Insulation detection method and battery management system based on signal estimation

Abstract

The invention discloses an insulation detection method and a battery management system based on signal estimation, comprising: acquiring historical observation data, thereby determining an actual observation signal and background noise, and constructing a Bayesian linear model; adding random noise to the historical observation data, and obtaining an estimated value of the actual observation signal according to the Bayesian linear model; determining a Bayesian mean square error matrix according to an error between the estimated value and the true value of the actual observation signal, with the goal of minimizing the trace of the Bayesian mean square error matrix, and solving the optimal probability density distribution of the random noise according to random noise constraints, and determining the current optimal weight, and constructing a Bayesian signal estimation model; generating random noise according to the optimal probability density distribution, adding the generated random noise to the observation data, and obtaining the actual observation signal by adopting the Bayesian signal estimation model; thereby performing insulation detection according to the actual observation signal, and enhancing the signal measurement and estimation capabilities of the insulation detection equipment.

Description

Technical Field

[0001] The present invention relates to the technical field of battery energy storage and electric vehicle insulation detection, and in particular to an insulation detection method based on signal estimation and a battery management system. Background Art

[0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art.

[0003] With the continuous development of the new energy industry, the voltage levels of battery energy storage systems and electric vehicles are constantly increasing. The voltage level of battery energy storage systems is as high as 1500V or more, and the voltage level of electric vehicles is as high as 900V or more. For safety reasons, high-voltage battery systems require that the total positive and negative terminals of the battery be electrically insulated from the shell. When the insulation is damaged, electric shock accidents may occur, which may cause personal injury or death in severe cases, and cause short circuits, fires, or even explosions in the battery pack, threatening the personal safety of energy storage system operators and electric vehicle passengers. Therefore, real-time monitoring of the insulation status of the battery pack, timely issuing early warning information and performing relevant protection processing based on the monitoring results are of great significance to the safe and stable operation of electric vehicles and battery energy storage systems.

[0004] At present, the insulation detection methods mainly include balanced bridge method, unbalanced bridge method, AC signal injection method and active insulation resistance detection method. The measured values ​​of balanced bridge method and unbalanced bridge method are related to multiple factors such as total voltage of battery pack and bridge arm voltage. The measured value of AC signal injection method is only related to the sampling value of detection point. It has simple structure and good operability, and has been widely used in battery energy storage system and electric vehicles.

[0005] Patent CN110109027B discloses a battery pack insulation impedance detection circuit and device, and a battery management device, which sets the AC signal injection point at the middle position of the battery pack voltage, reducing the impact of battery voltage fluctuations to a certain extent. However, since this method lacks an effective signal estimation method, in applications where the Y capacitor is large, its detection cycle is long, resulting in poor timeliness and low detection accuracy.

[0006] Patent CN107064638B discloses an insulation resistance detection device and method based on an injection cycle adaptive strategy, which uses an injection signal cycle adaptive method to reduce the impact of Y capacitors on insulation detection. However, this solution does not consider the impact of battery voltage fluctuations. Since battery voltage fluctuations can lead to inaccurate observation values, the detection accuracy is reduced in actual use. Summary of the invention

[0007] The traditional AC signal injection insulation detection method requires the measurement of weak voltage signals. Due to the strong electromagnetic interference in the energy storage system and the battery pack of electric vehicles, the measurement accuracy of weak voltage signals is poor. In order to solve the above problems, the present invention proposes an insulation detection method and battery management system based on signal estimation to achieve low-precision quantitative observation values, enhance the measurement and estimation capabilities of insulation detection equipment for signals, and improve the insulation detection accuracy of battery energy storage systems and electric vehicles.

[0008] In order to achieve the above object, the present invention adopts the following technical solution:

[0009] In a first aspect, the present invention provides an insulation detection method based on signal estimation, comprising:

[0010] Obtain historical observation data to determine the actual observation signal and background noise, and build a Bayesian linear model;

[0011] Add the same distributed random noise to the historical observation data, and assign an initial weight to each node of the historical observation data, so as to obtain an estimated value of the actual observation signal based on the Bayesian linear model for the historical observation data with added random noise;

[0012] According to the error between the estimated value and the true value of the actual observed signal, the Bayesian mean square error matrix is ​​determined, and the objective function is to minimize the trace of the Bayesian mean square error matrix. According to the constraints of the set random noise, the optimal probability density distribution of the random noise is solved, and the current optimal weight is determined, so as to construct a Bayesian signal estimation model;

[0013] Obtaining observation data, generating random noise according to the optimal probability density distribution, adding the generated random noise to the observation data, and using the Bayesian signal estimation model to obtain the actual observation signal for the observation data with the added random noise;

[0014] The insulation resistance is obtained according to the actual observation signal, and the insulation detection result is obtained according to the comparison between the insulation resistance and the preset threshold value.

[0015] As an optional implementation, when the actual observed signal and the background noise obey a joint Gaussian distribution, the Bayesian mean square error matrix is:

[0016]

[0017] The Bayesian signal estimation model is:

[0018]

[0019] Among them, E s (s) is the mean vector of the actual observed signal s, B is the Bayesian mean square error matrix, C sis the covariance matrix, C sy is the cross-correlation matrix between the actual observed signal s and the output vector y, C y is the cross-covariance matrix of the output vector y, C ys is the cross-correlation matrix C sy The transposed matrix of E is , s is the actual observed signal, y is the output vector of the sensor network, x,β (·) is the expectation operator.

[0020] As an optional implementation, when the actual observed signal is uniformly distributed, the Bayesian mean square error matrix is:

[0021]

[0022] The Bayesian signal estimation model is:

[0023]

[0024] Among them, E s (s) is the mean vector of the actual observed signal s, s is the actual observed signal, C s is the covariance matrix, C sy is the cross-correlation matrix between the actual observed signal s and the output vector y, E(y) is the mean vector of the output vector, y is the sensor network output vector, C y is the covariance matrix of the output vector y, and T is the number of iterations.

[0025] As an optional implementation, the optimal weight is Among them, C y is the cross-covariance matrix of the output vector y, C sy is the actual observed signal s k The cross-correlation matrix between the output vector y, C ys is the cross-correlation matrix C sy The transposed matrix of .

[0026] As an optional implementation, the probability density distribution of random noise is:

[0027]

[0028] Among them, λ k is the normalization coefficient, φ() is the Gaussian kernel function, μ k is the mean, σ k is the standard deviation, β is the random noise, and k is the number of observed signals.

[0029] As an optional implementation, the constraints of the random noise include:

[0030]

[0031] λ k ≥0,k=1,2,…K

[0032] σ k ≥0,k=1,2,…K.

[0033] In a second aspect, the present invention provides an insulation detection system based on signal estimation, comprising:

[0034] A linear model building module, configured to obtain historical observation data, thereby determining the actual observation signal and background noise, and building a Bayesian linear model;

[0035] The initial estimation module is configured to add the same distributed random noise to the historical observation data, and assign an initial weight to each node of the historical observation data, so as to obtain an estimated value of the actual observation signal according to the Bayesian linear model for the historical observation data with the added random noise;

[0036] The signal estimation model building module is configured to determine the Bayesian mean square error matrix according to the error between the estimated value and the true value of the actual observed signal, take minimizing the trace of the Bayesian mean square error matrix as the objective function, and solve the optimal probability density distribution of the random noise according to the set random noise constraint conditions, and determine the current optimal weight value, so as to build the Bayesian signal estimation model;

[0037] A signal estimation module is configured to obtain observation data, generate random noise according to an optimal probability density distribution, add the generated random noise to the observation data, and use a Bayesian signal estimation model to obtain an actual observation signal for the observation data to which the random noise is added;

[0038] The insulation detection module is configured to obtain the insulation resistance according to the actual observation signal, and obtain the insulation detection result according to the comparison between the insulation resistance and the preset threshold value.

[0039] In a third aspect, the present invention provides an electronic device comprising a memory and a processor, and computer instructions stored in the memory and executed on the processor, wherein when the computer instructions are executed by the processor, the method described in the first aspect is performed.

[0040] In a fourth aspect, the present invention provides a computer-readable storage medium for storing computer instructions, wherein when the computer instructions are executed by a processor, the method described in the first aspect is performed.

[0041] In a fifth aspect, the present invention provides a battery management system, including the insulation detection system with adaptive sampling frequency as described in the second aspect.

[0042] Compared with the prior art, the present invention has the following beneficial effects:

[0043] The present invention proposes an insulation detection method and a battery management system based on signal estimation, which realize low-precision quantitative observation values, enhance the measurement and estimation capabilities of insulation detection equipment for signals, and improve the insulation detection accuracy of battery energy storage systems and electric vehicles.

[0044] The present invention proposes an insulation detection method and a battery management system based on signal estimation. Random noise is artificially added, and a quantized low-precision output value is used as the actual observation signal. A Bayesian signal estimation model is designed. The mean square error between the output value of the estimation model and the actual value is used as the optimization objective function to solve the optimal probability density distribution of adding random noise. The optimal random noise is generated using the optimal probability density distribution. The estimation accuracy of the actual observation signal can be improved to the performance level of a minimum mean square error estimator constructed based on the original measurement data. The mean square error of the Bayesian signal estimation model reaches the Bayesian lower bound, which greatly improves the signal estimation capability of the low-precision sensor network output.

[0045] In order to improve the accuracy of insulation detection and reduce equipment power consumption and design complexity, the present invention proposes a Bayesian signal estimation model based on low-precision quantized observations. A sensor network node model composed of multiple quantizers is used for a single measurement data. The sum of the measurement data and independent and identically distributed random noise is input into each sensor network node. The outputs of the multiple quantizers are summed and averaged, and the result is used as the output signal of the network node. This can effectively reduce the impact of battery voltage fluctuations and Y capacitors and improve insulation detection accuracy.

[0046] Advantages of additional aspects of the present invention will be given in part in the following description, and in part will become obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] The accompanying drawings in the specification, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0048] Figure 1 A circuit schematic diagram of measuring insulation resistance using a signal injection method provided in Example 1 of the present invention;

[0049] Figure 2 A simplified schematic diagram of a circuit for measuring insulation resistance using a signal injection method provided in Example 1 of the present invention;

[0050] Figure 3 A flow chart of an insulation detection method based on signal estimation provided in Embodiment 1 of the present invention;

[0051] Figure 4 A schematic diagram of a Bayesian signal estimation model provided in Example 1 of the present invention;

[0052] Figure 5 A schematic diagram of the optimal probability density distribution function of random noise provided in Example 1 of the present invention;

[0053] Figure 6 A diagram of simulation results of mean square error as observed data increases provided in Example 1 of the present invention;

[0054] Figure 7 A diagram of simulation results of mean square error as the signal-to-noise ratio increases when the amplitude signal provided in Example 1 of the present invention is Gaussian distributed;

[0055] Figure 8 A diagram of simulation results of mean square error as the signal-to-noise ratio increases when the amplitude signal provided in Example 1 of the present invention is uniformly distributed. DETAILED DESCRIPTION

[0056] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0057] It should be noted that the following detailed descriptions are exemplary and are intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present invention belongs.

[0058] It should be noted that the terms used herein are only for describing specific embodiments, and are not intended to limit exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that the terms "include" and "have" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0059] In the absence of conflict, the embodiments of the present invention and the features of the embodiments may be combined with each other.

[0060] Example 1

[0061] This embodiment provides an insulation detection method based on signal estimation, including:

[0062] Obtain historical observation data to determine the actual observation signal and background noise, and build a Bayesian linear model;

[0063] Add the same distributed random noise to the historical observation data, and assign an initial weight to each node of the historical observation data, so as to obtain an estimated value of the actual observation signal based on the Bayesian linear model for the historical observation data with added random noise;

[0064] According to the error between the estimated value and the true value of the actual observed signal, the Bayesian mean square error matrix is ​​determined;

[0065] Taking the minimization of the trace of the Bayesian mean square error matrix as the objective function, and according to the set random noise constraints, solving the optimal probability density distribution of random noise and determining the current optimal weights, the Bayesian signal estimation model is constructed;

[0066] Obtaining observation data, generating random noise according to the optimal probability density distribution, adding the generated random noise to the observation data, and using the Bayesian signal estimation model to obtain the actual observation signal for the observation data with the added random noise;

[0067] The insulation resistance is obtained according to the actual observation signal, and the insulation detection result is obtained according to the comparison between the insulation resistance and the preset threshold value.

[0068] like Figure 1 The figure shows the circuit schematic diagram of measuring insulation resistance using the signal injection method. Rp and Rn are the positive insulation resistance and the negative insulation resistance respectively, R is the sampling resistor inside the insulation detection device, BAT is the battery pack, PE is the system chassis, and GEN is the AC signal generator. For AC power supply, the DC power supply with very small internal resistance in the circuit can be regarded as a short circuit, so Figure 1 Can be simplified as Figure 2 The circuit shown. Figure 2 Where Req is the equivalent parallel insulation resistance of Rp and Rn. The value of the insulation resistance Req is obtained by measuring the voltage V on the sampling resistor R inside the insulation detection device, so as to determine whether a short circuit fault occurs in the battery energy storage system and the electric vehicle, so as to complete the insulation detection.

[0069] However, due to the existence of battery voltage fluctuations, the observed data will be inaccurate. Therefore, in order to improve the measurement accuracy of the voltage V, this embodiment proposes a signal estimation method based on low-precision quantized observation values ​​to achieve insulation detection, such as Figure 3 As shown, the specific steps are as follows:

[0070] Step 1: Get the measured data x = [x 1 ,x 2 ,…,x N ] T , N represents the length of the measured data vector x, which can be used to determine the N×p-dimensional measurement matrix Q and the p×1-dimensional actual observation signal s=[s 1 ,s 2 ,…,s p ] T And the background noise with zero mean α=[α 1 ,α 2 ,…,αN ] T , where the mean vector of the actual observed signal s is E s (s), the covariance matrix is ​​C s , the covariance matrix of the background noise α is C α ;

[0071] The constructed Bayesian linear model is:

[0072] x=Qs+α (1)

[0073] Step 2: For each observation x n Make L copies and add independent and identically distributed random noise β nl ;

[0074] Several observation data nodes constitute a sensor network. A parallel quantizer array is introduced at the observation data node of the sensor network, that is, L identical binary quantizers h are used. n (·), the output of each quantizer is y nl =h n (x n +β nl ), sum them and take the average value to get the output signal of each observation data node:

[0075]

[0076] Where, l = 1, 2, ..., L, n = 1, 2, ..., N;

[0077] The quantizer is:

[0078]

[0079] Among them, γ is the threshold.

[0080] In this embodiment, the output signals of N observation nodes are each assigned a weight, then the kth actual observation signal s k The estimated value of is:

[0081]

[0082] Among them, w 0 is the bias weight coefficient, w=[w 1 ,w 2 ,…,w N ] T is a weight vector of N×1 dimension, y=[y 1 ,y 2 ,…,y N ] T is the output vector of the N×1 dimensional sensor network.

[0083] Step 3: Take the kth actual observation signal s k Estimated value of With the true value s k The error between but The Bayesian mean square error is:

[0084] E x,β (ε 2 )=E x,β [(w 0 +w T ys k ) 2 ] (5)

[0085] Among them, E x,β (·) represents the joint probability distribution f x,β The expectation operator of (x,β).

[0086] Using the quadratic function relationship between the Bayesian mean square error and the bias weight coefficient, let the partial derivative The optimal weight bias coefficient is obtained as:

[0087]

[0088] but The Bayesian mean square error is:

[0089]

[0090] Among them, C y =E x,β [(yE x,β (y))(yE x,β (y)) T ], and

[0091] Let the partial differential equation The optimal weight vector obtained is:

[0092]

[0093] Then the kth actual observed signal s k The Bayesian estimator of is:

[0094]

[0095] Its Bayesian mean square error is:

[0096]

[0097] Step 4: Transform k actual observation signals s k Estimated value of The combination is constructed into a Bayesian signal estimation model, such as Figure 4 As shown:

[0098]

[0099] Among them, C sy =E x,β [(sE s (s))(yE x,β (y)) T ],and but The Bayesian mean square error matrix is:

[0100]

[0101] In the above steps 2 to 4, due to the ease of artificially adding random noise, assuming that the number of parallel quantizers is large enough, the output signal y of the nth node n The expectation is:

[0102]

[0103] in, [Q] n is the nth row of the observation matrix Q, is the background noise α n The probability distribution function, f s (s) is the prior probability distribution of the actual observed signal s, f β (β) is the probability density distribution function of the added random noise.

[0104] The output signal y of the nth node n The second moment of is:

[0105]

[0106] The kth actual observed signal s k and the output signal y of the nth node n The cross-correlation is:

[0107]

[0108] The output signal y of the nth and jth nodes n and j The cross correlation is:

[0109]

[0110] Will and Substitute into the cross-correlation matrix C sy and the cross-covariance matrix C y , C sy The element in row k and column n of Where n=1,2,…,N,k=1,2,…,p;C y The diagonal elements of The off-diagonal elements are

[0111] Using the obtained C sy and C y , get the estimated value and its Bayesian mean square error matrix

[0112] Step 5: The probability density distribution of random noise is:

[0113]

[0114] Among them, λ k is the normalization coefficient, φ() is the Gaussian kernel function, and the Gaussian kernel function μ k is the mean, σ k is the standard deviation;

[0115] The objective function is to minimize the trace of the Bayesian mean square error matrix B, and the constraints of random noise are specifically:

[0116]

[0117] In this embodiment, the optimization problem of the probability density distribution of random noise can be solved using a sequential quadratic programming optimization algorithm; sequential quadratic programming optimization is a commonly used algorithm for solving nonlinear function constraint optimization problems, and can be solved with the help of commonly used optimization software packages such as Matlab.

[0118] Step 6: According to the optimal probability density distribution Generate N×L random noise β nl , and add it to the observed data, and use the Bayesian signal estimation model to get the actual observed signal, that is, the voltage V, so that the insulation resistance value can be obtained.

[0119] In this embodiment, the above process is when the actual observation signal and the background noise obey the joint Gaussian distribution. If the actual observation signal obeys the uniform distribution, the calculation of the second-order moment in the above calculation method needs to be approximated by the numerical method. The specific steps are:

[0120] The constructed Bayesian linear model and quantizer are:

[0121] x=Qs+α

[0122]

[0123] The output signal of the sensor network node is:

[0124]

[0125] After T iterations, find y n The expected mean of is:

[0126]

[0127] For each observation x n Perform the above steps and combine to get the output vector y=[y 1 ,y 2 ,…,y N ] T and the mean vector E(y) = [E(y 1 ),E(y 2 ),…,E(y N )] T ;

[0128] After T iterations, the covariance matrix of the output vector y is The cross-covariance matrix between the actual observed signal s and the output vector y is

[0129] The estimated value and the Bayesian mean square error matrix distribution are:

[0130]

[0131]

[0132] Taking minimizing the trace of the Bayesian mean square error matrix as the objective function, the optimal probability density distribution of random noise is solved;

[0133] The remaining steps are the same and will not be repeated here.

[0134] Experimental verification

[0135] Experimental verification 1:

[0136] It is known that the actual observed signal s = [s 1 ,s 2 ] T It obeys a zero-mean Gaussian distribution, and its covariance matrix is ​​C s = [1, 0; 0, 0.6], it is known that the background noise α of zero-mean Gaussian distribution has a covariance matrix C α=0.5I N , I N is an N-order unit matrix, and the deterministic measurement matrix is:

[0137] The Bayesian linear model is: x = Qs + α;

[0138] Each node in the sensor network uses the same quantizer: Here, the threshold γ=0.

[0139] The probability function distribution of adding random noise β is composed of K = 10 Gaussian kernel functions, and the initial parameters are randomly generated and meet the constraints;

[0140] According to the constraints, and taking the trace tr(B) of the Bayesian mean square error matrix B as the minimum objective function, the optimal probability density function of adding random noise is like Figure 5 As shown;

[0141] Let the number of network nodes N = 3, that is, the number of observation data, and the Bayesian mean square error matrix of the minimum mean square error estimator based on the observation data x is B MMSE =[0.2555,-0.1314;-0.1314,0.1533]; Note that when the joint distribution of the actual observed signal and the background noise is a Gaussian prior distribution, the minimum mean square error estimate based on the observed data x For the calculation of its mean square error matrix, please refer to the book [SM Kay, Fundamentals of Statistical Signal Processing: Estimation Theory, Vol. 1, Prentice-Hall, Upper Saddle River, New Jersey, 1998], which will not be repeated here.

[0142] Add content such as Figure 5 The random noise with the probability density distribution shown in the figure has a Bayesian mean square error matrix of B=[0.2555,-0.1314;-0.1314,0.1533]. It can be found that the Bayesian signal estimation model designed in this embodiment can reach the level of the minimum mean square error estimator based on the observed data x.

[0143] Figure 6 The Bayesian signal estimation model designed for this embodiment adds the optimal probability density function When the random noise is , the trace of the Bayesian mean square error matrix decreases as the number of observations N increases. For the convenience of comparison, the minimum mean square error estimator based on the original observation data x and the trace of its mean square error matrix are also shown in Figure 6It can be seen that the Bayesian signal estimation model designed in this embodiment can achieve the performance level of the minimum mean square error estimator based on the original measurement data x.

[0144] Experimental verification 2:

[0145] Given the number of observations N = 3, the Bayesian linear model x n =A cos(2πf(n-1))+α n , where the random signal frequency f = 1 / N is known, the amplitude A is unknown, and the background noise α n is a zero-mean Gaussian distribution with a variance of Add random noise β nm is a zero-mean Gaussian distribution with a variance root σ β For the design parameters to be optimized, the same quantizer is selected in the sensor network: The threshold γ=0.

[0146] like Figure 7 As shown, the amplitude A follows the mean of 0 and the variance is When the Gaussian distribution is used, the distribution form of the added noise takes the form of a Gaussian kernel function, that is, Gaussian noise, with the added noise variance root σ β The increase of signal amplitude is estimated For the convenience of comparison, Figure 7 The minimum mean square error estimate of the original observation data x is also shown in It can be seen that as the noise variance root σ is added β increases to the optimal value of 3.208. The Bayesian signal estimation model designed in this embodiment can achieve the performance level of the minimum mean square error estimator based on the original observation data x.

[0147] like Figure 8 The variance is shown when the amplitude A follows a uniform distribution on [0,1]. Number of quantizers L = 2 × 10 3 , the number of simulation experiments is T = 10 4 . With the variance root of the added noise σ β The increase of signal amplitude is estimated The simulation result diagram of the mean square error of Figure 8 The minimum mean square error estimate of the original measurement data x is also shown in It can be seen that with the addition of the optimal noise variance root σ β increases to the optimal value, the Bayesian signal estimation model designed in this embodiment can achieve the performance level of the minimum mean square error estimator based on the original observation data x.

[0148] Example 2

[0149] This embodiment provides an insulation detection system based on signal estimation, including:

[0150] A linear model building module, configured to obtain historical observation data, thereby determining the actual observation signal and background noise, and building a Bayesian linear model;

[0151] The initial estimation module is configured to add the same distributed random noise to the historical observation data, and assign an initial weight to each node of the historical observation data, so as to obtain an estimated value of the actual observation signal according to the Bayesian linear model for the historical observation data with the added random noise;

[0152] The signal estimation model building module is configured to determine the Bayesian mean square error matrix according to the error between the estimated value and the true value of the actual observed signal, take minimizing the trace of the Bayesian mean square error matrix as the objective function, and solve the optimal probability density distribution of the random noise according to the set random noise constraint conditions, and determine the current optimal weight value, so as to build the Bayesian signal estimation model;

[0153] A signal estimation module is configured to obtain observation data, generate random noise according to an optimal probability density distribution, add the generated random noise to the observation data, and use a Bayesian signal estimation model to obtain an actual observation signal for the observation data to which the random noise is added;

[0154] The insulation detection module is configured to obtain the insulation resistance according to the actual observation signal, and obtain the insulation detection result according to the comparison between the insulation resistance and the preset threshold value.

[0155] It should be noted that the above modules correspond to the steps described in Example 1, and the examples and application scenarios implemented by the above modules and the corresponding steps are the same, but are not limited to the contents disclosed in the above Example 1. It should be noted that the above modules, as part of the system, can be executed in a computer system such as a set of computer executable instructions.

[0156] In further embodiments, there is also provided:

[0157] A battery management system includes the insulation detection system based on signal estimation described in Example 2.

[0158] An electronic device includes a memory and a processor, and computer instructions stored in the memory and executed on the processor, wherein when the computer instructions are executed by the processor, the method described in Embodiment 1 is performed. For the sake of brevity, it will not be described in detail here.

[0159] It should be understood that in this embodiment, the processor may be a central processing unit CPU, and the processor may also be other general-purpose processors, digital signal processors DSP, application-specific integrated circuits ASIC, off-the-shelf programmable gate arrays FPGA or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0160] The memory may include a read-only memory and a random access memory, and provide instructions and data to the processor. A portion of the memory may also include a non-volatile random access memory. For example, the memory may also store information about the device type.

[0161] A computer-readable storage medium is used to store computer instructions, and when the computer instructions are executed by a processor, the method described in Example 1 is completed.

[0162] The method in Example 1 can be directly embodied as a hardware processor, or a combination of hardware and software modules in the processor. The software module can be located in a mature storage medium in the field such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory, or an electrically erasable programmable memory, a register, etc. The storage medium is located in the memory, and the processor reads the information in the memory and completes the steps of the above method in combination with its hardware. To avoid repetition, it is not described in detail here.

[0163] Those skilled in the art will appreciate that the units, i.e., algorithm steps, of the various examples described in conjunction with this embodiment can be implemented in electronic hardware or in a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0164] Although the above describes the specific implementation mode of the present invention in conjunction with the accompanying drawings, it is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art on the basis of the technical solution of the present invention without creative work are still within the scope of protection of the present invention.

Claims

1. An insulation detection method based on signal estimation, characterized in that: include: Obtain historical observation data to determine the actual observation signal and background noise, and build a Bayesian linear model; Add the same distributed random noise to the historical observation data, and assign an initial weight to each node of the historical observation data, so as to obtain an estimated value of the actual observation signal based on the Bayesian linear model for the historical observation data with added random noise; According to the error between the estimated value and the true value of the actual observed signal, the Bayesian mean square error matrix is ​​determined, and the objective function is to minimize the trace of the Bayesian mean square error matrix. According to the constraints of the set random noise, the optimal probability density distribution of the random noise is solved, and the current optimal weight is determined, so as to construct a Bayesian signal estimation model; Obtaining observation data, generating random noise according to the optimal probability density distribution, adding the generated random noise to the observation data, and using the Bayesian signal estimation model to obtain the actual observation signal for the observation data with the added random noise; The insulation resistance is obtained according to the actual observation signal, and the insulation detection result is obtained according to the comparison between the insulation resistance and the preset threshold value.

2. The insulation detection method based on signal estimation according to claim 1, characterized in that: When the actual observed signal and background noise obey the joint Gaussian distribution, the Bayesian mean square error matrix is: The Bayesian signal estimation model is: in, The actual observed signal The mean vector of is the Bayesian mean square error matrix, is the covariance matrix, The actual observed signal and the output vector The cross-correlation matrix between The output vector The cross-covariance matrix of is the cross-correlation matrix The transposed matrix of is the actual observed signal, is the sensor network output vector, is the expectation operator.

3. The insulation detection method based on signal estimation according to claim 1, characterized in that: When the actual observed signal is uniformly distributed, the Bayesian mean square error matrix is: The Bayesian signal estimation model is: in, The actual observed signal The mean vector of is the actual observed signal, is the covariance matrix, The actual observed signal and the output vector The cross-correlation matrix between The mean vector of the output vector, is the sensor network output vector, The output vector The covariance matrix of is the number of iterations.

4. The insulation detection method based on signal estimation according to any one of claims 1 to 3, characterized in that: The optimal weight is ,in, The output vector The cross-covariance matrix of The actual observed signal and the output vector The cross-correlation matrix between is the cross-correlation matrix The transposed matrix of .

5. The insulation detection method based on signal estimation according to any one of claims 1 to 3, characterized in that: The probability density distribution of random noise is: in, is the normalization coefficient, is the Gaussian kernel function, is the mean, is the standard deviation, is random noise, k is the number of observed signals.

6. The insulation detection method based on signal estimation according to claim 5, characterized in that: The constraints on random noise include: 。 7. An insulation detection system based on signal estimation, characterized in that: include: A linear model building module, configured to obtain historical observation data, thereby determining the actual observation signal and background noise, and building a Bayesian linear model; The initial estimation module is configured to add the same distributed random noise to the historical observation data, and assign an initial weight to each node of the historical observation data, so as to obtain an estimated value of the actual observation signal according to the Bayesian linear model for the historical observation data with the added random noise; The signal estimation model building module is configured to determine the Bayesian mean square error matrix according to the error between the estimated value and the true value of the actual observed signal, take minimizing the trace of the Bayesian mean square error matrix as the objective function, and solve the optimal probability density distribution of the random noise according to the set random noise constraint conditions, and determine the current optimal weight value, so as to build the Bayesian signal estimation model; A signal estimation module is configured to obtain observation data, generate random noise according to an optimal probability density distribution, add the generated random noise to the observation data, and use a Bayesian signal estimation model to obtain an actual observation signal for the observation data to which the random noise is added; The insulation detection module is configured to obtain the insulation resistance according to the actual observation signal, and obtain the insulation detection result according to the comparison between the insulation resistance and the preset threshold value.

8. A battery management system, characterized in that: The invention comprises the insulation detection system based on signal estimation as claimed in claim 7.

9. An electronic device, characterized in that: The method comprises a memory and a processor and computer instructions stored in the memory and executed on the processor, wherein when the computer instructions are executed by the processor, the method according to any one of claims 1 to 6 is completed.

10. A computer-readable storage medium, characterized in that: Used to store computer instructions, which, when executed by a processor, complete the method described in any one of claims 1 to 6.

Images