CN105954695A

CN105954695A - Synchronization-based homogeneous-sensor mutation parameter recognizing method and device

Info

Publication number: CN105954695A
Application number: CN201610248401.8A
Authority: CN
Inventors: 黄明山; 陈新春; 梅华; 王颖; 张灏; 李辉; 都正周; 歹志阳; 陈淘
Original assignee: Yinchuan Power Supply Company State Grid Ningxia Electric Power Co Ltd; State Grid Corp of China SGCC; Xuji Group Co Ltd; Henan Xuji Instrument Co Ltd
Current assignee: Yinchuan Power Supply Company State Grid Ningxia Electric Power Co Ltd; State Grid Corp of China SGCC; Xuji Group Co Ltd; Henan Xuji Instrument Co Ltd
Priority date: 2016-04-20
Filing date: 2016-04-20
Publication date: 2016-09-21
Anticipated expiration: 2036-04-20
Also published as: CN105954695B

Abstract

The present invention relates to a method and device for identifying mutation parameters of a homogeneous sensor based on synchronization. A point value is extracted from each time point of each cycle sequence to form a one-dimensional data sequence measured multiple times by a single sensor, and cycle data synchronized by cycle is constructed. sequence. The obtained data samples are preprocessed based on maximum likelihood estimation. The empirical mode decomposition is used to obtain the IMF modal components of each order based on the combination preprocessing of the sensor signal containing a sudden change; the Hilbert-Huang transformation is used to obtain the instantaneous frequency map and the instantaneous amplitude map of the dominant IMF component of the signal, according to the HHT map The sudden point, amplitude change, and instantaneous frequency trend feature information are used to detect and identify the abnormal signal of the sensor. In the present invention, different types of sensors are indiscriminately fused to obtain equal-period data samples, the differences between different sensors and random errors in the signal collection process are reduced, and the identification accuracy of sensor mutation parameters is improved.

Description

A synchronization-based homogeneous sensor mutation parameter identification method and device

技术领域technical field

本发明研究中电力设备常用的同质传感器信号为对象，研究了对传感器测量的电数据进行融合过程并给出工程计算和实现的方法后，给出一种识别传感器突变参数检测和识别的有效方法。In the research of the present invention, the commonly used homogeneous sensor signals of power equipment are taken as the object, and after studying the fusion process of the electrical data measured by the sensors and giving the engineering calculation and realization methods, an effective method for detecting and identifying sudden change parameters of the sensors is given. method.

背景技术Background technique

一方面：用电设备能实时地感知电网的有效供电能力是有序用电的基础,电网能实时获取用电设备用电真实状态是智能电网的前提.传统的供用电信息采集往往由单一的传感器来完成，即使采用多个(种)传感器也多是分时使用，所以是从多个侧面孤立地反映电网的信息。随着技术的进步，这些测量数据需要融合处理，即指利用多个传感器的输出推断出一个有效的信息。On the one hand: the real-time perception of the effective power supply capacity of the power grid by electrical equipment is the basis for orderly electricity consumption, and the real-time status of power consumption of electrical equipment by the grid is the premise of a smart grid. Traditional power supply and consumption information collection is often performed by a single Even if multiple (types) of sensors are used, they are mostly used in time-sharing, so the information of the power grid is reflected in isolation from multiple sides. With the advancement of technology, these measurement data need to be fused, that is, to use the output of multiple sensors to infer an effective information.

另一方面：处在电力系统的传感器受到生产厂家生产环境、采用技术的不同，不同厂家的仪器甚至同一厂家的不同批次的仪器测得同一对象的某个物理数据都有差异，尤其在幅度上差别明显，因此对海量数据对比、自动分析等造成困难。On the other hand: the sensor in the power system is affected by the manufacturer's production environment and technology. Different manufacturers' instruments or even different batches of the same manufacturer's instruments measure a certain physical data of the same object. There are differences, especially in the amplitude. Therefore, it is difficult for massive data comparison and automatic analysis.

电力系统中的各种传感器因上述原因成为同质传感器。同质传感器是指观测同一物理现象的若干个传感器，该若干个传感器可以不同时、不同位置，但是被检测或采集信号的特征相同。由于不同传感器的个体差异和信号采集过程中的随机误差，因此现在对同质传感器的海量数据进行检测和识别较为困难，数据质量差，传感器突变参数识别精度低。Various sensors in the power system become homogeneous sensors for the above reasons. Homogeneous sensors refer to several sensors that observe the same physical phenomenon. The several sensors may not be at the same time or at different locations, but the characteristics of the detected or collected signals are the same. Due to the individual differences of different sensors and random errors in the signal acquisition process, it is difficult to detect and identify massive data from homogeneous sensors, the data quality is poor, and the identification accuracy of sensor mutation parameters is low.

发明内容Contents of the invention

本发明的目的是提供一种基于同步的同质传感器突变参数识别方法与装置，用以消减不同传感器差异和信号采集过程中的随机误差，改善数据质量从而提高传感器突变参数识别精度。The purpose of the present invention is to provide a method and device for identifying mutation parameters of homogeneous sensors based on synchronization, which is used to reduce the difference between different sensors and random errors in the signal acquisition process, improve data quality and thereby improve the recognition accuracy of sensor mutation parameters.

为实现上述目的，本发明的方案包括：To achieve the above object, the solution of the present invention includes:

一种基于同步的同质传感器突变参数识别方法，步骤如下：A synchronization-based homogeneous sensor mutation parameter identification method, the steps are as follows:

步骤A：中央处理器以T为采样周期，同质传感器定时对系统被测信号进行采样和量化，并得到相同采样频率下的采样数据；Step A: The central processor takes T as the sampling period, and the homogeneous sensor regularly samples and quantifies the measured signal of the system, and obtains the sampling data at the same sampling frequency;

步骤B：将采样数据按周期进行截取，获得m个周期序列为：Y₁(t),Y₂(t),…Y_m(t)，每个周期样本中包含了N个数据点，即Y_i＝[X_i(1),X_i(2),…X_i(N)]，其中i＝1,2,…,m；Step B: Intercept the sampling data by cycle, and obtain m cycle sequences: Y ₁ (t), Y ₂ (t),...Y _m (t), and each cycle sample contains N data points, namely Y _i =[X _i (1),X _i (2),...X _i (N)], where i=1,2,...,m;

步骤C：构造按周期同步构建的周期数据序列：将每个周期数据中各个时刻点数据视为同步后的同一对象的多次测量结果，即各个周期对应时刻点数据为一个样本，多个周期的数据中相对应时刻的每个点值即构成一个数据序列，从而可视为单一传感器多次测量的一维数据序列，即构造了按周期同步构建的周期数据序列Y₁′(t),Y₂′(t),…Y′_N(t)，每组一维数据序列中包括了m个数据点，即Y_i′＝[X₁(1),X₂(2),…X_m(N)]；Step C: Construct a periodic data sequence constructed synchronously by cycle: treat the data at each time point in each cycle data as the multiple measurement results of the same object after synchronization, that is, the data at each time point corresponding to each cycle is a sample, and multiple cycles Each point value at the corresponding time in the data constitutes a data sequence, which can be regarded as a one-dimensional data sequence measured multiple times by a single sensor, that is, a periodic data sequence Y ₁ ′(t) constructed synchronously by cycle is constructed, Y ₂ ′(t),…Y′ _N (t), each set of one-dimensional data sequence includes m data points, that is, Y _i ′=[X ₁ (1),X ₂ (2),…X _m (N)];

步骤D：对数据样本Y_t'进行基于极大似然估计的优化组合预处理；Step D: Perform optimal combined preprocessing based on maximum likelihood estimation on the data sample Y _t ';

步骤E：对基于组合预处理的含有突变传感器信号采用EMD分解得到各阶IMF分量；Step E: EMD decomposition is used to obtain the IMF components of each order based on the combined preprocessing-based sensor signal containing mutations;

步骤F：对传感器异常信号的主导IMF分量采用HHT变换得到瞬时频率图和瞬时幅值图；Step F: HHT transformation is used for the dominant IMF component of the sensor abnormal signal to obtain an instantaneous frequency graph and an instantaneous amplitude graph;

步骤G：依据HHT变换图中的突变点、幅值变化、瞬时频率走势特征信息对传感器突变参数进行检测和识别。Step G: Detect and identify the sudden change parameters of the sensor according to the sudden change points, amplitude changes, and instantaneous frequency trend characteristic information in the HHT transformation diagram.

进一步的，所述步骤D中，对数据样本Y_t'进行基于极大似然估计的优化组合预处理的方式为：Further, in the step D, the method of performing the optimal combination preprocessing based on the maximum likelihood estimation on the data sample Y _t ' is as follows:

S101：假设在给定的时刻，待测环境特征为X，传感器的值为Y,则该传感器的测量模型为：Y＝f(X)+V其中，V是符合高斯分布的噪声项；所谓数据融合就是由N个传感器得到测量值Y₁、Y₂、…、Y_N,并按某种估计准则从这些测量值中得到特征参数X的最优估计；S101: Assuming that at a given moment, the feature of the environment to be measured is X, and the value of the sensor is Y, then the measurement model of the sensor is: Y=f(X)+V where, V is a noise item conforming to the Gaussian distribution; the so-called Data fusion is to obtain the measured values Y ₁ , Y ₂ ,..., Y _N from N sensors, and obtain the optimal estimation of the characteristic parameter X from these measured values according to a certain estimation criterion;

S102：寻找合适的准则函数，即当X被估计为X(Y)时所产生的损失最小；取损失函数为均匀损失：S102: Find a suitable criterion function, that is, when X is estimated to be X(Y), the loss generated is the smallest; take the loss function as uniform loss:

$L L [[\overset{^^}{X x} ((Y Y)),, X x]] = = \{\begin{matrix} 11 & | | \overset{^^}{X x} - - X x | | &GreaterEqual; &Greater Equal; \frac{ϵ ϵ}{22} \\ 00 & | | \overset{^^}{X x} - - X x | | < < \frac{ϵ ϵ}{22} \end{matrix}$

S103：在损失函数L的基础上，定义相应估计风险的函数R：S103: On the basis of the loss function L, define a function R corresponding to the estimated risk:

其中，p(x)、p(x|y)表示概率分布； Among them, p(x), p(x|y) represent the probability distribution;

S104：取风险最小为估计准则，即S104: Take the minimum risk as the estimation criterion, namely

$\frac{\partial \partial R R}{\partial \partial X x} {| |}_{X x = = \overset{^^}{X x} ((Y Y))} = = 00 - - - - - - ((11))$

可以得到符合式(1)的最优估计为It can be obtained that the optimal estimate conforming to formula (1) is

${\overset{^^}{X x}}_{o o p p t t} ((Y Y)) = = arg arg max max p p ((X x | | Y Y))$

S105：在具有N个传感器的系统中，相应的信息融合可以看作是在观测值Y₁、Y₂、…、Y_N下，值X具有最大后验的估计为 S105: In a system with N sensors, the corresponding information fusion can be regarded as under the observed values Y ₁ , Y ₂ , ..., Y _N , the value X has the maximum a posteriori estimate as

S106：为了求出最大后验估计，对所有可能参数X均采用p(x)＝1,通过公式推导得到：S106: In order to obtain the maximum a posteriori estimate, p(x)=1 is adopted for all possible parameters X, and obtained by formula derivation:

$p p ((X x | | {Y Y}_{11},, {Y Y}_{22},, ... ... ... ... {Y Y}_{N N})) &Proportional; &Proportional; {Π Π}_{i i = = 11}^{N N} p p ((X x | | {Y Y}_{i i})) - - - - - - ((22))$

此时最大后验估计即简化为极大似然估计，相应的融合计算公式为：At this time, the maximum a posteriori estimation is simplified to the maximum likelihood estimation, and the corresponding fusion calculation formula is:

${V V}_{f f}^{- - 11} = = {Σ Σ}_{i i = = 11}^{N N} {A A}_{i i}^{T T} {V V}_{i i}^{- - 11} {A A}_{i i} + + {V V}_{X x}^{- - 11} - - - - - - ((33))$

${X x}_{f f} = = {V V}_{f f} {Σ Σ}_{i i = = 11}^{N N} {A A}_{i i}^{T T} {V V}_{i i}^{- - 11} {Y Y}_{i i} - - - - - - ((44))$

当传感器为一维时，且不考虑坐标变换，则式(3)与(4)可简化为：When the sensor is one-dimensional and the coordinate transformation is not considered, the formulas (3) and (4) can be simplified as:

${σ σ}_{f f}^{- - 22} = = {Σ Σ}_{i i = = 11}^{N N} {σ σ}_{i i}^{22} - - - - - - ((55))$

${y the y}_{f f} = = {Σ Σ}_{i i = = 11}^{N N} \frac{{σ σ}_{f f}^{22}}{{σ σ}_{i i}^{22}} {Y Y}_{i i} - - - - - - ((66))$

进一步的，所述步骤E中对基于组合预处理的含有突变传感器信号采用EMD分解得到各阶IMF分量的过程为：Further, in the step E, the process of obtaining the IMF components of each order by using EMD decomposition based on the combined preprocessing of the sensor signal containing a sudden change is:

S201：对信号x[t]所有局部极大值点和所有局部极小值点用三次样条函数进行插值，并拟合上下包络线；x[t]为经过组合预处理的含有突变特性的传感器信号；S201: Interpolate all local maximum points and all local minimum points of the signal x[t] with a cubic spline function, and fit the upper and lower envelopes; sensor signal;

S202：求取上下包络线的平均值曲线M₁(t)(t＝1、2、3…m)，则采样信号x[t]与M₁(t)之差即为P₁(t)：P₁(t)＝x[t]-M₁(t)；S202: Calculate the average curve M ₁ (t) of the upper and lower envelopes (t=1, 2, 3...m), then the difference between the sampling signal x[t] and M ₁ (t) is P ₁ (t ): P ₁ (t)=x[t]-M ₁ (t);

S203：如果P₁(t)同时满足下述IMF的两个条件，则其为第一个IMF分量，否则将其作为原始信号重复S201到S202，得到P₁₁(t)：P₁₁(t)＝P₁(t)-M₁₁(t)，其中：M₁₁(t)为P₁(t)的上下包络线的平均曲线；S203: If P ₁ (t) satisfies the following two conditions of IMF at the same time, then it is the first IMF component, otherwise repeat S201 to S202 as the original signal to obtain P ₁₁ (t): P ₁₁ (t) =P ₁ (t)-M ₁₁ (t), wherein: M ₁₁ (t) is the average curve of the upper and lower envelopes of P ₁ (t);

所述IMF分量满足的两个条件为：整个时间历程内，穿越零点次数与极值点数相等或至多相差1；且信号上任意一点，由局部极大值定义的上包络线和局部极小值点定义的下包络线的均值为0，即信号关于时间轴局部对称；The two conditions that the IMF component satisfies are: in the whole time history, the number of zero crossings is equal to or at most differs from the number of extremum points; and at any point on the signal, the upper envelope defined by the local maximum and the local minimum The mean value of the lower envelope defined by the value points is 0, that is, the signal is locally symmetrical about the time axis;

S204：重复上述步骤筛选，直到第k次筛选时由式(1)得到的P_1k(t)满足IMF分量的两个条件：P_1k(t)＝P_1(1-k)(t)-M_1k(t) (1)S204: Repeat the above steps to screen until the P _1k (t) obtained by formula (1) meets the two conditions of the IMF component: P _1k (t)=P _1(1-k) (t)- M _1k (t) (1)

S205：除信号外，在实际计算时可以通过式(2)求取门限值S_D来判断每次筛选结果是否为IMF分量：S205: In addition to the signal, in the actual calculation, the threshold value _SD can be obtained by formula (2) to determine whether each screening result is an IMF component:

${S S}_{D D.} = = {Σ Σ}_{t t = = 00}^{r r} | | \frac{{P P}_{11 ((k k - - 11))} ((t t)) - - {P P}_{11 k k} ((t t))}{{P P}^{22}_{11 ((k k - - 11))} ((t t))} | |;; - - - - - - ((22))$

其中：r为传感器信号的采样点数，门限值S_D通常取0.2到0.3；Where: r is the number of sampling points of the sensor signal, and the threshold _SD is usually 0.2 to 0.3;

S206：令C₁(t)＝P_1k(t)，则C₁(t)即为第一个IMF分量，其包含了原信号x[t]中周期最短的IMF分量；将C₁(t)从x[t]中分离出来：R₁(t)＝x[t]-C₁(t)；S206: Let C ₁ (t)=P _1k (t), then C ₁ (t) is the first IMF component, which contains the IMF component with the shortest period in the original signal x[t]; C ₁ (t ) is separated from x[t]: R ₁ (t)=x[t]-C ₁ (t);

S207：将R₁(t)作为新的值重复以上步骤S201至S205n次，可获得信号x[t]的n个IMF分量：R_n(t)＝R_n-1(t)-C_n(t)；S207: Repeat the above steps S201 to S205n times with R ₁ (t) as a new value to obtain n IMF components of the signal x[t]: R _n (t)=R _n-1 (t)-C _n ( t);

S208：当R_n(t)为单调函数从信号x[t]不能再分解出其他分量时，整个分解过程结束；得到： S208: When R _n (t) is a monotone function and no other components can be decomposed from the signal x[t], the whole decomposition process ends; get:

进一步的，所述步骤F对传感器异常信号的主导IMF分量采用HHT变换得到瞬时频率图和瞬时幅值图的过程为：Further, the process of obtaining the instantaneous frequency graph and the instantaneous amplitude graph by HHT transforming the dominant IMF component of the sensor abnormal signal in the step F is as follows:

S301：将通过EMD分解后获得的所有IMF分量进行希尔伯特变换；给定C(t)的Hilbert形式为：S301: Perform Hilbert transform on all IMF components obtained after EMD decomposition; the Hilbert form of given C(t) is:

其中，λ为积分变量。 Among them, λ is the integral variable.

S302：构造一个解析信号Z(t)：S302: Construct an analytical signal Z(t):

Z(t)＝C(t)+iH(t)＝A(t)e^iθ(t)；Z(t)=C(t)+iH(t)=A(t)e ^iθ(t) ;

S303：幅值函数： S303: Amplitude function:

S304：幅角函数： S304: Argument function:

S305：瞬时频率： S305: Instantaneous frequency:

进一步的，所述步骤G中依据HHT变换图中的突变点、幅值变化、瞬时频率走势特征信息传感器突变参数进行检测和识别的过程为：Further, in the step G, the process of detecting and identifying sensor mutation parameters based on the mutation point, amplitude change, and instantaneous frequency trend feature information in the HHT transformation diagram is as follows:

S306：对传感器中突变参数进行分类：骤降、骤升、谐波、脉冲；S306: classify the sudden change parameters in the sensor: sudden drop, sudden rise, harmonic, pulse;

S307：依据HHT图中针对不同的传感器异常信号的突变点、幅值变化以及瞬时频率走势等特性信息不同，分类识别出传感器何时经历了何种突变。S307: Classify and identify when and what kind of mutation the sensor has experienced according to the different characteristic information such as the mutation point, amplitude change, and instantaneous frequency trend of different sensor abnormal signals in the HHT diagram.

本发明还提供了一种基于同步的同质传感器突变参数识别装置，包括：The present invention also provides a synchronization-based homogeneous sensor mutation parameter identification device, comprising:

模块A：中央处理器以T为采样周期，同质传感器定时对系统被测信号进行采样和量化，并得到相同采样频率下的采样数据；Module A: The central processor takes T as the sampling period, and the homogeneous sensor regularly samples and quantifies the measured signal of the system, and obtains the sampling data at the same sampling frequency;

模块B：将采样数据按周期进行截取，获得m个周期序列为：Y₁(t),Y₂(t),…Y_m(t)，每个周期样本中包含了N个数据点，即Y_i＝[X_i(1),X_i(2),…X_i(N)]，其中i＝1,2,…,m；Module B: Intercept the sampling data by cycle, and obtain m cycle sequences: Y ₁ (t), Y ₂ (t),...Y _m (t), and each cycle sample contains N data points, namely Y _i =[X _i (1),X _i (2),...X _i (N)], where i=1,2,...,m;

模块C：构造按周期同步构建的周期数据序列：将每个周期数据中各个时刻点数据视为同步后的同一对象的多次测量结果，即各个周期对应时刻点数据为一个样本，多个周期的数据中相对应时刻的每个点值即构成一个数据序列，从而可视为单一传感器多次测量的一维数据序列，即构造了按周期同步构建的周期数据序列Y₁′(t),Y₂′(t),…Y′_N(t)，每组一维数据序列中包括了m个数据点，即Y_i′＝[X₁(1),X₂(2),…X_m(N)]；Module C: Construct a periodic data sequence constructed synchronously by cycle: treat the data at each time point in each cycle data as the multiple measurement results of the same object after synchronization, that is, the data at each time point corresponding to each cycle is a sample, and multiple cycles Each point value at the corresponding time in the data constitutes a data sequence, which can be regarded as a one-dimensional data sequence measured multiple times by a single sensor, that is, a periodic data sequence Y ₁ ′(t) constructed synchronously by cycle is constructed, Y ₂ ′(t),…Y′ _N (t), each set of one-dimensional data sequence includes m data points, that is, Y _i ′=[X ₁ (1),X ₂ (2),…X _m (N)];

模块D：对数据样本Y_t'进行基于极大似然估计的优化组合预处理；Module D: Optimal combination preprocessing based on maximum likelihood estimation for data sample Y _t ';

模块E：对基于组合预处理的含有突变传感器信号采用EMD分解得到各阶IMF分量；Module E: EMD decomposition is used to obtain the IMF components of each order for the sensor signal containing mutation based on combined preprocessing;

模块F：对传感器异常信号的主导IMF分量采用HHT变换得到瞬时频率图和瞬时幅值图；Module F: use HHT transformation to obtain the instantaneous frequency map and instantaneous amplitude map of the dominant IMF component of the abnormal signal of the sensor;

模块G：依据HHT变换图中的突变点、幅值变化、瞬时频率走势特征信息对传感器突变参数进行检测和识别。Module G: Detect and identify sensor mutation parameters based on the mutation point, amplitude change, and instantaneous frequency trend characteristic information in the HHT transformation diagram.

进一步的，所述模块D中，对数据样本Y_t'进行基于极大似然估计的优化组合预处理的方式为：Further, in the module D, the method of performing the optimal combination preprocessing based on the maximum likelihood estimation on the data sample Y _t ' is as follows:

$L L ((\overset{^^}{X x} ((Y Y)),, X x)) = = \{\begin{matrix} 11 & | | \overset{^^}{X x} - - X x | | &GreaterEqual; &Greater Equal; \frac{ϵ ϵ}{22} \\ 00 & | | \overset{^^}{X x} - - X x | | < < \frac{ϵ ϵ}{22} \end{matrix}$

S106：为了求出最大后验估计，对所有可能参数X均采用p(x)＝1,通过公式推导可以得到：S106: In order to obtain the maximum a posteriori estimate, p(x)=1 is adopted for all possible parameters X, and it can be obtained by formula derivation:

当传感器为一维时，且不考虑坐标变换，则式(3)与(4)可简化为：When the sensor is one-dimensional, and the coordinate transformation is not considered, the equations (3) and (4) can be simplified as:

进一步的，所述模块E中对基于组合预处理的含有突变传感器信号采用EMD分解得到各阶IMF分量的过程为：Further, in the module E, the process of obtaining the IMF components of each order by using EMD decomposition to the sensor signal containing a sudden change based on the combined preprocessing is:

所述IMF分量满足的两个条件为：整个时间历程内，穿越零点次数与极值点数相等或至多相差1；信号上任意一点，由局部极大值定义的上包络线和局部极小值点定义的下包络线的均值为0，即信号关于时间轴局部对称；The two conditions that the IMF component satisfies are: in the whole time course, the number of crossing zero points is equal to or at most differs from the number of extremum points; at any point on the signal, the upper envelope defined by the local maximum value and the local minimum value The mean value of the lower envelope defined by the point is 0, that is, the signal is locally symmetrical about the time axis;

进一步的，所述模块F对传感器异常信号的主导IMF分量采用HHT变换得到瞬时频率图和瞬时幅值图的过程为：Further, the process of obtaining the instantaneous frequency graph and instantaneous amplitude graph by the module F using HHT transformation on the dominant IMF component of the sensor abnormal signal is as follows:

S301：将通过EMD分解后获得的所有IMF分量进行希尔伯特变换S301: Perform Hilbert transform on all IMF components obtained after EMD decomposition

(Hilbert)变换。给定C(t)的Hilbert形式为：(Hilbert) transform. Given the Hilbert form of C(t) is:

其中，λ为积分变量。 Among them, λ is the integral variable.

Z(t)＝C(t)+iH(t)＝A(t)e^iθ(t)；Z(t)=C(t)+iH(t)=A(t)e ^iθ(t) ;

S303：幅值函数： S303: Amplitude function:

S304：幅角函数： S304: Argument function:

S305：瞬时频率： S305: Instantaneous frequency:

进一步的，所述模块G中依据HHT变换图中的突变点、幅值变化、瞬时频率走势特征信息传感器突变参数进行检测和识别的过程为：Further, the process of detecting and identifying sensor mutation parameters based on the mutation point, amplitude change, and instantaneous frequency trend characteristic information in the module G is as follows:

本发明的一种基于同步的同质传感器突变参数识别方法与装置，中央处理器以T秒为采样周期，同质传感器(传感器观测的是同一物理现象)定时对被测信号进行采样和量化并获得关于时间的数据序列X(N)，将采样数据按周期进行截取，获得m个周期序列为：Y₁(t),Y₂(t),…Y_m(t)，每个周期样本中包含了N个数据点，即Y_i＝[X_i(1),X_i(2),…X_i(N)]，其中i＝1,2,…,m。从每个周期序列各时刻点中抽取一个点值组成单一传感器多次测量的一维数据序列，构造按周期同步构建的周期数据序列，即Y_i′(t),Y₂′(t),…Y′_N(t)，每组一维数据序列中包括了m个数据点，即Y_i′＝[X₁(1),X₂(2),…X_m(N)]。对得到的数据样本Y_t'进行基于极大似然估计和最小二乘法估计的优化组合预处理，用于指导电力系统周期采样数据指导。对基于组合预处理的含有突变传感器信号采用经验模态分解得到各阶IMF模态分量；对信号主导IMF分量采用希尔伯特—黄变换得到瞬时频率图和瞬时幅值图，依据HHT图中的突变点、幅值变化、瞬时频率走势特征信息对传感器异常信号进行参数检测和识别。本发明将不同类传感器进行无差别融合，得到等周期数据样本，消减不同传感器差异和信号采集过程中的随机误差，提高传感器突变参数识别精度。A method and device for identifying mutation parameters of a homogeneous sensor based on synchronization of the present invention, the central processing unit takes T seconds as the sampling period, and the homogeneous sensor (the sensor observes the same physical phenomenon) regularly samples and quantifies the measured signal and Obtain the data sequence X(N) about time, intercept the sampling data by cycle, and obtain m cycle sequences: Y ₁ (t), Y ₂ (t),...Y _m (t), in each cycle sample Contains N data points, that is, Y _i =[X _i (1),X _i (2),...X _i (N)], where i=1,2,...,m. Extract a point value from each time point of each periodic sequence to form a one-dimensional data sequence measured by a single sensor multiple times, and construct a periodic data sequence that is synchronously constructed according to the cycle, that is, Y _i ′(t), Y ₂ ′(t), ...Y′ _N (t), each group of one-dimensional data sequence includes m data points, that is, Y _i ′=[X ₁ (1), X ₂ (2), . . . X _m (N)]. The obtained data sample Y _t ' is preprocessed based on the optimal combination of maximum likelihood estimation and least square estimation, which is used to guide the periodic sampling data guidance of the power system. The empirical mode decomposition is used to obtain the IMF modal components of each order based on the combination preprocessing of the sensor signal containing a sudden change; the Hilbert-Huang transformation is used to obtain the instantaneous frequency map and the instantaneous amplitude map of the dominant IMF component of the signal, according to the HHT map The sudden point, amplitude change, and instantaneous frequency trend feature information are used to detect and identify the abnormal signal of the sensor. In the present invention, different types of sensors are indiscriminately fused to obtain equal-period data samples, the differences between different sensors and random errors in the signal collection process are reduced, and the identification accuracy of sensor mutation parameters is improved.

附图说明Description of drawings

图1是基于同步的同质传感器突变参数识别方法的流程图；Fig. 1 is the flow chart of the homogeneous sensor mutation parameter identification method based on synchronization;

图2是本发明模拟传感器骤降信号IMF1分量的HHT瞬时频率和幅值图；Fig. 2 is the HHT instantaneous frequency and amplitude figure of analog sensor dip signal IMF1 component of the present invention;

图3是本发明模拟传感器暂升信号IMF1分量的HHT瞬时频率和幅值图；Fig. 3 is the HHT instantaneous frequency and amplitude diagram of the analog sensor swell signal IMF1 component of the present invention;

图4是本发明模拟传感器脉冲信号IMF1分量的HHT瞬时频率和幅值图；Fig. 4 is the HHT instantaneous frequency and amplitude figure of analog sensor pulse signal IMF1 component of the present invention;

图5是本发明模拟传感器谐波信号IMF1分量的HHT瞬时频率和幅值图。Fig. 5 is a HHT instantaneous frequency and amplitude diagram of the harmonic signal IMF1 component of the analog sensor in the present invention.

具体实施方式detailed description

下面结合附图对本发明做进一步详细的说明。The present invention will be described in further detail below in conjunction with the accompanying drawings.

方法实施例method embodiment

如图1所示，一种基于同步的同质传感器突变参数识别方法，包括以下步骤：As shown in Figure 1, a synchronization-based homogeneous sensor mutation parameter identification method includes the following steps:

步骤A：中央处理器以T(秒)为采样周期，同质传感器定时对系统被测信号进行采样和量化，并得到相同采样频率下的数据样本X(N)的具体步骤为：Step A: The central processing unit takes T (second) as the sampling period, and the homogeneous sensor regularly samples and quantifies the measured signal of the system, and the specific steps to obtain the data sample X(N) at the same sampling frequency are:

中央处理器采样周期为T(秒)，包含在系统内部的m个传感器(可以不同时、不同位置，但是被检测或采集信号的特征相同)定时对系统内部的被测信号进行采样和量化，进而得到相同采样频率下的数据样本X_i(N)，其中i＝1,2,…,m。The sampling period of the central processing unit is T (seconds), and the m sensors included in the system (can be at different times and different positions, but the characteristics of the detected or collected signals are the same) regularly sample and quantify the measured signals inside the system, Further, data samples X _i (N) at the same sampling frequency are obtained, where i=1, 2, . . . , m.

进一步的，所述步骤C：构造按周期同步构建的周期数据序列，即Y₁′(t),Y₂′(t),…Y′_N(t)，每组一维数据序列中包括了m个数据点，即Y_i′＝[X₁(1),X₂(2),…X_m(N)]的具体步骤为：Further, the step C: Construct periodic data sequences constructed synchronously according to the period, namely Y ₁ ′(t), Y ₂ ′(t),...Y′ _N (t), each group of one-dimensional data sequences includes m data points, that is, Y _i ′=[X ₁ (1), X ₂ (2),…X _m (N)] The specific steps are:

将采集到的电力系统数据在按周期进行截取后，将每个周期数据中各个时刻点数据视为同步后的同一对象的多次测量结果，即各个周期对应时刻点数据为一个样本，多个周期的数据中相对应时刻的每个点值即构成一个数据序列，从而可视为单一传感器多次测量的一维数据序列，即构造了按周期同步构建的周期数据序列，也即Y₁′(t),Y₂′(t),…Y′_N(t)，每组一维数据序列中包括了m个数据点，即Y_i′＝[X₁(1),X₂(2),…X_m(N)]。After the collected power system data is intercepted by cycle, the data at each time point in each cycle data is regarded as the multiple measurement results of the same object after synchronization, that is, the data at each time point corresponding to each cycle is a sample, and multiple Each point value at the corresponding moment in the periodic data constitutes a data sequence, which can be regarded as a one-dimensional data sequence measured multiple times by a single sensor, that is, a periodic data sequence constructed synchronously by cycle is constructed, that is, Y ₁ ′ (t), Y ₂ ′(t),…Y′ _N (t), each set of one-dimensional data sequence includes m data points, that is, Y _i ′=[X ₁ (1),X ₂ (2) ,...X _m (N)].

进一步的，所述步骤D：对数据样本Y_t'进行基于极大似然估计预处理的具体步骤为：Further, the step D: the specific steps of preprocessing the data sample Y _t ' based on maximum likelihood estimation are:

S101：假设在给定的时刻，待测环境特征为X，传感器的值为Y,则该传感器的测量模型为：Y＝f(X)+V其中，V是符合高斯分布的噪声项。所谓数据融合就是由N个传感器得到测量值Y₁、Y₂、…、Y_N,并按某种估计准则从这些测量值中得到特征参数X的最优估计。S101: Assuming that at a given moment, the feature of the environment to be measured is X, and the value of the sensor is Y, then the measurement model of the sensor is: Y=f(X)+V where, V is a noise item conforming to Gaussian distribution. The so-called data fusion is to obtain the measured values Y ₁ , Y ₂ ,..., Y _N from N sensors, and obtain the optimal estimation of the characteristic parameter X from these measured values according to a certain estimation criterion.

S102：寻找合适的准则函数，即当X被估计为X(Y)时所产生的损失最小。取损失函数为均匀损失：S102: Find an appropriate criterion function, that is, when X is estimated to be X(Y), the resulting loss is the smallest. Take the loss function as uniform loss:

可以得到符合式(1)的最优估计(最大后验估计)为The optimal estimate (maximum a posteriori estimate) conforming to formula (1) can be obtained as

S106：为了求出最大后验估计，在某些条件下，我们无法决定特征参数X的先验分布，采用“模糊先验”的概念，即对所有可能参数X均采用p(x)＝1,通过公式推导可以得到：S106: In order to obtain the maximum a posteriori estimate, under certain conditions, we cannot determine the prior distribution of the characteristic parameter X, and adopt the concept of "fuzzy prior", that is, use p(x)=1 for all possible parameters X , can be obtained by formula derivation:

进一步的，所述步骤E：Further, the step E:

注释：x[t]为经过组合预处理的含有突变特性的传感器信号。Note: x[t] is the combined preprocessed sensor signal containing mutation characteristics.

S201：对信号x[t]所有局部极大值点和所有局部极小值点用三次样条函数进行插值，并拟合上下包络线；S201: Interpolate all local maximum points and all local minimum points of the signal x[t] with a cubic spline function, and fit the upper and lower envelopes;

所述IMF分量满足的两个条件为：整个时间历程内，穿越零点次数与极值点数相等或至多相差1；信号上任意一点，由局部极大值定义的上包络线和局部极小值点定义的下包络线的均值为0，即信号关于时间轴局部对称。The two conditions that the IMF component satisfies are: in the whole time course, the number of crossing zero points is equal to or at most differs from the number of extremum points; at any point on the signal, the upper envelope defined by the local maximum value and the local minimum value The mean value of the lower envelope defined by the points is 0, that is, the signal is locally symmetrical about the time axis.

其中：r为电力系统信号的采样点数，门限值S_D通常取0.2到0.3；Where: r is the number of sampling points of the power system signal, and the threshold _SD is usually 0.2 to 0.3;

S206：令C₁(t)＝P_1k(t)，则C₁(t)即为第一个IMF分量，其包含了原信号x[t]中周期最短的IMF分量。将C₁(t)从x[t]中分离出来：R₁(t)＝x[t]-C₁(t)；S206: Let C ₁ (t)=P _1k (t), then C ₁ (t) is the first IMF component, which includes the IMF component with the shortest period in the original signal x[t]. Separate C ₁ (t) from x[t]: R ₁ (t)=x[t]-C ₁ (t);

S208：当R_n(t)为单调函数从信号x[t]不能再分解出其他分量时，整个分解过程结束。得到： S208: When R _n (t) is a monotone function and no other components can be decomposed from the signal x[t], the whole decomposition process ends. get:

进一步的，所述步骤F：对传感器异常信号的主导IMF分量采用HHT变换得到瞬时频率图和瞬时幅值图的具体步骤按照两种方案进行：Further, the step F: the specific steps of obtaining the instantaneous frequency map and the instantaneous amplitude map by using HHT transformation for the dominant IMF component of the abnormal signal of the sensor are carried out according to two schemes:

其中，λ为积分变量。 Among them, λ is the integral variable.

Z(t)＝C(t)+iH(t)＝A(t)e^iθ(t)；Z(t)=C(t)+iH(t)=A(t)e ^iθ(t) ;

S303：幅值函数： S303: Amplitude function:

S304：幅角函数： S304: Argument function:

S305：瞬时频率： S305: Instantaneous frequency:

进一步的，所述步骤G：依据HHT变换图中的突变点、幅值变化、瞬时频率走势特征信息传感器突变参数进行检测和识别的过程为：Further, the step G: the process of detecting and identifying the sudden change parameters of the sensor based on the sudden change point, amplitude change, and instantaneous frequency trend characteristic information in the HHT transformation diagram is as follows:

S306：对传感器中突变参数进行分类：骤降、骤升、谐波、脉冲。S306: Classify the sudden change parameters in the sensor: sudden drop, sudden rise, harmonic, and pulse.

依据仿真试验对本发明进行介绍：利用MATLAB产生各种传感器突变信号(骤降、骤升、谐波、脉冲)，采样频率为1kHz，采样点数1000，电压基波频率50Hz，绘制成表1。According to simulation test, the present invention is introduced: utilize MATLAB to produce various sensor mutation signals (sag, swell, harmonic, pulse), sampling frequency is 1kHz, sampling point number 1000, voltage fundamental wave frequency 50Hz, is drawn into Table 1.

表1电压扰动类型Table 1 Types of voltage disturbances

依据表1不同传感器突变类型，并参照本发明进行仿真实验，图2至图5是本案例算法的检测结果。Based on different sensor mutation types in Table 1, and referring to the present invention to carry out simulation experiments, Fig. 2 to Fig. 5 are the detection results of the algorithm in this case.

从结果图2和图3可以看出：幅值和瞬时频率在0.45s和0.55s处发生跃变，瞬时频率经历先缓慢上升再下降后维持一段平稳时间，然后再缓慢上升再下降，正常时刻瞬时频率维持在50Hz左右；幅值先大幅度上升后经历一段平稳过程再大幅下降，由此表明此种异常信号为骤降。而骤升异常信号特征信息刚好与骤降异常信号相反。利用上述不同的信息可以很好的区别骤降、骤升传感器异常信号。It can be seen from the results in Figure 2 and Figure 3 that the amplitude and instantaneous frequency jump at 0.45s and 0.55s, and the instantaneous frequency rises slowly and then decreases for a period of time, and then rises slowly and then decreases. The instantaneous frequency is maintained at about 50Hz; the amplitude first rises sharply and then goes through a period of stability and then drops sharply, which indicates that the abnormal signal is a sudden drop. However, the characteristic information of the sudden rise abnormal signal is just opposite to that of the sudden drop abnormal signal. Utilizing the above different information can well distinguish the abnormal signal of the sag and swell sensor.

从结果图4至图5可以看出：瞬时频率和幅值在某一时刻出现尖峰信息，对应于脉冲异常信号；瞬时频率在某一段时间内；瞬时频率在某一时间段内，经历剧烈抖动，加之幅值在同一时间段内先经历暂升后幅值剧烈变化后回归稳态，对应于谐波异常信号。It can be seen from the results in Figures 4 to 5 that peak information appears at a certain moment in the instantaneous frequency and amplitude, corresponding to an abnormal pulse signal; the instantaneous frequency is within a certain period of time; the instantaneous frequency experiences severe jitter within a certain period of time , and the amplitude first experienced a sudden rise and then a sharp change in the amplitude in the same period of time, and then returned to a steady state, corresponding to the harmonic abnormal signal.

综上所述：针对于不同的传感器异常信号，本算法可以检测出不同的瞬时频率和幅值特征信息对应于不同的异常信号；可以将传感器异常信号精准分类，实现传感器突变特性的自动分析。此算法系统辨识性好，便于硬件实现。To sum up: for different sensor abnormal signals, this algorithm can detect different instantaneous frequency and amplitude characteristic information corresponding to different abnormal signals; it can accurately classify sensor abnormal signals and realize automatic analysis of sensor mutation characteristics. This algorithm has good system identification and is convenient for hardware implementation.

本发明公开了一种基于同步的同质传感器突变参数识别方法，中央处理器以T秒为采样周期，同质传感器(传感器观测的是同一物理现象)定时对被测信号进行采样和量化并获得关于时间的数据序列X(N)，将采样数据按周期进行截取，获得m个周期序列为：Y₁(t),Y₂(t),…Y_m(t)，每个周期样本中包含了N个数据点，即Y_i＝[X_i(1),X_i(2),…X_i(N)]，其中i＝1,2,…,m。从每个周期序列各时刻点中抽取一个点值组成单一传感器多次测量的一维数据序列，构造按周期同步构建的周期数据序列，即Y₁′(t),Y₂′(t),…Y′_N(t)，每组一维数据序列中包括了m个数据点，即Y_i′＝[X₁(1),X₂(2),…X_m(N)]。对得到的数据样本Y_t'进行基于极大似然估计和最小二乘法估计的优化组合预处理，用于指导电力系统周期采样数据指导。对基于组合预处理的含有突变传感器信号采用经验模态分解得到各阶IMF模态分量；对信号主导IMF分量采用希尔伯特—黄变换得到瞬时频率图和瞬时幅值图，依据HHT图中的突变点、幅值变化、瞬时频率走势特征信息对传感器异常信号进行参数检测和识别。本发明将不同类传感器进行无差别融合，得到等周期数据样本，消减不同传感器差异和信号采集过程中的随机误差，提高传感器突变参数识别精度。The invention discloses a method for identifying mutation parameters of a homogeneous sensor based on synchronization. The central processing unit takes T seconds as the sampling period, and the homogeneous sensor (the sensor observes the same physical phenomenon) regularly samples and quantifies the measured signal and obtains Regarding the data sequence X(N) of time, the sampled data is intercepted by cycle, and m cycle sequences are obtained as: Y ₁ (t), Y ₂ (t),...Y _m (t), each cycle sample contains N data points are obtained, that is, Y _i =[X _i (1),X _i (2),...X _i (N)], where i=1,2,...,m. Extract a point value from each time point of each cycle sequence to form a one-dimensional data sequence measured multiple times by a single sensor, and construct a cycle data sequence constructed synchronously by cycle, that is, Y ₁ ′(t), Y ₂ ′(t), ...Y′ _N (t), each group of one-dimensional data sequence includes m data points, that is, Y _i ′=[X ₁ (1), X ₂ (2), . . . X _m (N)]. The obtained data sample Y _t ' is preprocessed based on the optimal combination of maximum likelihood estimation and least square estimation, which is used to guide the periodic sampling data guidance of the power system. The empirical mode decomposition is used to obtain the IMF modal components of each order based on the combination preprocessing of the sensor signal containing a sudden change; the Hilbert-Huang transformation is used to obtain the instantaneous frequency map and the instantaneous amplitude map of the dominant IMF component of the signal, according to the HHT map The sudden point, amplitude change, and instantaneous frequency trend feature information are used to detect and identify the abnormal signal of the sensor. In the present invention, different types of sensors are indiscriminately fused to obtain equal-period data samples, the differences between different sensors and random errors in the signal collection process are reduced, and the identification accuracy of sensor mutation parameters is improved.

装置实施例Device embodiment

一种基于同步的同质传感器突变参数识别装置，包括：A synchronization-based homogeneous sensor mutation parameter identification device, comprising:

模块D：对数据样本Y_t'进行基于极大似然估计预处理；Module D: Preprocessing the data sample Y _t ' based on maximum likelihood estimation;

本实施例中所指的装置实际上是实现上述方法实施例的软件构架，其中的各种模块均为软件功能模块，存储与存储器中，由处理器执行。The device referred to in this embodiment is actually a software framework for realizing the above-mentioned method embodiment, and various modules therein are software functional modules, stored in a memory, and executed by a processor.

以上给出了本发明涉及的具体实施方式，但本发明不局限于所描述的实施方式。在本发明给出的思路下，采用对本领域技术人员而言容易想到的方式对上述实施例中的技术手段进行变换、替换、修改，并且起到的作用与本发明中的相应技术手段基本相同、实现的发明目的也基本相同，这样形成的技术方案是对上述实施例进行微调形成的，这种技术方案仍落入本发明的保护范围内。The specific embodiments related to the present invention are given above, but the present invention is not limited to the described embodiments. Under the idea given by the present invention, the technical means in the above-mentioned embodiments are transformed, replaced, and modified in ways that are easy for those skilled in the art, and the functions played are basically the same as those of the corresponding technical means in the present invention. 1. The purpose of the invention realized is also basically the same, and the technical solution formed in this way is formed by fine-tuning the above-mentioned embodiments, and this technical solution still falls within the protection scope of the present invention.

Claims

1. A method for identifying sudden change parameters of a homogeneous sensor based on synchronization, characterized in that the steps are as follows:

Step A: The central processor takes T as the sampling period, and the homogeneous sensor regularly samples and quantifies the measured signal of the system, and obtains the sampling data at the same sampling frequency;

Step B: Intercept the sampling data by cycle, and obtain m cycle sequences: Y ₁ (t), Y ₂ (t),...Y _m (t), and each cycle sample contains N data points, namely Y _i =[X _i (1),X _i (2),...X _i (N)], where i=1,2,...,m;

Step C: Construct a periodic data sequence constructed synchronously by cycle: treat the data at each time point in each cycle data as the multiple measurement results of the same object after synchronization, that is, the data at each time point corresponding to each cycle is a sample, and multiple cycles Each point value in the corresponding time in the data constitutes a data sequence, which can be regarded as a one-dimensional data sequence measured multiple times by a single sensor, that is, a periodic data sequence Y′ ₁ (t) constructed synchronously by cycle is constructed, Y′ ₂ (t),…Y′ _N (t), each group of one-dimensional data sequence includes m data points, that is, Y′ _i =[X ₁ (1),X ₂ (2),…X _m (N)];

Step D: Preprocessing the data sample Y′ _t based on maximum likelihood estimation;

Step E: EMD decomposition is used to obtain the IMF components of each order based on the combined preprocessing-based sensor signal containing mutations;

Step F: HHT transformation is used for the dominant IMF component of the sensor abnormal signal to obtain an instantaneous frequency graph and an instantaneous amplitude graph;

Step G: Detect and identify the sudden change parameters of the sensor according to the sudden change points, amplitude changes, and instantaneous frequency trend characteristic information in the HHT transformation diagram.

2. a kind of homogeneous sensor mutation parameter identification method based on synchronization according to claim 1, is characterized in that, in described step D, carries out the mode of preprocessing based on maximum likelihood estimation to data sample Y ' _t for:

S101: Assuming that at a given moment, the feature of the environment to be measured is X, and the value of the sensor is Y, then the measurement model of the sensor is: Y=f(X)+V where, V is a noise item conforming to the Gaussian distribution; the so-called Data fusion is to obtain the measured values Y ₁ , Y ₂ ,..., Y _N from N sensors, and obtain the optimal estimation of the characteristic parameter X from these measured values according to a certain estimation criterion;

S102: Find a suitable criterion function, that is, when X is estimated to be X(Y), the loss generated is the smallest; take the loss function as uniform loss:

L L [[\overset{^^}{X x} ((Y Y)),, X x]] = = \{\begin{matrix} 11 & | | \overset{^^}{X x} - - X x | | &GreaterEqual; &Greater Equal; \frac{ϵ ϵ}{22} \\ 00 & | | \overset{^^}{X x} - - X x | | < < \frac{ϵ ϵ}{22} \end{matrix}

S103: On the basis of the loss function L, define a function R corresponding to the estimated risk:

Among them, p(x), p(x|y) represent the probability distribution;

S104: Take the minimum risk as the estimation criterion, namely

\frac{\partial \partial R R}{\partial \partial X x} {| |}_{X x = = \overset{^^}{X x} ((Y Y))} = = 00 - - - - - - ((11))

The best estimate in accordance with formula (1) is obtained as

{\overset{^^}{X x}}_{o o p p t t} ((Y Y)) = = arg arg max max p p ((X x | | Y Y))

S105: In a system with N sensors, the corresponding information fusion can be regarded as under the observed values Y ₁ , Y ₂ , ..., Y _N , the value X has the maximum a posteriori estimate as

S106: In order to obtain the maximum a posteriori estimate, p(x)=1 is adopted for all possible parameters X, and it can be obtained by formula derivation:

p p ((X x | | {Y Y}_{11},, {Y Y}_{22},, ... ... ... ... {Y Y}_{N N})) &Proportional; &Proportional; {Π Π}_{i i = = 11}^{N N} p p ((X x | | {Y Y}_{i i})) - - - - - - ((22))

At this time, the maximum a posteriori estimation is simplified to the maximum likelihood estimation, and the corresponding fusion calculation formula is:

{V V}_{f f}^{- - 11} = = {Σ Σ}_{i i = = 11}^{N N} {A A}_{i i}^{T T} {V V}_{i i}^{- - 11} {A A}_{i i} + + {V V}_{X x}^{- - 11} - - - - - - ((33))

{X x}_{f f} = = {V V}_{f f} {Σ Σ}_{i i = = 11}^{N N} {A A}_{i i}^{T T} {V V}_{i i}^{- - 11} {Y Y}_{i i} - - - - - - ((44))

When the sensor is one-dimensional and the coordinate transformation is not considered, the formulas (3) and (4) are simplified as:

{σ σ}_{f f}^{- - 22} = = {Σ Σ}_{i i = = 11}^{N N} {σ σ}_{i i}^{22} - - - - - - ((55))

{y the y}_{f f} = = {Σ Σ}_{i i = = 11}^{N N} \frac{{σ σ}_{f f}^{22}}{{σ σ}_{i i}^{22}} {Y Y}_{i i} - - - - - - ((66))

Suppose Y ₁ , Y ₂ , ..., Y _N are the measured values corresponding to N sensors respectively, according to the least squares estimation criterion, is to make the error function: To achieve the minimum Y value, we get:

3. a kind of homogeneous sensor mutation parameter recognition method based on synchronization according to claim 2, is characterized in that, adopts EMD decomposition to obtain each order IMF component to containing mutation sensor signal based on combined pretreatment in described step E The process is:

S201: Interpolate all local maximum points and all local minimum points of the signal x[t] with a cubic spline function, and fit the upper and lower envelopes; sensor signal;

S202: Find the average curve M ₁ (t) of the upper and lower envelopes, t=1, 2, 3...m, then the difference between the sampling signal x[t] and M ₁ (t) is P ₁ (t) : P ₁ (t)=x[t]-M ₁ (t);

S203: If P ₁ (t) satisfies the following two conditions of IMF at the same time, then it is the first IMF component, otherwise repeat S201 to S202 as the original signal to obtain P ₁₁ (t): P ₁₁ (t) =P ₁ (t)-M ₁₁ (t), wherein: M ₁₁ (t) is the average curve of the upper and lower envelopes of P ₁ (t);

The two conditions that the IMF component satisfies are: in the whole time course, the number of crossing zero points is equal to or at most differs from the number of extremum points; at any point on the signal, the upper envelope defined by the local maximum value and the local minimum value The mean value of the lower envelope defined by the point is 0, that is, the signal is locally symmetrical about the time axis;

S204: Repeat the above steps to screen until the P _1k (t) obtained by formula (1) meets the two conditions of the IMF component: P _1k (t)=P _1(1-k) (t)- M _1k (t) (1)

S205: Determine whether each screening result is an IMF component by calculating the threshold value _SD through formula (2):

{S S}_{D D.} = = {Σ Σ}_{t t = = 00}^{r r} | | \frac{{P P}_{11 ((k k - - 11))} ((t t)) - - {P P}_{11 k k} ((t t))}{{P P}^{22}_{11 ((k k - - 11))} ((t t))} | |;; - - - - - - ((22))

Where: r is the number of sampling points of the sensor signal, and the threshold _SD is usually 0.2 to 0.3;

S206: Let C ₁ (t)=P _1k (t), then C ₁ (t) is the first IMF component, which contains the IMF component with the shortest period in the original signal x[t]; C ₁ (t ) is separated from x[t]: R ₁ (t)=x[t]-C ₁ (t);

S207: Repeat the above steps S201 to S205n times with R ₁ (t) as a new value to obtain n IMF components of the signal x[t]: R _n (t)=R _n-1 (t)-C _n (t );

S208: When R _n (t) is a monotone function and no other components can be decomposed from the signal x[t], the whole decomposition process ends; get:

4. a kind of method based on synchronous homogeneous sensor sudden change parameter identification method according to claim 3, it is characterized in that, described step F adopts HHT transformation to obtain instantaneous frequency graph and instantaneous amplitude graph to the leading IMF component of sensor abnormal signal The process is:

S301: Perform Hilbert transform on all IMF components obtained after EMD decomposition; the Hilbert form of given C(t) is:

Among them, λ is the integral variable;

S302: Construct an analytical signal Z(t):

Z(t)=C(t)+iH(t)=A(t)e ^iθ(t) ;

S303: Amplitude function:

S304: Argument function:

S305: Instantaneous frequency:

5. according to a kind of synchronous homogeneous sensor mutation parameter recognition method based on any one of claim 1-4, it is characterized in that, in the described step G, according to the mutation point, amplitude change, The process of detection and identification of sudden frequency trend characteristic information sensor mutation parameters is as follows:

S306: classify the sudden change parameters in the sensor: sudden drop, sudden rise, harmonic, pulse;

S307: Classify and identify when and what kind of mutation the sensor has experienced, according to the different characteristic information such as the mutation point, amplitude change, and instantaneous frequency trend of different sensor abnormal signals in the HHT diagram.

6. A synchronization-based homogeneous sensor mutation parameter identification device, characterized in that it comprises:

Module A: The central processor takes T as the sampling period, and the homogeneous sensor regularly samples and quantifies the measured signal of the system, and obtains the sampling data at the same sampling frequency;

Module B: Intercept the sampling data by cycle, and obtain m cycle sequences: Y ₁ (t), Y ₂ (t),...Y _m (t), and each cycle sample contains N data points, namely Y _i =[X _i (1),X _i (2),...X _i (N)], where i=1,2,...,m;

Module C: Construct a periodic data sequence constructed synchronously by cycle: treat the data at each time point in each cycle data as the multiple measurement results of the same object after synchronization, that is, the data at each time point corresponding to each cycle is a sample, and multiple cycles Each point value in the corresponding time in the data constitutes a data sequence, which can be regarded as a one-dimensional data sequence measured multiple times by a single sensor, that is, a periodic data sequence Y′ ₁ (t) constructed synchronously by cycle is constructed, Y′ ₂ (t),…Y′ _N (t), each group of one-dimensional data sequence includes m data points, that is, Y′ _i =[X ₁ (1),X ₂ (2),…X _m (N)];

Module D: preprocessing the data sample Y′ _t based on maximum likelihood estimation;

Module E: EMD decomposition is used to obtain the IMF components of each order for the sensor signal containing mutation based on combined preprocessing;

Module F: use HHT transformation to obtain the instantaneous frequency map and instantaneous amplitude map of the dominant IMF component of the abnormal signal of the sensor;

Module G: Detect and identify sensor mutation parameters based on the mutation point, amplitude change, and instantaneous frequency trend characteristic information in the HHT transformation diagram.

7. A kind of synchronous based homogeneous sensor sudden change parameter recognition device according to claim 6, is characterized in that, in described module D, carries out the optimized combination preprocessing based on maximum likelihood estimation to data sample Y ' _t The way is:

L L [[\overset{^^}{X x} ((Y Y)),, X x]] = = \{\begin{matrix} 11 & | | \overset{^^}{X x} - - X x | | &GreaterEqual; &Greater Equal; \frac{ϵ ϵ}{22} \\ 00 & | | \overset{^^}{X x} - - X x | | < < \frac{ϵ ϵ}{22} \end{matrix}

Among them, p(x), p(x|y) represent the probability distribution;

S104: Take the minimum risk as the estimation criterion, namely

\frac{\partial \partial R R}{\partial \partial X x} {| |}_{X x = = \overset{^^}{X x} ((Y Y))} = = 00 - - - - - - ((11))

The optimal estimate (maximum a posteriori estimate) conforming to formula (1) can be obtained as

{\overset{^^}{X x}}_{o o p p t t} ((Y Y)) = = arg arg max max p p ((X x | | Y Y))

p p ((X x | | {Y Y}_{11},, {Y Y}_{22},, ... ... ... ... {Y Y}_{N N})) &Proportional; &Proportional; {Π Π}_{i i = = 11}^{N N} p p ((X x | | {Y Y}_{i i})) - - - - - - ((22))

{V V}_{f f}^{- - 11} = = {Σ Σ}_{i i = = 11}^{N N} {A A}_{i i}^{T T} {V V}_{i i}^{- - 11} {A A}_{i i} + + {V V}_{X x}^{- - 11} - - - - - - ((33))

{X x}_{f f} = = {V V}_{f f} {Σ Σ}_{i i = = 11}^{N N} {A A}_{i i}^{T T} {V V}_{i i}^{- - 11} {Y Y}_{i i} - - - - - - ((44))

When the sensor is one-dimensional, and the coordinate transformation is not considered, the equations (3) and (4) can be simplified as:

{σ σ}_{f f}^{- - 22} = = {Σ Σ}_{i i = = 11}^{N N} {σ σ}_{i i}^{22} - - - - - - ((55))

{y the y}_{f f} = = {Σ Σ}_{i i = = 11}^{N N} \frac{{σ σ}_{f f}^{22}}{{σ σ}_{i i}^{22}} {Y Y}_{i i} - - - - - - ((66))

8. a kind of homogeneous sensor mutation parameter identification device based on synchronization according to claim 7, is characterized in that, adopts EMD decomposition to obtain the IMF components of each order to contain the mutation sensor signal based on combination pretreatment in the described module E The process is:

S205: In addition to the signal, in the actual calculation, the threshold value _SD can be obtained by formula (2) to determine whether each screening result is an IMF component:

{S S}_{D D.} = = {Σ Σ}_{t t = = 00}^{r r} | | \frac{{P P}_{11 ((k k - - 11))} ((t t)) - - {P P}_{11 k k} ((t t))}{{P P}^{22}_{11 ((k k - - 11))} ((t t))} | |;; - - - - - - ((22))

9. A kind of synchronous based homogeneous sensor mutation parameter identification device according to claim 8, characterized in that, said module F adopts HHT transformation to obtain instantaneous frequency graph and instantaneous amplitude graph for the dominant IMF component of sensor abnormal signal The process is:

Among them, λ is the integral variable;

S302: Construct an analytical signal Z(t):

Z(t)=C(t)+iH(t)=A(t)e ^iθ(t) ;

S303: Amplitude function:

S304: Argument function:

S305: Instantaneous frequency:

10. A kind of homogeneous sensor mutation parameter identification device based on synchronization according to any one of claims 6-9, characterized in that, in the module G, according to the mutation point, the amplitude change, The process of detection and identification of sudden frequency trend characteristic information sensor mutation parameters is as follows:

S30: According to the different characteristic information such as the mutation point, the amplitude change and the instantaneous frequency trend of different sensor abnormal signals in the HHT diagram, classify and identify when and what kind of mutation the sensor has experienced.