Method and device for monitoring state of electric heating unit

Abstract

The invention provides an electric heating unit state monitoring method and device, and relates to the technical field of electric heating unit state monitoring. The method comprises the following steps: performing feature extraction on operation data based on a neural network model obtained by pre-training through a whale optimization algorithm to obtain a first data feature; performing feature dimension reduction on the first data feature by adopting an auto-encoder to obtain a second data feature; and inputting the second data features into a pre-trained adaptive threshold pruning random forest model for feature classification to obtain state type information. The key features can be quickly and accurately identified in a large amount of complex electric heating unit operation data, the complexity of data processing is reduced, and the processing speed is increased. State type classification is carried out on the obtained features through the random forest, the classification accuracy and response speed are improved, and the method is suitable for the state monitoring scene of the electric heating unit. In addition, the robustness is high, and the data in the noise environment can be effectively processed.

Description

Technical Field

[0001] The present disclosure relates to the technical field of state monitoring of electric heating units, and in particular to a state monitoring method and device for an electric heating unit. Background Art

[0002] In the field of condition monitoring of electric thermal units, the accuracy of monitoring is closely related to the reliability and safe operation of related systems. Traditional monitoring methods mostly rely on direct reading of sensor data and threshold judgment. This method has certain limitations when dealing with complex systems. For example, in the scenario where an electric thermal unit is coupled with a new energy storage system, the electric thermal unit needs to adjust its output according to the state of the energy storage system to provide additional support for the energy storage system. In this scenario, the data is relatively complex, and the existing condition monitoring methods for electric thermal units are still insufficient in terms of data diversity, model generalization ability, and real-time processing capabilities. The computational burden is heavy when processing nonlinear high-dimensional data, and the robustness to noise and outliers is insufficient. Summary of the invention

[0003] The present disclosure aims to solve one of the technical problems in the related art at least to some extent.

[0004] To this end, the first embodiment of the present disclosure proposes a state monitoring method for an electric heating unit, comprising the following steps:

[0005] Obtaining the operating data of the electric heating unit to be tested;

[0006] Extracting features from the operating data based on a neural network model to obtain a first data feature; the neural network model is pre-trained using a whale optimization algorithm;

[0007] Using an autoencoder to perform feature dimension reduction on the first data feature to obtain a second data feature;

[0008] The second data feature is input into a pre-trained random forest model with adaptive threshold pruning to perform feature classification, so as to obtain the state type information of the electric heating unit to be tested.

[0009] In some embodiments of the present disclosure, the neural network model is pre-trained through the following steps: obtaining the operating training data of the electric heating unit and the data feature labels corresponding to the operating training data; inputting the operating training data into the initial neural network model for feature extraction to obtain a third data feature output by the initial neural network model; using the whale optimization algorithm to adjust the learning rate, weight and number of nodes of the hidden layer of the neural network of the initial neural network model based on the error value between the third data feature and the data feature label to obtain the neural network model.

[0010] In some embodiments of the present disclosure, the obtaining of operation training data of the electric thermal generator includes: obtaining initial operation training data of the electric thermal generator; and performing data expansion on the initial operation training data based on a generative adversarial network based on lag decomposition to obtain the operation training data.

[0011] In some embodiments of the present disclosure, the random forest model is pre-trained through the following steps: obtaining multiple training data of the random forest model, each of the training data includes a data feature sample of the electric heating unit and a state type information label corresponding to the data feature sample; performing self-service sampling in the multiple data feature samples to obtain a sample set of each decision tree in the random forest model; applying a dynamic feature selection strategy to perform feature selection in the sample set of each decision tree to obtain a feature subset used for training of each decision tree; using the sample set and the feature subset to train the decision tree, and during the training process, pruning decisions are made on the decision tree according to the pruning threshold of the random forest model, and the pruning threshold is adjusted according to the depth of the decision tree and the error rate of the random forest model; the error rate is determined according to the state type information label and the predicted state type information output by the random forest model.

[0012] The second aspect of the present disclosure provides a state monitoring device for an electric heating unit, comprising:

[0013] An acquisition module is used to acquire the operating data of the electric heating unit to be tested;

[0014] A feature extraction module, used for extracting features from the operating data based on a neural network model to obtain a first data feature; the neural network model is pre-trained by a whale optimization algorithm;

[0015] A feature processing module, configured to perform feature dimension reduction on the first data feature by using an autoencoder to obtain a second data feature;

[0016] A classification module is used to input the second data feature into a pre-trained random forest model with adaptive threshold pruning to perform feature classification, so as to obtain the state type information of the electric heating unit to be tested.

[0017] In some embodiments of the present disclosure, the device also includes a first training module; wherein the first training module is used to: obtain operation training data of the electric heating unit and data feature labels corresponding to the operation training data; input the operation training data into the initial neural network model for feature extraction to obtain a third data feature output by the initial neural network model; use a whale optimization algorithm to adjust the learning rate, weight and number of nodes of the hidden layer of the neural network of the initial neural network model based on the error value between the third data feature and the data feature label to obtain the neural network model.

[0018] In some embodiments of the present disclosure, the first training module is specifically used to: obtain initial operation training data of the electric heating unit; perform data expansion on the initial operation training data based on a generative adversarial network with lag decomposition to obtain the operation training data.

[0019] In some embodiments of the present disclosure, the device also includes a second training module; wherein the second training module is used to: obtain multiple training data of the random forest model, each of the training data includes a data feature sample of the electric heating unit and a state type information label corresponding to the data feature sample; perform self-service sampling in the multiple data feature samples to obtain a sample set of each decision tree in the random forest model; apply a dynamic feature selection strategy to perform feature selection in the sample set of each decision tree to obtain a feature subset used by each decision tree to participate in training; use the sample set and the feature subset to train the decision tree, and during the training process, make pruning decisions on the decision tree according to the pruning threshold of the random forest model, and adjust the pruning threshold according to the depth of the decision tree and the error rate of the random forest model; the error rate is determined according to the state type information label and the predicted state type information output by the random forest model.

[0020] A third aspect of the present disclosure provides an electronic device, comprising: a processor, and a memory communicatively connected to the processor;

[0021] The memory stores computer-executable instructions;

[0022] The processor executes the computer-executable instructions stored in the memory to implement the method described in the first aspect.

[0023] The fourth aspect of the present disclosure provides a computer-readable storage medium, characterized in that the computer-readable storage medium stores computer execution instructions, and when the computer execution instructions are executed by a processor, they are used to implement the method described in the first aspect.

[0024] The state monitoring method of the electric thermal unit provided by the present disclosure uses a neural network model obtained through training with a whale optimization algorithm to perform feature extraction on the electric thermal unit to be tested, and can quickly and accurately identify key features in a large amount of complex electric thermal unit operation data. The complexity of data processing can be reduced and the processing speed can be improved through dimensionality reduction processing. The state type classification of the obtained features is performed through a random forest with adaptive threshold pruning, which improves the accuracy and response speed of classification. The present disclosure is suitable for the state monitoring scenario of electric thermal units, which can effectively improve the monitoring performance, reduce the demand for computing resources, and enhance the model's adaptability to new and unseen data, thereby improving the overall efficiency and economic benefits of the monitoring system. In addition, the present disclosure has high robustness and can effectively process data under various operating conditions, including stable operation in a noisy environment.

[0025] Additional aspects and advantages of the present disclosure will be given in part in the following description and in part will be obvious from the following description or learned through practice of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] The above and / or additional aspects and advantages of the present disclosure will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0027] Figure 1 A schematic flow chart of a method for monitoring the state of an electric heating unit provided in an embodiment of the present disclosure;

[0028] Figure 2 A flowchart of a training method for a neural network model provided by an embodiment of the present disclosure;

[0029] Figure 3 A schematic diagram of a state monitoring device for an electric heating unit provided in an embodiment of the present disclosure. DETAILED DESCRIPTION

[0030] Embodiments of the present disclosure are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present disclosure, and should not be construed as limiting the present disclosure.

[0031] Specifically, the state monitoring method and device of the electric heating unit according to the embodiment of the present disclosure are described below with reference to the accompanying drawings.

[0032] Figure 1 The following is a flow chart of a method for monitoring the state of an electric heating unit provided by an embodiment of the present disclosure. Figure 1 As shown, the state monitoring method of the electric heating unit may include the following steps:

[0033] Step 101, obtaining the operating data of the electric heating unit to be tested.

[0034] In some embodiments of the present disclosure, the storage format of the operating data of the electric heating unit to be tested may adopt a structured JSON format, and the operating data may include one or more of the following:

[0035] R a1 : Temperature, indicating the real-time temperature of the unit;

[0036] R a2 : Pressure, indicating the working pressure of the unit;

[0037] R a3 : Current, indicating the real-time current of the unit;

[0038] R a4 : Voltage, indicating the working voltage of the unit;

[0039] R a5 : Vibration frequency, records the vibration condition of the unit during operation;

[0040] R a6 : Energy storage capacity, showing the current power status of the energy storage unit;

[0041] R a7 : Energy consumption, the energy consumption of a computer group in a specific time;

[0042] R a8 : Fault code, recording any fault type that occurs;

[0043] R a9 : Running time, recording the running time of the unit from startup to the current time;

[0044] R a10 :Maintenance records, recording the historical maintenance data of the unit.

[0045] Step 102: extracting features from the running data based on the neural network model to obtain a first data feature. The neural network model is pre-trained using the whale optimization algorithm.

[0046] In some embodiments of the present disclosure, a 6-layer fully connected neural network model can be used to extract features from operating data. In order to avoid the problems of gradient vanishing, gradient exploding or falling into local optimal solutions that may be encountered when using neural networks to a certain extent, affecting the stability of training and the performance of the model, the embodiments of the present disclosure use a whale optimization algorithm to optimize the parameters of the neural network model, train the neural network model, and use the self-organization and adaptability of individual whales to find the optimal solution by simulating the hunting behavior of whale groups. A self-regulating mechanism is used in the training process, allowing the network to dynamically adjust its weight update strategy based on the error of the previous training cycle, and using the sensitivity in chaos theory to enhance the model's responsiveness to initial conditions and parameter changes. The use of deep learning technology for feature extraction can automatically identify the features most relevant to the device status and improve the efficiency and accuracy of data processing.

[0047] In one implementation, the operation training data of the electric heating unit and the data feature labels corresponding to the operation training data can be obtained. The data feature labels are, for example, "normal", "warning", "fault", or "normal", "abnormal", etc. The weights and biases of the initial neural network model are initialized, which can be expressed as follows:

[0048] W p,0 =μ p +σ p N(0,I)

[0049] b p,0 =v p N(0,I)

[0050] Where W p,0 is the initial value of the weight of the neural network model, b p,0 is the initial value of the bias of the neural network model, μ p is the mean shift of the weights of the neural network model, ν p is the mean shift of the weight bias of the neural network model, σ p is the standard deviation of the weight initialization of the neural network model, N(0,I) means the mean is 0, the standard deviation is the normal distribution of the identity matrix, and I is the identity matrix. Optionally, μ p Set to 0.01, ν p Set to 0.05, σ p Set to 0.001.

[0051] The running training data is input into the initial neural network model for feature extraction to obtain the third data feature output by the initial neural network model. The learning rate, weight and number of nodes in the hidden layer of the initial neural network model are adjusted based on the error value between the third data feature and the data feature label using the whale optimization algorithm to obtain the neural network model.

[0052] In the process of training the neural network model using the whale optimization algorithm, the output error of the neural network model is evaluated, and the learning rate is adjusted through the chaotic feedback mechanism to avoid overfitting or underfitting. The dynamic update method of the self-adjusting learning rate is expressed as:

[0053] η p,t =η p,t-1 ·exp(-γ p ΔE p,t-1 )

[0054] Where η p,t is the learning rate of the neural network model at the tth iteration, η p,t-1 is the learning rate of the neural network model at the t-1th iteration, γ p It is the adjustment factor of the neural network model, controlling the sensitivity of the learning rate change, ΔE p,t-1 It is the error change of the neural network model in the previous cycle, that is, the change between the neural network model error of the t-1th iteration and the t-2th iteration.

[0055] Furthermore, the weights are updated according to the updated learning rate, which is expressed as:

[0056]

[0057] Where W p,t is the weight of the neural network model at the tth iteration, W p,t-1 is the weight of the neural network model at the t-1th iteration, E p is the error of the neural network model, which is calculated by the preset Softmax function on the output features of the last layer of the neural network. is the symbol of partial derivative.

[0058] During the training process, a regularization strategy is used to prevent overfitting and ensure the generalization ability of the model. Specifically, regularization and sparseness are performed by dynamically adjusting the number of nodes in the hidden layer of the neural network, which can be expressed as:

[0059] N p,t =N p,t-1 +sgn(ΔE p,t-1 )·β p

[0060] Where N p,t is the number of nodes in the hidden layer of the neural network at the tth iteration, N p,t-1 is the number of nodes in the hidden layer of the neural network at the t-1th iteration, β p is the node adjustment step size of the hidden layer of the neural network, sgn() is the sign function of the error change, which is used to decide whether to increase or decrease the number of nodes, that is, for sgn(ΔEp,t-1 ), when ΔE p,t-1 When sgn(ΔE p,t-1 ) is 1, when ΔE p,t-1 When sgn(ΔE p,t-1 ) is -1.

[0061] In one embodiment, the node adjustment step size of the hidden layer of the neural network depends on a function of the network level, and the calculation method is expressed as:

[0062]

[0063] Among them, Int cs () is the rounding operation, α p is the influence coefficient of adjusting the adjustment step size, d p is the depth of the current network layer. Optionally, d p The value is 6, α p Set to 2.5.

[0064] Repeat the above steps to adjust the parameters of the neural network model until the preset stop iteration condition is met, which means that the model training is completed. In one embodiment, the preset stop iteration condition is to reach a preset maximum number of iterations. Optionally, the preset maximum number of iterations can be set to 1000 times. For the training process of the neural network model, please refer to Figure 2 Schematic diagram.

[0065] In some embodiments of the present disclosure, in order to make up for the deficiency of original training data and avoid monitoring errors caused by insufficient data, the initial operation training data of the electric thermal unit can be expanded based on a generative adversarial network with lag decomposition to obtain operation training data and enhance the effect of model training.

[0066] Step 103: Use an autoencoder to perform feature dimension reduction on the first data feature to obtain a second data feature.

[0067] The first data feature is reduced in dimension, which optimizes the efficiency of data processing and the operation speed of the model. The neural network model trained by the whale optimization algorithm and the feature dimension reduction of the data can quickly and accurately extract the most representative features from a large amount of monitoring data.

[0068] Step 104: input the second data feature into a pre-trained random forest model with adaptive threshold pruning to perform feature classification, and obtain the state type information of the electric heating unit to be tested.

[0069] Random forest generation uses traditional self-service aggregation methods to build multiple decision trees, while the adaptive threshold pruning mechanism applies a dynamically adjusted threshold based on model complexity and error rate to each decision tree, thereby optimizing the structure and performance of the entire forest. The present invention adopts a model sparsification mechanism to automatically adjust the pruning threshold according to the performance of the decision tree to cope with different data distributions and noise levels.

[0070] In some embodiments of the present disclosure, the parameters of the random forest model may be initialized:

[0071]

[0072] Among them, T u represents the number of trees in the forest, D u Indicates the maximum depth of the tree, S min,u represents the minimum number of samples of a node, λ u,0 is the initial threshold for pruning, γ u Adjust the amplitude for the threshold.

[0073] Obtain multiple training data of the random forest model. Each training data includes a data feature sample of the electric heating unit (the total data feature sample of each training data includes multiple features) and a state type information label corresponding to the data feature sample. Perform self-service sampling in multiple data feature samples to obtain a sample set for each decision tree in the random forest model. Among them, the self-service sampling is expressed as:

[0074]

[0075] In the formula, represents the sample set used by the kth tree, Boot() is the self-service sampling function, X u is the data feature sample, and n is the number of samples.

[0076] The dynamic feature selection strategy is applied to perform feature selection in the sample set of each decision tree to obtain the feature subset used by each decision tree in training. Dynamic feature selection is expressed as:

[0077]

[0078] in, represents the feature subset selected for the kth tree, X u is a data feature sample, Y u is the data feature sample X u The corresponding state type information label, α u is the feature importance threshold.

[0079] Furthermore, the feature selection function Sele() is expressed as:

[0080] Sele(X u ,Y u ,α u )={f|f∈X u ,I(f;Y u )>α u}

[0081] Among them, I(f;Y u ) represents the feature f and the target variable Y u The mutual information between them, which reflects the correlation between the features and the targets, α u is a threshold, and the feature will be selected only when the mutual information between the feature and the target is greater than this threshold.

[0082] The decision tree is trained using the selected sample set and feature subset. During the training process, the decision tree is pruned according to the pruning threshold of the random forest model. The generation and pruning of the decision tree are expressed as:

[0083]

[0084] in, represents the kth tree, Grow() is the function of decision tree growth and pruning, λ u is the pruning threshold.

[0085] The function GrowTree can be expanded as follows:

[0086]

[0087] Among them, sp() represents the data segmentation function based on a certain selection criterion (such as information gain, Gini impurity), maf() is a function to create leaf nodes, which will be called when the error of the current branch is lower than the threshold λ, and Err(X) calculates the data error of the current node and decides whether to continue segmentation.

[0088] The pruning threshold is adjusted according to the depth of the decision tree and the error rate of the random forest model. The error rate is determined based on the state type information label and the predicted state type information output by the random forest model, which can be expressed as follows:

[0089]

[0090] in, is the error rate of the random forest model, N is the number of data feature samples, y n is the state type information label corresponding to the nth data feature sample, is the feature vector of the nth data feature sample, f u Use the current weight The model prediction function.

[0091] The threshold adjustment strategy setting can be expressed as:

[0092]

[0093] Among them, λ u,i+1 is the threshold for the i+1th iteration, λ u,i is the threshold of the i-th iteration, γ u is the threshold adjustment amplitude, is the error of the current model, which depends on the data characteristics and model weights

[0094] After completing the training of all decision trees, the second data feature of the electric heating unit to be tested is input into the pre-trained random forest model with adaptive threshold pruning for feature classification. In one example, the state type information may include "normal", "warning", and "fault". The outputs of each tree are integrated through an ensemble learning method, and the state type information of the electric heating unit to be tested is determined by majority voting or average probability method. The entire forest is finally optimized and adjusted to further remove trees with poor performance, optimize redundancy and generalization ability, and the forest integration is expressed as:

[0095]

[0096] in, The state type information predicted by the random forest model, is the kth tree, is the second data feature, and Pre() is the prediction function.

[0097] By implementing the embodiments of the present disclosure, the neural network model obtained by training with the whale optimization algorithm can be used to extract features from the electric thermal unit to be tested, and key features can be quickly and accurately identified in a large amount of complex electric thermal unit operation data. The dimensionality reduction process can reduce the complexity of data processing and improve the processing speed. The state type classification of the obtained features is performed through the random forest with adaptive threshold pruning, which improves the accuracy and response speed of classification. The present disclosure is applicable to the state monitoring scenario of electric thermal units, which can effectively improve the monitoring performance, reduce the demand for computing resources, and enhance the model's adaptability to new and unseen data, thereby improving the overall efficiency and economic benefits of the monitoring system. In addition, the present disclosure has high robustness and can effectively process data under various operating conditions, including stable operation in a noisy environment.

[0098] It should be noted that the disclosed embodiments are applicable to scenarios where electric thermal units are coupled with new energy storage systems. By real-time monitoring of equipment status, equipment failures can be effectively prevented to ensure the stability and safety of system operation. In addition, equipment maintenance efficiency can be significantly improved, and maintenance and repair time can be optimized through accurate status prediction, thereby reducing operating costs.

[0099] Figure 3 Schematic diagram of a state monitoring device for an electric heating unit provided by an embodiment of the present disclosure. Figure 3 As shown, the device includes: an acquisition module 301, a feature extraction module 302, a feature processing module 303 and a classification module 304.

[0100] The acquisition module 301 is used to acquire the operating data of the electric heating unit to be tested.

[0101] The feature extraction module 302 is used to extract features from the operating data based on the neural network model to obtain the first data feature. The neural network model is pre-trained by the whale optimization algorithm.

[0102] The feature processing module 303 is used to use an autoencoder to perform feature dimension reduction on the first data feature to obtain a second data feature.

[0103] The classification module 304 is used to input the second data feature into a pre-trained random forest model with adaptive threshold pruning to perform feature classification, so as to obtain the state type information of the electric heating unit to be tested.

[0104] In some embodiments of the present disclosure, Figure 3 On the basis of the embodiment shown, the state monitoring device of the electric heating unit further includes a first training module. The first training module is used to: obtain the operation training data of the electric heating unit and the data feature labels corresponding to the operation training data; input the operation training data into the initial neural network model for feature extraction to obtain the third data feature output by the initial neural network model; and use the whale optimization algorithm to adjust the learning rate, weight and number of nodes of the hidden layer of the initial neural network model based on the error value between the third data feature and the data feature label to obtain the neural network model.

[0105] In some embodiments of the present disclosure, the first training module is specifically used to: obtain initial operation training data of the electric heating unit; perform data expansion on the initial operation training data based on a generative adversarial network based on lag decomposition to obtain operation training data.

[0106] In some embodiments of the present disclosure, Figure 3On the basis of the illustrated embodiment, the state monitoring device of the electric heating unit also includes a second training module. The second training module is used to: obtain multiple training data of the random forest model, each training data includes a data feature sample of the electric heating unit and a state type information label corresponding to the data feature sample; perform self-service sampling in multiple data feature samples to obtain a sample set of each decision tree in the random forest model; apply a dynamic feature selection strategy to perform feature selection in the sample set of each decision tree to obtain a feature subset used for each decision tree to participate in training; use the sample set and feature subset to train the decision tree, perform pruning decisions on the decision tree according to the pruning threshold of the random forest model during the training process, and adjust the pruning threshold according to the depth of the decision tree and the error rate of the random forest model; the error rate is determined according to the state type information label and the predicted state type information output by the random forest model.

[0107] Regarding the device in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be elaborated here.

[0108] In order to implement the above embodiments, the present disclosure also proposes an electronic device, including: a processor, and a memory communicatively connected to the processor; the memory stores computer-executable instructions; the processor executes the computer-executable instructions stored in the memory to implement the method provided by the above embodiments.

[0109] In order to implement the above embodiments, the present disclosure further proposes a computer-readable storage medium, in which computer-executable instructions are stored. When the computer-executable instructions are executed by a processor, they are used to implement the methods provided by the above embodiments.

[0110] In order to implement the above embodiments, the present disclosure also proposes a computer program product, including a computer program, which implements the methods provided by the above embodiments when executed by a processor.

[0111] In the description of the aforementioned embodiments, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present disclosure. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, unless they are contradictory.

[0112] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of the features. In the description of the present disclosure, "plurality" means at least two, such as two, three, etc., unless otherwise clearly and specifically defined.

[0113] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, segment or portion of code that includes one or more executable instructions for implementing the steps of a custom logical function or process, and the scope of the preferred embodiments of the present disclosure includes additional implementations in which functions may not be performed in the order shown or discussed, including performing functions in a substantially simultaneous manner or in reverse order depending on the functions involved, which should be understood by technicians in the technical field to which the embodiments of the present disclosure belong.

[0114] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute the instructions), or in combination with these instruction execution systems, devices or apparatuses. For the purpose of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in combination with these instruction execution systems, devices or apparatuses. More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk box (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), a fiber optic device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium and then editing, interpreting or processing in other suitable ways if necessary, and then stored in a computer memory.

[0115] It should be understood that the various parts of the present disclosure can be implemented in hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, multiple steps or methods can be implemented in software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, any one of the following technologies known in the art or a combination thereof can be used to implement: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0116] A person skilled in the art may understand that all or part of the steps in the method for implementing the above-mentioned embodiment may be completed by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, which, when executed, includes one or a combination of the steps of the method embodiment.

[0117] In addition, each functional unit in each embodiment of the present disclosure may be integrated into a processing module, or each unit may exist physically separately, or two or more units may be integrated into one module. The above-mentioned integrated module may be implemented in the form of hardware or in the form of a software functional module. If the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it may also be stored in a computer-readable storage medium.

[0118] The storage medium mentioned above may be a read-only memory, a magnetic disk or an optical disk, etc. Although the embodiments of the present disclosure have been shown and described above, it can be understood that the above embodiments are exemplary and cannot be understood as limitations of the present disclosure. A person of ordinary skill in the art may change, modify, replace and modify the above embodiments within the scope of the present disclosure.

Claims

1. A method for monitoring the state of an electric heating unit, characterized in that: The following steps are involved: Obtaining the operating data of the electric heating unit to be tested; Extracting features from the operating data based on a neural network model to obtain a first data feature; the neural network model is pre-trained using a whale optimization algorithm; Using an autoencoder to perform feature dimension reduction on the first data feature to obtain a second data feature; The second data feature is input into a pre-trained random forest model with adaptive threshold pruning to perform feature classification, so as to obtain the state type information of the electric heating unit to be tested.

2. The method according to claim 1, characterized in that The neural network model is pre-trained by the following steps: Acquire operation training data of the electric heating unit and data feature labels corresponding to the operation training data; Inputting the running training data into an initial neural network model for feature extraction to obtain a third data feature output by the initial neural network model; The whale optimization algorithm is used to adjust the learning rate, weight and number of nodes in the hidden layer of the neural network of the initial neural network model based on the error value between the third data feature and the data feature label to obtain the neural network model.

3. The method according to claim 2, characterized in that The step of obtaining the operation training data of the electric heating unit comprises: Acquiring initial operation training data of the electric heating unit; The initial running training data is expanded based on a generative adversarial network with lag decomposition to obtain the running training data.

4. The method according to any one of claims 1 to 3, characterized in that The random forest model is pre-trained by the following steps: Acquire multiple training data of the random forest model, each of the training data includes a data feature sample of an electric heating unit and a state type information label corresponding to the data feature sample; Performing self-service sampling among the plurality of data feature samples to obtain a sample set for each decision tree in the random forest model; Applying a dynamic feature selection strategy to perform feature selection in the sample set of each decision tree to obtain a feature subset used by each decision tree in training; The decision tree is trained using the sample set and the feature subset. During the training process, pruning decisions are made on the decision tree according to a pruning threshold of the random forest model, and the pruning threshold is adjusted according to the depth of the decision tree and the error rate of the random forest model; the error rate is determined according to the state type information label and the predicted state type information output by the random forest model.

5. A state monitoring device for an electric heating unit, characterized in that: include: An acquisition module is used to acquire the operating data of the electric heating unit to be tested; A feature extraction module, used for extracting features from the operating data based on a neural network model to obtain a first data feature; the neural network model is pre-trained by a whale optimization algorithm; A feature processing module, configured to perform feature dimension reduction on the first data feature by using an autoencoder to obtain a second data feature; A classification module is used to input the second data feature into a pre-trained random forest model with adaptive threshold pruning to perform feature classification, so as to obtain the state type information of the electric heating unit to be tested.

6. The device according to claim 5, characterized in that The device further comprises a first training module; wherein the first training module is used for: Acquire operation training data of the electric heating unit and data feature labels corresponding to the operation training data; Inputting the running training data into an initial neural network model for feature extraction to obtain a third data feature output by the initial neural network model; The whale optimization algorithm is used to adjust the learning rate, weight and number of nodes in the hidden layer of the neural network of the initial neural network model based on the error value between the third data feature and the data feature label to obtain the neural network model.

7. The device according to claim 6, characterized in that The first training module is specifically used for: Acquiring initial operation training data of the electric heating unit; The initial running training data is expanded based on a generative adversarial network with lag decomposition to obtain the running training data.

8. The device according to any one of claims 5 to 7, characterized in that: The device further comprises a second training module; wherein the second training module is used for: Acquire multiple training data of the random forest model, each of the training data includes a data feature sample of an electric heating unit and a state type information label corresponding to the data feature sample; Performing self-service sampling among the plurality of data feature samples to obtain a sample set for each decision tree in the random forest model; Applying a dynamic feature selection strategy to perform feature selection in the sample set of each decision tree to obtain a feature subset used by each decision tree in training; The decision tree is trained using the sample set and the feature subset. During the training process, pruning decisions are made on the decision tree according to a pruning threshold of the random forest model, and the pruning threshold is adjusted according to the depth of the decision tree and the error rate of the random forest model; the error rate is determined according to the state type information label and the predicted state type information output by the random forest model.

9. An electronic device, characterized in that: include: A processor, and a memory communicatively connected to the processor; The memory stores computer-executable instructions; The processor executes the computer-executable instructions stored in the memory to implement the method according to any one of claims 1 to 4.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-executable instructions, which are used to implement the method according to any one of claims 1 to 4 when executed by a processor.

Images