CN115099272A - Method and device for processing time sequence signal, equipment and readable medium - Google Patents
Method and device for processing time sequence signal, equipment and readable medium Download PDFInfo
- Publication number
- CN115099272A CN115099272A CN202210702746.1A CN202210702746A CN115099272A CN 115099272 A CN115099272 A CN 115099272A CN 202210702746 A CN202210702746 A CN 202210702746A CN 115099272 A CN115099272 A CN 115099272A
- Authority
- CN
- China
- Prior art keywords
- signal
- sensors
- training
- time
- sample data
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Abstract
The invention provides a method, a device, equipment and a readable medium for processing a time sequence signal. The method comprises the following steps: generating prediction sample data according to time sequence signals acquired by a plurality of sensors in real time; performing feature extraction processing on the prediction sample data to obtain prediction sample features; and inputting the predicted sample characteristics into a trained signal processing model, and obtaining an output result of the signal processing model. According to the technical scheme, the time sequence of each sensor is considered, the spatial relationship of the sensors is also considered, the spatial relationship of the sensors is effectively reserved, and the accuracy and the processing efficiency of time sequence signal processing can be effectively improved.
Description
The present application is a divisional application of patent applications entitled "method and apparatus for processing a timing signal, device, and readable medium" with an application date of 2019, 9/5/2019 and an application number of 201910838480.1.
Technical Field
The present invention relates to the field of computer application technologies, and in particular, to a method, an apparatus, a device, and a readable medium for processing a timing signal.
Background
In the prior art, signals generated in many scenes are timing signals. Timing signal processing such as anomaly detection, classification or numerical prediction, etc. are all very important tasks in the study of this class of signals.
In the prior art, a system is usually provided with a plurality of distributed sensors, and the signal of each sensor can be regarded as a timing signal. The prior art method mainly processes signals on the time axis of a single sensor signal to perform some relevant prediction judgments, but the single sensor signal fluctuates with time and has certain randomness, so that the judgment is inaccurate.
Disclosure of Invention
The invention provides a method, a device, equipment and a readable medium for processing a time sequence signal, which are used for improving the accuracy of time sequence signal processing.
The invention provides a method for processing a time sequence signal, which comprises the following steps:
generating prediction sample data according to time sequence signals acquired by a plurality of sensors in real time;
performing feature extraction processing on the prediction sample data to obtain prediction sample features;
and inputting the predicted sample characteristics into a trained signal processing model, and obtaining an output result of the signal processing model.
The invention also provides a training method of the signal processing model, which comprises the following steps:
generating a training sample data set according to time sequence signals acquired by a plurality of sensors in a target environment in a historical manner and corresponding historical state information in the target environment;
performing feature extraction processing on the training sample data set to obtain a training sample feature set;
and training a signal processing model based on the training sample feature set, the collected historical state information and a machine learning algorithm.
The present invention also provides a device for processing a timing signal, comprising:
the generating module is used for generating prediction sample data according to the time sequence signals acquired by the sensors in real time;
the extraction module is used for carrying out feature extraction processing on the prediction sample data to obtain the features of the prediction sample;
and the processing module is used for inputting the predicted sample characteristics to the trained signal processing model and acquiring the output result of the signal processing model.
The invention also provides a training device of the signal processing model, which comprises:
the generating module is used for generating a training sample data set according to time sequence signals historically acquired by a plurality of sensors in a target environment and corresponding historical state information in the target environment;
the extraction module is used for carrying out feature extraction processing on the training sample data set to obtain a training sample feature set;
and the training module is used for training a signal processing model based on the training sample characteristic set, the collected historical state information and a machine learning algorithm.
The present invention also provides a computing device comprising:
a processor; and
a memory having executable code stored thereon which, when executed by the processor, causes the processor to perform a method as described in any one of the above.
The invention also provides a non-transitory machine-readable storage medium having stored thereon executable code which, when executed by a processor of an electronic device, causes the processor to perform a method as any one of the above.
According to the time sequence signal processing method, the time sequence signal processing device, the time sequence signal processing equipment and the readable medium, through the scheme, the time sequence signals of the plurality of sensors can be processed in real time by adopting the pre-trained signal processing model.
Drawings
The above and other objects, features and advantages of the present disclosure will become more apparent by describing in greater detail exemplary embodiments thereof with reference to the attached drawings, in which like reference numerals generally represent like parts throughout.
FIG. 1 is a flowchart illustrating a method for processing a timing signal according to an embodiment of the present invention.
FIG. 2 is a flowchart of a method for training a signal processing model according to an embodiment of the present invention.
Fig. 3 is a schematic diagram of a frame structure of a CNN model used in the present invention.
Fig. 4 is a schematic diagram of two functions Sigmoid and Tanh provided by the present invention.
FIG. 5 is a code logic diagram of a batch normalization layer in the CNN model of the present invention.
Fig. 6 is an operation diagram of the pooling layer in the CNN model of the present invention.
Fig. 7 is a diagram comparing a DropOut layer with a previous fully connected layer in the CNN model of the present invention.
Fig. 8 is a block diagram of an embodiment of a device for processing timing signals according to the present invention.
FIG. 9 is a block diagram of an embodiment of a training apparatus for a signal processing model according to the present invention.
FIG. 10 shows a schematic block diagram of a computing device that can be used to implement the method described above according to an embodiment of the invention.
Detailed Description
Preferred embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. While the preferred embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be embodied in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.
The inventors of the present application found that the existing time series signal analysis classification method is usually solved by using a traditional statistical regression model (such as ARIMA) or a custom extracted time window aggregation feature. The methods can analyze the historical distribution of a single time sequence signal from the perspective of the single time sequence signal so as to predict the future, but in reality, a plurality of time sequence signal sensors are usually distributed in a system, the signal of the single sensor has randomness along with the fluctuation of time, and the current method cannot well utilize the spatial relationship of the signals of the adjacent sensors, so that the information of one dimension is lost, and the problem of inaccurate judgment is often caused.
FIG. 1 is a flowchart illustrating a method for processing a timing signal according to an embodiment of the present invention. As shown in fig. 1, the method for processing a timing signal according to this embodiment may specifically include the following steps:
s100, generating prediction sample data according to time sequence signals acquired by a plurality of sensors in real time;
the execution main body of the method for processing the time sequence signals in the embodiment is a time sequence signal processing device, and the time sequence signal processing device processes the time sequence signals acquired by a plurality of sensors by adopting a trained signal processing model.
For example, the processing of the time-series signal of the present embodiment may include processing such as abnormality detection of the time-series signal, classification of the time-series signal, or numerical prediction of the time-series signal. The classification of the time sequence signal can be two or more.
The principle of the timing signal processing method according to the present embodiment can be implemented regardless of which processing is performed on the timing signal. The signals of each of the plurality of sensors of the present embodiment have a time sequence relationship, and a spatial relative relationship may also be preset between different sensors.
In particular, in practical applications, in order to process the time-series signals, a plurality of sensors may be disposed in the detection environment to acquire the time-series signals in real time. For example, in a scenario of detecting whether it rains in the future, a plurality of environment sensors, such as at least two of a humidity sensor, a temperature sensor, a wind speed sensor, and an air pressure sensor, may be provided, and a pre-trained signal processing model may be used to predict whether it rains in the future according to time sequence signals of the plurality of sensors, where an output result of the signal processing model is an environment detection result.
When detecting whether a certain designated machine has the problem of operation overheating, a temperature sensor can be arranged on each part of the designated machine, and then a pre-trained signal processing model is adopted to judge whether the designated machine has the danger of operation overheating and the like according to the time sequence signals of the temperature sensors of each part. The output result of the corresponding signal processing model at this time is the result of the determination as to whether or not the specified machine is running too hot.
In practice, for each of the plurality of sensors. A sensor with a sampling rate of 500 can be set, and 500 values per second can be obtained as the signal for this second. In the actual sampling process, signals in the sampling period can be selected according to the preset sampling period and used as signals collected by the sensor in real time. The sampling period of this embodiment may be 5 seconds, 3 seconds, or other preset time lengths according to actual requirements, which is not described herein in detail for example. The signals collected by each sensor within the current preset time span are timing signals collected in real time. And taking the time sequence signals of the plurality of sensors within each preset time length as real-time prediction sample data. For example, when the preset time length is 5 seconds, each sensor in the prediction sample data corresponding to the sampling rate of 500 includes 2500 sample data.
The above signal representation for each sensor, but in the present embodiment, at least two or even a plurality of sensors are involved, and a spatial order may be configured for the sensors in advance, so that, for signals acquired by the plurality of sensors in real time within each preset time length, the signals may be arranged according to the spatial order configured for the plurality of sensors in advance, and a two-dimensional signal array corresponding to time and space may be generated as one piece of prediction sample data. For example, in this embodiment, each of the n sensors acquires m timing signal values within a preset time duration, and the corresponding two-dimensional signal array is n × m.
S101, performing feature extraction processing on the prediction sample data to obtain prediction sample features;
the prediction sample data of the embodiment comprises all the sampling data which are acquired by the plurality of sensors in real time within a preset time span, the data is rich, a part of features can be extracted from all the sampling data, then the extracted part of features is subjected to statistical processing, and the prediction sample features are generated based on the processed values of the statistics. In this embodiment, in order to enrich the predicted sample characteristics, values of two or more kinds of statistics may be acquired. For example, in this embodiment, the step S101 "perform feature extraction processing on the prediction sample data to obtain the prediction sample feature", which may specifically include: and (3) counting the values of k statistics of each time window in d2 adjacent time windows of each sensor in d1 adjacent sensors in the two-dimensional signal array, and taking the values of the statistics as pixel values to obtain k signal pictures of d1 × d2 as predicted sample characteristics, wherein d1, d2 and k are positive integers larger than 1.
In this embodiment, the preset time length may be 10s, 20s or other continuous time lengths, and the time window may be a time unit smaller than the preset time length, for example, each second may be taken as one time window in this embodiment.
When the two-dimensional signal in this embodiment is n × m, at this time d1 is a positive integer greater than 1 and smaller than n, and the range of d2 in this embodiment is greater than 1 and smaller than the number of the largest time windows included in the preset time length. For example, in practical applications, each second may be used as a time window.
The k statistics of the present embodiment may include portions of signal range, signal variance, signal mean, signal maximum difference, and the like.
According to each piece of predicted sample data, the values of k statistics in each time window of d2 time windows of each sensor in adjacent d1 sensors in the corresponding two-dimensional signal array can be counted, and k pieces of d1 × d2 signal pictures are obtained by taking the values of the statistics as pixel values, so that k pieces of d1 × d2 pictures in the same time space can be constructed to serve as predicted sample characteristics.
For example, in a certain scene, for a signal map with prediction sample data of n × m, k is 3, and signal polar difference, signal variance and signal mean value are respectively taken, at this time, three signal pictures of d1 × d2 can be obtained, and in one signal picture of d1 × d2, the pixel value of each position is the value of the signal polar difference of the sensor corresponding to the abscissa of the corresponding position within the time window corresponding to the ordinate of the corresponding position; in another d1 × d2 signal picture, the pixel value of each position is the value of the signal variance of the sensor corresponding to the abscissa of the corresponding position within the time window corresponding to the ordinate of the corresponding position; in the last signal picture of d1 × d2, the pixel value of each position is the average value of the signals of the sensor corresponding to the abscissa of the corresponding position in the time window corresponding to the ordinate of the corresponding position.
And S102, inputting the predicted sample characteristics into the trained signal processing model, and obtaining an output result of the signal processing model.
The obtained k signal pictures of d1 × d2 can be input to the signal processing model as the signals of d1 × d2 of k channels, that is, a three-dimensional signal picture of d1 × d2 × k is input to the signal processing model. In this embodiment, each prediction sample feature further corresponds to one label, and the time period corresponding to the prediction sample feature may be used for identification. If it is predicted whether the machine part near the certain position at the time t is in danger of overheating, the time series signals of a plurality of sensors labeled near the certain position at the time t need to be collected first and processed as prediction sample data to obtain the prediction sample characteristics at the time t, and according to the above-mentioned manner of the embodiment, the prediction sample characteristics of d1 × d2 of one k channels are obtained. Finally, the predicted sample characteristics of d1 x d2 of k channels are input into a pre-trained signal processing model, and the signal processing model can input whether the machine part near a certain position at the time t is in danger of overheating or not, for example, the output value can be a probability value between 0 and 1, and for example, the closer to 1, the more dangerous, the closer to 0 and the safer, and vice versa, the output value can be set. In this case, the signal processing model of the present embodiment is trained in advance to be a detection model of overheating danger. The signal processing model of this embodiment may be trained by machine learning based on a Convolutional Neural Network (CNN) model.
In this embodiment, under different usage scenarios, the number of k channels and the specific statistics may be different, and the k channels and the specific statistics are configured in advance according to the requirements of the scenarios.
Compared with the prior art, in the technical scheme of the embodiment, the time sequence of each sensor is considered, and the spatial relationship of the sensors is also considered, so that the spatial relationship of the sensors is effectively reserved, and the accuracy and the processing efficiency of time sequence signal processing can be effectively improved.
FIG. 2 is a flowchart illustrating a method for training a signal processing model according to an embodiment of the present invention. As shown in fig. 2, the training method of the signal processing model of this embodiment may specifically include the following steps:
s200, generating a training sample data set according to time sequence signals acquired by a plurality of sensors in a target environment in a historical manner and corresponding historical state information in the target environment;
the main execution body of the training method of the signal processing model of the present embodiment is a training device of the signal processing model, and the training device of the signal processing model is used for training the signal processing model of the embodiment shown in fig. 1.
First, the generation process of the training sample data set in step S200 may specifically include the following steps: cutting time sequence signals historically acquired by a plurality of sensors into a plurality of signal segments according to a preset time length; for each signal segment, time sequence signals of a plurality of sensors in the signal segment are arranged according to a spatial sequence configured for the plurality of sensors in advance, a two-dimensional signal array corresponding to time and space is generated to be used as a sample, a historical state in the corresponding time of the signal segment is used as a mark, and a piece of training sample data is obtained and put into a training sample data set.
Each piece of training sample data in the training sample data set of this embodiment is different from the prediction sample data of the embodiment shown in fig. 1, and further includes a corresponding historical state, so that the signal processing model is trained based on the two-dimensional signal array and the corresponding historical state in each piece of training sample data.
The historical state of the present embodiment may differ based on the signal processing model to be trained. For example, if the signal processing model of the present embodiment is an abnormality detection model of a time sequence signal, the corresponding historical state may be whether the state at the corresponding time is abnormal. If the signal processing model is a classification model of a time sequence signal, the corresponding historical state may be a classification to which the corresponding time belongs. If the signal processing model is a numerical prediction model of a time sequence signal, the corresponding historical state is a historical numerical value of the corresponding moment. In practical applications, the signal processing model of this embodiment may also be a processing model of other time sequence signals, and the corresponding historical states may also be other parameters, which are not described in detail herein for example.
In addition, the two-dimensional signal array in the training sample data of this embodiment is generated in the same manner as the two-dimensional signal array in the prediction sample data of the embodiment shown in fig. 1, and reference may be made to the related description of the embodiment shown in fig. 1 for details, which is not repeated herein.
S201, performing feature extraction processing on a training sample data set to obtain a training sample feature set;
for example, when the step S201 is implemented, the method may specifically include: for each training sample data in the training sample data set, counting the values of k statistics of each time window in d2 adjacent time windows of each sensor in d1 adjacent sensors in the two-dimensional signal array, and taking the values of the statistics as pixel values to obtain k signal pictures of d1 × d2 as training sample features, wherein d1, d2 and k are positive integers larger than 1.
In this embodiment, based on each piece of training sample data, a corresponding training sample feature is extracted, and the process of obtaining the feature of the prediction sample is completely the same as that of performing the feature extraction processing on the prediction sample data in step S101 in the embodiment shown in fig. 1, and reference may be made to the related description of the embodiment shown in fig. 1 in detail, which is not repeated herein.
Similarly, the statistics of the present embodiment may include signal range, signal variance, signal mean, or signal maximum difference.
Similarly, if the embodiment trains a signal processing model for predicting whether it will rain in the future, the plurality of sensors in the corresponding target environment may include at least two of a humidity sensor, a temperature sensor, a wind speed sensor, and a pressure sensor, and the output result of the signal processing model is an environment detection result.
Alternatively, if the present embodiment trains a signal processing model that predicts whether a given machine is running too hot, the sensors in the target environment may include temperature sensors for the temperature of various components of the given machine. At this time, the output result of the signal processing model is a judgment result as to whether or not the specified machine is running overheated.
Similarly, the signal processing model of the present embodiment may also be another model for performing time sequence signal processing, and details are not repeated herein.
S202, training a signal processing model based on the training sample feature set, the collected historical state information and a machine learning algorithm.
Through the processing of the steps, a training sample feature set can be obtained, and the training sample feature set can comprise a plurality of training sample features. In addition, each training sample feature corresponds to a segment of signal corresponding to a plurality of sensors in the target environment, and each segment of signal corresponds to one piece of historical state information, so each training sample feature can correspond to one piece of historical state information. Before training, each training sample characteristic and one piece of historical state information corresponding to the training sample characteristic are used as one piece of training data. Before training, an untrained signal processing model, such as a CNN model, may be obtained, and then initial values are assigned to parameters of the model. During training, a piece of training data is taken, the characteristics of a training sample in the training data are input into the signal processing model, and the signal processing model outputs a predicted state. And then judging whether the predicted state is consistent with the acquired historical state, if not, adjusting parameters of the signal processing model to enable the predicted state to be consistent with the acquired historical state. And continuously training the signal processing model by adopting all training sample characteristics in the collected training sample characteristic set and corresponding historical state information according to the training mode until the predicted state is consistent with the collected historical state, finishing training, and determining parameters of the signal processing model so as to determine the signal processing model.
The more training sample data in the training sample data set is collected in this embodiment, the more accurate the trained signal processing model is. For example, the number of training sample data in a training sample data set collected in practical applications may reach over a million orders.
The training process of the signal processing model of this embodiment is similar to the principle of the processing process of the time sequence signal in the embodiment shown in fig. 1, and reference may be made to the description of the relevant steps in the embodiment shown in fig. 1 for details, which are not repeated herein.
The CNN model used in the signal processing model of this embodiment may formulate different learning targets according to different problem definitions after taking the two-dimensional timing signal as input, so as to solve the problems of two-classification, multi-classification, regression, and the like. For the classification problem, the result output by the corresponding model is the probability of each class; for the regression problem, the model result is the predicted value. Because the CNN model structure has good learning and anti-overfitting capabilities on the data of the picture type, the CNN model is used as a prototype in the invention to train a corresponding signal processing model.
During specific training, after a picture sample, namely a training sample characteristic, is constructed, signal pictures of a plurality of channels are input to the CNN model for training.
For example, when the input sample dimension is 16 and the number of channels is 2, the corresponding frame structure of the CNN model may be as shown in fig. 3 when solving the three-classification problem.
As can be seen from the model structure shown in fig. 3, the main components of the model at this time include a 2-dimensional convolution layer (conv2D), an Activation layer (Activation), a batch normalization layer (batch normalization), a pooling layer (MaxPooling), and a DropOut layer. The computational logic and functions of each component are as follows:
2-dimensional convolutional layer: dividing the numerical value in the input picture sample into small windows, and performing convolution operation, wherein the convolution formula is as follows:
the convolution operation can extract local information in the image through parameter sharing and small window operation, and the local perception capability is enhanced.
An active layer: the active layer is used for mapping the output result of the network calculation to another function, so that the method is more beneficial to model learning, and the functions such as Sigmoid, Tanh, ReLU and the like are common. Fig. 4 is a schematic diagram of two functions Sigmoid and Tanh provided in this embodiment.
Batch normalization layer: the batch normalization calculation logic is to perform one-time normalization of standard normal distribution on each batch of samples entering the model training, so that the overfitting is resisted, the training speed is improved, and the requirement on initialization data is reduced. The code logic of the corresponding batch normalization layer may be as shown in FIG. 5.
A pooling layer: like convolution, Pooling also has a sliding kernel, which may be referred to as a sliding window, and in fig. 6, the size of the sliding window is 2 × 2, and the step is 2, and the maximum value is taken as output every time an area is slid, which is called Max Pooling. The purpose is to abstract some global features, reduce the dimensionality of data and accelerate calculation.
DropOut layer: during training, some network neurons are abandoned, and the robustness of the model is enhanced by a random abandon method in an attempt to resist overfitting. On the right of fig. 7 is a comparison of the DropOut network with the previous full connection.
According to the training method of the signal processing model, through the scheme, an accurate signal processing model applied to a target environment with a plurality of sensors can be trained, and then time sequence signals collected by the plurality of sensors in the target environment can be processed based on the signal processing model.
Fig. 8 is a block diagram of an embodiment of a timing signal processing apparatus according to the present invention. As shown in fig. 8, the apparatus for processing a timing signal according to this embodiment may specifically include:
the generation module 10 is configured to generate prediction sample data according to the timing signals acquired by the plurality of sensors in real time;
the extraction module 11 is configured to perform feature extraction processing on the prediction sample data generated by the generation module 10 to obtain prediction sample features;
the processing module 12 is configured to input the prediction sample features extracted by the extraction module 11 into the trained signal processing model, and obtain an output result of the signal processing model;
further optionally, wherein the generating module 10 is configured to:
and arranging the signals acquired by the plurality of sensors in real time within each preset time length according to a spatial sequence preset for the plurality of sensors to generate a two-dimensional signal array corresponding to time and space to serve as a piece of prediction sample data.
Further optionally, wherein the extracting module 11 is configured to:
and (3) counting the values of k statistics of each time window in d2 adjacent time windows of each sensor in d1 adjacent sensors in the two-dimensional signal array, and taking the values of the statistics as pixel values to obtain k signal pictures of d1 × d2 as predicted sample characteristics, wherein d1, d2 and k are positive integers larger than 1.
Further optionally, wherein the statistics comprise signal range, signal variance, signal mean, or signal maximum difference.
Further optionally, wherein the processing of the timing signal comprises: abnormality detection of a timing signal, classification of a timing signal, or numerical prediction of a timing signal.
Further optionally, wherein the plurality of sensors includes at least two of a humidity sensor, a temperature sensor, a wind speed sensor, and a barometric pressure sensor, the output result of the signal processing model is an environmental detection result;
alternatively, the plurality of sensors are a plurality of temperature sensors for detecting the temperatures of the respective components of the designated machine, respectively, and the output result of the signal processing model is a result of determination as to whether or not the designated machine is running overheated.
The processing apparatus for timing signals in this embodiment implements processing of timing signals by using the modules, and has the same implementation principle and technical effect as those of the related method embodiments, and reference may be made to the related descriptions of the method embodiments in detail, which are not repeated herein.
FIG. 9 is a block diagram of an embodiment of a training apparatus for a signal processing model according to the present invention. As shown in fig. 9, the training apparatus of the signal processing model in this embodiment may specifically include:
the generating module 20 is configured to generate a training sample data set according to time sequence signals historically acquired by a plurality of sensors in a target environment and corresponding historical state information in the target environment;
the extraction module 21 is configured to perform feature extraction processing on the training sample data set generated by the generation module 20 to obtain a training sample feature set;
the training module 22 is configured to train a signal processing model based on the training sample feature set extracted by the extraction module 21, the historical state information collected in the training sample data set generated by the generation module 20, and a machine learning algorithm.
Further optionally, wherein the generating module 20 is configured to:
cutting time sequence signals historically acquired by a plurality of sensors into a plurality of signal segments according to a preset time length;
for each signal segment, time sequence signals of a plurality of sensors in the signal segment are arranged according to a spatial sequence configured for the plurality of sensors in advance, a two-dimensional signal array corresponding to time and space is generated to be used as a sample, a historical state in the corresponding time of the signal segment is used as a mark, and a piece of training sample data is obtained and put into a training sample data set.
Further optionally, wherein the extracting module 21 is configured to:
for each training sample data in the training sample data set, counting the values of k statistics of each time window in d2 adjacent time windows of each sensor in d1 adjacent sensors in the two-dimensional signal array, and taking the values of the statistics as pixel values to obtain k signal pictures of d1 × d2 as training sample features, wherein d1, d2 and k are positive integers larger than 1.
Further optionally, wherein the statistics comprise signal range, signal variance, signal mean, or signal maximum difference.
Further optionally, wherein the signal processing model comprises: an abnormality detection model for a time series signal, a classification model for a time series signal, or a numerical prediction model for a time series signal.
Further optionally, wherein the plurality of sensors includes at least two of a humidity sensor, a temperature sensor, a wind speed sensor, and a barometric pressure sensor, the output result of the signal processing model is an environmental detection result;
alternatively, the plurality of sensors may be a plurality of temperature sensors for detecting the temperatures of the respective components of the designated machine, respectively, and the output result of the signal processing model may be a result of determination as to whether or not the designated machine is overheated.
The training device for the signal processing model of the above embodiment implements training of the signal processing model by using the above module, and has the same implementation principle and technical effect as those of the above related method embodiment, and reference may be made to the related description of the above method embodiment in detail, which is not repeated herein.
FIG. 10 shows a schematic structural diagram of a computing device that can be used to implement the above method according to an embodiment of the invention. For example, the computing device may be used to implement the above-described timing signal processing method or training method of a signal processing model.
Referring to fig. 10, the computing device 1000 includes a memory 1010 and a processor 1020.
The processor 1020 may be a multi-core processor or may include multiple processors. In some embodiments, processor 1020 may include a general-purpose host processor and one or more special purpose coprocessors such as a Graphics Processor (GPU), Digital Signal Processor (DSP), or the like. In some embodiments, processor 1020 may be implemented using custom circuits, such as an Application Specific Integrated Circuit (ASIC) or a Field Programmable Gate Array (FPGA).
The memory 1010 may include various types of storage units, such as system memory, Read Only Memory (ROM), and permanent storage. Wherein the ROM may store static data or instructions that are needed by the processor 1020 or other modules of the computer. The persistent storage device may be a read-write storage device. The persistent storage may be a non-volatile storage device that does not lose stored instructions and data even after the computer is powered off. In some embodiments, the persistent storage device employs a mass storage device (e.g., magnetic or optical disk, flash memory) as the persistent storage device. In other embodiments, the permanent storage may be a removable storage device (e.g., floppy disk, optical drive). The system memory may be a read-write memory device or a volatile read-write memory device, such as a dynamic random access memory. The system memory may store instructions and data that some or all of the processors require at runtime. Further, the memory 1010 may comprise any combination of computer-readable storage media, including various types of semiconductor memory chips (DRAM, SRAM, SDRAM, flash, programmable read only memory), magnetic and/or optical disks may also be employed. In some embodiments, memory 1010 may include a removable storage device that is readable and/or writable, such as a Compact Disc (CD), a read-only digital versatile disc (e.g., DVD-ROM, dual layer DVD-ROM), a read-only Blu-ray disc, an ultra-density optical disc, a flash memory card (e.g., SD card, min SD card, Micro-SD card, etc.), a magnetic floppy disc, or the like. Computer-readable storage media do not contain carrier waves or transitory electronic signals transmitted by wireless or wired means.
The memory 1010 stores executable code, which when processed by the processor 1020, causes the processor 1020 to perform the above-mentioned method for processing a time-series signal or training a signal processing model.
The processing method of a time series signal or the training method of a signal processing model according to the present invention has been described in detail above with reference to the accompanying drawings.
Furthermore, the method according to the invention may also be implemented as a computer program or computer program product comprising computer program code instructions for carrying out the above-mentioned steps defined in the above-mentioned method of the invention.
Alternatively, the invention may also be embodied as a non-transitory machine-readable storage medium (or computer-readable storage medium, or machine-readable storage medium) having stored thereon executable code (or a computer program, or computer instruction code) which, when executed by a processor of an electronic device (or computing device, server, etc.), causes the processor to perform the steps of the above-described method according to the invention.
Those of skill would further appreciate that the various illustrative logical blocks, modules, circuits, and algorithm steps described in connection with the disclosure herein may be implemented as electronic hardware, computer software, or combinations of both.
The flowchart and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of systems and methods according to various embodiments of the present invention. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that, in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and/or flowchart illustration, and combinations of blocks in the block diagrams and/or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.
Having described embodiments of the present invention, the foregoing description is intended to be exemplary, not exhaustive, and not limited to the embodiments disclosed. Many modifications and variations will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described embodiments. The terminology used herein is chosen in order to best explain the principles of the embodiments, the practical application, or improvements made to the technology in the marketplace, or to enable others of ordinary skill in the art to understand the embodiments disclosed herein.
Claims (10)
1. A method for processing a timing signal, comprising:
generating prediction sample data according to time sequence signals acquired by a plurality of sensors in real time;
performing feature extraction processing on the prediction sample data to obtain prediction sample features;
and inputting the predicted sample characteristics into a trained signal processing model, and obtaining an output result of the signal processing model.
2. The method of claim 1, wherein said generating prediction sample data from time-series signals acquired in real-time by a plurality of sensors comprises:
and arranging the signals acquired by the sensors in real time within each preset time length according to a spatial sequence preset for the sensors, and generating a two-dimensional signal array corresponding to time and space to serve as a piece of prediction sample data.
3. The method according to claim 2, wherein performing feature extraction processing on the prediction sample data to obtain prediction sample features comprises:
counting the values of k statistics of each time window of d2 adjacent time windows of each sensor in d1 adjacent sensors in the two-dimensional signal array, and taking the values of the statistics as pixel values to obtain k signal pictures of d1 × d2 as the predicted sample features, wherein d1, d2 and k are positive integers larger than 1.
4. The method of claim 3, wherein the statistics comprise signal range, signal variance, signal mean, or signal maximum difference.
5. The method of claim 1, wherein the processing of the timing signal comprises: abnormality detection of a timing signal, classification of a timing signal, or numerical prediction of a timing signal.
6. A method of training a signal processing model, comprising:
generating a training sample data set according to time sequence signals acquired by a plurality of sensors in a target environment in a historical manner and corresponding historical state information in the target environment;
performing feature extraction processing on the training sample data set to obtain a training sample feature set;
and training a signal processing model based on the training sample feature set, the collected historical state information and a machine learning algorithm.
7. An apparatus for processing a timing signal, comprising:
the generating module is used for generating prediction sample data according to the time sequence signals acquired by the sensors in real time;
the extraction module is used for carrying out feature extraction processing on the prediction sample data to obtain the features of the prediction sample;
and the processing module is used for inputting the predicted sample characteristics to the trained signal processing model and acquiring the output result of the signal processing model.
8. A training apparatus for a signal processing model, comprising:
the generating module is used for generating a training sample data set according to time sequence signals acquired by a plurality of sensors in a target environment and corresponding historical state information in the target environment;
the extraction module is used for carrying out feature extraction processing on the training sample data set to obtain a training sample feature set;
and the training module is used for training a signal processing model based on the training sample characteristic set, the collected historical state information and a machine learning algorithm.
9. A computing device, comprising:
a processor; and
a memory having executable code stored thereon, which when executed by the processor, causes the processor to perform the method of any one of claims 1-5; or to perform the method of claim 6.
10. A non-transitory machine-readable storage medium having stored thereon executable code that, when executed by a processor of an electronic device, causes the processor to perform the method of any one of claims 1-5; or to perform the method of claim 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210702746.1A CN115099272A (en) | 2019-09-05 | 2019-09-05 | Method and device for processing time sequence signal, equipment and readable medium |
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910838480.1A CN110555413B (en) | 2019-09-05 | 2019-09-05 | Method and device for processing time sequence signal, equipment and readable medium |
| CN202210702746.1A CN115099272A (en) | 2019-09-05 | 2019-09-05 | Method and device for processing time sequence signal, equipment and readable medium |
Related Parent Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910838480.1A Division CN110555413B (en) | 2019-09-05 | 2019-09-05 | Method and device for processing time sequence signal, equipment and readable medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115099272A true CN115099272A (en) | 2022-09-23 |
Family
ID=68739102
Family Applications (2)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910838480.1A Active CN110555413B (en) | 2019-09-05 | 2019-09-05 | Method and device for processing time sequence signal, equipment and readable medium |
| CN202210702746.1A Pending CN115099272A (en) | 2019-09-05 | 2019-09-05 | Method and device for processing time sequence signal, equipment and readable medium |
Family Applications Before (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910838480.1A Active CN110555413B (en) | 2019-09-05 | 2019-09-05 | Method and device for processing time sequence signal, equipment and readable medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (2) | CN110555413B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112016760A (en) * | 2020-09-04 | 2020-12-01 | 平安科技（深圳）有限公司 | Combined prediction method and device of multi-dimensional information, computer equipment and medium |
| CN113095258A (en) * | 2021-04-20 | 2021-07-09 | 深圳力维智联技术有限公司 | Directional signal extraction method, system, device and storage medium |
Family Cites Families (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| DE112013004129T5 (en) * | 2012-08-22 | 2015-05-21 | General Electric Company | Wireless system and method for measuring an operating condition of a machine |
| CN108256626A (en) * | 2016-12-28 | 2018-07-06 | 中国科学院深圳先进技术研究院 | The Forecasting Methodology and device of time series |
| US11080533B2 (en) * | 2017-05-01 | 2021-08-03 | Sensormatic Electronics, LLC | Surveillance system with human behavior prediction by human action recognition |
| CN108730776B (en) * | 2018-06-22 | 2020-07-14 | 大连理工大学 | Oil pipeline leakage detection method based on extreme learning machine information fusion |
| CN109522600B (en) * | 2018-10-16 | 2020-10-16 | 浙江大学 | Complex equipment residual service life prediction method based on combined deep neural network |
| CN109800627B (en) * | 2018-12-03 | 2020-06-09 | 第四范式（北京）技术有限公司 | Petroleum pipeline signal abnormity detection method and device, equipment and readable medium |
| CN110084286A (en) * | 2019-04-10 | 2019-08-02 | 武汉理工大学 | A kind of human motion recognition method of sensor-based ECOC technology |
-
2019
- 2019-09-05 CN CN201910838480.1A patent/CN110555413B/en active Active
- 2019-09-05 CN CN202210702746.1A patent/CN115099272A/en active Pending
Also Published As
| Publication number | Publication date |
|---|---|
| CN110555413B (en) | 2022-05-31 |
| CN110555413A (en) | 2019-12-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| EP3410351B1 (en) | Learning program, learning method, and object detection device | |
| CN106529511B (en) | image structuring method and device | |
| CN111104925B (en) | Image processing method, image processing apparatus, storage medium, and electronic device | |
| KR20140028809A (en) | Adaptive image processing apparatus and method in image pyramid | |
| Wang et al. | A novel fire detection approach based on CNN-SVM using tensorflow | |
| CN112906647B (en) | Method and device for monitoring load of small-span bridge, computer equipment and storage medium | |
| CN109829371B (en) | Face detection method and device | |
| CN110555413B (en) | Method and device for processing time sequence signal, equipment and readable medium | |
| CN109977848B (en) | Training method and device for pornography detection model, computer equipment and readable medium | |
| CN113343985B (en) | License plate recognition method and device | |
| CN112912888A (en) | Apparatus and method for identifying video activity | |
| CN116453067B (en) | Sprinting timing method based on dynamic visual identification | |
| CN110866428B (en) | Target tracking method, device, electronic equipment and storage medium | |
| CN112541394A (en) | Black eye and rhinitis identification method, system and computer medium | |
| CN111753658A (en) | Post sleep warning method and device and computer equipment | |
| US20200151492A1 (en) | Feature determination apparatus and method adapted to multiple object sizes | |
| CN114140663A (en) | Multi-scale attention and learning network-based pest identification method and system | |
| CN111738194B (en) | Method and device for evaluating similarity of face images | |
| CN117095198A (en) | Remote sensing image retrieval network training method, application method, electronic equipment and medium | |
| CN115984671A (en) | Model online updating method and device, electronic equipment and readable storage medium | |
| EP4332910A1 (en) | Behavior detection method, electronic device, and computer readable storage medium | |
| Sladojević et al. | Integer arithmetic approximation of the HoG algorithm used for pedestrian detection | |
| CN115774784A (en) | Text object identification method and device | |
| CN117083621A (en) | Detector training method, device and storage medium | |
| JP7383435B2 (en) | Image processing device, image processing method, and program |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |