Time sequence analysis method, computer equipment and storage medium
The eigenfeatures of the time series are extracted through the morphological entropy calculation method, which solves the parameter dependence problem of traditional methods when dealing with complex time series, achieves the improvement of stability and reliability, and conducts efficient analysis in a real-time environment.
Patent Information
- Application Number
- CN202411334528.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-24
- Publication Date
- 2024-12-20
AI Technical Summary
Traditional time series analysis methods have significant limitations in processing nonlinear and high complexity time series data, and are highly dependent on parameter selection, resulting in uncertainty and complexity of analysis results.
The morphological entropy calculation method is adopted, and by standardizing the time series signals and dividing them into bandwidth, stating the fluctuation distribution probability and phase bias of the sampling point set, morphological entropy is defined to extract the eigenfeatures of complex time series and reduce the dependence of parameter settings.
It improves the stability and reliability of the analysis results, realizes a comprehensive analysis of time series data under low computational complexity, captures the intrinsic morphological characteristics of the time series, and is suitable for real-time and online environments, with high efficiency and robustness.
Smart Images
Figure CN119167064A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of time series analysis, and in particular to a time series analysis method, a computer device and a storage medium. Background Art
[0002] Time series analysis has important application value in many fields, such as financial market forecasting, acoustic signal processing, industrial and mechanical data and physiological signal analysis. Time series data records the information of system status changes over time, and usually has the characteristics of non-stationarity, noise and complexity. Therefore, accurate and efficient analysis of it is a challenging task.
[0003] Traditional time series analysis methods mainly rely on statistical models, such as autoregressive models, moving average models, and autoregressive integrated moving average models. These methods are effective in capturing the basic characteristics of time series, but have significant limitations when dealing with nonlinear and highly complex time series data. In addition, these methods are highly dependent on parameter selection, and different parameter settings may lead to significant differences in analysis results, increasing the complexity and uncertainty of the application. Summary of the invention
[0004] The purpose of the present invention is to provide a time series analysis method, a computer device and a storage medium, which utilize morphological entropy calculation to realize the extraction of intrinsic features of complex time series, greatly reducing the dependence on parameter settings, which is conducive to improving the stability and reliability of the analysis results.
[0005] The technical solution provided by the present invention is as follows:
[0006] In a first aspect, the present invention provides a time series analysis method, comprising the steps of:
[0007] Get non-stationary time series signals of preset length;
[0008] The amplitude distribution of the non-stationary time series signal is limited to a range of 0 to 1 by a standardization technique to obtain a normalized time series signal;
[0009] Dividing the normalized time series signal into a plurality of bandwidths, and counting the sampling point sets within each of the bandwidths;
[0010] Calculating the fluctuation distribution probability corresponding to each of the sampling point sets, and calculating the entropy description of the fluctuation pattern based on the time series morphology based on the fluctuation distribution probability;
[0011] Taking the bandwidth corresponding to the sampling point set with the most sampling points as the baseline bandwidth, and defining a wavelet set based on the baseline bandwidth, wherein the wavelet set is at least three sampling point sets that are on the same side of the baseline bandwidth and are continuous;
[0012] Calculating the phase offset corresponding to the wavelet set to form an entropy description of the phase pattern based on the time series morphology;
[0013] The morphological entropy is defined according to the entropy description of the fluctuation pattern based on the time series morphology and the entropy description of the phase pattern based on the time series morphology, and the morphological entropy value corresponding to the non-stationary time series signal is calculated.
[0014] The above implementation method normalizes the time series signal and divides the time series signal into several bandwidths, and counts the sampling point sets within each bandwidth, so as to calculate the fluctuation distribution probability corresponding to each sampling point set, and calculate the entropy description of the fluctuation pattern based on the time series morphology based on the fluctuation distribution probability; at the same time, a baseline bandwidth is defined, and a wavelet set is defined based on the baseline bandwidth, so as to calculate the phase offset corresponding to each wavelet set, thereby forming an entropy description of the phase pattern based on the time series morphology; and then, according to the entropy description of the fluctuation pattern based on the time series morphology and the entropy description of the phase pattern based on the time series morphology, the morphological entropy can be defined, so as to calculate the morphological entropy value corresponding to the non-stationary time series signal. The use of morphological entropy calculation to extract the intrinsic features of complex time series greatly reduces the dependence on parameter settings, which is beneficial to improving the stability and reliability of the analysis results; at the same time, by standardizing the time series signals and extracting morphological features, it is possible to achieve a comprehensive analysis of time series data under the premise of low computational complexity, and effectively mine the potential information of time series at different scales and frequency domains, thereby providing a more accurate feature representation; and the use of morphological entropy calculation methods can efficiently capture the intrinsic morphological features of time series without the need for complex parameter settings, with high computational efficiency and robustness, which is conducive to efficient time series analysis in real-time and online environments.
[0015] In some embodiments, the fluctuation distribution probability P k The calculation formula is:
[0016]
[0017] Among them, m is the number of divided bandwidths, B k For each set of sampling points of the bandwidth, B k ={δ1,δ2,...,δ ε}, ε∈[0,N], δ represents a single sampling point, k represents the bandwidth index, and satisfies k∈[1,m];
[0018] The calculation formula of Amplitude (m), which describes the entropy of the fluctuation pattern based on the time series morphology, is:
[0019]
[0020] in, is the midline amplitude value of the kth bandwidth; is used to uniformly scale and quantize the entropy value. In some embodiments, the calculation formula of the baseline bandwidth Ψ is:
[0021] ψ=argmax ε (B k ),k∈[1,m]
[0022] Among them, Ψ satisfies Ψ upper -Ψ lower =1 / m,Ψ upper and lower Respectively represent the upper and lower bounds of the baseline bandwidth.
[0023] In some embodiments, the wavelet set W s ={δ1,δ2,...,δ ξ},ξ∈[3,N];
[0024] The calculating of the phase offset corresponding to the wavelet set to form an entropy description of the phase pattern based on the time series morphology specifically includes:
[0025] The amplitude of the sampling point set is obtained by subtracting (Ψ upper +Ψ lower ) / 2 to get the relative amplitude;
[0026] Determine whether the wavelet set is located below the baseline bandwidth, and if so, flip the wavelet set so that the relative amplitude calculated by the sampling points included in the wavelet set is positive;
[0027] Define the overall energy position θ for each of the wavelet sets s ,θ s The wavelet set W is defined as s The time coordinate of the area bisector;
[0028] Then, the mapping function f(x; α) is used to generate each wavelet set W s Assigning Phase Offset
[0029]
[0030] Among them, Φ s is the wavelet set W s The average amplitude of the overall energy position θ s The importance of is scaled, N is the total length of the time series, and α represents the parameter configuration of the mapping function;
[0031] According to the phase offset of each of the sub-wave sets, an entropy description Phase(m) of the phase pattern based on the time series morphology is formed,
[0032]
[0033] Among them, P s is the distribution probability corresponding to the s-th wavelet set, which is the ratio of the number of sampling points of the wavelet set to the total length N of the time series; It is used to uniformly quantify entropy values.
[0034] In some embodiments, the morphological entropy ME(m) is calculated as follows:
[0035]
[0036] In some embodiments, the non-stationary time series signal includes any one or more of: physiological signals, industrial and mechanical signals, environmental and meteorological data signals, financial and economic data signals, audio and acoustic signals.
[0037] This scheme is applicable to various types of time series data, including finance, acoustics, industry, physiological signals and other fields. It provides broad application prospects and practical significance, and helps to achieve efficient and accurate quantification of time series complexity in various practical application scenarios.
[0038] In some embodiments, after obtaining a non-stationary time series signal of a preset length, the method further includes:
[0039] A filtering algorithm is designed according to the signal sampling rate and the frequency range of the analyzable signal in the non-stationary time series signal, and the non-stationary time series signal is filtered by the filtering algorithm.
[0040] In some embodiments, the normalization technique is one or more of maximum-minimum normalization and normal cumulative distribution function normalization.
[0041] In a second aspect, the present application provides a computer device, comprising a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the time series analysis method of the first aspect.
[0042] In a third aspect, the present application provides a computer storage medium having a computer program or instructions stored thereon, which, when executed by a processor, implements the steps of the time series analysis method of the first aspect.
[0043] The time series analysis method, computer device and storage medium provided by the present invention have at least the following beneficial effects:
[0044] 1) This application uses morphological entropy calculation to extract the intrinsic features of complex time series, which greatly reduces the dependence on parameter settings and is conducive to improving the stability and reliability of analysis results;
[0045] 2) This application can achieve comprehensive analysis of time series data at a lower computational complexity by standardizing time series signals and extracting morphological features, effectively mining the potential information of time series at different scales and frequency domains, and thus providing more accurate feature representation;
[0046] 3) This application uses the morphological entropy calculation method to efficiently capture the intrinsic morphological characteristics of the time series without complex parameter settings, has high computational efficiency and robustness, and is conducive to efficient time series analysis in real-time and online environments;
[0047] 4) This scheme is applicable to various types of time series data, including finance, acoustics, industry, physiological signals and other fields, providing broad application prospects and practical significance, and helping to achieve efficient and accurate quantification of time series complexity in various practical application scenarios;
[0048] 5) This application uses the morphological entropy method to measure the complexity and uncertainty of time series, which can provide important analytical information, facilitate further time series modeling and prediction, and improve the accuracy and effectiveness of the overall analysis;
[0049] 6) This scheme has the advantages of high efficiency, no parameters, and wide applicability. It can play an important role in the processing of various complex and non-stationary time series data. It provides a practical and efficient time series signal processing tool, which helps to promote technological progress and application development in related fields. BRIEF DESCRIPTION OF THE DRAWINGS
[0050] The preferred implementation scheme will be described below in a clear and understandable manner with reference to the accompanying drawings to further illustrate the above-mentioned characteristics, technical features, advantages and implementation methods of the present solution.
[0051] Figure 1 It is a schematic diagram of the overall process of an embodiment of the present invention;
[0052] Figure 2 It is a schematic diagram of the effect of morphological entropy on the quantification of entropy estimation value under different parameter settings;
[0053] Figure 3 It is a schematic diagram comparing the processing time of permutation entropy and morphological entropy for time series of different lengths;
[0054] Figure 4 Schematic diagram of quantitative estimation value of morphological entropy for real-time detection of arrhythmia in an embodiment of the present invention;
[0055] Figure 5 It is a schematic diagram of the quantitative estimation value of the morphological entropy real-time capture of poor quality areas in an embodiment of the present invention. DETAILED DESCRIPTION
[0056] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the specific implementation methods of the present invention will be described below with reference to the accompanying drawings. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings and other implementation methods can be obtained based on these drawings without creative work.
[0057] In order to simplify the drawings, only the parts related to the present invention are schematically shown in each figure, and they do not represent the actual structure of the product. In addition, in order to simplify the drawings and facilitate understanding, in some figures, only one of the parts with the same structure or function is schematically drawn or marked. In this article, "one" not only means "only one", but also means "more than one".
[0058] Time series analysis has important application value in many fields, such as financial market forecasting, acoustic signal processing, industrial and mechanical data and physiological signal analysis. Time series data records the information of system status changes over time, and usually has the characteristics of non-stationarity, noise and complexity. Therefore, it is a challenging task to analyze it accurately and efficiently.
[0059] Traditional time series analysis methods mainly rely on statistical models, such as autoregressive models, moving average models, and autoregressive integrated moving average models. These methods are effective in capturing the basic characteristics of time series, but have significant limitations when dealing with nonlinear and highly complex time series data. In addition, these methods are highly dependent on parameter selection, and different parameter settings may lead to significant differences in analysis results, increasing the complexity and uncertainty of the application.
[0060] In recent years, the rapid development of machine learning technology has provided new methods and tools for time series analysis. Deep learning models such as convolutional neural networks, recurrent neural networks, and long short-term memory networks have performed well in time series feature extraction and prediction. However, these models usually have a large number of parameters and high computational costs, and are highly sensitive to hyperparameter settings, which brings many challenges in practical applications. In addition, traditional deep learning models have certain limitations when dealing with long-range dependencies in time series data, which affects the accuracy and stability of predictions.
[0061] Therefore, existing time series analysis methods usually require a lot of computing resources and time, and are difficult to apply efficiently in real-time and online environments. Especially for complex and non-stationary time series data, it is often difficult to strike a balance between ensuring computing efficiency and analysis accuracy. With the increase in data volume and the complexity of application scenarios, it is particularly important to develop an efficient, accurate and easy-to-implement time series analysis method. This scheme introduces morphological entropy and realizes the extraction of intrinsic features of complex time series based on morphological entropy calculation, thereby greatly reducing the dependence on parameter settings, which is conducive to improving the stability and reliability of the analysis results. The following is a detailed description of this scheme in conjunction with the accompanying drawings:
[0062] In one embodiment, referring to the accompanying drawings Figure 1 The present invention provides a time series analysis method, comprising the steps of:
[0063] S100, obtaining a non-stationary time series signal of a preset length;
[0064] S200, limiting the amplitude distribution of the non-stationary time series signal to a range of 0 to 1 through a standardization technique to obtain a normalized time series signal;
[0065] S300, dividing the normalized time series signal into a number of bandwidths, and counting the sampling point sets within each bandwidth;
[0066] S400, calculating the fluctuation distribution probability corresponding to each sampling point set, and calculating the entropy description of the fluctuation pattern based on the time series morphology based on the fluctuation distribution probability;
[0067] S500: taking the bandwidth corresponding to the sampling point set with the most sampling points as the baseline bandwidth, and defining a wavelet set based on the baseline bandwidth, wherein the wavelet set is a set of at least three sampling points that are on the same side of the baseline bandwidth and are continuous;
[0068] S600, calculating the phase offset corresponding to the wavelet set to form an entropy description of the phase pattern based on the time series morphology;
[0069] S700, defining morphological entropy according to the entropy description of the fluctuation pattern based on the time series morphology and the entropy description of the phase pattern based on the time series morphology, and calculating the morphological entropy value corresponding to the non-stationary time series signal.
[0070] In some specific implementations, a non-stationary time series signal with a sampling length of N is obtained by an acquisition device, X = {x1, ..., x i ,...,x N}. After obtaining a non-stationary time series signal of a preset length, it also includes: designing a filtering algorithm according to the signal sampling rate and the frequency range of the analyzable signal in the non-stationary time series signal, and filtering the non-stationary time series signal through the filtering algorithm. After obtaining a non-stationary time series signal, the amplitude distribution of the non-stationary time series signal is limited to the range of 0 to 1 through a standardization technique. The standardization technique is one or more of maximum and minimum value normalization and normal cumulative distribution function standardization, which are not limited here. For the obtained normalized sequence, set the standard bandwidth division parameters, and the specific parameters can be adjusted according to the actual scenario. The normalized amplitude domain is equally divided into m bandwidths, and the set of sampling points in each bandwidth is counted. When calculating the morphological entropy, it can be performed from two aspects: the entropy description of the fluctuation pattern based on the time series morphology and the entropy description of the phase pattern based on the time series morphology. By calculating the fluctuation distribution probability corresponding to each sampling point set, the entropy description of the fluctuation pattern based on the time series morphology can be calculated based on the fluctuation distribution probability; and through the sampling point set, the bandwidth corresponding to the sampling point set with the most sampling points is taken as the baseline bandwidth, and the sub-wave set is defined based on the baseline bandwidth, and the phase offset corresponding to the wavelet set is calculated, and the entropy description of the phase pattern based on the time series morphology can be further calculated; the morphological entropy can be defined through the entropy description of the fluctuation pattern based on the time series morphology and the entropy description of the phase pattern based on the time series morphology, thereby calculating the morphological entropy value corresponding to the non-stationary time series signal.
[0071] This scheme normalizes the time series signal and divides the time series signal into several bandwidths, and counts the sampling point sets within each bandwidth. It can calculate the fluctuation distribution probability corresponding to each sampling point set, and calculate the entropy description of the fluctuation pattern based on the time series morphology based on the fluctuation distribution probability; at the same time, define the baseline bandwidth, and define the sub-wave set based on the baseline bandwidth, and calculate the phase offset corresponding to each sub-wave set, thereby forming an entropy description of the phase pattern based on the time series morphology; and then define the morphological entropy based on the entropy description of the fluctuation pattern based on the time series morphology and the entropy description of the phase pattern based on the time series morphology, thereby calculating the morphological entropy value corresponding to the non-stationary time series signal. The use of morphological entropy calculation to extract the intrinsic features of complex time series greatly reduces the dependence on parameter settings, which is beneficial to improving the stability and reliability of the analysis results. At the same time, by standardizing the time series signals and extracting morphological features, it is possible to achieve a comprehensive analysis of the time series data under the premise of low computational complexity, and effectively mine the potential information of the time series at different scales and frequency domains, thereby providing a more accurate feature representation. The morphological entropy calculation method can efficiently capture the intrinsic morphological features of the time series without the need for complex parameter settings, and has high computational efficiency and robustness, which is conducive to efficient time series analysis in real-time and online environments.
[0072] In one embodiment, based on the above embodiment, the fluctuation distribution probability P k The calculation formula is:
[0073]
[0074] Among them, m is the number of divided bandwidths, B k For each bandwidth, B is the set of sampling points. k ={δ1,δ2,...,δ ε}, ε∈[0,N], δ represents a single sampling point, k represents the bandwidth index, and satisfies k∈[1,m].
[0075] The calculation formula of Amplitude (m), which describes the entropy of the fluctuation pattern based on the time series morphology, is:
[0076]
[0077] in, is the midline amplitude value of the kth bandwidth, for example, when m is equal to 4, =0.125; It is used to uniformly scale the entropy value.
[0078] In some implementations, based on the above embodiments, the calculation formula of the baseline bandwidth Ψ is:
[0079] ψ=argmax ε (B k ),k∈[1,m]
[0080] Among them, Ψ satisfies Ψ upper -Ψ lower =1 / m,Ψ upper and lower Represent the upper and lower bounds of the baseline bandwidth respectively.
[0081] Wavelet set W s ={δ1,δ2,...,δ ξ},ξ∈[3,N]; Calculate the phase offset corresponding to the wavelet set to form an entropy description of the phase pattern based on the time series morphology, including:
[0082] S610, the amplitude of the sampling point set is subtracted from (Ψ upper +Ψ lower ) / 2 to get the relative amplitude;
[0083] S620, determining whether the wavelet set is located below the baseline bandwidth, and if so, flipping the wavelet set so that the relative amplitude calculated by the sampling points included in the wavelet set is positive;
[0084] S630, defining the overall energy position θ for each wavelet set s ,θ s is defined as the wavelet set W s The time coordinate of the area bisector; θ s Calculated based on each wavelet set separately, not affected by the wavelet set W s Positional effects that appear on the time series;
[0085] S640, then use the mapping function f(x; α) to generate each wavelet set W s Assigning Phase Offset
[0086]
[0087] Among them, Φ s is the wavelet set W s The average amplitude of the overall energy position θ s The importance of is scaled, N is the total length of the time series, α represents the parameter configuration of the mapping function, and the linear form f(x; α) = 10x is used here;
[0088] S650, according to the phase offset of each sub-wave set, an entropy description Phase(m) of the phase pattern based on the time series morphology is formed,
[0089]
[0090] Among them, P s is the distribution probability corresponding to the s-th wavelet set, which is the ratio of the number of sampling points of the wavelet set to the total length N of the time series; It is used to uniformly quantify entropy values.
[0091] In one embodiment, based on the above embodiment, the calculation formula of morphological entropy ME(m) is:
[0092]
[0093] In this calculation formula, the differences between the fluctuation pattern entropy and phase pattern entropy estimates of different samples are amplified by the square operation, thereby enhancing the detectability of outliers, while the average operation allows the two indicators to participate in morphological measurement simultaneously.
[0094] In one embodiment, based on the foregoing embodiment, the non-stationary time series signal includes any one or more of physiological signals, industrial and mechanical signals, environmental and meteorological data signals, financial and economic data signals, audio and acoustic signals.
[0095] This solution is applicable to various types of time series data, including finance, acoustics, industry, physiological signals and other fields. It provides broad application prospects and practical significance, and helps to achieve efficient and accurate time series processing in various practical application scenarios.
[0096] Take the time series analysis method provided in this application for real-time arrhythmia detection and analysis of ECG signals as an example. The development of this method aims to overcome the parameter sensitivity (such as Figure 2 ) and computational efficiency (e.g. Figure 3 ) and provide a solution suitable for real-time monitoring and data collection quality control. The workflow is as follows:
[0097] The real-time calculation characteristics of morphological entropy are used to monitor ECG signals in real time to detect arrhythmia events. The data used in this embodiment comes from the MIT arrhythmia database, in which each record contains data from two channels, the duration of each record is about 0.5h, and the sampling frequency is 360Hz. This scheme selects the first channel data of one of the records and uses single-lead ECG data containing ventricular ectopic beats (V type) and supraventricular ectopic beats (S type) for analysis, that is, to obtain a non-stationary time series signal of a preset length.
[0098] A 0.5-45 Hz Butterworth bandpass filter was used to obtain the preprocessed signal, and then the calculation window was set to 1 second for time series sliding calculation. For each 1 second time series obtained in each window, the maximum and minimum values were normalized to obtain the normalized time series.
[0099] The bandwidth hyperparameter m is selected to be equal to 100, and the above-mentioned method of calculating the morphological entropy is adopted to calculate from two aspects: the entropy description of the fluctuation pattern based on the time series morphology and the entropy description of the phase pattern based on the time series morphology, to obtain the morphological entropy corresponding to the ECG signal.
[0100] In this embodiment, the sliding step size is set to 0.1 second to obtain a continuous entropy value estimate. Figure 4As shown, a 60-second arrhythmia segment is displayed, and each heartbeat is labeled, including Class N and abnormal Class S and Class V heartbeats. It can be observed from the figure that the entropy estimation based on morphological entropy has excellent morphological detectability. It has a significantly fluctuating entropy value at the abnormal heartbeat, while the Class N heartbeat tends to have a low and stable entropy value, indicating that the typical detection mode designed based on the method of the present invention has excellent arrhythmia detectability. In addition, this embodiment provides an application possibility for evaluating the quality of ECG signals during real-time monitoring. The data used in this embodiment comes from the PhysioNet database, in which each record contains data from one channel, and the duration of each record ranges from 10 seconds to 60 seconds, with a sampling frequency of 300Hz. This embodiment selected three ECG records marked by experts as containing noise segments for analysis. As shown Figure 5 As shown in the figure, three ECG signal segments containing local poor quality areas are shown. Figure 5 The test observed that for areas with poor signal quality ( Figure 5 The gray area of the original ECG signal), the morphological entropy value calculated by sliding shows significant abnormal fluctuations in the poor quality area, indicating that the method based on the present invention can timely discover and warn of ECG signal quality problems by real-time monitoring of time series changes.
[0101] The time series analysis method provided by the present invention has at least the following beneficial effects:
[0102] 1) This application uses morphological entropy calculation to extract the intrinsic features of complex time series, which greatly reduces the dependence on parameter settings and is conducive to improving the stability and reliability of analysis results;
[0103] 2) This application can achieve comprehensive analysis of time series data at a lower computational complexity by performing standardization processing and morphological feature extraction on time series signals, effectively mining the potential information of time series at different scales and frequency domains, and thus providing more accurate feature representation;
[0104] 3) This application uses the morphological entropy calculation method to efficiently capture the intrinsic morphological characteristics of the time series without complex parameter settings, has high computational efficiency and robustness, and is conducive to efficient time series analysis in real-time and online environments;
[0105] 4) This scheme is applicable to various types of time series data, including finance, acoustics, industry, physiological signals and other fields, providing broad application prospects and practical significance, and helping to achieve efficient and accurate quantification of time series complexity in various practical application scenarios;
[0106] 5) This application uses the morphological entropy method to measure the complexity and uncertainty of time series, which can provide important analytical information, facilitate further time series modeling and prediction, and improve the accuracy and effectiveness of the overall analysis;
[0107] 6) This scheme has the advantages of high efficiency, no parameters, and wide applicability. It can play an important role in the processing of various complex and non-stationary time series data. It provides a practical and efficient time series signal processing tool, which helps to promote technological progress and application development in related fields.
[0108] In one embodiment, the present application further provides a computer device, including a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the time series analysis of the above embodiment.
[0109] In one embodiment, the present application further provides a computer storage medium on which a computer program or instruction is stored. When the computer program or instruction is executed by a processor, the steps of the time series analysis of the above embodiment are implemented.
[0110] The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc.
[0111] The memory may be an internal storage unit of the simulation system, such as a hard disk or memory of an intelligent device. The memory may also be an external storage device of the intelligent device, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the intelligent device. Further, the memory is used to store the computer program and other programs and data required by the verification method of the FPGA. The memory may also be used to temporarily store data that has been output or is to be output.
[0112] The communication bus is a circuit that connects the elements described and implements transmission between these elements. For example, the central processing unit receives commands from other elements through the communication bus, decrypts the received commands, and performs calculations or data processing according to the decrypted commands. The memory may include program modules, such as a kernel, middleware, an application programming interface (API) and an application. The program module may be composed of software, firmware or hardware, or at least two of them. The input / output interface forwards commands or data entered by the user through the input / output interface (such as a sensor, keyboard, touch screen). The communication interface connects the speed measurement device of the artificial heart with other network devices, user devices, and networks. For example, the communication interface can be connected to the network via wired or wireless connection to connect to other external network devices or user devices. Wireless communication may include at least one of the following: wireless fidelity (WiFi), Bluetooth (BT), near field communication technology (NFC), global satellite positioning system (GPS) and cellular communication, etc. Wired communication may include at least one of the following: universal serial bus (USB), high-definition multimedia interface (HDMI), asynchronous transmission standard interface (RS-232), etc. The network may be a telecommunication network and a communication network. The communication network may be a computer network, the Internet, the Internet of Things, or a telephone network. The verification device of the FPGA may be connected to the network via a communication interface, and the protocol used by the speed measuring device of the artificial heart and other network devices to communicate may be supported by at least one of an application, an application programming interface (API), a middleware, a kernel, and a communication interface.
[0113] The transmission line electromagnetic simulation method of the present application can be implemented with program codes executable by a computing device, so that they can be stored in a storage device and executed by the computing device, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. Thus, the present invention is not limited to any specific combination of hardware and software.
[0114] It should be noted that the above embodiments can be freely combined as needed. The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered as the protection scope of the present invention.
Claims
1. A time series analysis method, characterized in that: Includes steps: Get non-stationary time series signals of preset length; The amplitude distribution of the non-stationary time series signal is limited to a range of 0 to 1 by a standardization technique to obtain a normalized time series signal; Dividing the normalized time series signal into a plurality of bandwidths, and counting the sampling point sets within each of the bandwidths; Calculating the fluctuation distribution probability corresponding to each of the sampling point sets, and calculating the entropy description of the fluctuation pattern based on the time series morphology based on the fluctuation distribution probability; Taking the bandwidth corresponding to the sampling point set with the most sampling points as the baseline bandwidth, and defining a wavelet set based on the baseline bandwidth, wherein the wavelet set is at least three sampling point sets that are on the same side of the baseline bandwidth and are continuous; Calculating the phase offset corresponding to the wavelet set to form an entropy description of the phase pattern based on the time series morphology; The morphological entropy is defined according to the entropy description of the fluctuation pattern based on the time series morphology and the entropy description of the phase pattern based on the time series morphology, and the morphological entropy value corresponding to the non-stationary time series signal is calculated.
2. The time series analysis method according to claim 1, characterized in that: The fluctuation distribution probability P k The calculation formula is: Among them, m is the number of divided bandwidths, B k For each set of sampling points of the bandwidth, B k ={δ1,δ2,...,δ ε }, ε∈[0,N], δ represents a single sampling point, k represents the bandwidth index, and satisfies k∈[1,m]; The calculation formula of Amplitude (m), which describes the entropy of the fluctuation pattern based on the time series morphology, is: in, is the midline amplitude value of the kth bandwidth; It is used to uniformly scale the entropy value.
3. The time series analysis method according to claim 2, characterized in that: The calculation formula of the baseline bandwidth Ψ is: ψ=argmax ε (B k ),k∈[1,m] Among them, Ψ satisfies Ψ upper -Ψ lower =1 / m,Ψ upper and lower Respectively represent the upper and lower bounds of the baseline bandwidth.
4. The time series analysis method according to claim 3, characterized in that: The wavelet set W s ={δ1,δ2,...,δ ξ },ξ∈[3,N]; The calculating of the phase offset corresponding to the wavelet set to form an entropy description of the phase pattern based on the time series morphology specifically includes: The amplitude of the sampling point set is obtained by subtracting (Ψ upper +Ψ lower ) / 2 to get the relative amplitude; Determine whether the wavelet set is located below the baseline bandwidth, and if so, flip the wavelet set so that the relative amplitude calculated by the sampling points included in the wavelet set is positive; Define the overall energy position θ for each of the wavelet sets s ,θ s The wavelet set W is defined as s The time coordinate of the area bisector; Then, the mapping function f(x; α) is used to generate each wavelet set W s Assigning Phase Offset Among them, Φ s is the wavelet set W s The average amplitude of the overall energy position θ s The importance of is scaled, N is the total length of the time series, and α represents the parameter configuration of the mapping function; According to the phase offset of each of the sub-wave sets, an entropy description Phase(m) of the phase pattern based on the time series morphology is formed, Among them, P s is the distribution probability corresponding to the s-th wavelet set, which is the ratio of the number of sampling points of the wavelet set to the total length N of the time series; It is used to uniformly quantify entropy values.
5. The time series analysis method according to claim 4, characterized in that: The calculation formula of the morphological entropy ME(m) is:
6. The time series analysis method according to claim 1, characterized in that: The non-stationary time series signals include any one or more of physiological signals, industrial and mechanical signals, environmental and meteorological data signals, financial and economic data signals, audio and acoustic signals.
7. The time series analysis method according to claim 1, characterized in that: After obtaining the non-stationary time series signal of the preset length, the method further includes: A filtering algorithm is designed according to the signal sampling rate and the frequency range of the analyzable signal in the non-stationary time series signal, and the non-stationary time series signal is filtered by the filtering algorithm.
8. The time series analysis method according to claim 1, characterized in that: The standardization technology is one or more of maximum and minimum value normalization and normal cumulative distribution function standardization.
9. A computer device comprising a memory, a processor and a computer program stored in the memory, characterized in that: The processor executes the computer program to implement the steps of the time series analysis method according to any one of claims 1 to 8.
10. A computer storage medium having a computer program or instruction stored thereon, characterized in that: When the computer program or instruction is executed by a processor, the steps of the time series analysis method according to any one of claims 1 to 8 are implemented.