A multi-dimensional fusion method and system based on perception data

Abstract

The present invention provides a multi-dimensional fusion method and system based on perception data, the method comprising: selecting a reference sensor, selecting a number of effective sensors corresponding to a target acquisition area based on the reference sensor, and selecting the reference sensor and the number of effective sensors as the first standby sensor; obtaining past perception tasks to select a number of second standby sensors from the first standby sensor; obtaining a number of real-time perception data, obtaining the temporal and spatial distances between the real-time perception data to obtain a screening threshold, and obtaining a standby data group; selecting a final data group from the number of standby data groups, and fusing the data in the final data group into fused data. Through the above method, the sensors are screened multiple times, while considering the position correlation and temporal and spatial correlation between different sensors, avoiding data fusion errors caused by low-correlation data, ensuring the accuracy of the fused data, and thus improving the accuracy of the prediction results.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a multi-dimensional fusion method and system based on perception data. Background Art

[0002] The Internet of Things originated in the field of media and is the third revolution in the information technology industry. The Internet of Things refers to connecting any object to the network through information sensing devices according to agreed protocols, and objects exchange and communicate information through information dissemination media to achieve intelligent recognition, prediction and other functions.

[0003] With the rapid development of the Internet of Things, many applications need to rely on accurate and reliable perception data as support. Perception data is generally obtained by setting up sensors for collection, but the amount of data provided by a single sensor is limited and it is difficult to use it as an accurate basis. The fusion of multiple perception data has improved the accuracy of the data to a certain extent. Taking weather forecasting as an example, after the collection is completed, the weather forecast is completed by fusing several perception data collected by several sensors.

[0004] However, the correlation between different sensors is different. Therefore, if all the perception data are directly fused, the data with weaker correlation will cause errors in the fused data, thus affecting the accuracy of the prediction results. Summary of the invention

[0005] In view of the shortcomings of the prior art, the purpose of the present invention is to provide a multidimensional fusion method and system based on perception data, aiming to solve the technical problem in the prior art that after directly fusing all the perception data, data with weak correlation will cause errors in the fused data, thereby affecting the accuracy of the prediction results.

[0006] In order to achieve the above objectives, in a first aspect, an embodiment of the present application provides a multi-dimensional fusion method based on perception data, comprising the following steps:

[0007] A reference sensor corresponding to a target acquisition area is selected from a plurality of sensors, and a plurality of valid sensors corresponding to the target acquisition area are selected from the remaining sensors based on the reference sensor, wherein the remaining sensors are sensors other than the reference sensor from the plurality of sensors, and the reference sensor and the plurality of valid sensors are selected as first standby sensors;

[0008] Acquire past sensing tasks, and extract past working data from the first standby sensor based on the past sensing tasks, and select a plurality of second standby sensors from a plurality of the first standby sensors based on the past working data;

[0009] Acquire a plurality of real-time perception data through a plurality of the second standby sensors, select one of the real-time perception data as the reference perception data, and select the remaining real-time perception data as the comparison perception data, acquire the spatiotemporal distance between the reference perception data and the comparison perception data, determine the screening threshold of the real-time perception data based on the spatiotemporal distance, select a plurality of standby perception data from the plurality of real-time perception data, and combine the plurality of standby perception data into a standby data group;

[0010] A final data group is selected from the plurality of standby data groups, and standby perception data corresponding to the final data group is selected as final perception data, and a fusion operation is performed on the plurality of final perception data to obtain fused data.

[0011] Compared with the prior art, the beneficial effects of the present invention are as follows: by selecting the reference sensor and obtaining the effective sensor based on the reference sensor, the positional relationship between different sensors is taken into consideration, and a screening of sensors is completed before data fusion, thereby improving the accuracy of subsequent data fusion to a certain extent; the past work data is extracted through the past perception tasks, and the sensors are screened for the second time before data fusion based on the data collection conditions of the sensors themselves, thereby further improving the accuracy of data fusion; after completing the screening of sensors, the spatiotemporal correlation between different sensors is taken into consideration by obtaining the spatiotemporal distance between the real-time perception data, thereby completing the elimination of the real-time perception data with weak spatiotemporal correlation, thereby avoiding data fusion errors caused by low-correlation data, ensuring the accuracy of the fused data, and thereby improving the accuracy of the prediction results.

[0012] Further, the step of selecting a plurality of valid sensors corresponding to the target acquisition area from the remaining sensors based on the reference sensor includes:

[0013] Acquire the reference space coordinates and reference detection distance of the reference sensor, and acquire the comparison space coordinates of the remaining sensors;

[0014] A plurality of valid sensors corresponding to the target acquisition area are selected from the remaining sensors based on the reference space coordinates, the comparison space coordinates and the reference detection distance.

[0015] Furthermore, the formula for obtaining the effective sensor is:

[0016] ,

[0017] in, , , They respectively represent the reference space x-axis coordinate, reference space y-axis coordinate, and reference space z-axis coordinate of the reference sensor in the spatial coordinate system. , , They respectively represent the x-axis coordinate, y-axis coordinate, and z-axis coordinate of the remaining i-th sensor in the spatial coordinate system, Indicates the reference detection distance.

[0018] Furthermore, the past perception tasks include past subordinate tasks corresponding to different acquisition areas, the past subordinate tasks include a plurality of past subtasks, the past work data include the number of past area participations and the number of past task participations, and the steps of extracting the past work data from the first standby sensor based on the past perception tasks, and selecting a plurality of second standby sensors from a plurality of the first standby sensors based on the past work data include:

[0019] Extracting the past collected data of the first standby sensor, and comparing the past collected data with the past subtasks in the past hanging tasks, respectively, to determine whether the first standby sensor participated in the past hanging tasks, and then obtaining the past regional participation times of the first standby sensor;

[0020] Compare the past collection data with all the past subtasks to obtain the number of past task sub-participation of the first standby sensor in different collection areas, and obtain the number of past task participation of the first standby sensor based on several of the past task sub-participation numbers;

[0021] Determine, based on the past task sub-participation times and the past regional participation times, a first weight component corresponding to the past task participation times and a second weight component corresponding to the past regional participation times;

[0022] Obtaining a real-name score of the first stand-by sensor through the first weight component, the number of past task participations, the second weight component, and the number of past regional participations;

[0023] A plurality of second sensors to be used are selected from the plurality of first sensors to be used based on the real-name score.

[0024] Furthermore, the calculation formula of the first weight component is:

[0025] ,

[0026] in, represents the first weight component of the i-th first standby sensor, represents the number of past task sub-participations of the i-th first standby sensor in the first collection area, represents the number of past task sub-participations of the i-th first standby sensor in the second acquisition area, represents the number of past task sub-participations of the i-th first standby sensor in the j-th acquisition area, Indicates the total number of past subtasks in the first collection area. Indicates the total number of past subtasks in the second collection area. represents the total number of past subtasks in the jth collection area, represents the number of past region participations of the i-th first standby sensor;

[0027] The calculation formula of the second weight component is:

[0028] ,

[0029] in, represents the second weight component of the i-th first standby sensor;

[0030] The formula for obtaining the real-name rating is:

[0031] ,

[0032] in, represents the real-name score of the i-th first standby sensor, Indicates the number of past tasks participated by the i-th first standby sensor.

[0033] Furthermore, the calculation formula of the space-time distance is:

[0034] ,

[0035] in, represents the spatiotemporal distance between the a-th benchmark perception data and the b-th comparison perception data corresponding to the a-th benchmark perception data in t sampling periods, represents the Euclidean distance between the mth time series data in the ath benchmark perception data and the nth time series data in the bth comparison perception data corresponding to the ath benchmark perception data, represents the spatiotemporal distance between the a-th benchmark perception data in the t-1 sampling period and the b-th comparison perception data corresponding to the a-th benchmark perception data in the t sampling period, represents the spatiotemporal distance between the a-th benchmark perception data in the t sampling period and the b-th comparison perception data corresponding to the a-th benchmark perception data in the t-1 sampling period, It represents the spatiotemporal distance between the a-th benchmark perception data in the t-1 sampling period and the b-th comparison perception data corresponding to the a-th benchmark perception data in the t-1 sampling period.

[0036] Furthermore, the step of determining the screening threshold of the real-time perception data based on the spatiotemporal distance includes:

[0037] numerically sorting the plurality of space-time distances to select a median distance from the plurality of space-time distances;

[0038] The screening threshold of the real-time sensing data is calculated by using a plurality of the spatiotemporal distances and the median distance.

[0039] Furthermore, the calculation formula of the screening threshold is:

[0040] ,

[0041] in, represents the screening threshold, represents the spatiotemporal distance between the a-th benchmark perception data and the b-th comparison perception data corresponding to the a-th benchmark perception data in t sampling periods, Indicates the median operation.

[0042] Furthermore, the step of fusing the plurality of final perception data to obtain fused data includes:

[0043] Selecting starting perception data from a plurality of the final perception data, and selecting the remaining final perception data as perception data to be fused;

[0044] The spatiotemporal correlation between the initial perception data and the perception data to be integrated is obtained, and data weights of the initial perception data and the perception data to be integrated are determined based on the spatiotemporal correlation, so as to integrate the initial perception data and the perception data to be integrated into integrated data through the data weights.

[0045] In a second aspect, an embodiment of the present application provides a multi-dimensional fusion system based on perception data, which is applied to the multi-dimensional fusion method based on perception data as described in the first aspect above, and the system includes:

[0046] A first screening module is used to select a reference sensor corresponding to a target acquisition area from a plurality of sensors, select a plurality of valid sensors corresponding to the target acquisition area from the remaining sensors based on the reference sensor, the remaining sensors are sensors other than the reference sensor from the plurality of sensors, and select the reference sensor and the plurality of valid sensors as first standby sensors;

[0047] a second screening module, configured to obtain past sensing tasks, extract past working data from the first standby sensor based on the past sensing tasks, and select a plurality of second standby sensors from a plurality of the first standby sensors based on the past working data;

[0048] an analysis module, configured to obtain a plurality of real-time perception data through a plurality of the second standby sensors, select one of the real-time perception data as the reference perception data, select the remaining real-time perception data as the comparison perception data, obtain the spatiotemporal distance between the reference perception data and the comparison perception data, determine a screening threshold of the real-time perception data based on the spatiotemporal distance, select a plurality of standby perception data from the plurality of real-time perception data, and combine the plurality of standby perception data into a standby data group;

[0049] The combination module is used to select a final data group from the plurality of standby data groups, select the standby perception data corresponding to the final data group as final perception data, and perform a fusion operation on the plurality of final perception data to obtain fused data.

[0050] In a third aspect, an embodiment of the present application provides a computer, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the multidimensional fusion method based on perception data as described in the first aspect above is implemented.

[0051] In a fourth aspect, an embodiment of the present application provides a storage medium on which a computer program is stored. When the computer program is executed by a processor, the multi-dimensional fusion method based on perception data as described in the first aspect above is implemented. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 Flow chart of a multi-dimensional fusion method based on perception data in a first embodiment of the present invention;

[0053] Figure 2 It is a structural block diagram of a multi-dimensional fusion system based on perception data in the second embodiment of the present invention;

[0054] The following specific implementation manner will further illustrate the present invention in conjunction with the above-mentioned drawings. DETAILED DESCRIPTION

[0055] In order to facilitate the understanding of the present invention, the present invention will be described more fully below with reference to the relevant drawings. Several embodiments of the present invention are given in the drawings. However, the present invention can be implemented in many different forms and is not limited to the embodiments described herein. On the contrary, the purpose of providing these embodiments is to make the disclosure of the present invention more thorough and comprehensive.

[0056] It should be noted that when an element is referred to as being "fixed to" another element, it may be directly on the other element or there may be a central element. When an element is considered to be "connected to" another element, it may be directly connected to the other element or there may be a central element at the same time. The terms "vertical", "horizontal", "left", "right" and similar expressions used herein are for illustrative purposes only.

[0057] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which the present invention belongs. The terms used herein in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The term "and / or" used herein includes any and all combinations of one or more of the related listed items.

[0058] See also Figure 1 The multi-dimensional fusion method based on perception data provided by the first embodiment of the present invention includes the following steps:

[0059] S10: selecting a reference sensor corresponding to a target acquisition area from a plurality of sensors, and selecting a plurality of valid sensors corresponding to the target acquisition area from the remaining sensors based on the reference sensor, wherein the remaining sensors are sensors other than the reference sensor from the plurality of sensors, and selecting the reference sensor and the plurality of valid sensors as first standby sensors;

[0060] The step S10 comprises:

[0061] S110: Acquire the reference space coordinates and reference detection distance of the reference sensor, and acquire the comparison space coordinates of the remaining sensors;

[0062] It can be understood that by constructing a spatial coordinate system in the target acquisition area, the reference spatial coordinates, the reference detection distance and the comparison spatial coordinates can be obtained.

[0063] S120: selecting a number of valid sensors corresponding to the target acquisition area from the remaining sensors based on the reference space coordinates, the comparison space coordinates and the reference detection distance;

[0064] The acquisition formula of the effective sensor is:

[0065] ,

[0066] in, , , They respectively represent the reference space x-axis coordinate, reference space y-axis coordinate, and reference space z-axis coordinate of the reference sensor in the spatial coordinate system. , , They respectively represent the x-axis coordinate, y-axis coordinate, and z-axis coordinate of the remaining i-th sensor in the spatial coordinate system, It can be understood that when the relationship between the reference space coordinate and the comparison space coordinate satisfies the above formula, the sensor corresponding to the comparison space coordinate is selected as the effective sensor.

[0067] S20: acquiring past sensing tasks, and extracting past working data from the first standby sensor based on the past sensing tasks, and selecting a plurality of second standby sensors from a plurality of the first standby sensors based on the past working data;

[0068] The past perception tasks include past subtasks corresponding to different collection areas, the past subtasks include several past subtasks, and the past work data include the number of past area participations and the number of past task participations. The past subtasks are essentially a collection of several past subtasks under a certain collection area, and each past subtask corresponds to a collection area.

[0069] The step S20 comprises:

[0070] S210: extracting the past collected data of the first standby sensor, and comparing the past collected data with the past subtasks in the past hanging tasks, so as to determine whether the first standby sensor participates in the past hanging tasks, and further obtain the past regional participation times of the first standby sensor;

[0071] The essence of comparing the past collected data with the past subtasks in the past hanging tasks is to query whether the past collected data exists in the basic data corresponding to the past subtasks. If so, it is determined that the first standby sensor has participated in the previous hanging tasks, that is, the number of past area participations is recorded as 1. It can be understood that the number of past area participations is not greater than the number of collection areas.

[0072] S220: Compare the past collection data with all the past subtasks to obtain the number of past task sub-participation of the first standby sensor in different collection areas, and obtain the number of past task sub-participation of the first standby sensor based on the number of past task sub-participation numbers;

[0073] It can be understood that in a collection area, that is, in a previous subtask, it is determined whether the previous collection data exists in the basic data corresponding to each previous subtask. If so, the previous task sub-participation times are recorded once. For different collection areas, the previous task sub-participation times are different, and then the previous task participation times are obtained by superimposing several previous subtask participation times.

[0074] S230: Determine a first weight component corresponding to the number of past task participations and a second weight component corresponding to the number of past regional participations based on the number of past task sub-participations and the number of past regional participations;

[0075] The calculation formula of the first weight component is:

[0076] ,

[0077] in, represents the first weight component of the i-th first standby sensor, represents the number of past task sub-participations of the i-th first standby sensor in the first collection area, represents the number of past task sub-participations of the i-th first standby sensor in the second acquisition area, represents the number of past task sub-participations of the i-th first standby sensor in the j-th acquisition area, Indicates the total number of past subtasks in the first collection area. Indicates the total number of past subtasks in the second collection area. represents the total number of past subtasks in the jth collection area, represents the number of past region participations of the i-th first standby sensor;

[0078] The calculation formula of the second weight component is:

[0079] ,

[0080] in, represents the second weight component of the i-th first stand-by sensor.

[0081] S240: Obtaining a real-name score of the first stand-by sensor through the first weight component, the number of past task participations, the second weight component, and the number of past area participations;

[0082] The formula for obtaining the real-name rating is:

[0083] ,

[0084] in, represents the real-name score of the i-th first standby sensor, represents the number of past task participations of the i-th first standby sensor. It can be understood that the real-name score of each of the first standby sensors needs to be obtained.

[0085] S250: selecting a plurality of second standby sensors from the plurality of first standby sensors based on the real-name score;

[0086] The real-name score is compared with the score threshold, and the first stand-by sensor corresponding to the real-name score greater than the score threshold is selected as the second stand-by sensor. It can be understood that this step takes into account the participation and scope of the sensor in different perception tasks when it worked in the past, and then judges the working efficiency of the sensor through the real-name score, filters out inefficient sensors, and avoids them affecting the accuracy of the data.

[0087] S30: acquiring a plurality of real-time perception data through a plurality of the second standby sensors, selecting one of the real-time perception data as the reference perception data, and selecting the remaining real-time perception data as the comparison perception data, acquiring the spatiotemporal distance between the reference perception data and the comparison perception data, determining a screening threshold of the real-time perception data based on the spatiotemporal distance, so as to select a plurality of standby perception data from the plurality of real-time perception data, and combining the plurality of standby perception data into a standby data group;

[0088] It should be noted that the real-time sensing data is time series data. The calculation formula of the spatiotemporal distance is:

[0089] ,

[0090] in, represents the spatiotemporal distance between the a-th benchmark perception data and the b-th comparison perception data corresponding to the a-th benchmark perception data in t sampling periods, represents the Euclidean distance between the mth time series data in the ath benchmark perception data and the nth time series data in the bth comparison perception data corresponding to the ath benchmark perception data, represents the spatiotemporal distance between the a-th benchmark perception data in the t-1 sampling period and the b-th comparison perception data corresponding to the a-th benchmark perception data in the t sampling period, represents the spatiotemporal distance between the a-th benchmark perception data in the t sampling period and the b-th comparison perception data corresponding to the a-th benchmark perception data in the t-1 sampling period, It represents the spatiotemporal distance between the ath reference perception data in the t-1 sampling period and the bth comparison perception data corresponding to the ath reference perception data in the t-1 sampling period. In the actual acquisition process, the number of time series sub-data of different sensors should be the same and correspond to the sampling period one by one. It can be seen from the above formula that in this embodiment, the situation that the sensor loses data during the process of collecting perception data is also considered. In the case of data loss, the data of adjacent sampling periods can be used as a reference to still obtain the spatiotemporal distance.

[0091] The step S30 comprises:

[0092] S310: numerically sorting the plurality of space-time distances to select a median distance from the plurality of space-time distances;

[0093] S320: Calculating a screening threshold of the real-time sensing data by using a plurality of the spatiotemporal distances and the median distance;

[0094] The calculation formula of the screening threshold is:

[0095] ,

[0096] in, represents the screening threshold, represents the spatiotemporal distance between the a-th benchmark perception data and the b-th comparison perception data corresponding to the a-th benchmark perception data in t sampling periods, Indicates the median operation.

[0097] Preferably, the step of selecting a plurality of standby perception data from a plurality of real-time perception data comprises:

[0098] Comparing a plurality of the spatiotemporal distances with the screening thresholds respectively;

[0099] If the spatiotemporal distance is greater than the screening threshold, the contrast perception data corresponding to the spatiotemporal distance greater than the screening threshold is selected as data to be deleted, and the data to be deleted is eliminated;

[0100] If the spatiotemporal distance is smaller than the screening threshold, the contrast perception data corresponding to the spatiotemporal distance smaller than the screening threshold is selected as the stand-by perception data.

[0101] S40: selecting a final data group from the plurality of standby data groups, selecting the standby sensing data corresponding to the final data group as final sensing data, and performing a fusion operation on the plurality of final sensing data to obtain fused data;

[0102] It can be understood that by selecting each of the real-time perception data as the reference perception data in turn, the standby data groups can be obtained in turn, and the number of the standby data groups is consistent with the number of the real-time perception data. The step of selecting the final data group from the plurality of standby data groups is specifically: obtaining the number of standby perception data in the standby data groups, and selecting the standby data group with the largest number as the final data group. By selecting the largest number of standby perception data for fusion, the accuracy of the fused data can be improved through strongly correlated multiple data.

[0103] The step S40 comprises:

[0104] S410: Selecting starting perception data from a plurality of the final perception data, and selecting the remaining final perception data as perception data to be fused;

[0105] S420: Acquire the spatiotemporal correlation between the initial perception data and the perception data to be fused, and determine data weights of the initial perception data and the perception data to be fused based on the spatiotemporal correlation, so as to fuse the initial perception data and the perception data to be fused into fused data by using the data weights;

[0106] It should be noted that the essential meanings of the initial perception data, the perception data to be integrated, the baseline perception data and the comparative perception data are all perception data. Therefore, there is also the spatiotemporal distance between the initial perception data and the perception data to be integrated. The spatiotemporal correlation can be obtained by quantifying the spatiotemporal distance through an exponential function. After obtaining the spatiotemporal correlation, several spatiotemporal correlations are summed to obtain the total correlation, and the proportion of a single spatiotemporal correlation in the total correlation is obtained to obtain the data weight, and then the fused data is formed by summing the product of the initial perception data and the corresponding data weight, and the perception data to be integrated and the corresponding data weight. By obtaining the spatiotemporal correlation, the correlation between the data is further considered, and the accuracy of data fusion is improved.

[0107] By selecting the reference sensor and obtaining the effective sensor based on the reference sensor, the positional relationship between different sensors is taken into consideration, and a screening of sensors is completed before data fusion, thereby improving the accuracy of subsequent data fusion to a certain extent; the past work data is extracted through the past perception tasks, and the sensors are screened for the second time before data fusion based on the data collection conditions of the sensors themselves, thereby further improving the accuracy of data fusion; after completing the screening of sensors, the spatiotemporal correlation between different sensors is taken into consideration by obtaining the spatiotemporal distance between the real-time perception data, thereby completing the elimination of the real-time perception data with weak spatiotemporal correlation, thereby avoiding data fusion errors caused by low-correlation data, ensuring the accuracy of the fused data, and thereby improving the accuracy of the prediction results.

[0108] See also Figure 2 The second embodiment of the present invention provides a multi-dimensional fusion system based on perception data, which is applied to the multi-dimensional fusion method based on perception data in the above embodiment, and will not be repeated here. As used below, the terms "module", "unit", "sub-unit", etc. can be a combination of software and / or hardware that implements predetermined functions. Although the devices described in the following embodiments are preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceivable.

[0109] The system comprises:

[0110] A first screening module 10 is used to select a reference sensor corresponding to a target acquisition area from a plurality of sensors, select a plurality of valid sensors corresponding to the target acquisition area from the remaining sensors based on the reference sensor, the remaining sensors are sensors other than the reference sensor from the plurality of sensors, and select the reference sensor and the plurality of valid sensors as first standby sensors;

[0111] The first screening module 10 comprises:

[0112] A first unit is used to obtain the reference space coordinates and the reference detection distance of the reference sensor, and to obtain the comparison space coordinates of the remaining sensors;

[0113] A second unit is used to select a number of valid sensors corresponding to the target acquisition area from the remaining sensors based on the reference space coordinates, the comparison space coordinates and the reference detection distance;

[0114] The second screening module 20 is used to obtain past sensing tasks, extract past working data from the first standby sensor based on the past sensing tasks, and select a plurality of second standby sensors from a plurality of the first standby sensors based on the past working data;

[0115] The second screening module 20 comprises:

[0116] The third unit is used to extract the past collection data of the first standby sensor, and compare the past collection data with the past subtasks in the past hanging tasks to determine whether the first standby sensor participates in the past hanging tasks, and then obtain the past regional participation times of the first standby sensor;

[0117] The fourth unit is used to compare the past collection data with all the past subtasks to obtain the number of past task sub-participation of the first standby sensor in different collection areas, and obtain the number of past task participation of the first standby sensor based on several past task sub-participation numbers;

[0118] A fifth unit is used to determine a first weight component corresponding to the number of past task participations and a second weight component corresponding to the number of past regional participations based on the number of past task sub-participations and the number of past regional participations;

[0119] A sixth unit is used to obtain the real-name score of the first stand-by sensor through the first weight component, the number of past task participations, the second weight component and the number of past regional participations;

[0120] A seventh unit, configured to select a plurality of second standby sensors from the plurality of first standby sensors based on the real-name score;

[0121] The analysis module 30 is used to obtain a plurality of real-time perception data through a plurality of the second standby sensors, select one of the real-time perception data as the reference perception data, and select the remaining real-time perception data as the comparison perception data, obtain the spatiotemporal distance between the reference perception data and the comparison perception data, determine the screening threshold of the real-time perception data based on the spatiotemporal distance, so as to select a plurality of standby perception data from the plurality of real-time perception data, and combine the plurality of standby perception data into a standby data group;

[0122] The analysis module 30 includes:

[0123] An eighth unit is used for numerically sorting the plurality of said space-time distances to select a median distance from the plurality of said space-time distances;

[0124] A ninth unit, configured to calculate a screening threshold of the real-time sensing data by using a plurality of the spatiotemporal distances and the median distance;

[0125] A combining module 40 is used to select a final data group from the plurality of the standby data groups, select the standby sensing data corresponding to the final data group as the final sensing data, and perform a fusion operation on the plurality of the final sensing data to obtain fused data;

[0126] The combined module 40 comprises:

[0127] A tenth unit is used to select starting perception data from a plurality of the final perception data, and select the remaining final perception data as perception data to be fused;

[0128] The eleventh unit is used to obtain the spatiotemporal correlation between the initial perception data and the perception data to be integrated, and determine the data weights of the initial perception data and the perception data to be integrated based on the spatiotemporal correlation, so as to integrate the initial perception data and the perception data to be integrated into integrated data through the data weights.

[0129] The present invention also provides a computer, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the multidimensional fusion method based on perception data as described in the above technical solution is implemented.

[0130] The present invention also provides a storage medium on which a computer program is stored. When the computer program is executed by a processor, the multi-dimensional fusion method based on perception data as described in the above technical solution is implemented.

[0131] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "examples", "specific examples", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0132] The above-mentioned embodiments only express several implementation methods of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the scope of the patent of the present invention. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.

Claims

1. A multi-dimensional fusion method based on perception data, characterized in that: The following steps are involved: A reference sensor corresponding to a target acquisition area is selected from a plurality of sensors, and a plurality of valid sensors corresponding to the target acquisition area are selected from the remaining sensors based on the reference sensor, wherein the remaining sensors are sensors other than the reference sensor from the plurality of sensors, and the reference sensor and the plurality of valid sensors are selected as first standby sensors; Acquire past sensing tasks, and extract past working data from the first standby sensor based on the past sensing tasks, and select a plurality of second standby sensors from a plurality of the first standby sensors based on the past working data; Acquire a plurality of real-time perception data through a plurality of the second standby sensors, select one of the real-time perception data as the reference perception data, and select the remaining real-time perception data as the comparison perception data, acquire the spatiotemporal distance between the reference perception data and the comparison perception data, determine the screening threshold of the real-time perception data based on the spatiotemporal distance, select a plurality of standby perception data from the plurality of real-time perception data, and combine the plurality of standby perception data into a standby data group; The calculation formula of the space-time distance is: ， in, represents the spatiotemporal distance between the a-th benchmark perception data and the b-th comparison perception data corresponding to the a-th benchmark perception data in t sampling periods, represents the Euclidean distance between the mth time series data in the ath benchmark perception data and the nth time series data in the bth comparison perception data corresponding to the ath benchmark perception data, represents the spatiotemporal distance between the a-th benchmark perception data in the t-1 sampling period and the b-th comparison perception data corresponding to the a-th benchmark perception data in the t sampling period, represents the spatiotemporal distance between the a-th benchmark perception data in the t sampling period and the b-th comparison perception data corresponding to the a-th benchmark perception data in the t-1 sampling period, represents the spatiotemporal distance between the a-th benchmark perception data in the t-1 sampling period and the b-th comparison perception data corresponding to the a-th benchmark perception data in the t-1 sampling period; A final data group is selected from the plurality of standby data groups, and standby perception data corresponding to the final data group is selected as final perception data, and a fusion operation is performed on the plurality of final perception data to obtain fused data.

2. The multidimensional fusion method based on perception data according to claim 1 is characterized in that: The step of selecting a plurality of valid sensors corresponding to the target acquisition area from the remaining sensors based on the reference sensor comprises: Acquire the reference space coordinates and reference detection distance of the reference sensor, and acquire the comparison space coordinates of the remaining sensors; A plurality of valid sensors corresponding to the target acquisition area are selected from the remaining sensors based on the reference space coordinates, the comparison space coordinates and the reference detection distance.

3. The multi-dimensional fusion method based on perception data according to claim 2 is characterized in that: The acquisition formula of the effective sensor is: ， in, , , They respectively represent the reference space x-axis coordinate, reference space y-axis coordinate, and reference space z-axis coordinate of the reference sensor in the spatial coordinate system. , , They respectively represent the x-axis coordinate, y-axis coordinate, and z-axis coordinate of the comparison space of the remaining i-th sensor in the spatial coordinate system, Indicates the reference detection distance.

4. The multi-dimensional fusion method based on perception data according to claim 1, characterized in that: The past perception tasks include past subordinate tasks corresponding to different acquisition areas, the past subordinate tasks include a plurality of past subtasks, the past work data include the number of past area participations and the number of past task participations, and the steps of extracting the past work data from the first standby sensor based on the past perception tasks, and selecting a plurality of second standby sensors from a plurality of the first standby sensors based on the past work data include: Extracting the past collected data of the first standby sensor, and comparing the past collected data with the past subtasks in the past hanging tasks, respectively, to determine whether the first standby sensor participated in the past hanging tasks, and then obtaining the past regional participation times of the first standby sensor; Compare the past collection data with all the past subtasks to obtain the number of past task sub-participation of the first standby sensor in different collection areas, and obtain the number of past task participation of the first standby sensor based on several of the past task sub-participation numbers; Determine, based on the past task sub-participation times and the past regional participation times, a first weight component corresponding to the past task participation times and a second weight component corresponding to the past regional participation times; Obtaining a real-name score of the first stand-by sensor through the first weight component, the number of past task participations, the second weight component, and the number of past regional participations; A plurality of second sensors to be used are selected from the plurality of first sensors to be used based on the real-name score.

5. The multi-dimensional fusion method based on perception data according to claim 4 is characterized in that: The calculation formula of the first weight component is: ， in, represents the first weight component of the i-th first standby sensor, represents the number of past task sub-participations of the i-th first standby sensor in the first collection area, represents the number of past task sub-participations of the i-th first standby sensor in the second acquisition area, represents the number of past task sub-participations of the i-th first standby sensor in the j-th acquisition area, Indicates the total number of past subtasks in the first collection area. Indicates the total number of past subtasks in the second collection area. represents the total number of past subtasks in the jth collection area, represents the number of past region participations of the i-th first standby sensor; The calculation formula of the second weight component is: ， in, represents the second weight component of the i-th first standby sensor; The formula for obtaining the real-name rating is: ， in, represents the real-name score of the i-th first standby sensor, Indicates the number of past tasks participated by the i-th first standby sensor.

6. The multi-dimensional fusion method based on perception data according to claim 1, characterized in that: The step of determining the screening threshold of the real-time perception data based on the spatiotemporal distance comprises: numerically sorting the plurality of space-time distances to select a median distance from the plurality of space-time distances; The screening threshold of the real-time sensing data is calculated by using a plurality of the spatiotemporal distances and the median distance.

7. The multi-dimensional fusion method based on perception data according to claim 6 is characterized in that: The calculation formula of the screening threshold is: ， in, represents the screening threshold, represents the spatiotemporal distance between the a-th benchmark perception data and the b-th comparison perception data corresponding to the a-th benchmark perception data in t sampling periods, Indicates the median operation.

8. The multi-dimensional fusion method based on perception data according to claim 1, characterized in that: The step of fusing the plurality of final perception data to obtain fused data comprises: Selecting starting perception data from a plurality of the final perception data, and selecting the remaining final perception data as perception data to be fused; The spatiotemporal correlation between the initial perception data and the perception data to be integrated is obtained, and data weights of the initial perception data and the perception data to be integrated are determined based on the spatiotemporal correlation, so as to integrate the initial perception data and the perception data to be integrated into integrated data through the data weights.

9. A multi-dimensional fusion system based on perception data, applied to the multi-dimensional fusion method based on perception data as claimed in any one of claims 1 to 8, characterized in that: The system comprises: A first screening module is used to select a reference sensor corresponding to a target acquisition area from a plurality of sensors, select a plurality of valid sensors corresponding to the target acquisition area from the remaining sensors based on the reference sensor, the remaining sensors are sensors other than the reference sensor from the plurality of sensors, and select the reference sensor and the plurality of valid sensors as first standby sensors; a second screening module, configured to obtain past sensing tasks, extract past working data from the first standby sensor based on the past sensing tasks, and select a plurality of second standby sensors from a plurality of the first standby sensors based on the past working data; an analysis module, configured to obtain a plurality of real-time perception data through a plurality of the second standby sensors, select one of the real-time perception data as the reference perception data, select the remaining real-time perception data as the comparison perception data, obtain the spatiotemporal distance between the reference perception data and the comparison perception data, determine a screening threshold of the real-time perception data based on the spatiotemporal distance, select a plurality of standby perception data from the plurality of real-time perception data, and combine the plurality of standby perception data into a standby data group; The calculation formula of the space-time distance is: ， in, represents the spatiotemporal distance between the a-th benchmark perception data and the b-th comparison perception data corresponding to the a-th benchmark perception data in t sampling periods, represents the Euclidean distance between the mth time series data in the ath benchmark perception data and the nth time series data in the bth comparison perception data corresponding to the ath benchmark perception data, represents the spatiotemporal distance between the a-th benchmark perception data in the t-1 sampling period and the b-th comparison perception data corresponding to the a-th benchmark perception data in the t sampling period, represents the spatiotemporal distance between the a-th benchmark perception data in the t sampling period and the b-th comparison perception data corresponding to the a-th benchmark perception data in the t-1 sampling period, represents the spatiotemporal distance between the a-th benchmark perception data in the t-1 sampling period and the b-th comparison perception data corresponding to the a-th benchmark perception data in the t-1 sampling period; The combination module is used to select a final data group from the plurality of standby data groups, select the standby perception data corresponding to the final data group as final perception data, and perform a fusion operation on the plurality of final perception data to obtain fused data.

Images