A multi-sensor fusion processing method based on distributed optoelectronic system

Abstract

The present invention discloses a multi-sensor fusion processing method based on a distributed optoelectronic system, and relates to the technical field of data analysis. The method of the present invention comprises: extracting historical data of optoelectronic system sensors, establishing an error model, evaluating error propagation coefficients, and matching error correction schemes; acquiring raw sensor data, unifying the format, and cleaning the data to form a preliminary target data set; performing similarity analysis on the target data set, matching the error correction scheme, correcting the data, analyzing the error propagation coefficients, and outputting prompt information; evaluating data quality, enhancing the data, and storing the enhanced data; extracting sensor data features, fusing features, generating comprehensive data feature representations, and realizing multi-sensor fusion processing.

Description

Technical Field

[0001] The present invention relates to the technical field of data analysis, and in particular to a multi-sensor fusion processing method based on a distributed optoelectronic system. Background Art

[0002] With the rapid development of science and technology, optoelectronic sensing technology has shown great application potential in many fields such as autonomous driving, industrial detection, environmental monitoring and intelligent security. Distributed optoelectronic systems achieve comprehensive coverage and real-time monitoring of the target environment by deploying multiple optoelectronic sensor nodes in the network. Since a single sensor is limited by its perception range, accuracy and reliability, it is often difficult to provide complete and accurate environmental information. Therefore, in the current distributed optoelectronic system, multi-sensor fusion technology has become a key means to improve the overall performance of the system. This technology integrates data from different sensors to achieve information complementarity and enhancement, thereby improving the perception accuracy, robustness and real-time performance of the system.

[0003] However, in existing distributed optoelectronic systems, multi-sensor fusion processing methods often face some key challenges. For example, sensor error and data quality are urgent issues to be solved, and the measurement error of the sensor itself may be amplified during the fusion process, thus affecting the final fusion result. Summary of the invention

[0004] The purpose of the present invention is to provide a multi-sensor fusion processing method based on a distributed optoelectronic system to solve the problems raised in the above background technology.

[0005] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0006] A multi-sensor fusion processing method based on a distributed optoelectronic system, the method comprising the following steps:

[0007] Step S100. Extract historical data of each photoelectric sensor node in the distributed photoelectric system from the database, and establish an error model for each sensor based on the historical data; evaluate the historical error propagation coefficient of each sensor error in the fusion process in combination with the error model and historical data, and match the corresponding error correction scheme for each historical error propagation coefficient in combination with the database;

[0008] Step S200. Acquire the original target data of each photoelectric sensor node from the distributed photoelectric system, and convert the original target data of each sensor into a unified format; perform data cleaning on the converted original target data to form a preliminarily processed target data set;

[0009] Step S300. For the initially processed target data set, a similarity analysis is performed for each element in the target data set, and a corresponding error correction scheme is matched according to the similarity analysis result; the corresponding target data is corrected based on the matched error correction scheme, the target error propagation coefficient corresponding to each element is calculated, and the target error propagation coefficient is analyzed with the corresponding historical error propagation coefficient, so as to output corresponding prompt information;

[0010] Step S400: Perform data quality assessment on the corrected target data, and according to the result of the data quality assessment, perform data enhancement on the corresponding target data, and store the enhanced target data instead of the corrected target data;

[0011] Step S500. Based on the target data of step S400, extract data features from the corresponding target data of each sensor; based on the data features, for each photoelectric sensor node, fuse the corresponding data features to generate a comprehensive data feature representation, thereby realizing multi-sensor fusion processing.

[0012] Furthermore, step S100 includes:

[0013] S101. Extract historical data of each photoelectric sensor node in the distributed photoelectric system from the database, and the historical data of each photoelectric sensor node is represented as Hi, where Hi = {hi1, hi2, ..., him}, Hi represents the historical data of the i-th photoelectric sensor node, hi1 represents the first sensor historical data of the i-th photoelectric sensor node, hi2 represents the second sensor historical data of the i-th photoelectric sensor node, and so on, him represents the m-th sensor historical data of the i-th photoelectric sensor node, i represents the data number of the photoelectric sensor node, and m represents the sensor data number corresponding to the photoelectric sensor node;

[0014] S102. For the sensor of each photoelectric sensor node, an error model is established in combination with the corresponding historical data, and the error model is selected by relevant personnel based on the corresponding historical data, so that the corresponding error model expression is obtained as follows: h'ij = hij + Eij, where h'ij represents the jth corrected sensor historical data of the i-th photoelectric sensor node, hij represents the jth original sensor historical data of the i-th photoelectric sensor node; Eij represents the error term, which is determined by different error models;

[0015] Among them, common error models include Gaussian error model, linear regression error model and nonlinear regression error model, etc. Relevant personnel select the most appropriate error model based on the characteristics of the original sensor historical data.

[0016] S103. Input the original sensor historical data and the corrected sensor historical data into the corresponding data fusion model respectively, so as to obtain the fusion output results Ro and Rc respectively; calculate the errors Eo and Ec of the fusion results respectively through the fusion output results Ro and Rc, and Eo = |Ro-T|, Ec = |Rc-T|, where T represents the theoretical or reference fusion result; calculate the difference Ed between the fusion results of the original data and the corrected data, and Ed = Ro-Rc; calculate the historical error propagation coefficient C, and C = Ed / (Eo-Ec); according to the historical error propagation coefficient of each photoelectric sensor node, search and match the corresponding error correction scheme in the database, and the historical error propagation coefficient of each photoelectric sensor node corresponds to an error correction scheme.

[0017] Further, step S200 includes:

[0018] S201. Acquire the original target data of each photoelectric sensor node from the distributed photoelectric system, and the original target data is represented by Di, wherein Di = {di1, di2, ..., dim}, Di represents the original target data of the i-th photoelectric sensor node, di1 represents the first sensor data of the i-th photoelectric sensor node, di2 represents the second sensor data of the i-th photoelectric sensor node, and so on, dim represents the m-th sensor data of the i-th photoelectric sensor node;

[0019] S202. The original target data of each photoelectric sensor node is converted into a unified format, and the converted data is expressed as Di_u=f(Di), where f() represents the data conversion function; based on the converted original target data, data cleaning is performed for each photoelectric sensor node to remove noise and outliers, and the cleaned data of each sensor are merged to form a preliminarily processed target data set A, and A={a1,a2,...,an}, where a1 represents the target data preliminarily processed by the first photoelectric sensor node, a2 represents the target data preliminarily processed by the second photoelectric sensor node, and so on, an represents the target data preliminarily processed by the nth photoelectric sensor node, and n represents the number of photoelectric sensor nodes in the photoelectric system.

[0020] Furthermore, step S300 includes:

[0021] S301. For each element in the initially processed target data set, the corresponding target data is obtained, and according to the photoelectric sensor node data number of the current target data, the historical data of all the same photoelectric sensor node data number is obtained; the current target data is analyzed with the historical data of all the same photoelectric sensor node data number, and error correction is performed on each element in the target data set according to the similarity analysis result;

[0022] S302. Analyze the corrected target data according to the processing method of the historical data in S102 and S103, so as to obtain the target error propagation coefficient C2; according to the photoelectric sensor node data number of the corrected target data, obtain all the historical error propagation coefficients C of the same photoelectric sensor node data number, and form a historical error propagation coefficient set B, and B = {b1(C), b2(C), ..., bv(C)}, wherein b1(C) represents the first historical error propagation coefficient C of the photoelectric sensor node data number that is the same as the target error propagation coefficient C2; b2(C) represents the second historical error propagation coefficient C of the photoelectric sensor node data number that is the same as the target error propagation coefficient C2, and so on, bv(C) represents the vth historical error propagation coefficient C of the photoelectric sensor node data number that is the same as the target error propagation coefficient C2, and v represents the number of historical error propagation coefficients C of the photoelectric sensor node data number that is the same as the target error propagation coefficient C2;

[0023] S302. If the corrected target data corresponding to the target error propagation coefficient C2 has an approximate error propagation coefficient C1, the difference Q between the target error propagation coefficient C2 and the approximate error propagation coefficient C1 is calculated, and Q = |C2-C1| / C2. If the difference Q is less than or equal to the preset difference Q0, no prompt information is output; if the difference Q is greater than the preset difference Q0, the photoelectric sensor node data number of the current corrected target data is output to the relevant personnel, and the relevant personnel perform corresponding processing;

[0024] If the corrected target data corresponding to the target error propagation coefficient C2 does not have an approximate error propagation coefficient C1, the target error propagation coefficient C2 is compared with the elements in the corresponding historical error propagation coefficient set B. If the historical error propagation coefficient set B contains a situation where the historical error propagation coefficient C is equal to the current target error propagation coefficient C1, the photoelectric sensor node data number of the current corrected target data is output to relevant personnel, who perform corresponding processing. If the historical error propagation coefficient set B does not contain a situation where the historical error propagation coefficient C is equal to the current target error propagation coefficient C1, the error correction scheme corresponding to the current target error propagation coefficient C1 is associated and stored.

[0025] Furthermore, in S301, error correction is performed on each element in the target data set according to the similarity analysis result. The specific process is as follows:

[0026] Perform similarity analysis on the current target data and all historical data with the same photoelectric sensor node data number. If the average similarity between the current target data and all historical data with the same photoelectric sensor node data number is greater than or equal to the similarity threshold, select the historical error propagation coefficient of the historical data with the largest similarity as the approximate error propagation coefficient C1, and correct the current target data according to the error correction scheme corresponding to the current approximate error propagation coefficient C1;

[0027] If the average similarity between the current target data and all historical data with the same photoelectric sensor node data number is less than the similarity threshold, the current target data will be output to relevant personnel, who will perform corresponding correction processing.

[0028] Furthermore, step S400 includes:

[0029] S401. Obtain the corrected target data and the corresponding reference data in the database, calculate the accuracy index Z of the corrected target data, and Z=1-(|Rs-Tc| / Tc), where Rs represents the corrected target data and Tc represents the corresponding reference data; combine the corrected target data and the corresponding reference data in the database to calculate the accuracy index J, and J=1-[V(Rs) / V(T)], where V(Rs) represents the variance of the corrected target data and V(T) represents the variance of the corresponding reference data; obtain the timestamp of the corrected target data and the expected update frequency W, calculate the timeliness index S, and S=1-(t0-t1) / W, where t0 represents the current timestamp and t1 represents the timestamp of the correction data;

[0030] S402. Combined with the above indicators, calculate the comprehensive quality score P, and P = αZ + βJ + γS, where α, β and γ represent the weight coefficients of the above indicators respectively, and α + β + γ = 1; according to the comprehensive quality score P, if the comprehensive quality score P is less than or equal to the quality score threshold P0, the corresponding corrected target data is marked as data to be enhanced, and the data to be enhanced is subjected to corresponding data enhancement processing, and the target data after data enhancement replaces the corrected target data for storage; if the comprehensive quality score P is greater than the quality score threshold P0, the corresponding corrected target data does not need to be data enhanced.

[0031] The decision on whether to enhance the data is based on the value of the comprehensive quality score P. This method ensures that only data with a quality score lower than the set threshold P0 will be enhanced, thereby optimizing the data processing process, avoiding blind enhancement of all data, saving computing resources, and improving processing efficiency.

[0032] Furthermore, in step S402, the following analysis is performed on the target data after data enhancement:

[0033] For each photoelectric sensor node, the number of sensors m1 corresponding to the target data undergoing data enhancement processing is counted, and the corresponding data enhancement ratio U is calculated by combining the number of sensors m1 and the number of all sensors m of the corresponding photoelectric sensor node, and U = m1 / m; if the data enhancement ratio U of a certain photoelectric sensor node is greater than or equal to 0.5, the data number of the corresponding photoelectric sensor node is output to relevant personnel, and the relevant personnel perform corresponding maintenance on the sensor equipment of the current photoelectric sensor node.

[0034] The analysis of data enhancement ratio U is introduced to monitor the device status of photoelectric sensor nodes. If the enhancement ratio of a photoelectric sensor node is higher than 50%, it indicates that there may be systemic problems in the node. Through this method, equipment failures can be discovered and handled in a timely manner, the strategy of equipment maintenance and overhaul can be optimized, and the long-term stability and reliability of the system can be improved.

[0035] Furthermore, step S500 includes:

[0036] S501. Extract corresponding data features from the enhanced target data, obtain corresponding feature vectors according to the data features, and perform standardization on the feature vectors corresponding to different sensors, and each sensor corresponds to a feature vector F;

[0037] S502. For each photoelectric sensor node, perform weighted fusion on the features of all sensors, so that the standardized feature vectors are fused into a comprehensive feature representation, expressed as: F_fused=Σ k∈[1,m] wkFk, where k represents the sensor number corresponding to the photoelectric sensor node, wk represents the weight coefficient of the kth sensor corresponding to the photoelectric sensor node, and Fk represents the feature vector of the kth sensor corresponding to the photoelectric sensor node.

[0038] Compared with the prior art, the beneficial effects of the present invention are as follows: by extracting error models from historical data and combining real-time data for error correction, the present invention can more accurately identify and correct sensor errors, reduce the amplification effect of errors in the data fusion process, and thus improve the accuracy of the final result; by performing unified format conversion, data cleaning, quality assessment and data enhancement on the original data, the present invention can effectively deal with data missing and incompleteness problems, ensure higher data quality in the data fusion process, and enhance the reliability of the final data; the present invention introduces the analysis of historical error propagation coefficients and target error propagation coefficients, performs more detailed error processing on the data of each sensor, can effectively apply historical experience to the error correction of current data, and optimize the error propagation process; by performing error correction and data enhancement on the target data of each sensor, and then fusing the data features, the present invention can generate a more accurate comprehensive data feature representation, thereby improving the accuracy and stability of multi-sensor fusion processing. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0040] Figure 1 It is a schematic diagram of the steps of a multi-sensor fusion processing method based on a distributed optoelectronic system of the present invention. DETAILED DESCRIPTION

[0041] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0042] See also Figure 1 , the present invention provides a technical solution:

[0043] A multi-sensor fusion processing method based on a distributed optoelectronic system, the method comprising the following steps:

[0044] Step S100. Extract historical data of each photoelectric sensor node in the distributed photoelectric system from the database, and establish an error model for each sensor based on the historical data; evaluate the historical error propagation coefficient of each sensor error in the fusion process in combination with the error model and historical data, and match the corresponding error correction scheme for each historical error propagation coefficient in combination with the database;

[0045] Step S200. Acquire the original target data of each photoelectric sensor node from the distributed photoelectric system, and convert the original target data of each sensor into a unified format; perform data cleaning on the converted original target data to form a preliminarily processed target data set;

[0046] Step S300. For the initially processed target data set, a similarity analysis is performed for each element in the target data set, and a corresponding error correction scheme is matched according to the similarity analysis result; the corresponding target data is corrected based on the matched error correction scheme, the target error propagation coefficient corresponding to each element is calculated, and the target error propagation coefficient is analyzed with the corresponding historical error propagation coefficient, so as to output corresponding prompt information;

[0047] Step S400: Perform data quality assessment on the corrected target data, and according to the result of the data quality assessment, perform data enhancement on the corresponding target data, and store the enhanced target data instead of the corrected target data;

[0048] Step S500. Based on the target data of step S400, extract data features from the corresponding target data of each sensor; based on the data features, for each photoelectric sensor node, fuse the corresponding data features to generate a comprehensive data feature representation, thereby realizing multi-sensor fusion processing.

[0049] Step S100 includes:

[0050] S101. Extract historical data of each photoelectric sensor node in the distributed photoelectric system from the database, and the historical data of each photoelectric sensor node is represented as Hi, where Hi = {hi1, hi2, ..., him}, Hi represents the historical data of the i-th photoelectric sensor node, hi1 represents the first sensor historical data of the i-th photoelectric sensor node, hi2 represents the second sensor historical data of the i-th photoelectric sensor node, and so on, him represents the m-th sensor historical data of the i-th photoelectric sensor node, i represents the data number of the photoelectric sensor node, and m represents the sensor data number corresponding to the photoelectric sensor node;

[0051] S102. For the sensor of each photoelectric sensor node, an error model is established in combination with the corresponding historical data, and the error model is selected by relevant personnel based on the corresponding historical data, so that the corresponding error model expression is obtained as follows: h'ij = hij + Eij, where h'ij represents the jth corrected sensor historical data of the i-th photoelectric sensor node, hij represents the jth original sensor historical data of the i-th photoelectric sensor node; Eij represents the error term, which is determined by different error models;

[0052] Among them, common error models include Gaussian error model, linear regression error model and nonlinear regression error model, etc. Relevant personnel select the most appropriate error model based on the characteristics of the original sensor historical data.

[0053] S103. Input the original sensor historical data and the corrected sensor historical data into the corresponding data fusion model respectively, so as to obtain the fusion output results Ro and Rc respectively; calculate the errors Eo and Ec of the fusion results respectively through the fusion output results Ro and Rc, and Eo = |Ro-T|, Ec = |Rc-T|, where T represents the theoretical or reference fusion result; calculate the difference Ed between the fusion results of the original data and the corrected data, and Ed = Ro-Rc; calculate the historical error propagation coefficient C, and C = Ed / (Eo-Ec); according to the historical error propagation coefficient of each photoelectric sensor node, search and match the corresponding error correction scheme in the database, and the historical error propagation coefficient of each photoelectric sensor node corresponds to an error correction scheme.

[0054] In this embodiment, it is assumed that: the fusion result Ro obtained using the original data is 50, the fusion result Rc obtained using the corrected data is 45, and the theoretical fusion result T is 48;

[0055] The fusion error of the original data is Eo=|50-48|=2; the fusion error of the corrected data is Ec=|45-48|=3; then, the fusion result difference is Ed=50-45=5;

[0056] The error propagation coefficient C is: C=Ed / (Eo-Ec)=5 / 2-3=-5.

[0057] Step S200 includes:

[0058] S201. Acquire the original target data of each photoelectric sensor node from the distributed photoelectric system, and the original target data is represented by Di, wherein Di = {di1, di2, ..., dim}, Di represents the original target data of the i-th photoelectric sensor node, di1 represents the first sensor data of the i-th photoelectric sensor node, di2 represents the second sensor data of the i-th photoelectric sensor node, and so on, dim represents the m-th sensor data of the i-th photoelectric sensor node;

[0059] S202. The original target data of each photoelectric sensor node is converted into a unified format, and the converted data is expressed as Di_u=f(Di), where f() represents the data conversion function; based on the converted original target data, data cleaning is performed for each photoelectric sensor node to remove noise and outliers, and the cleaned data of each sensor are merged to form a preliminarily processed target data set A, and A={a1,a2,...,an}, where a1 represents the target data preliminarily processed by the first photoelectric sensor node, a2 represents the target data preliminarily processed by the second photoelectric sensor node, and so on, an represents the target data preliminarily processed by the nth photoelectric sensor node, and n represents the number of photoelectric sensor nodes in the photoelectric system.

[0060] Step S300 includes:

[0061] S301. For each element in the initially processed target data set, the corresponding target data is obtained, and according to the photoelectric sensor node data number of the current target data, the historical data of all the same photoelectric sensor node data number is obtained; the current target data is analyzed with the historical data of all the same photoelectric sensor node data number, and error correction is performed on each element in the target data set according to the similarity analysis result;

[0062] S302. Analyze the corrected target data according to the processing method of the historical data in S102 and S103, so as to obtain the target error propagation coefficient C2; according to the photoelectric sensor node data number of the corrected target data, obtain all the historical error propagation coefficients C of the same photoelectric sensor node data number, and form a historical error propagation coefficient set B, and B = {b1(C), b2(C), ..., bv(C)}, wherein b1(C) represents the first historical error propagation coefficient C of the photoelectric sensor node data number that is the same as the target error propagation coefficient C2; b2(C) represents the second historical error propagation coefficient C of the photoelectric sensor node data number that is the same as the target error propagation coefficient C2, and so on, bv(C) represents the vth historical error propagation coefficient C of the photoelectric sensor node data number that is the same as the target error propagation coefficient C2, and v represents the number of historical error propagation coefficients C of the photoelectric sensor node data number that is the same as the target error propagation coefficient C2;

[0063] S302. If the corrected target data corresponding to the target error propagation coefficient C2 has an approximate error propagation coefficient C1, the difference Q between the target error propagation coefficient C2 and the approximate error propagation coefficient C1 is calculated, and Q = |C2-C1| / C2. If the difference Q is less than or equal to the preset difference Q0, no prompt information is output; if the difference Q is greater than the preset difference Q0, the photoelectric sensor node data number of the current corrected target data is output to the relevant personnel, and the relevant personnel perform corresponding processing;

[0064] If the corrected target data corresponding to the target error propagation coefficient C2 does not have an approximate error propagation coefficient C1, the target error propagation coefficient C2 is compared with the elements in the corresponding historical error propagation coefficient set B. If the historical error propagation coefficient set B contains a situation where the historical error propagation coefficient C is equal to the current target error propagation coefficient C1, the photoelectric sensor node data number of the current corrected target data is output to relevant personnel, who perform corresponding processing. If the historical error propagation coefficient set B does not contain a situation where the historical error propagation coefficient C is equal to the current target error propagation coefficient C1, the error correction scheme corresponding to the current target error propagation coefficient C1 is associated and stored.

[0065] In S301, error correction is performed on each element in the target data set according to the similarity analysis result. The specific process is as follows:

[0066] Perform similarity analysis on the current target data and all historical data with the same photoelectric sensor node data number. If the average similarity between the current target data and all historical data with the same photoelectric sensor node data number is greater than or equal to the similarity threshold, select the historical error propagation coefficient of the historical data with the largest similarity as the approximate error propagation coefficient C1, and correct the current target data according to the error correction scheme corresponding to the current approximate error propagation coefficient C1;

[0067] If the average similarity between the current target data and all historical data with the same photoelectric sensor node data number is less than the similarity threshold, the current target data will be output to relevant personnel, who will perform corresponding correction processing.

[0068] Step S400 includes:

[0069] S401. Obtain the corrected target data and the corresponding reference data in the database, calculate the accuracy index Z of the corrected target data, and Z=1-(|Rs-Tc| / Tc), where Rs represents the corrected target data and Tc represents the corresponding reference data; combine the corrected target data and the corresponding reference data in the database to calculate the accuracy index J, and J=1-[V(Rs) / V(T)], where V(Rs) represents the variance of the corrected target data and V(T) represents the variance of the corresponding reference data; obtain the timestamp of the corrected target data and the expected update frequency W, calculate the timeliness index S, and S=1-(t0-t1) / W, where t0 represents the current timestamp and t1 represents the timestamp of the correction data;

[0070] S402. Combined with the above indicators, calculate the comprehensive quality score P, and P = αZ + βJ + γS, where α, β and γ represent the weight coefficients of the above indicators respectively, and α + β + γ = 1; according to the comprehensive quality score P, if the comprehensive quality score P is less than or equal to the quality score threshold P0, the corresponding corrected target data is marked as data to be enhanced, and the data to be enhanced is subjected to corresponding data enhancement processing, and the target data after data enhancement replaces the corrected target data for storage; if the comprehensive quality score P is greater than the quality score threshold P0, the corresponding corrected target data does not need to be data enhanced.

[0071] The decision on whether to enhance the data is based on the value of the comprehensive quality score P. This method ensures that only data with a quality score lower than the set threshold P0 will be enhanced, thereby optimizing the data processing process, avoiding blind enhancement of all data, saving computing resources, and improving processing efficiency.

[0072] In step S402, the following analysis is further performed on the target data after data enhancement:

[0073] For each photoelectric sensor node, the number of sensors m1 corresponding to the target data undergoing data enhancement processing is counted, and the corresponding data enhancement ratio U is calculated by combining the number of sensors m1 and the number of all sensors m of the corresponding photoelectric sensor node, and U = m1 / m; if the data enhancement ratio U of a certain photoelectric sensor node is greater than or equal to 0.5, the data number of the corresponding photoelectric sensor node is output to relevant personnel, and the relevant personnel perform corresponding maintenance on the sensor equipment of the current photoelectric sensor node.

[0074] The analysis of data enhancement ratio U is introduced to monitor the device status of photoelectric sensor nodes. If the enhancement ratio of a photoelectric sensor node is higher than 50%, it indicates that there may be systemic problems in the node. Through this method, equipment failures can be discovered and handled in a timely manner, the strategy of equipment maintenance and overhaul can be optimized, and the long-term stability and reliability of the system can be improved.

[0075] Step S500 includes:

[0076] S501. Extract corresponding data features from the enhanced target data, obtain corresponding feature vectors according to the data features, and perform standardization on the feature vectors corresponding to different sensors, and each sensor corresponds to a feature vector F;

[0077] S502. For each photoelectric sensor node, perform weighted fusion on the features of all sensors, so that the standardized feature vectors are fused into a comprehensive feature representation, expressed as: F_fused=Σ k∈[1,m] wkFk, where k represents the sensor number corresponding to the photoelectric sensor node, wk represents the weight coefficient of the kth sensor corresponding to the photoelectric sensor node, and Fk represents the feature vector of the kth sensor corresponding to the photoelectric sensor node.

[0078] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0079] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A multi-sensor fusion processing method based on a distributed optoelectronic system, characterized in that: The method comprises the following steps: Step S100. Extract historical data of each photoelectric sensor node in the distributed photoelectric system from the database, and establish an error model for each sensor based on the historical data; evaluate the historical error propagation coefficient of the error of each sensor in the fusion process in combination with the error model and historical data, and match the corresponding error correction scheme for each historical error propagation coefficient in combination with the database; Step S200. Acquire the original target data of each photoelectric sensor node from the distributed photoelectric system, and convert the original target data of each sensor into a unified format; perform data cleaning on the converted original target data to form a preliminarily processed target data set; Step S300. For the initially processed target data set, a similarity analysis is performed for each element in the target data set, and a corresponding error correction scheme is matched according to the similarity analysis result; the corresponding target data is corrected based on the matched error correction scheme, the target error propagation coefficient corresponding to each element is calculated, and the target error propagation coefficient is analyzed with the corresponding historical error propagation coefficient, so as to output corresponding prompt information; Step S400: Perform data quality assessment on the corrected target data, and according to the result of the data quality assessment, perform data enhancement on the corresponding target data, and store the enhanced target data instead of the corrected target data; Step S500. Based on the target data of step S400, extract data features from the corresponding target data of each sensor; based on the data features, for each photoelectric sensor node, fuse the corresponding data features to generate a comprehensive data feature representation, thereby realizing multi-sensor fusion processing.

2. The multi-sensor fusion processing method based on a distributed optoelectronic system according to claim 1, characterized in that: The step S100 includes: S101. Extract historical data of each photoelectric sensor node in the distributed photoelectric system from the database, and the historical data of each photoelectric sensor node is represented as Hi, where Hi = {hi1, hi2, ..., him}, Hi represents the historical data of the i-th photoelectric sensor node, hi1 represents the first sensor historical data of the i-th photoelectric sensor node, hi2 represents the second sensor historical data of the i-th photoelectric sensor node, and so on, him represents the m-th sensor historical data of the i-th photoelectric sensor node, i represents the data number of the photoelectric sensor node, and m represents the sensor data number corresponding to the photoelectric sensor node; S102. For the sensor of each photoelectric sensor node, an error model is established in combination with the corresponding historical data, and the error model is selected by relevant personnel based on the corresponding historical data, so that the corresponding error model expression is obtained as follows: h'ij = hij + Eij, where h'ij represents the jth corrected sensor historical data of the i-th photoelectric sensor node, hij represents the jth original sensor historical data of the i-th photoelectric sensor node; Eij represents the error term, which is determined by different error models; S103. Input the original sensor historical data and the corrected sensor historical data into the corresponding data fusion model respectively, so as to obtain the fusion output results Ro and Rc respectively; calculate the errors Eo and Ec of the fusion results respectively through the fusion output results Ro and Rc, and Eo = |Ro-T|, Ec = |Rc-T|, where T represents the theoretical or reference fusion result; calculate the difference Ed between the fusion results of the original data and the corrected data, and Ed = Ro-Rc; calculate the historical error propagation coefficient C, and C = Ed / (Eo-Ec); according to the historical error propagation coefficient of each photoelectric sensor node, search and match the corresponding error correction scheme in the database, and the historical error propagation coefficient of each photoelectric sensor node corresponds to an error correction scheme.

3. The multi-sensor fusion processing method based on a distributed optoelectronic system according to claim 1, characterized in that: The step S200 includes: S201. Acquire the original target data of each photoelectric sensor node from the distributed photoelectric system, and the original target data is represented by Di, wherein Di = {di1, di2, ..., dim}, Di represents the original target data of the i-th photoelectric sensor node, di1 represents the first sensor data of the i-th photoelectric sensor node, di2 represents the second sensor data of the i-th photoelectric sensor node, and so on, dim represents the m-th sensor data of the i-th photoelectric sensor node; S202. The original target data of each photoelectric sensor node is converted into a unified format, and the converted data is expressed as Di_u=f(Di), where f() represents the data conversion function; based on the converted original target data, data cleaning is performed for each photoelectric sensor node to remove noise and outliers, and the cleaned data of each sensor are merged to form a preliminarily processed target data set A, and A={a1,a2,...,an}, where a1 represents the target data preliminarily processed by the first photoelectric sensor node, a2 represents the target data preliminarily processed by the second photoelectric sensor node, and so on, an represents the target data preliminarily processed by the nth photoelectric sensor node, and n represents the number of photoelectric sensor nodes in the photoelectric system.

4. The multi-sensor fusion processing method based on a distributed optoelectronic system according to claim 2, characterized in that: The step S300 includes: S301. For each element in the initially processed target data set, the corresponding target data is obtained, and according to the photoelectric sensor node data number of the current target data, the historical data of all the same photoelectric sensor node data number is obtained; the current target data is analyzed with the historical data of all the same photoelectric sensor node data number, and error correction is performed on each element in the target data set according to the similarity analysis result; S302. Analyze the corrected target data according to the processing method of the historical data in S102 and S103, so as to obtain the target error propagation coefficient C2; according to the photoelectric sensor node data number of the corrected target data, obtain all the historical error propagation coefficients C of the same photoelectric sensor node data number, and form a historical error propagation coefficient set B, and B = {b1(C), b2(C), ..., bv(C)}, wherein b1(C) represents the first historical error propagation coefficient C of the photoelectric sensor node data number that is the same as the target error propagation coefficient C2; b2(C) represents the second historical error propagation coefficient C of the photoelectric sensor node data number that is the same as the target error propagation coefficient C2, and so on, bv(C) represents the vth historical error propagation coefficient C of the photoelectric sensor node data number that is the same as the target error propagation coefficient C2, and v represents the number of historical error propagation coefficients C of the photoelectric sensor node data number that is the same as the target error propagation coefficient C2; S302. If the corrected target data corresponding to the target error propagation coefficient C2 has an approximate error propagation coefficient C1, the difference Q between the target error propagation coefficient C2 and the approximate error propagation coefficient C1 is calculated, and Q = |C2-C1| / C2. If the difference Q is less than or equal to the preset difference Q0, no prompt information is output; if the difference Q is greater than the preset difference Q0, the photoelectric sensor node data number of the current corrected target data is output to the relevant personnel, and the relevant personnel perform corresponding processing; If the corrected target data corresponding to the target error propagation coefficient C2 does not have an approximate error propagation coefficient C1, the target error propagation coefficient C2 is compared with the elements in the corresponding historical error propagation coefficient set B. If the historical error propagation coefficient set B contains a situation where the historical error propagation coefficient C is equal to the current target error propagation coefficient C1, the photoelectric sensor node data number of the current corrected target data is output to relevant personnel, who perform corresponding processing. If the historical error propagation coefficient set B does not contain a situation where the historical error propagation coefficient C is equal to the current target error propagation coefficient C1, the error correction scheme corresponding to the current target error propagation coefficient C1 is associated and stored.

5. The multi-sensor fusion processing method based on a distributed optoelectronic system according to claim 4, characterized in that: In S301, error correction is performed on each element in the target data set according to the similarity analysis result. The specific process is as follows: Perform similarity analysis on the current target data and all historical data with the same photoelectric sensor node data number. If the average similarity between the current target data and all historical data with the same photoelectric sensor node data number is greater than or equal to the similarity threshold, select the historical error propagation coefficient of the historical data with the largest similarity as the approximate error propagation coefficient C1, and correct the current target data according to the error correction scheme corresponding to the current approximate error propagation coefficient C1; If the average similarity between the current target data and all historical data with the same photoelectric sensor node data number is less than the similarity threshold, the current target data will be output to relevant personnel, who will perform corresponding correction processing.

6. The multi-sensor fusion processing method based on a distributed optoelectronic system according to claim 5, characterized in that: The step S400 includes: S401. Obtain the corrected target data and the corresponding reference data in the database, calculate the accuracy index Z of the corrected target data, and Z=1-(|Rs-Tc| / Tc), where Rs represents the corrected target data and Tc represents the corresponding reference data; combine the corrected target data and the corresponding reference data in the database to calculate the accuracy index J, and J=1-[V(Rs) / V(T)], where V(Rs) represents the variance of the corrected target data and V(T) represents the variance of the corresponding reference data; obtain the timestamp of the corrected target data and the expected update frequency W, calculate the timeliness index S, and S=1-(t0-t1) / W, where t0 represents the current timestamp and t1 represents the timestamp of the correction data; S402. Combined with the above indicators, calculate the comprehensive quality score P, and P = αZ + βJ + γS, where α, β and γ represent the weight coefficients of the above indicators respectively, and α + β + γ = 1; according to the comprehensive quality score P, if the comprehensive quality score P is less than or equal to the quality score threshold P0, the corresponding corrected target data is marked as data to be enhanced, and the data to be enhanced is subjected to corresponding data enhancement processing, and the target data after data enhancement replaces the corrected target data for storage; if the comprehensive quality score P is greater than the quality score threshold P0, the corresponding corrected target data does not need to be data enhanced.

7. The multi-sensor fusion processing method based on a distributed optoelectronic system according to claim 6, characterized in that: In step S402, the following analysis is further performed on the target data after data enhancement: For each photoelectric sensor node, the number of sensors m1 corresponding to the target data to be processed for data enhancement is counted, and the corresponding data enhancement ratio U is calculated by combining the number of sensors m1 and the number of all sensors m of the corresponding photoelectric sensor node, and U = m1 / m; If the data enhancement ratio U of a certain photoelectric sensor node is greater than or equal to 0.5, the data number of the corresponding photoelectric sensor node is output to relevant personnel, and the relevant personnel perform corresponding maintenance on the sensor equipment of the current photoelectric sensor node.

8. The multi-sensor fusion processing method based on a distributed optoelectronic system according to claim 1, characterized in that: The step S500 includes: S501. Extract corresponding data features from the enhanced target data, obtain corresponding feature vectors according to the data features, and perform standardization on the feature vectors corresponding to different sensors, and each sensor corresponds to a feature vector F; S502. For each photoelectric sensor node, perform weighted fusion on the features of all sensors, so that the standardized feature vectors are fused into a comprehensive feature representation, expressed as: F_fused=Σ k∈[1,m] wkFk, where k represents the sensor number corresponding to the photoelectric sensor node, wk represents the weight coefficient of the kth sensor corresponding to the photoelectric sensor node, and Fk represents the feature vector of the kth sensor corresponding to the photoelectric sensor node.

Images