Equipment state prediction method and device, electronic equipment and computer storage medium
The method enhances power equipment status prediction by using deep learning models with linear and polynomial interpolation to preprocess trend components, addressing the limitations of traditional moving averages and improving prediction accuracy.
Patent Information
- Application Number
- CN202411161612.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-22
- Publication Date
- 2024-12-13
AI Technical Summary
Traditional moving average methods in power equipment status prediction fail to flexibly capture complex dynamic changes in time series data, are affected by uneven fluctuations, and cannot effectively predict future trends due to equal weighting and simple averaging.
A method involving deep learning models that combine basic linear transformation and piecewise polynomial interpolation, preprocessing historical data to separate trend components into history-based and derivative-based components, and using B-spline basis functions for enhanced prediction accuracy.
Improves the accuracy of power equipment status prediction by flexibly capturing dynamic changes and reducing the impact of fluctuations, providing reliable forecasting for real-time decision-making and preventive maintenance.
Smart Images
Figure CN119128429A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the technical field of power equipment management, and in particular to a method, device, electronic device and computer storage medium for predicting equipment status. Background Art
[0002] Power equipment status prediction is a crucial task in the power system. The prediction task for equipment status monitoring can enable maintainers to know and evaluate the operating status of the equipment in advance and handle it in time, which can effectively improve the operating efficiency and reliability of the power system.
[0003] Traditional forecasting methods use sliding average decomposition to distinguish between trend and seasonal components in time series, and eliminate short-term fluctuations in data through smoothing to highlight long-term trends.
[0004] However, there are some significant limitations of the method based on the moving average. First, the moving average method gives the same weight to the data in all windows during the smoothing process, which cannot flexibly capture the potential complex dynamic changes in the time series. Second, the moving average method is easily affected by uneven fluctuations, resulting in errors in the trend components obtained. In addition, the moving average is essentially a simple average of historical data and cannot effectively predict future trends based on historical data. Summary of the invention
[0005] In view of this, the present application provides a method, apparatus, electronic device and computer storage medium for predicting device status, which improves the limitations of the sliding average method in trend item decomposition, thereby improving the accuracy of device status prediction.
[0006] The first aspect of the present application provides a method for predicting a device state, comprising:
[0007] Receive the device's historical data time series;
[0008] Decomposing the historical data time series of the device to obtain an initial trend component and a seasonal component;
[0009] Preprocessing the initial trend component to obtain a mixed trend component;
[0010] The mixed trend component and the seasonal component are processed by an improved deep learning model to obtain a device status prediction result of the device; wherein the improved deep learning model combines basic linear transformation and piecewise polynomial interpolation mechanism.
[0011] Optionally, decomposing the historical data time series of the device to obtain an initial trend component and a seasonal component includes:
[0012] The historical data time series of the equipment is processed by a sliding average method to obtain an initial trend component;
[0013] The seasonal component is obtained by subtracting the initial trend component from the historical data time series of the device.
[0014] Optionally, preprocessing the initial trend component to obtain a mixed trend component includes:
[0015] Decomposing the initial trend component to obtain a first trend component and a second trend component; wherein the first trend component is a trend component based on history; and the second trend component is a trend component based on derivatives;
[0016] A weighted average is performed on the first trend component and the second trend component to obtain a mixed trend component.
[0017] Optionally, decomposing the initial trend component to obtain a first trend component and a second trend component includes:
[0018] The initial trend component is processed by a sliding average method to obtain a first trend component;
[0019] The initial trend component is processed by linear regression to obtain a second trend component.
[0020] Optionally, the mixed trend component and the seasonal component are processed by an improved deep learning model to obtain a device state prediction result of the device, including:
[0021] After receiving the mixed trend component and the seasonal component, the improved deep learning model initializes a set of uniformly distributed grid nodes; wherein the grid nodes define at least one local area in the feature space; each of the local areas is described by a set of B-spline basis functions;
[0022] The B-spline basis function performs piecewise polynomial transformation on the mixed trend component and the seasonal component to obtain a first characteristic result;
[0023] Performing a linear transformation on the mixed trend component and the seasonal component to obtain a second characteristic result;
[0024] The first characteristic result and the second characteristic result are combined to obtain a device state prediction result of the device.
[0025] A second aspect of the present application provides a device for predicting a device state, comprising:
[0026] A first receiving unit, used for receiving a time series of historical data of a device;
[0027] A first decomposition unit is used to decompose the historical data time series of the device to obtain an initial trend component and a seasonal component;
[0028] A preprocessing unit, used for preprocessing the initial trend component to obtain a mixed trend component;
[0029] A prediction unit is used to process the mixed trend component and the seasonal component through an improved deep learning model to obtain a device status prediction result of the device; wherein the improved deep learning model combines basic linear transformation and piecewise polynomial interpolation mechanism.
[0030] Optionally, the first decomposition unit includes:
[0031] A first sliding average unit, used for processing the historical data time series of the device by using a sliding average method to obtain an initial trend component;
[0032] The first decomposition subunit is used to subtract the initial trend component from the historical data time series of the device to obtain the seasonal component.
[0033] Optionally, the preprocessing unit includes:
[0034] A second decomposition unit is used to decompose the initial trend component to obtain a first trend component and a second trend component; wherein the first trend component is a history-based trend component; and the second trend component is a derivative-based trend component;
[0035] The weighted averaging unit is used to perform weighted averaging on the first trend component and the second trend component to obtain a mixed trend component.
[0036] Optionally, the second decomposition unit includes:
[0037] A second sliding average unit is used to process the initial trend component by a sliding average method to obtain a first trend component;
[0038] The linear regression unit is used to process the initial trend component by linear regression to obtain a second trend component.
[0039] Optionally, the prediction unit includes:
[0040] An initialization unit, configured to initialize a set of uniformly distributed grid nodes after the improved deep learning model receives the mixed trend component and the seasonal component; wherein the grid nodes define at least one local region in the feature space; and each of the local regions is described by a set of B-spline basis functions;
[0041] A piecewise polynomial transformation unit, used for the B-spline basis function to perform piecewise polynomial transformation on the mixed trend component and the seasonal component to obtain a first characteristic result;
[0042] A linear transformation unit, used for performing a linear transformation on the mixed trend component and the seasonal component to obtain a second characteristic result;
[0043] A merging unit is used to merge the first characteristic result and the second characteristic result to obtain a device state prediction result of the device.
[0044] A third aspect of the present application provides an electronic device, including:
[0045] one or more processors;
[0046] a storage device having one or more programs stored thereon;
[0047] When the one or more programs are executed by the one or more processors, the one or more processors implement the device status prediction method as described in any one of the first aspects.
[0048] A fourth aspect of the present application provides a computer storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the method for predicting the device state as described in any one of the first aspects is implemented.
[0049] It can be seen from the above scheme that the present application provides a prediction method, device, electronic device and computer storage medium for equipment status. The method decomposes the equipment's historical data time series after receiving the equipment's historical data time series to obtain an initial trend component and a seasonal component; before inputting the initial trend component and the seasonal component into the improved deep learning model for prediction, the initial trend component is preprocessed to obtain a mixed trend component; then, the mixed trend component and the seasonal component are processed by the improved deep learning model to obtain the equipment status prediction result of the equipment; wherein the improved deep learning model combines the basic linear transformation and the piecewise polynomial interpolation mechanism. Compared with the prior art, the limitations of the sliding average method in the decomposition of trend terms are improved, thereby improving the accuracy of the prediction of the equipment status. BRIEF DESCRIPTION OF THE DRAWINGS
[0050] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.
[0051] Figure 1 A specific flow chart of a method for predicting device status provided in an embodiment of the present application;
[0052] Figure 2 A flowchart for decomposing a time series of historical data of a device provided by another embodiment of the present application;
[0053] Figure 3 A flowchart of preprocessing an initial trend component to obtain a mixed trend component is provided in another embodiment of the present application;
[0054] Figure 4 A flow chart for decomposing initial trend components provided in another embodiment of the present application;
[0055] Figure 5 A flowchart of prediction using an improved deep learning model provided in another embodiment of the present application;
[0056] Figure 6 A schematic diagram of a device state prediction apparatus provided by another embodiment of the present application;
[0057] Figure 7 A schematic diagram of an electronic device for implementing a method for predicting device status provided in another embodiment of the present application. DETAILED DESCRIPTION
[0058] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.
[0059] The term "including" and its variations used herein are open inclusions, i.e., "including but not limited to". The term "based on" means "based at least in part on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments". The relevant definitions of other terms will be given in the following description.
[0060] It is understandable that before using the technical solutions disclosed in the embodiments of the present disclosure, the types, scope of use, usage scenarios, etc. of the personal information involved in the present disclosure should be informed to the user and the user's authorization should be obtained in an appropriate manner in accordance with relevant laws and regulations.
[0061] It should be noted that the concepts such as "first" and "second" mentioned in this application are only used to distinguish different devices, modules or units, and are not used to limit the order or interdependence of the functions performed by these devices, modules or units.
[0062] It should be noted that the modifications of "one" and "plurality" mentioned in the present application are illustrative rather than restrictive, and those skilled in the art should understand that unless otherwise clearly indicated in the context, it should be understood as "one or more".
[0063] The present application embodiment provides a method for predicting device status, such as Figure 1 As shown, the specific steps include:
[0064] S101, receiving a time series of historical data of a device.
[0065] Taking a transformer as an example, the historical data of the equipment includes but is not limited to oil temperature, etc., which is not limited here.
[0066] S102. Decompose the historical data time series of the equipment to obtain an initial trend component and a seasonal component.
[0067] In the specific implementation process of the present application, the historical data time series of the device can be decomposed by the DLinear model but is not limited to obtain the initial trend component and the seasonal component, which is not limited here.
[0068] Among them, the DLinear model pays special attention to decomposing data into two main components: trend and seasonality when dealing with time series. This decomposition method helps to better understand and predict the dynamic changes of time series data. The trend part reflects the change trend of data over time, while the seasonal part captures the cyclical changes in the data.
[0069] In the DLinear model, the trend part is usually estimated by sliding average or other smoothing techniques, while the seasonal part is determined by analyzing the periodic patterns in the data. This decomposition method is not only suitable for time series forecasting, but also helps in anomaly detection and other related analysis. For example, by analyzing the stability of the trend part, any sudden changes or anomalies can be detected, which is particularly important for real-time monitoring and early warning systems.
[0070] In addition, when processing time series data, the DLinear model also considers data preprocessing, such as zero mean, to further optimize the performance of the model. These preprocessing steps help reduce data noise and irrelevant variables, allowing the model to focus more on capturing key features related to the prediction target.
[0071] Optionally, in another embodiment of the present application, an implementation of step S102 is as follows: Figure 2 As shown, including:
[0072] S201. The historical data time series of the equipment is processed by a sliding average method to obtain an initial trend component.
[0073] Among them, the Moving Average method is a time series analysis technique used to analyze trends and periodic changes in data. It smoothes random fluctuations in the data by taking the arithmetic average of continuous subintervals in the time series data, thereby revealing the development trend of the data.
[0074] Specifically, the sliding average method smoothes the data by calculating the average value within a fixed-size data window. This window slides over the data set, and a new average value is calculated each time, resulting in a smoothed data series.
[0075] S202. Subtract the initial trend component from the device's historical data time series to obtain a seasonal component.
[0076] Traditional moving average decomposition methods are used to distinguish between trend and seasonal components in time series, and to eliminate short-term fluctuations in data through smoothing to highlight long-term trends. However, methods based on moving averages have some significant limitations. First, the moving average method gives the same weight to data in all windows during smoothing, and cannot flexibly capture the potential complex dynamic changes in time series. Second, the moving average method is easily affected by uneven fluctuations, resulting in errors in the trend components obtained. In addition, the moving average is essentially a simple average of historical data and cannot effectively predict future trends based on historical data.
[0077] S103, preprocessing the initial trend component to obtain a mixed trend component.
[0078] In response to the above limitations, this application proposes a trend decomposition method with a hybrid strategy, in which the initial trend component and the seasonal component are preprocessed before being input into the improved deep learning model to obtain a hybrid trend component, wherein the hybrid trend component can more effectively and accurately capture the trend changes in the time series.
[0079] Optionally, in another embodiment of the present application, an implementation of step S103 is as follows: Figure 3 As shown, including:
[0080] S301. Decompose the initial trend component to obtain a first trend component and a second trend component.
[0081] Among them, the first trend component is the trend component based on history, and the second trend component is the trend component based on derivatives. The trend component based on history represents the long-term trend in the time series, and the trend component based on derivatives helps the model grasp the future change law of the trend. At the same time, the learnable parameters increase the flexibility of the subsequent model for the selection of historical data, which helps the model avoid the impact of historical fluctuations.
[0082] Optionally, in another embodiment of the present application, an implementation of step S301 is as follows: Figure 4 As shown, including:
[0083] S401. Process the initial trend component by a sliding average method to obtain a first trend component.
[0084] In the specific implementation process of the present application, the initial trend component is firstly processed by sliding average to obtain a history-based trend component (first trend component) to represent the long-term trend in the time series.
[0085] S402: Process the initial trend component by linear regression to obtain a second trend component.
[0086] In the specific implementation process of the present application, a shared linear layer may be used but not limited to calculate the initial trend component in a linear regression manner to obtain a derivative-based trend component (second trend component) to grasp the future change law of the trend.
[0087] S302 , performing weighted averaging on the first trend component and the second trend component to obtain a mixed trend component.
[0088] By combining sliding average and linear regression, this application can more accurately capture trend changes in time series, overcome the limitations of traditional sliding average methods in trend term decomposition, and at the same time reduce the impact of model fluctuations on trend term decomposition. Secondly, the method adopted in this application can effectively separate the trend and residual parts, provide more detailed feature inputs, and thus improve the predictive performance of the model. Thirdly, in the subsequent application scenarios of equipment monitoring and maintenance, by capturing and analyzing subtle fluctuations in time series, it can provide maintenance personnel with more reliable forecasting information to support real-time decision-making and preventive maintenance.
[0089] S104. Process the mixed trend component and the seasonal component through the improved deep learning model to obtain the equipment status prediction result of the equipment.
[0090] Among them, the improved deep learning model combines basic linear transformation and piecewise polynomial interpolation mechanism.
[0091] In the specific implementation process of this application, the improved deep learning model can be, but is not limited to, an improved KAN (Kolmogorv-ArnoldNet) model, which is not limited here. Among them, the improved KAN model includes a linear layer built based on KAN, which can be called KANLInear. KANLinear achieves high-precision modeling of complex temporal relationships in time series data by combining basic linear transformations and piecewise polynomial interpolation mechanisms.
[0092] Optionally, in another embodiment of the present application, an implementation of step S104 is as follows: Figure 5 As shown, including:
[0093] S501. After receiving the mixed trend component and the seasonal component, the improved deep learning model initializes a set of uniformly distributed grid nodes.
[0094] The grid nodes define at least one local region in the feature space; each local region is represented by a set of B-spline basis functions (B i (x)) for description.
[0095] S502. Perform piecewise polynomial transformation on the mixed trend component and the seasonal component using a B-spline basis function to obtain a first characteristic result.
[0096] In the specific implementation process of this application, the following formula can be used to perform piecewise polynomial transformation on the mixed trend component and seasonal component of the input to finely capture the nonlinear characteristics of the input data:
[0097]
[0098] Among them, w i Represents the weight of each piecewise polynomial, which enables the model to smoothly transition in the continuous feature space, thereby enhancing the nonlinear expression ability and generalization of the model.
[0099] S503, performing linear transformation on the mixed trend component and the seasonal component to obtain a second characteristic result.
[0100] Continuing with the above example, in addition to piecewise polynomial interpolation, KANLinear also includes a base linear transformation part, which first preprocesses the input mixed trend component and seasonal component through an activation function (such as SiLU) to introduce preliminary nonlinear characteristics, and then applies the standard linear transformation Base(x):
[0101] Base(x)=W base f(x);
[0102] Among them, W baseIt is a weighted activation function, f(x) is the activation function, and x refers to the mixed trend component and seasonal component of the input.
[0103] This step provides the model with the necessary linear processing capabilities and enhances the stability and efficiency of the model when processing linear relationships.
[0104] S504: Combine the first characteristic result and the second characteristic result to obtain a device state prediction result of the device.
[0105] In the specific implementation process of this application, KANLinear combines the first characteristic result and the second characteristic result to obtain the device state prediction result y of the device:
[0106] y=Base(x)+Spline(x)
[0107] It should be noted that a key feature of time series data is its inherent nonlinear structure, which originates from cyclical changes, sudden events, or shifts in long-term trends. KANLinear provides a powerful mechanism through piecewise polynomial interpolation to accurately fit the nonlinear patterns of data in small local areas. KANLinear creates multiple segmented regions in the feature space and uses polynomial function approximation, which allows KANLinear to capture short-term fluctuations in time series in a local range.
[0108] At the same time, the linear part of KANLinear provides direct modeling capabilities for the linear trend of time series data, making the model have better predictive capabilities for linear trends. KAN's unique design enables complex nonlinear modeling with fewer layers, providing a more efficient and intuitive way to simulate nonlinear relationships, thus outperforming traditional MLP models in terms of model accuracy, interpretability, and computational efficiency.
[0109] We use the Electricity dataset to conduct experiments. The proposer of this dataset believes that the operating condition of the transformer can be reflected from the oil temperature. Therefore, this dataset is well suited for tasks aimed at monitoring the operating status of equipment.
[0110] At the same time, we adopt the evaluation method, set the time window to 72, and predict the information of the next 24 steps. At the same time, in order to obtain the model's prediction for each step, we use the evaluation strategy of step-by-step evaluation within the prediction window, and calculate the performance indicators step by step for each step. For the convenience of display, we select 1-3, 11-13, and 22-24 steps for evaluation to measure the performance of the model in the early, middle, and late stages of the prediction interval.
[0111] In the designed experiment, KANLinear and DLinear of this application are compared. As shown in Table 1, the experiment observes that under the same loss function, the model of this application still has advanced performance.
[0112]
[0113] Table 1
[0114] It can be observed that the model of the present application has achieved good performance in the entire interval, which shows that the model of the present application has better performance in the implementation based on the linear model.
[0115] Next, we conduct ablation experiments to determine the feasibility of using KANLinear to replace MLP. The leading indicators of the improved model in this application are bolded to facilitate observation of the performance of the model. It can be seen that using KANLinear to replace MLP has achieved significant improvement.
[0116]
[0117] Table 2
[0118] At the same time, the effectiveness of the hybrid strategy of trend decomposition is explored through experiments. As shown in Table 3, most of the indicators of the model are improved when the hybrid strategy of trend decomposition is used.
[0119]
[0120]
[0121] Table 3
[0122] From the above scheme, it can be seen that the present application provides a method for predicting the state of an equipment. After receiving the time series of historical data of the equipment, the time series of historical data of the equipment is decomposed to obtain the initial trend component and the seasonal component; before the initial trend component and the seasonal component are input into the improved deep learning model for prediction, the initial trend component is preprocessed to obtain the mixed trend component; then, the mixed trend component and the seasonal component are processed by the improved deep learning model to obtain the prediction result of the equipment state of the equipment; wherein, the improved deep learning model combines the basic linear transformation and the piecewise polynomial interpolation mechanism. Compared with the prior art, the limitations of the sliding average method in the decomposition of trend items are improved, thereby improving the accuracy of the prediction of the equipment state.
[0123] The present application embodiment provides a device for predicting the state of a device, such as Figure 6 As shown, specifically including:
[0124] The first receiving unit 601 is used to receive a time series of historical data of a device.
[0125] The first decomposition unit 602 is used to decompose the historical data time series of the device to obtain an initial trend component and a seasonal component.
[0126] Optionally, in another embodiment of the present application, an implementation of the first decomposition unit 601 includes:
[0127] The moving average unit is used to process the historical data time series of the equipment using the sliding average method to obtain the initial trend component.
[0128] The first decomposition subunit is used to subtract the initial trend component from the historical data time series of the equipment to obtain the seasonal component.
[0129] For the specific working process of the units disclosed in the above embodiments of the present application, please refer to the corresponding method embodiments, such as Figure 2 As shown, no further description is given here.
[0130] The preprocessing unit 603 is used to preprocess the initial trend component to obtain a mixed trend component.
[0131] Optionally, in another embodiment of the present application, an implementation of the preprocessing unit 603 includes:
[0132] The second decomposition unit is used to decompose the initial trend component to obtain a first trend component and a second trend component.
[0133] Among them, the first trend component is a trend component based on history; the second trend component is a trend component based on derivative.
[0134] The weighted averaging unit is used to perform weighted averaging on the first trend component and the second trend component to obtain a mixed trend component.
[0135] For the specific working process of the units disclosed in the above embodiments of the present application, please refer to the corresponding method embodiments, such as Figure 3 As shown, no further description is given here.
[0136] Optionally, in another embodiment of the present application, an implementation of the second decomposition unit includes:
[0137] The sliding average unit is used to process the initial trend component by a sliding average method to obtain a first trend component.
[0138] The linear regression unit is used to process the initial trend component by linear regression to obtain a second trend component.
[0139] For the specific working process of the units disclosed in the above embodiments of the present application, please refer to the corresponding method embodiments, such as Figure 4 As shown, no further description is given here.
[0140] The prediction unit 604 is used to process the mixed trend component and the seasonal component through an improved deep learning model to obtain a device status prediction result of the device.
[0141] Among them, the improved deep learning model combines basic linear transformation and piecewise polynomial interpolation mechanism.
[0142] For the specific working process of the units disclosed in the above embodiments of the present application, please refer to the corresponding method embodiments, such as Figure 1 As shown, no further description is given here.
[0143] Optionally, in another embodiment of the present application, an implementation of the prediction unit 604 includes:
[0144] The initialization unit is used to initialize a set of uniformly distributed grid nodes after the improved deep learning model receives the mixed trend component and the seasonal component.
[0145] The grid nodes define at least one local region in the feature space; each local region is described by a set of B-spline basis functions.
[0146] The piecewise polynomial transformation unit is used for performing piecewise polynomial transformation on the mixed trend component and the seasonal component using the B-spline basis function to obtain the first characteristic result.
[0147] The linear transformation unit is used to perform linear transformation on the mixed trend component and the seasonal component to obtain the second characteristic result.
[0148] The merging unit is used to merge the first characteristic result and the second characteristic result to obtain a device state prediction result of the device.
[0149] For the specific working process of the units disclosed in the above embodiments of the present application, please refer to the corresponding method embodiments, such as Figure 5 As shown, no further description is given here.
[0150] As can be seen from the above scheme, the present application provides a prediction device for equipment status. After receiving the equipment's historical data time series through the first receiving unit 601, the first decomposition unit 602 decomposes the equipment's historical data time series to obtain an initial trend component and a seasonal component; before inputting the initial trend component and the seasonal component into the improved deep learning model for prediction, the preprocessing unit 603 preprocesses the initial trend component to obtain a mixed trend component; then, the prediction unit 604 processes the mixed trend component and the seasonal component through the improved deep learning model to obtain the equipment status prediction result of the equipment; wherein, the improved deep learning model combines the basic linear transformation and the piecewise polynomial interpolation mechanism. Compared with the prior art, the limitations of the sliding average method in the decomposition of trend items are improved, thereby improving the accuracy of the prediction of the equipment status.
[0151] The functions described above herein may be performed at least in part by one or more hardware logic components. For example, without limitation, exemplary types of hardware logic components that may be used include: field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), complex programmable logic devices (CPLDs), and the like.
[0152] Another embodiment of the present application provides an electronic device, such as Figure 7 As shown, including:
[0153] One or more processors 701.
[0154] The storage device 702 stores one or more programs.
[0155] When the one or more programs are executed by the one or more processors 701, the one or more processors 701 implement the device status prediction method as described in any one of the above embodiments.
[0156] Another embodiment of the present application provides a computer storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the method for predicting the device state as described in any one of the above embodiments is implemented.
[0157] In the context of the present application, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.
[0158] It should be noted that the computer-readable medium mentioned above in the present application may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present application, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, device or device. In the present application, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, in which a computer-readable program code is carried. This propagated data signal may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. The computer readable signal medium may also be any computer readable medium other than a computer readable storage medium, which may send, propagate or transmit a program for use by or in conjunction with an instruction execution system, apparatus or device. The program code contained on the computer readable medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (radio frequency), etc., or any suitable combination of the above.
[0159] The computer-readable medium may be included in the electronic device, or may exist independently without being incorporated into the electronic device.
[0160] Another embodiment of the present application provides a computer program product. When the computer program product is executed, it is used to execute any of the above-mentioned device status prediction methods.
[0161] In particular, according to an embodiment of the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present application includes a computer program product, which includes a computer program carried on a non-transitory computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network via a communication device, or installed from a storage device, or installed from a ROM. When the computer program is executed by a processing device, the above-mentioned functions defined in the method of the embodiment of the present application are executed.
[0162] Although the subject matter has been described in language specific to structural features and / or methodological logical actions, it should be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or actions described above. On the contrary, the specific features and actions described above are merely example forms of implementing the claims.
[0163] Although several specific implementation details are included in the above discussion, these should not be interpreted as limiting the scope of the present application. Certain features described in the context of a separate embodiment can also be implemented in a single embodiment in combination. On the contrary, the various features described in the context of a single embodiment can also be implemented in multiple embodiments individually or in any suitable sub-combination.
[0164] The above description is only a preferred embodiment of the present application and an explanation of the technical principles used. Those skilled in the art should understand that the scope of application involved in the present application is not limited to the technical solution formed by a specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the above application concept. For example, the above features are replaced with (but not limited to) technical features with similar functions applied in the present application.
Claims
1. A method for predicting equipment status, characterized in that: include: Receive the device's historical data time series; Decomposing the historical data time series of the device to obtain an initial trend component and a seasonal component; Preprocessing the initial trend component to obtain a mixed trend component; The mixed trend component and the seasonal component are processed by an improved deep learning model to obtain a device status prediction result of the device; wherein the improved deep learning model combines basic linear transformation and piecewise polynomial interpolation mechanism.
2. The device status prediction method according to claim 1, characterized in that: Decomposing the historical data time series of the device to obtain an initial trend component and a seasonal component includes: The historical data time series of the equipment is processed by a sliding average method to obtain an initial trend component; The seasonal component is obtained by subtracting the initial trend component from the historical data time series of the device.
3. The device status prediction method according to claim 1, characterized in that: The preprocessing of the initial trend component to obtain a mixed trend component includes: Decomposing the initial trend component to obtain a first trend component and a second trend component; wherein the first trend component is a trend component based on history; and the second trend component is a trend component based on derivatives; A weighted average is performed on the first trend component and the second trend component to obtain a mixed trend component.
4. The method for predicting the equipment status according to claim 3, characterized in that: Decomposing the initial trend component to obtain a first trend component and a second trend component includes: The initial trend component is processed by a sliding average method to obtain a first trend component; The initial trend component is processed by linear regression to obtain a second trend component.
5. The method for predicting the equipment status according to claim 1, characterized in that: The mixed trend component and the seasonal component are processed by the improved deep learning model to obtain the equipment state prediction result of the equipment, including: After receiving the mixed trend component and the seasonal component, the improved deep learning model initializes a set of uniformly distributed grid nodes; wherein the grid nodes define at least one local area in the feature space; each of the local areas is described by a set of B-spline basis functions; The B-spline basis function performs piecewise polynomial transformation on the mixed trend component and the seasonal component to obtain a first characteristic result; Performing a linear transformation on the mixed trend component and the seasonal component to obtain a second characteristic result; The first characteristic result and the second characteristic result are combined to obtain a device state prediction result of the device.
6. A device for predicting equipment status, characterized in that: include: A first receiving unit, used for receiving a time series of historical data of a device; A first decomposition unit is used to decompose the historical data time series of the device to obtain an initial trend component and a seasonal component; A preprocessing unit, used for preprocessing the initial trend component to obtain a mixed trend component; A prediction unit is used to process the mixed trend component and the seasonal component through an improved deep learning model to obtain a device status prediction result of the device; wherein the improved deep learning model combines basic linear transformation and piecewise polynomial interpolation mechanism.
7. The device for predicting the state of an equipment according to claim 6, characterized in that: The first decomposition unit comprises: A first sliding average unit, used for processing the historical data time series of the device by using a sliding average method to obtain an initial trend component; The first decomposition subunit is used to subtract the initial trend component from the historical data time series of the device to obtain the seasonal component.
8. The device for predicting equipment status according to claim 6, characterized in that: The pre-processing unit comprises: A second decomposition unit is used to decompose the initial trend component to obtain a first trend component and a second trend component; wherein the first trend component is a history-based trend component; and the second trend component is a derivative-based trend component; The weighted averaging unit is used to perform weighted averaging on the first trend component and the second trend component to obtain a mixed trend component.
9. An electronic device, characterized in that: include: one or more processors; a storage device having one or more programs stored thereon; When the one or more programs are executed by the one or more processors, the one or more processors are enabled to implement the device status prediction method as described in any one of claims 1 to 5.
10. A computer storage medium, characterized in that: A computer program is stored thereon, wherein when the computer program is executed by a processor, the method for predicting the device status as claimed in any one of claims 1 to 5 is implemented.
Citation Information
Patent Citations
Transfer equipment consumption optimal combination prediction method based on uncertain weight
CN114548561A