An array IMU data fusion method and system based on two-stage weight values

Abstract

The application provides a kind of array IMU data fusion method and system based on two-stage weight value.Belongs to the field of IMU data fusion, the method includes determining the error calibration of array IMU original data;Based on the time series data of each IMU in the static state after calibration, the zero drift instability coefficient of each IMU is obtained;According to the zero drift instability coefficient, the fixed weight value of each IMU in data fusion is calculated in inverse proportion relationship;Real-time acquisition of the current data of each IMU after calibration, the consistency degree of each IMU is obtained, and the measurement weight value of each IMU in data fusion is calculated in inverse proportion relationship according to the consistency degree;Based on fixed weight value and measurement weight value, according to the preset synthesis rule, the fusion weight value is distributed, and the fused IMU data is output.The method can ensure the data fusion accuracy on the basis of low computing power demand.

Description

Technical Field

[0001] This application relates to the field of array IMU data fusion technology, and more specifically, to an array IMU data fusion method and system based on dual-stage weights. Background Technology

[0002] With the widespread adoption of microelectromechanical inertial measurement units (IMUs) in wearable devices, unmanned platforms, and vehicle systems due to their small size, low power consumption, and low cost, single IMUs are no longer sufficient to meet the requirements for high precision and robustness in many scenarios. To improve the accuracy and reliability of positioning / gait detection / motion estimation, researchers and engineers are increasingly employing array IMU schemes. This involves placing several inertial sensing units on the same platform or in close proximity and performing multi-sensor fusion to achieve noise reduction, error term suppression, and improved usability and robustness. Therefore, to achieve high-precision array IMU data fusion, further processing of the raw array IMU data is required using array IMU data fusion methods.

[0003] Currently, array IMU data fusion methods can be mainly divided into weighted array IMU data fusion methods based on fixed weights and learning-based array IMU data fusion methods. For example, the published invention patent CN118623874A provides a method that pre-calculates and stores calibration coefficients and fusion weights for each sensor in the IMU array, and directly uses these weights for weighted fusion at the device end. However, this method only calculates the fusion weights based on historical data and does not consider real-time data conditions, thus limiting the accuracy of data fusion. The published invention patent CN118410455A provides an array IMU data fusion method based on convolutional neural networks, but learning-based array IMU data fusion methods require high computing power, which increases the hardware cost of the system in real-time operating systems.

[0004] Therefore, in order to achieve high-precision array IMU data fusion in real-time systems, it is necessary to comprehensively consider the data fusion accuracy and algorithm complexity to achieve high-precision array IMU data fusion with low computing power requirements. Summary of the Invention

[0005] The purpose of this application is to provide a data fusion method and system for array IMUs based on dual-stage weights, which can ensure data fusion accuracy with low computing power requirements.

[0006] This application is implemented as follows:

[0007] In a first aspect, this application provides a method for array IMU data fusion based on dual-stage weights, comprising the following steps:

[0008] S1: Perform deterministic error calibration on the raw data of the array IMU;

[0009] S2: Based on the time series data of each IMU in a static state after calibration, obtain the zero-bias instability coefficient of each IMU; calculate the fixed weight of each IMU in data fusion according to the zero-bias instability coefficient in an inverse relationship.

[0010] S3: Real-time acquisition of current data of each IMU after calibration, acquisition of the consistency degree of each IMU, and calculation of the measurement weight of each IMU in data fusion based on the consistency degree in an inverse relationship;

[0011] S4: Based on fixed weights and measurement weights, allocate fusion weights according to preset synthesis rules and output the fused IMU data.

[0012] Based on the first aspect, the steps for deterministic error calibration of the raw data from the array IMU include:

[0013] The expression for the deterministic error calibration of the array IMU is set as follows:

[0014] ;

[0015] in, For the first in the array IMU i The output vector of the accelerometer before calibration. For the first in the array IMU i The calibrated output vector of each accelerometer For the first in the array IMU i The constant zero-bias vector of each accelerometer. For the first in the array IMU i Calibration matrix for each accelerometer, For the first in the array IMU i The output vector of each gyroscope before calibration For the first in the array IMU i The output vector after gyroscope calibration For the first in the array IMU i The constant zero-bias vector of each gyroscope. For the first in the array IMU i Calibration matrix for each gyroscope;

[0016] Based on the calibration models of the accelerometer and gyroscope, the calibration coefficients of the accelerometer are determined by the six-position method, and the calibration coefficients of the gyroscope are calculated by the rate method.

[0017] Based on the first aspect, the steps of obtaining the zero-bias instability coefficient of each IMU based on the time-series data of each IMU in a static state after calibration, and calculating the fixed weight of each IMU in data fusion according to the zero-bias instability coefficient in an inverse relationship, include:

[0018] The expression for the fixed weight allocation matrix used in gyroscope data fusion is defined as follows:

[0019] ;

[0020] in, A fixed weight allocation matrix for gyroscope data fusion. For the first in the array IMU i The zero-bias instability coefficient of a three-axis gyroscope n This represents the number of independent IMUs in the array IMU;

[0021] The fixed weight allocation matrix used for accelerometer data fusion is defined as follows:

[0022] ;

[0023] in, A fixed weighting matrix for accelerometer data fusion. For the first in the array IMU i The zero-bias instability coefficient of a triaxial accelerometer n This represents the number of independent IMUs in the array IMU.

[0024] Based on the first aspect, the steps of acquiring the current data of each calibrated IMU in real time, obtaining the consistency level of each IMU, and calculating the measurement weight of each IMU in data fusion according to the consistency level in an inverse relationship include:

[0025] The first in the array IMU i The expression defining the consistency level between one IMU and other IMUs is set as follows:

[0026] ;

[0027] in, For the first in the array IMU i The degree of consistency between one axis of each IMU and the corresponding axes of other IMUs. For the first in the array IMU i One axis of an IMU n This represents the number of independent IMUs in the array IMU;

[0028] The expression for the measurement weight allocation matrix used in gyroscope data fusion is defined as follows:

[0029] ;

[0030] in, Assignment matrix for measurement weights used in gyroscope data fusion. For the first in the array IMUi The degree of consistency between this triaxial gyroscope and other triaxial gyroscopes. n This represents the number of independent IMUs in the array IMU;

[0031] The expression for the fixed weight allocation matrix used in accelerometer data fusion is defined as follows:

[0032] ;

[0033] in, The measurement weight assignment matrix used for accelerometer data fusion. For the first in the array IMU i The degree of consistency between this triaxial accelerometer and other triaxial accelerometers. n This represents the number of independent IMUs in the array IMU.

[0034] Based on the first aspect, the steps of allocating fusion weights according to preset synthesis rules based on fixed weights and measurement weights, and outputting the fused IMU data include:

[0035] The fusion weight matrix expression for the array IMU is defined as follows:

[0036] ;

[0037] in, This is the fusion weight matrix for array IMU data fusion. For the fixed weight matrix of array IMU data fusion, This is the measurement weight matrix for array IMU data fusion.

[0038] Based on the first aspect, the steps for obtaining the zero-bias instability coefficients of each IMU include:

[0039] Based on time-series data under static conditions, the zero-bias instability coefficient of each IMU is calculated using Allen's analysis of variance.

[0040] Based on the first aspect, the steps to obtain the consistency level of each IMU include:

[0041] The degree of consistency of each IMU is obtained by calculating the deviation or variance between the current data of each IMU and the average data of all IMUs.

[0042] Secondly, this application provides an array IMU data fusion system based on dual-stage weights, comprising:

[0043] Deterministic error calibration module: It is configured to perform deterministic error calibration on the raw data of the array IMU;

[0044] Fixed weight calculation module: It is configured to obtain the zero-bias instability coefficient of each IMU based on the time series data of each IMU in the static state after calibration; and calculate the fixed weight of each IMU in data fusion according to the zero-bias instability coefficient in an inverse relationship.

[0045] Measurement weight calculation module: It is configured to acquire the current data of each IMU after calibration in real time, obtain the consistency degree of each IMU, and calculate the measurement weight of each IMU in data fusion based on the consistency degree in an inverse relationship.

[0046] Fusion and output module: It is configured to allocate fusion weights according to preset synthesis rules based on fixed weights and measurement weights, and output the fused IMU data.

[0047] Thirdly, this application provides an electronic device, comprising:

[0048] Memory, used to store one or more programs;

[0049] processor;

[0050] The above method is implemented when one or more programs are executed by the processor.

[0051] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-described method.

[0052] Compared with the prior art, this application has at least the following advantages or beneficial effects:

[0053] This invention provides a method and system for array IMU data fusion based on dual-stage weights. By performing deterministic error calibration on the raw array IMU data, all IMU units are placed on a unified and accurate measurement baseline, so that subsequent data fusion is no longer a simple averaging of chaotic data, but an effective synthesis of precise information, thus laying a solid foundation for accuracy, stability and reliability.

[0054] By using time-series data of each calibrated IMU in a static state, the zero-bias instability coefficient of each IMU is obtained. Based on the zero-bias instability coefficient, a fixed weight for each IMU in data fusion is calculated using an inverse proportional relationship. Utilizing the long-term intrinsic noise characteristics (zero-bias instability) exhibited by each IMU in a static state, a fixed weight reflecting its essential performance is assigned to it. This establishes a stable and reliable performance foundation for the entire fusion process. It ensures that IMU units with inherently better performance and higher quality can continue to play a more important role during the fusion process, thus laying the foundation for high accuracy and high stability.

[0055] By acquiring the current data of each calibrated IMU in real time, the consistency level of each IMU is determined. Based on the consistency level, the measurement weight of each IMU in data fusion is calculated inversely proportionally; the measurement weight does not depend on the initial calibration but is based on the latest data. Therefore, it can capture this dynamic change and reward IMUs that perform better at the current moment in real time. It can also respond to and suppress short-term faults, transient interference, or abnormal data in the array in an instant, preventing these bad data from contaminating the final fusion result.

[0056] By using fixed weights and measurement weights, fusion weights are allocated according to preset synthesis rules, and the fused IMU data is output. The fixed and measurement weights of the IMUs in data fusion complement each other, constructing a stable yet flexible intelligent fusion method. For example, a high-quality IMU with high fixed weights may experience a decrease in measurement weights if subjected to momentary interference. However, because of its high fixed weights, its overall influence will not drop to zero, avoiding the "false positive" killing of core units due to a single interference. After the interference, its measurement weights can quickly recover. Conversely, an IMU with low fixed weights may have its measurement weights increase if its data happens to be highly consistent with the group at a certain moment, allowing it to make a correct contribution to the fusion result and fully utilize hardware resources. This mechanism of long-term reputation (fixed weights) + short-term performance (measurement weights) is more intelligent and robust than any single-weight strategy. By allocating fusion weights, more accurate data selection can be made in complex scenarios, directly improving the accuracy of the output data. Attached Figure Description

[0057] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0058] Figure 1 This is a flowchart of an embodiment of an array IMU data fusion method based on dual-stage weights according to this application;

[0059] Figure 2 The results of gyroscope simulation in the array IMU data fusion simulation using the method described in this application;

[0060] Figure 3 The acceleration simulation results in the array IMU data fusion simulation using the method described in this application;

[0061] Figure 4 This is a schematic diagram of the structure of an array IMU data fusion system based on dual-stage weights according to this application;

[0062] Figure 5 This is a schematic diagram of the structure of an electronic device according to this application.

[0063] icon:

[0064] 1. Deterministic error calibration module; 2. Fixed weight calculation module; 3. Measurement weight calculation module; 4. Fusion and output module; 5. Processor; 6. Memory; 7. Communication interface. Detailed Implementation

[0065] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0066] The following detailed description of some embodiments of this application is provided in conjunction with the accompanying drawings. Unless otherwise specified, the various embodiments and features described below can be combined with each other. Example

[0067] Through long-term research and practice, the inventors have discovered that existing array IMU data fusion methods can be mainly divided into two categories: weighted array IMU data fusion methods based on fixed weights and learning-based array IMU data fusion methods. For example, published invention patent CN118623874A provides a method that pre-calculates and stores calibration coefficients and fusion weights for each sensor in the IMU array, and directly uses these weights for weighted fusion at the device end. However, this method only calculates fusion weights based on historical data and does not consider real-time data conditions, thus limiting the accuracy of data fusion. Published invention patent CN118410455A provides an array IMU data fusion method based on convolutional neural networks, but learning-based array IMU data fusion methods require high computing power, increasing the hardware cost of the system in real-time operation.

[0068] Therefore, this application provides a data fusion method for array IMUs based on dual-stage weights, which can guarantee data fusion accuracy with low computing power requirements.

[0069] Please refer to Figure 1 This method for fusion of array IMU data based on dual-stage weights includes the following steps:

[0070] S1: Perform deterministic error calibration on the raw data of the array IMU;

[0071] Specifically, one purpose of deterministic error calibration of the array IMUs is to unify the benchmark and eliminate internal friction. By placing all IMU units on a unified and accurate measurement benchmark, subsequent data fusion is no longer a simple averaging of chaotic data, but an effective synthesis of precise information, thus laying a solid foundation for the accuracy, stability, and reliability of the entire system. This ensures that every IMU data point processed by the subsequent fusion algorithm is a measurement based on its actual physical motion, rather than a distorted signal containing large inherent errors. Considering the vast differences in the deterministic error characteristics (such as zero bias) of different IMU units, precise individual calibration allows for "seeking commonalities while reserving differences," calibrating each unit to a unified standard. This eliminates the need for the costly process of selecting a batch of IMUs with inherently very similar zero biases; instead, the calibration algorithm can compensate for individual hardware differences, greatly improving the flexibility and cost-effectiveness of hardware selection.

[0072] As one implementation method, this step can be achieved in the following way:

[0073] The expression for the deterministic error calibration of the array IMU is set as follows:

[0074] ;

[0075] in, For the first in the array IMU i The output vector of the accelerometer before calibration. For the first in the array IMU i The calibrated output vector of each accelerometer For the first in the array IMU i The constant zero-bias vector of each accelerometer. For the first in the array IMU i Calibration matrix for each accelerometer, For the first in the array IMU i The output vector of each gyroscope before calibration For the first in the array IMU i The output vector after gyroscope calibration For the first in the array IMU i The constant zero-bias vector of each gyroscope. For the first in the array IMU i Calibration matrix for each gyroscope;

[0076] Based on the calibration models of the accelerometer and gyroscope, the calibration coefficients of the accelerometer are determined by the six-position method, and the calibration coefficients of the gyroscope are calculated by the rate method.

[0077] S2: Based on the time series data of each IMU in a static state after calibration, obtain the zero-bias instability coefficient of each IMU; calculate the fixed weight of each IMU in data fusion according to the zero-bias instability coefficient in an inverse relationship.

[0078] Specifically, zero-bias instability is a key indicator of an IMU's intrinsic quality. It represents the magnitude of fluctuation in the IMU output around its zero-bias value when at rest. A lower value indicates better long-term stability and more reliable data. Calculating fixed weights in an inverse relationship, IMUs with lower zero-bias instability receive higher fixed weights. This is equivalent to automatically identifying and reusing higher-performing IMU units at the algorithm level. This design ensures that, over long-term operation, the fusion results naturally tilt towards the better-performing sensors, thus embedding high-precision inherent in the system architecture. This fundamentally improves the long-term accuracy and stability of data fusion results.

[0079] As one implementation method, this step can be achieved in the following way:

[0080] The expression for the fixed weight allocation matrix used in gyroscope data fusion is defined as follows:

[0081] ;

[0082] in, A fixed weight allocation matrix for gyroscope data fusion. For the first in the array IMU i The zero-bias instability coefficient of a three-axis gyroscope n This represents the number of independent IMUs in the array IMU;

[0083] The fixed weight allocation matrix used for accelerometer data fusion is defined as follows:

[0084] ;

[0085] in, A fixed weighting matrix for accelerometer data fusion. For the first in the array IMU i The zero-bias instability coefficient of a triaxial accelerometer n This represents the number of independent IMUs in the array IMU.

[0086] S3: Real-time acquisition of current data of each IMU after calibration, acquisition of the consistency degree of each IMU, and calculation of the measurement weight of each IMU in data fusion based on the consistency degree in an inverse relationship;

[0087] Specifically, the consistency level measures how closely each IMU's real-time data approximates the consensus of the entire IMU group. If an IMU's output deviates significantly from the group due to sudden vibration, shock, electromagnetic interference, or internal transient faults, its consistency level immediately deteriorates. The measurement weights are then calculated inversely, meaning that lagging IMUs are immediately assigned a very low weight, significantly weakening their influence in the fusion process. This setup makes the scheme highly immune to transient hardware failures and external interference, ensuring the reliability of the output at every moment. Furthermore, the measurement weights are not dependent on initial calibration but are based on the latest data. Therefore, it can capture these dynamic changes and reward IMUs that perform better at the current moment in real time. It can respond instantly to and suppress short-term faults, transient interference, or anomalous data in the array, preventing these bad data from contaminating the final fusion result.

[0088] As one implementation method, this step can be achieved in the following way:

[0089] The first in the array IMU i The expression defining the consistency level between one IMU and other IMUs is set as follows:

[0090] ;

[0091] in, For the first in the array IMU i The degree of consistency between one axis of each IMU and the corresponding axes of other IMUs. For the first in the array IMU i One axis of an IMU n This represents the number of independent IMUs in the array IMU;

[0092] The expression for the measurement weight allocation matrix used in gyroscope data fusion is defined as follows:

[0093] ;

[0094] in, Assignment matrix for measurement weights used in gyroscope data fusion. For the first in the array IMU i The degree of consistency between this triaxial gyroscope and other triaxial gyroscopes. n This represents the number of independent IMUs in the array IMU;

[0095] The expression for the fixed weight allocation matrix used in accelerometer data fusion is defined as follows:

[0096] ;

[0097] in, The measurement weight assignment matrix used for accelerometer data fusion. For the first in the array IMU i The degree of consistency between this triaxial accelerometer and other triaxial accelerometers. n This represents the number of independent IMUs in the array IMU.

[0098] S4: Based on fixed weights and measurement weights, allocate fusion weights according to preset synthesis rules and output the fused IMU data.

[0099] Specifically, the fixed weights and measurement weights of the IMU in data fusion complement each other, constructing a stable yet flexible intelligent fusion method. Fixed weights act like a "long-term reputation record," preventing high-performance units from being completely discarded due to accidental momentary interference. Measurement weights act like a "real-time performance score," preventing low-performance or faulty units from continuously polluting the output due to their "long service life." After fusion, a high-performance IMU subjected to momentary interference will have its final weight decrease due to the reduction in measurement weights, but it will not drop to zero (because its fixed weights are high); a low-performance IMU with consistently poor performance will have its final weight become very low due to the persistently low measurement weights. This mechanism achieves the fairest and most reasonable dynamic management of sensors. For example, a high-quality IMU with high fixed weights will have its measurement weights decrease if subjected to momentary interference. However, because its fixed weights are high, its overall impact will not drop to zero, avoiding the "false kill" of core units due to a single interference. After the interference, its measurement weights can quickly recover. For example, an IMU with low fixed weights might have its measurement weights increased if its data happens to be highly consistent with the population at a certain moment. This allows it to make a correct contribution to the fusion result, achieving full utilization of hardware resources. This mechanism of long-term reputation (fixed weights) + short-term performance (measurement weights) is more intelligent and robust than any single-weight strategy. It can make more accurate data selections in complex scenarios, thereby directly improving the accuracy of the output data.

[0100] As one implementation method, this step can be achieved in the following way:

[0101] The fusion weight matrix expression for the array IMU is defined as follows:

[0102] ;

[0103] in, This is the fusion weight matrix for array IMU data fusion. For the fixed weight matrix of array IMU data fusion, This is the measurement weight matrix for array IMU data fusion.

[0104] Preferably, the steps for obtaining the zero-bias instability coefficients of each IMU include:

[0105] Based on time-series data under static conditions, the zero-bias instability coefficient of each IMU is calculated using Allen's analysis of variance.

[0106] Preferably, the steps for obtaining the consistency level of each IMU include:

[0107] The degree of consistency of each IMU is obtained by calculating the deviation or variance between the current data of each IMU and the average data of all IMUs.

[0108] To verify the correctness of the present invention, a field test was conducted. The performance parameters of a single IMU in the experiment are listed in Table 1.

[0109] Table 1: Performance parameters of a single IMU in the actual experiment

[0110] index range Zero bias stability Data update rate Initial zero bias gyroscope ±500 ° / s 10 ° / h 200 Hz 3 ° / s accelerometer ±6 G 80 ug 200 Hz 25 mg

[0111] Three sets of experimental measurements were conducted using static data collected from an 8-array IMU system (as shown in Table 1). The array IMU data were fused using both the traditional fixed-weight data fusion method and the method proposed in this invention. Allan variance analysis was then performed on the collected data. Table 2 shows the reduction factor of the noise figure along the Z-axis of the array IMU by the method of this invention compared to the traditional fixed-weight data fusion method.

[0112] Table 2: Comparison of Z-axis noise figure reduction factors for IMU arrays with different algorithms

[0113]

[0114] As shown in Table 2, the method of the present invention reduces the noise figure of an 8-array IMU by approximately 1.8 times. Compared with the data fusion method using fixed weights, it improves performance by approximately 25.7%, 26.2%, 52.5%, and 37.7% in terms of gyroscope bias instability, gyroscope random walk, accelerometer bias instability, and accelerometer random walk, respectively. Furthermore, it does not require training through deep learning or similar methods; high-precision array IMU data fusion can be achieved through time-domain signal processing. Therefore, the effectiveness and correctness of the method provided in this invention are verified.

[0115] Furthermore, in this embodiment, the method of this application is used for array IMU data fusion simulation. For the gyroscope simulation results, please refer to... Figure 2 Please refer to the accelerometer simulation results. Figure 3 .

[0116] from Figure 2 and Figure 3It can be seen that the simulation results of array IMU data fusion using the method of the present invention have significantly improved noise performance compared with the unfused single sensor.

[0117] Please refer to Figure 4 This embodiment also provides an array IMU data fusion system based on dual-stage weights, including:

[0118] Deterministic error calibration module 1: It is configured to perform deterministic error calibration on the raw data of the array IMU;

[0119] Fixed weight calculation module 2: It is configured to obtain the zero-bias instability coefficient of each IMU based on the time series data of each IMU in the static state after calibration; and calculate the fixed weight of each IMU in data fusion according to the zero-bias instability coefficient in an inverse relationship.

[0120] Measurement weight calculation module 3: It is configured to acquire the current data of each IMU after calibration in real time, obtain the consistency degree of each IMU, and calculate the measurement weight of each IMU in data fusion based on the consistency degree in an inverse relationship.

[0121] Fusion and output module 4: It is configured to allocate fusion weights according to preset synthesis rules based on fixed weights and measurement weights, and output the fused IMU data.

[0122] For a detailed implementation of the array IMU data fusion system based on dual-stage weights, please refer to the above-mentioned implementation of the array IMU data fusion method based on dual-stage weights, which will not be elaborated further here.

[0123] Please refer to Figure 5 This embodiment also provides an electronic device, including:

[0124] Memory 6 is used to store one or more programs;

[0125] Processor 5; Processor 5 and memory 6 are connected via communication interface 7;

[0126] The above method is implemented when one or more programs are executed by processor 5.

[0127] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor 5, implements the above-described method.

[0128] It will be apparent to those skilled in the art that this application is not limited to the details of the exemplary embodiments described above, and that this application can be implemented in other specific forms without departing from the spirit or essential characteristics of this application. Therefore, the embodiments should be considered illustrative and non-limiting in all respects, and the scope of this application is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within this application. No reference numerals in the claims should be construed as limiting the scope of the claims.

Claims

1. A two-stage weight-based array IMU data fusion method, characterized in that, The method comprises the following steps: S1: deterministic error calibration of array IMU raw data; S2: based on the timing data of each IMU in the static state after calibration, the zero drift instability coefficient of each IMU is obtained; According to the zero drift instability coefficient, the fixed weight value of each IMU in data fusion is calculated in inverse proportion; comprising: The fixed weight assignment matrix definition expression used in the fusion of gyro data is set as: ; wherein, a fixed weight assignment matrix for use in fusion of gyro data, is a bias instability coefficient for the jth three-axis gyro in the array IMU, i is a bias instability coefficient for the jth three-axis gyro in the array IMU, n is a number of individual IMUs in the array IMU; The fixed weight assignment matrix used in the accelerometer data fusion defines the expression as: ; wherein, a fixed weight assignment matrix for accelerometer data fusion, a zero offset instability coefficient for the i th three-axis accelerometer in the array IMU, n a number of independent IMUs in the array IMU; S3: real-time acquisition of the current data of each IMU after calibration, the consistency degree of each IMU is obtained, and the measurement weight value of each IMU in data fusion is calculated in inverse proportion according to the consistency degree; comprising: The degree of consistency of the i-th IMU in the array IMU with the other IMUs is defined by the expression: i ​ ； wherein, a degree of consistency of one axis of the i-th IMU in the array IMU with corresponding axes of other IMUs, i a degree of consistency of one axis of the i-th IMU in the array IMU with corresponding axes of other IMUs, an axis of the i-th IMU in the array IMU, i an axis of the i-th IMU in the array IMU, n a number of independent IMUs in the array IMU; The definition expression of the measurement weight distribution matrix used in the fusion of gyro data is set as: ; wherein, a measurement weight distribution matrix used for gyro data fusion, a consistency degree of a i third-axis gyro in the array IMU with other third-axis gyros, n a number of independent IMUs in the array IMU; The fixed weight assignment matrix defining expression for using accelerometer data fusion is set as: ; wherein, a measurement weight assignment matrix used for accelerometer data fusion, a consistency degree of the i-th tri-axial accelerometer in the array IMU with other tri-axial accelerometers, i a consistency degree of the i-th tri-axial accelerometer in the array IMU with other tri-axial accelerometers, n a number of independent IMUs in the array IMU; S4: based on the fixed weight value and the measurement weight value, the fusion weight value is distributed according to the preset synthesis rule, and the fused IMU data is output; comprising: The fusion weight matrix expression of the defined array IMU is: ; wherein, is a fusion weight matrix for array IMU data fusion, is a fixed weight matrix for array IMU data fusion, is a measurement weight matrix for array IMU data fusion.

2. The dual-stage weight-based array IMU data fusion method according to claim 1, wherein, The step of deterministic error calibration of array IMU raw data comprises: The expression of the array IMU deterministic error calibration is set as: ; in, For the first in the array IMU i The output vector of the accelerometer before calibration. For the first in the array IMU i The calibrated output vector of each accelerometer For the first in the array IMU i The constant zero-bias vector of each accelerometer. For the first in the array IMU i Calibration matrix for each accelerometer, For the first in the array IMU i The output vector of each gyroscope before calibration For the first in the array IMU i The output vector after gyroscope calibration For the first in the array IMU i The constant zero-bias vector of each gyroscope. For the first in the array IMU i Calibration matrix for each gyroscope; On the basis of accelerometer and gyroscope calibration model, six position method is used to determine the calibration coefficient of accelerometer, and rate method is used to calculate the calibration coefficient of gyroscope.

3. The dual-stage weight-based array IMU data fusion method according to claim 1, wherein, The step of obtaining the zero drift instability coefficient of each IMU comprises: Based on the timing data in the static state, the zero drift instability coefficient of each IMU is calculated by Allan variance analysis method.

4. The dual-stage weight-based array IMU data fusion method according to claim 1, wherein, The step of obtaining the consistency degree of each IMU comprises: The deviation or variance of the current data of each IMU and the average value of all IMU data is calculated to obtain the consistency degree of each IMU.

5. A two-stage weight-based array IMU data fusion system, characterized in that, Comprising: Deterministic error calibration module: configured to calibrate the array IMU raw data; Fixed weight value calculation module: configured to obtain the zero drift instability coefficient of each IMU based on the timing data of each IMU in the static state after calibration; Based on the zero-bias instability coefficient, the fixed weights of each IMU in data fusion are calculated inversely proportionally; including: defining the expression of the fixed weight allocation matrix used for gyroscope data fusion as: ;in, A fixed weight allocation matrix for gyroscope data fusion. For the first in the array IMU i The zero-bias instability coefficient of a three-axis gyroscope n The number of independent IMUs in the array IMU; the fixed weight allocation matrix used for accelerometer data fusion is defined as follows: ;in, A fixed weighting matrix for accelerometer data fusion. For the first in the array IMU i The zero-bias instability coefficient of a triaxial accelerometer n This represents the number of independent IMUs in the array IMU; Measurement weight calculation module: It is configured to acquire the current data of each calibrated IMU in real time, obtain the consistency degree of each IMU, and calculate the measurement weight of each IMU in data fusion according to the consistency degree in an inverse relationship; including: the measurement weight of the IMU in the array IMU is calculated in real time. i The expression defining the consistency level between one IMU and other IMUs is set as follows: ;in For the first in the array IMU i The degree of consistency between one axis of each IMU and the corresponding axes of other IMUs. For the first in the array IMU i One axis of an IMU n Let be the number of independent IMUs in the array IMU; define the expression for the measurement weight allocation matrix used for gyroscope data fusion as follows: ;in, Assignment matrix for measurement weights used in gyroscope data fusion. For the first in the array IMU i The degree of consistency between this triaxial gyroscope and other triaxial gyroscopes. n Let be the number of independent IMUs in the array IMU; the expression for the fixed weight allocation matrix used for accelerometer data fusion is defined as: ;in, The measurement weight assignment matrix used for accelerometer data fusion. For the first in the array IMU i The degree of consistency between this triaxial accelerometer and other triaxial accelerometers. n This represents the number of independent IMUs in the array IMU; The fusion and output module is configured to assign a fusion weight according to a preset synthesis rule based on a fixed weight and a measurement weight, and output fused IMU data; comprising: defining an array IMU fusion weight matrix expression as: ; wherein, is a fusion weight matrix of array IMU data fusion, is a fixed weight matrix of array IMU data fusion, is a measurement weight matrix of array IMU data fusion.

6. An electronic device, comprising: Comprising: Memory, for storing one or more programs; Processor; When the one or more programs are executed by the processor, the method as claimed in any one of claims 1-4 is implemented.

7. A computer-readable storage medium having stored thereon a computer program, characterized in that The computer program is executed by the processor to implement the method as claimed in any one of claims 1-4.

Images