, and "potential fault" historical operation data.
[0091] S6. According to the training sets and the test sets, train a preset decision tree to obtain a trained decision tree, and integrate the trained decision trees to obtain a random forest model;
[0092] In a preferred embodiment, the training of the preset decision tree according to the training sets and the test sets to obtain a trained decision tree includes:
[0093] For each training set, input the training set into a preset untrained decision tree, so that the untrained decision tree performs iterative training according to the historical operation data and the corresponding operation status labels in the training set until the preset convergence condition is reached, and stop the iterative training to obtain a preliminarily trained decision tree;
[0094] For each preliminarily trained decision tree, use the test set as the input of the decision tree, and determine the evaluation accuracy of the decision tree according to the evaluation result output by the decision tree;
[0095] According to the evaluation accuracies of the preliminarily trained decision trees, assign corresponding weights to the preliminarily trained decision trees to obtain each trained decision tree;
[0096] Schematically, for each training set, use it as the training sample of a preset untrained decision tree;
[0097] Specifically, the training set is input into the corresponding untrained decision tree. During the training process, the untrained decision tree calculates the Gini index of each feature at the splitting step of each node and selects the feature that can maximize the data purity (i.e., minimize the Gini index) for splitting. The above splitting operation is continuously performed until the decision tree reaches a predetermined depth. The calculation formula of the Gini index Gini(x) is as follows:
[0098]
[0099] In the formula: P j is the probability that the sample X contains the feature attribute j. j refers to the feature attribute used to divide the sample at each node of the decision tree, which refers to the temperature signal, vibration signal, pressure signal, and voltage signal in this embodiment. m is the upper limit value of the feature attribute;
[0100] It should be noted that the preset untrained decision tree is constructed based on recursive splitting;
[0101] Specifically, after obtaining each preliminarily trained decision tree, it is necessary to calculate the out-of-bag error OOB_error (Out-of-Bag Error) of each preliminarily trained decision tree to evaluate the performance of each preliminarily trained decision tree. The calculation formula is as follows:
[0102]
[0103] In the formula: represents the label result obtained by predicting the sample i using all decision trees that have not used the sample i during the training process. N represents the total number of samples in the test set, and I is the Indicator Function, which is used to judge whether the condition holds.
[0104] Schematically, for each preliminarily trained decision tree, the test set is used as the input of each preliminarily trained decision tree, and according to the evaluation results output by each preliminarily trained decision tree, its corresponding evaluation accuracy is determined;
[0105] Specifically, for each preliminarily trained decision tree, evaluate its performance in the classification task, and then assign different weights according to the classification ability of each tree to ensure the effect of each tree. The specific calculation formula of the weight is as follows:
[0106]
[0107] In the formula: X′ is the total number of test samples in the test set, X′ correct,i is the number of samples correctly classified by the hth decision tree, ω his the weight of the h-th decision tree, and n is the total number of decision trees;
[0108] Meanwhile, calculate the Gini importance of each feature, evaluate its contribution, and optimize the feature selection of the decision tree:
[0109]
[0110] In the formula: Gini_importance(j) represents the Gini importance of feature attribute j; ΔGini(h,j) represents the change in the Gini index caused by feature attribute j in the h-th decision tree;
[0111] Finally, integrate the trained decision trees into the random forest model.
[0112] S7. Evaluate the operating status of the electromechanical equipment according to the random forest model;
[0113] In a preferred embodiment, the evaluating the operating status of the electromechanical equipment according to the random forest model includes:
[0114] Obtain the real-time operating data of the electromechanical equipment;
[0115] Input the real-time operating data into the random forest model, so that each trained decision tree in the random forest model evaluates the real-time operating data respectively, obtains the evaluation results output by each trained decision tree, uses the weights of each trained decision tree to perform weighted analysis on the evaluation results output by each trained decision tree, obtains a weighted analysis result, and determines the operating status of the real-time operating data according to the weighted analysis result;
[0116] Schematically, integrate the trained decision trees into the random forest model. Each trained decision tree will perform classification prediction on the real-time input real-time operating data, and each decision tree will output the evaluation result corresponding to the real-time operating data. Among them, this evaluation result represents the most likely operating status of the electromechanical equipment corresponding to the real-time operating data. Then, the random forest model performs weighted voting based on the outputs of each decision tree and combines the weights of each decision tree. The decision tree with a larger weight will have a greater impact on the final result. Thus, according to the weighted voting result, obtain the operating status (normal, abnormal, or potential fault) of the electromechanical equipment output by the random forest model. The output expression of the random forest model is specifically as follows:
[0117]
[0118] In the formula, I represents the indicator function, and I(Y h =c) represents the evaluation result Y of the h-th decision treeh Whether it is the c-th operating state. If so, it is 1; if not, it is 0; c L Indicates the total number of categories of the operating states of the electromechanical equipment; n is the total number of decision trees; P(c|x t ) represents the probability value that the electromechanical equipment corresponding to the real-time operation data x t is in the c-th operating state; ω h is the weight of the h-th decision tree; Represents the operating state output by the random forest model;
[0119] It should be noted that in this embodiment, there are three categories of operating states. Therefore, L in C L is 3. The operating state "normal" corresponds to c1, the operating state "abnormal" corresponds to c2, and the operating state "potential fault" corresponds to c3.
[0120] See Figure 2 and Figure 3 , Figure 2 is a schematic diagram of the prediction accuracy of the traditional random forest for predicting the operating state of the electromechanical equipment based on the real-time operation data, Figure 3 is a schematic diagram of the prediction evaluation accuracy of the random forest model in this embodiment for predicting the operating state of the electromechanical equipment based on the real-time operation data; it can be seen that Figure 3 the coincidence degree of each prediction point in Figure 2 with the actual operating state is higher than that in
[0121] See Figure 4 , in this embodiment, by integrating multiple decision trees into the random forest model and performing weighted voting based on the evaluation results output by each decision tree and in combination with the weights of each decision tree, compared with the traditional random forest, this method greatly improves the accuracy when predicting the operating state and has significant advantages:
[0122] See Figure 5, in this embodiment, through data balancing processing, after the obtained initial historical operation dataset undergoes data balancing processing, the problem of unbalanced data categories is solved. As a result, during the subsequent model training process, the model can learn more comprehensive feature information, effectively avoiding the model being biased towards majority-class data. And during the decision tree training stage, by sampling the dataset with replacement, that is, Bootstrap resampling, the construction and training of the decision tree are carried out based on the sampled training set and test set. When training the decision tree, different decision trees can learn different feature combinations and distribution situations in the dataset, increasing the diversity among the decision trees; the capture capabilities of different decision trees for various types of data are different. The traditional random forest model treats all decision trees equally, while weighted voting can highlight the role of the decision tree that is more accurate in identifying specific data categories; if a certain decision tree is extremely sensitive to the temperature signal characteristics in the abnormal state and is given a higher weight, the random forest model will be more accurate when judging the abnormal state. This method can effectively improve the prediction accuracy of the operation state, more accurately evaluate the operation state of the electromechanical equipment, and by setting the alarm trigger, when the final evaluation result is "abnormal" or "potential fault", the system can promptly send an alarm signal so that the operation and maintenance personnel can quickly take measures to reduce the losses caused by equipment failures.
[0123] In a preferred embodiment, the method for evaluating the operation state of the electromechanical equipment further includes:
[0124] When determining that the operation state of the real-time operation data is abnormal or a potential fault, trigger an alarm;
[0125] Specifically, when determining that the operation state of the real-time operation data is abnormal or a potential fault, immediately trigger an alarm. The power system connected to the electromechanical equipment will quickly call the pre-set alarm module to send a warning signal in various ways such as sound and light. At the same time, the alarm information will be synchronously transmitted to the remote monitoring terminal so that the operation and maintenance personnel can promptly determine the operation state of the electromechanical equipment.
[0126] Refer to Figure 6 , which is an apparatus for evaluating the operation state of an electromechanical equipment provided by an embodiment of the present invention, including: an initial historical operation dataset acquisition module, a first historical operation dataset acquisition module, a second historical operation dataset acquisition module, a third historical operation dataset acquisition module, a training set and test set acquisition module, a model training module, and an operation state evaluation module;
[0127] The initial historical operation dataset acquisition module is used to acquire the initial historical operation dataset; wherein, the initial historical operation dataset includes the historical operation data of several electromechanical equipments;
[0128] The first historical operation dataset acquisition module is configured to assign corresponding operation status tags to each piece of historical operation data in the initial historical operation dataset to obtain a first historical operation dataset;
[0129] The second historical operation dataset acquisition module is configured to classify each piece of historical operation data in the first historical operation dataset based on the operation status tags to obtain a second historical operation dataset; wherein, the second historical operation dataset includes several sample subsets for characterizing different operation statuses;
[0130] The third historical operation dataset acquisition module is configured to balance the number of samples in each sample subset in the second historical operation dataset to obtain a third historical operation dataset;
[0131] The training set and test set acquisition module is configured to perform sampling with replacement on the third historical operation dataset to obtain several training sets and test sets;
[0132] The model training module is configured to train a preset decision tree according to the training set and the test set to obtain a trained decision tree, and integrate the trained decision trees to obtain a random forest model;
[0133] The operation status evaluation module is configured to evaluate the operation status of the electromechanical device according to the random forest model.
[0134] See Figure 7 , an embodiment of the present application further provides a terminal device, including:
[0135] One or more processors;
[0136] A memory coupled to the processor for storing one or more programs;
[0137] When the one or more programs are executed by the one or more processors, the one or more processors implement the electromechanical device operation status evaluation method as described above.
[0138] The processor is used to control the overall operation of the terminal device to complete all or part of the steps of the above-mentioned electromechanical device operation status evaluation method. The memory is used to store various types of data to support the operation of the terminal device. Such data may include, for example, instructions for any application or method operating on the terminal device, as well as application-related data. The memory can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk.
[0139] In an exemplary embodiment, the terminal device can be implemented by one or more application specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors or other electronic components, and is used to execute the electromechanical device operation status evaluation method described in any of the above embodiments, and achieve the same technical effects as the above method.
[0140] In another exemplary embodiment, there is also provided a computer-readable storage medium including a computer program. When the computer program is executed by a processor, it implements the steps of the electromechanical device operation status evaluation method described in any of the above embodiments. For example, the computer-readable storage medium can be the above-mentioned memory including the computer program, and the above computer program can be executed by the processor of the terminal device to complete the electromechanical device operation status evaluation method described in any of the above embodiments, and achieve the same technical effects as the above method.
[0141] The above is the preferred embodiment of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the principle of the present invention, several improvements and modifications can be made, and these improvements and modifications are also regarded as the protection scope of the present invention.
Claims
1. A method for evaluating the operating state of a mechanical and electrical device, characterized in that including: Obtain an initial historical operation dataset; wherein, the initial historical operation dataset includes historical operation data of a number of electromechanical devices; Assign corresponding operation status labels to each piece of historical operation data in the initial historical operation dataset to obtain a first historical operation dataset; Classify each piece of historical operation data in the first historical operation dataset based on the operation status label to obtain a second historical operation dataset; wherein, the second historical operation dataset includes a number of sample subsets for characterizing different operation states; Balance the number of samples in each sample subset in the second historical operation dataset to obtain a third historical operation dataset; Perform sampling with replacement on the third historical operation dataset to obtain a number of training sets and test sets; Train a preset decision tree according to the training set and the test set to obtain a trained decision tree, and integrate the trained decision trees to obtain a random forest model; Evaluate the operation state of the electromechanical device according to the random forest model.
2. The method for evaluating the operating state of the electromechanical equipment according to claim 1, wherein, The categories of the operation status labels include normal, abnormal, and potential fault; The step of classifying each piece of historical operation data in the first historical operation dataset based on the operation status label to obtain a second historical operation dataset includes: Classify each piece of historical operation data according to the operation status label of each piece of historical operation data in the first historical operation dataset to obtain a number of sample subsets; wherein, the sample subsets include a normal sample subset, an abnormal sample subset, and a potential fault sample subset.
3. The operation state evaluation method of the electromechanical equipment according to claim 2, characterized in that The step of balancing the number of samples in each sample subset in the second historical operation dataset to obtain a third historical operation dataset includes: Identify the number of samples in the normal sample subset, the abnormal sample subset, and the potential fault sample subset respectively to obtain a sample subset identified as a minority class, a sample subset identified as a majority class, and a sample subset identified as a class that does not need to be adjusted; For the sample subset identified as a minority class, randomly generate a number of synthetic samples according to the distances between the samples in the sample subset to obtain an expanded sample subset; For the sample subset identified as a majority class, randomly delete a number of samples in the sample subset to obtain a reduced sample subset; Integrate the expanded sample subset, the reduced sample subset, and the sample subset identified as a class that does not need to be adjusted into the third historical operation dataset.
4. The method for evaluating the operating state of the electromechanical equipment according to claim 3, characterized in that, The step of training a preset decision tree according to the training set and the test set to obtain a trained decision tree includes: For each training set, input the training set into a preset untrained decision tree, so that the untrained decision tree performs iterative training according to the historical operation data and the corresponding operation status label in the training set until a preset convergence condition is reached, and stop the iterative training to obtain a preliminarily trained decision tree; For each preliminarily trained decision tree, use the test set as the input of the decision tree, and determine the evaluation accuracy of the decision tree according to the evaluation result output by the decision tree; According to the evaluation accuracy of each decision tree completed by the preliminary training, assign corresponding weights to each decision tree completed by the preliminary training to obtain each trained decision tree.
5. The method for evaluating the operating state of an electromechanical device according to claim 4, wherein, Evaluating the operating status of the electromechanical equipment according to the random forest model includes: Obtain the real-time operating data of the electromechanical equipment; Input the real-time operating data into the random forest model, so that each trained decision tree in the random forest model evaluates the real-time operating data respectively, obtain the evaluation results output by each trained decision tree, use the weights of each trained decision tree to perform weighted analysis on the evaluation results output by each trained decision tree, obtain a weighted analysis result, and determine the operating status of the real-time operating data according to the weighted analysis result.
6. The method for evaluating the operating state of the electromechanical equipment according to claim 5, wherein The electromechanical equipment includes a compressor and a DAC device; the initial historical operating data set includes the historical vibration signal of the compressor, the historical temperature signal of the compressor, the historical pressure signal of the compressor, and the historical voltage signal of the DAC device.
7. The method for evaluating the operating state of the electromechanical equipment according to claim 6, characterized in that It also includes: When determining that the operating status of the real-time operating data is abnormal or a potential fault, trigger an alarm.
8. An operating state evaluation device for a mechanical and electrical equipment, characterized in that, It includes: an initial historical operating data set acquisition module, a first historical operating data set acquisition module, a second historical operating data set acquisition module, a third historical operating data set acquisition module, a training set and test set acquisition module, a model training module, and an operating status evaluation module; The initial historical operating data set acquisition module is used to acquire an initial historical operating data set; wherein, the initial historical operating data set includes the historical operating data of several electromechanical devices; The first historical operating data set acquisition module is used to assign corresponding operating status labels to each historical operating data in the initial historical operating data set to obtain a first historical operating data set; The second historical operating data set acquisition module is used to classify each historical operating data in the first historical operating data set based on the operating status label to obtain a second historical operating data set; wherein, the second historical operating data set includes several sample subsets for representing different operating statuses; The third historical operating data set acquisition module is used to balance the number of samples in each sample subset in the second historical operating data set to obtain a third historical operating data set; The training set and test set acquisition module is used to perform sampling with replacement on the third historical operating data set to obtain several training sets and test sets; The model training module is used to train a preset decision tree according to the training set and the test set to obtain a trained decision tree, and integrate the trained decision trees to obtain a random forest model; The operating status evaluation module is used to evaluate the operating status of the electromechanical equipment according to the random forest model.
9. A terminal device, characterized in that, It includes: One or more processors; A memory, coupled to the processor, for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the electromechanical device operating state evaluation method according to any one of claims 1-7.
10. A storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the electromechanical device operating state evaluation method according to any one of claims 1-7.