Mobile voice target system based on multi-dimensional perception technology

Abstract

The invention belongs to the technical field of target training, and provides a mobile voice target system based on a multi-dimensional perception technology, which comprises a target end, a shooting end hardware module and a shooting target module, the target end integrates various parts, can be placed as required, and can simulate training, target scoring and the like; the shooting end simulates shooting; the training score calculation module calculates non-sound-listening position distinguishing scores and sound-listening position distinguishing scores, and the sound-listening position distinguishing scores are calculated in combination with factors such as time and ring number; the sound listening and position distinguishing score analysis module processes the data and establishes a deep learning score analysis model, through data processing and standardized normalization, linear regression modeling and coefficient calculation by a least square method are performed, the influence of characteristic factors on scores is analyzed according to coefficient absolute values, and the established model is utilized to scientifically guide, train and improve the scores.

Description

Technical Field

[0001] The invention belongs to the technical field of target training, in particular to a mobile voice target system based on multi-dimensional perception technology. Background Art

[0002] In traditional shooting training, target systems mostly use live ammunition shooting, which not only has high training costs, but also poses great safety risks in the transportation, storage and use of live ammunition. At the same time, traditional targets are difficult to simulate complex and changeable training scenarios and cannot meet diverse training needs. For example, the training methods for the ability to identify positions by sound are extremely limited. In addition, during night training, traditional targets lack an effective way to display shooting results, which affects the training effect. Some existing simulated shooting training systems have single functions and are not scientific and accurate enough in score calculation and analysis, making it difficult to provide effective training guidance to trainees. To this end, the present invention provides a mobile voice target system based on multi-dimensional perception technology. Summary of the invention

[0003] In order to make up for the deficiencies of the prior art, at least one technical problem raised in the background technology is solved.

[0004] The technical solution adopted by the present invention to solve its technical problem is: In a first aspect, the present invention provides a mobile voice target system based on multi-dimensional perception technology, comprising the following modules: Shooting end module: uses invisible infrared laser to simulate the shooting process; Target module: receives the invisible infrared laser emitted by the shooting module and forms physical interaction, while simulating training sounds; Shooting training module: Use the shooting end and the target end to build a dual training mode, namely, the sound identification training and the non-sound identification training mode; Score calculation module: Process the data obtained from shooting training, determine the geometric partition of the target board, determine the number of rings, and introduce the area normalization coefficient to calculate the score; derive the sound identification training score based on the non-sound identification training score, and use the accumulation method to calculate the total score; Sound identification performance analysis module: processes the sound identification performance and establishes a training performance analysis model; Sound identification training guidance module: Use linear regression model for training guidance.

[0005] In a second aspect, the present invention provides a mobile speech target method based on multi-dimensional perception technology, comprising the following steps: S1: Use invisible infrared laser to simulate the shooting process; S2: Receives the invisible infrared laser emitted by the shooting end module and forms physical interaction, while simulating training sounds; S3: Use the shooting end and the target end to build a dual training mode, namely, the sound identification training and the non-sound identification training mode; S4: Process the data obtained from the shooting training, determine the geometric partition of the target board, determine the number of rings, and introduce the area normalization coefficient to calculate the score; obtain the sound identification training score based on the non-sound identification training score, and use the accumulation method to calculate the total score; S5: Process the sound location identification results and establish a training results analysis model; S6: Use linear regression model for training guidance.

[0006] The beneficial effects of the present invention are as follows: compared with the traditional live-fire shooting method, the use of laser simulated targets and shooting process can significantly reduce training costs, and can simulate complex and changeable training scenes to meet diversified training needs; the integrated voice module and the realization of multiple target linkage can generate irregular sound signals, effectively training the shooter's ability to hear and identify positions; the design of the breathing indicator light meets the needs of night training and intuitively displays the shooting results; it has a scientific and accurate performance calculation module, which can calculate the non-audio positioning and audio positioning training results respectively; the performance analysis module establishes a model by processing a large amount of historical data to optimize the training parameters. BRIEF DESCRIPTION OF THE DRAWINGS

[0007] The present invention will be further described below in conjunction with the accompanying drawings.

[0008] Figure 1 It is a system module diagram of a mobile voice target system based on multi-dimensional perception technology of the present invention; Figure 2 It is a flow chart of steps of a mobile voice target method based on multi-dimensional perception technology of the present invention; Figure 3 is a front view schematic diagram of the target model in Example 1 of the present invention; Figure 4 It is a schematic diagram of the back side of the target model in Example 1 of the present invention. DETAILED DESCRIPTION

[0009] In order to make the technical means, creative features, objectives and effects achieved by the present invention easy to understand, the present invention is further explained below in conjunction with specific implementation methods.

[0010] Embodiment 1: Figure 1 As shown, a mobile voice target system based on multi-dimensional perception technology according to an embodiment of the present invention includes: Shooting end module: uses laser diode as the core element for emitting laser to simulate the shooting process; driving circuit: provides appropriate driving current for laser diode; power supply module: provides stable power for laser emission; microcontroller: used for burst mode in shooting simulation according to preset program logic; switch: used to control the opening and closing of the shooting end; Target end module: uses a photoelectric sensor array to receive invisible infrared lasers and convert optical signals into electrical signals; signal processing circuit: includes amplifiers, filters, and analog-to-digital converters ADC to remove noise interference in electrical signals and improve signal quality and reliability; microcontroller STM32: used to determine the hit of the laser beam; breathing indicator light: used to intuitively display the shooting results; voice module: used to simulate training sounds and voice target reporting; communication module: used to transmit training data; power supply: uses a 3.7-volt 18650 battery to power the entire system; switch: used to control the opening and closing of the target; like Figure 3 and Figure 4 As shown, all the above hardware parts are integrated into a humanoid target of standard imperial size; it should be noted that, according to actual training needs, the target can be placed on a desktop, or can be placed on the ground outdoors with a special bracket; a magnetic device is integrated behind the target, and the magnetic technology can be used to place the target in a suitable position for training; In addition, if Figure 3 As shown, the breathing indicator light is integrated on the target panel of the humanoid target to intuitively display the shooting results; the voice module is integrated at the bottom of the target to simulate the training sound and voice target reporting; Shooting training module: When conducting simulated shooting training, first arrange the targets in advance according to the training content; if it is non-sound identification training, the appropriate target shooting distance and number of targets can be set according to the training requirements; if it is sound identification training, it is necessary to set several targets at different distances and directions according to the training requirements, and these targets are linked and adjusted; after linkage adjustment, different targets will alternately emit training sounds according to irregular instructions, which is convenient for shooters to practice the ability of sound identification; after completing the target arrangement, after turning the switch on the back of the target, the target indicator light flashes green for 1S and then goes out, indicating that the humanoid target is successfully started; at the same time, press the switch integrated in the shooting end to start the shooting end equipment and officially start the simulation training; It should be noted that in the dynamic target shooting training mode, the target shooting must be carried out within 5 seconds after the training sound is emitted. The red light of the indicator light that hits the target will flash for 1 second and then go out, and the number of rings will be reported. The blue light of the indicator light that does not hit the target will flash for 1 second and then go out, and the next shooting training will be carried out. The interval between each training is 1 minute. In the technical solution of this embodiment, laser is used to simulate the target and shooting process in traditional technology; this innovative measure can effectively reduce the training cost and ensure the safety of training; the design of the target incorporates a breathing indicator light, which can fully meet the needs of night training; in addition, the target also integrates a voice module, and multiple targets are linked to each other to generate irregular sound signals, which helps to train the shooter's ability to hear and identify positions.

[0011] Embodiment 2: Based on Embodiment 1, the present invention provides a mobile voice target system based on multi-dimensional perception technology, further comprising: Score calculation module: process the training data obtained during training; obtain the training data during training and determine the geometric partition data of the target board; determine the position coordinates of the hit point and calculate the distance from the coordinates to the center of the target through the formula, and compare the calculated distance with the radius range of each ring area to determine the number of rings; First, calculate the results of the non-auditory positioning training. Specifically, since the target ring is designed like a circular runway, the target surface is divided into specific areas for calculation. Let the radius of the semicircles at both ends be R, and the distance between the centers of the circles be L, that is, the length of the middle rectangular part. Then the length of the middle rectangle is L, and the width of the rectangle is 2R; let the number of rings be N, and it is divided into N concentric rings from the center of the bull's eye. Then the width of each ring is ; The score processing process for non-auditory position identification training is as follows: Set the bull's eye at the origin , the middle rectangle extends along the x-axis, where x ranges from ; then the range of y is The center origins of the radii at both ends are located at and ; Partition determination ring number, for the middle rectangular area ; Determination of vertical distance: As a benchmark, the boundary of the kth ring is: ; ; Then the expression for calculating the number of rings in the middle rectangular area is: ; For semicircular areas ; Calculate shooting point The distance d to the center of the nearest semicircle is calculated using the Euclidean distance formula: ; The sign is the same as the sign of x; then the expression for calculating the number of rings in the semicircular area is: ; Since the rings of the standard British-sized human-shaped target are designed like a circular runway, the difficulty of hitting the rings in different areas is also different. Therefore, the present invention provides a score weight correction, which specifically quantifies the score difficulty through the weight coefficient; and operates it by introducing the area normalization coefficient; assuming that the area of ​​the kth ring is ; Then the area of ​​the kth ring can be expressed as , where k∈(k1, k2); the score of the kth ring The available formula is: ;in, It is the preset reference value of the corresponding number of rings; For the calculation of scores for multiple shots, the determined scores can be added up by the above method to obtain the total score; that is, if there are n shots, the scores of each shot are recorded as Total score: ;in, Indicates the single score of the nth shooting; Based on the calculation of the non-audio positioning training results, the present invention also provides a method for calculating the score of the audio positioning training, which is as follows: according to the training content, the training data is obtained, and the training data includes: a time factor T, the time factor refers to the time from the training sound to the target being hit, wherein the shorter the time, the higher the score; a ring number factor R, the ring number factor refers to the hit ring number score in the non-audio positioning training, wherein the higher the ring number score, the higher the score; a distance factor D, the distance factor is the straight-line distance between the shooter and the target, wherein the longer the straight-line distance, the greater the shooting difficulty; the number of targets N, which is the number of targets set, wherein the more targets, the greater the interference, the greater the shooting difficulty, and the more significant the score attenuation; Substitute the obtained sound position identification training data into the formula: ; Calculate its score; where T is the time factor, R is the ring number factor, D is the distance factor, and N is the number of targets; The basic score part reflects the shooting accuracy and difficulty. The farther the distance and the higher the number of rings, the higher the basic score. is the attenuation coefficient of the number of targets; the natural logarithm function is used to balance the interference of multiple targets. When the number of targets increases, the score attenuation slows down, which is consistent with the nonlinear characteristics of human perception; To normalize the constant, adjust the score to a readable range, usually The value of is set to 1000; Sound location analysis module: obtain a large amount of historical sound location training data, analyze and process a large amount of historical results to establish a training results analysis model, and optimize the training parameters based on the training results analysis model to improve the training results; Based on the training results of listening to sounds and identifying positions, the present invention establishes a performance analysis module based on a deep learning model to scientifically guide trainees to improve their training results. The specific operation steps are to process a large number of dynamic training results and fill in the missing data with the mean value, specifically through the formula: ;in, The table shows a training performance data set, including: time factor T, number of rings factor R, distance factor D, number of targets N, For the data that needs to be filled, is the jth non-missing data in the training score data set, is the total number of non-missing data in the training performance data set, calculated by this formula It is the data used to fill in missing values; Median filling, using the formula: ; median function means to find the median; if n is an odd number, then , ,arrive The median is the value in the middle after these data are arranged in ascending order; if n is an even number, the average of the two middle numbers is taken as the median; where n represents the total number of data, is the nth data in the data; The outliers are processed using the box plot method. First, the quartiles are calculated, and Q1 is calculated. Q1 is the 25th percentile, which means the data value in the top 25% position after sorting the data from small to large. The calculation method is that if the number of data is n, when n×0.25 is not an integer, the position index is rounded up, and the data at this position is Q1. Calculate Q3, which is the 75th percentile. Similarly, after sorting the data from small to large, the data value in the top 75% position; when n×0.75 is not an integer, round up to get the position index, and the data at that position is Q3; Further, the interquartile range is calculated; using the formula: ; IQR reflects the degree of dispersion of the middle 50% of the data; for the constant value range, the upper limit is: ; The lower limit is: ; Any data less than the lower limit or greater than the upper limit is considered an outlier; Data standardization, using Z-score standardization to standardize the data, using the formula: ; Where x is the original data, μ is the mean of the data in the training performance data set, and σ is the standard deviation of the data in the training performance data set; Through this formula, the original data x is converted into standardized data x' centered on the mean and scaled by the standard deviation; First calculate the mean μ and standard deviation σ of the data, and then for each data in the data column, calculate according to the above formula to obtain the corresponding standardized data; All training performance data sets are standardized according to the above method, and all standardized feature columns are normalized using the formula: ; Here Xmin is the minimum value in the feature column data, and Xmax is the maximum value; through this formula, the standardized data X is mapped to the range of 0 to 1; Establish a performance analysis model and use the linear regression model to establish a performance analysis model. The formula is: ;in, is the model intercept, , , , is the proportional coefficient; T is the time factor, R is the ring number factor, D is the distance factor, and N is the number of targets. is the error term, which represents the part that cannot be explained by the model; Using the least squares method, the formula is: ; Here D is a matrix composed of the eigenvalues ​​T, R, D, and N in the training performance data set, and each row represents the eigenvalue of a training. is the transposed matrix of D, yes The inverse matrix of are the coefficients of the model; Sound identification training guidance module: Use linear regression model for training guidance; For example, in a linear regression model, the larger the absolute value of the proportionality coefficient is, the greater the influence of the corresponding characteristic factor on the training performance is. When other characteristic factors remain unchanged, the influence of the characteristic factor on the performance is analyzed to guide the training performance; For example, If it is equal to -15, then with other characteristic factors unchanged, the score will decrease by 15 points for every 1 meter increase in distance; if the current training results do not match the expected target results, you can , , , The specific value of is used to adjust the non-auditory position identification; The technical solution of this embodiment is as follows: first, based on the training calculation method of non-sound positioning results, the sound positioning training results are obtained; then, the sound positioning training results are standardized and normalized, and the sound positioning training results are deeply analyzed with the help of a linear regression model; through this series of operations, guidance can be provided to trainees from a scientific level to help them improve their training results.

[0012] Embodiment 3: Figure 2 As shown, based on Example 1 and Example 2, the present invention also provides a mobile voice target method based on multi-dimensional perception technology, comprising the following steps: S1: Use laser diode as the core element for emitting laser to simulate the shooting process; drive circuit: provide appropriate drive current for laser diode; power module: provide stable power for laser emission; microcontroller: used for burst mode in shooting simulation according to preset program logic; switch: used to control the opening and closing of the shooting end; S2: Uses a photoelectric sensor array to receive invisible infrared lasers and convert optical signals into electrical signals; signal processing circuit: includes amplifiers, filters, and analog-to-digital converters ADC to remove noise interference in electrical signals and improve signal quality and reliability; microcontroller STM32: used to determine the hit of the laser beam; breathing indicator light: used to intuitively display the shooting results; voice module: used to simulate training sounds and voice target reporting; communication module: used to transmit training data; power supply: uses a 3.7-volt 18650 battery to power the entire system; switch: used to control the opening and closing of the target; S3: When conducting simulated shooting and target shooting training, first arrange the targets in advance according to the training content; if it is non-sound identification training, the appropriate target shooting distance and number of targets can be set according to the training requirements; if it is sound identification training, it is necessary to set several targets at different distances and directions according to the training requirements, and these targets are linked and adjusted; after linkage adjustment, different targets will alternately emit training sounds according to irregular instructions, which is convenient for shooters to practice the ability of sound identification; after completing the target arrangement, after turning the switch on the back of the target, the target indicator light flashes green for 1S and then goes out, indicating that the humanoid target is successfully started; at the same time, press the switch integrated in the shooting end to start the shooting end equipment and officially start the simulation training; S4: Processing the training data obtained during training; obtaining the training data during training, determining the geometric partition data of the target plate surface; determining the position coordinates of the hit point, calculating the distance from the coordinates to the center of the target through a formula, and comparing the calculated distance with the radius range of each ring area to determine the number of rings; Since the target ring is designed like a circular runway, the target surface is divided into specific areas for calculation. Let the radius of the semicircles at both ends be R, and the distance between the centers of the circles be L, that is, the length of the middle rectangular part. Then the length of the middle rectangle is L, and the width of the rectangle is 2R. Let the number of rings be N, and it is divided into N concentric rings from the center of the target. Then the width of each ring is ; Set the bullseye at the origin , the middle rectangle extends along the y-axis, where the range of y is ; then the range of x is The center origins of the radii at both ends are located at and ; Partition determination ring number, for the middle rectangular area Determination of vertical distance: As a benchmark, the boundary of the kth ring is: ; ; Then the expression for calculating the number of rings is: ; For semicircular areas ; Calculate shooting point The distance d to the center of the nearest semicircle is calculated using the Euclidean distance formula: ; where the sign is opposite to the sign of x; then the expression for calculating the number of rings is: ; Introduce the area normalization coefficient to operate it; suppose the area of ​​the kth ring is ; Then the area of ​​the kth ring can be expressed as ; Score of the kth round The available formula is: ;in, It is the preset reference value of the corresponding number of rings; For the calculation of scores for multiple shots, the determined scores can be added up by the above method to obtain the total score; that is, if there are n shots, the scores of each shot are recorded as Total score: ;in, Indicates the single score of the nth shooting; Based on the non-auditory positioning training results, the audio positioning training results are calculated as follows: Substitute the obtained sound position identification training data into the formula: ; Calculate its score; where T is the time factor, R is the ring number factor, D is the distance factor, and N is the number of targets; The basic score part reflects the shooting accuracy and difficulty. The farther the distance and the higher the number of rings, the higher the basic score. is the attenuation coefficient of the number of targets; the natural logarithm function is used to balance the interference of multiple targets. When the number of targets increases, the score attenuation slows down, which is consistent with the nonlinear characteristics of human perception; To normalize the constant, adjust the score to a readable range, usually The value of is set to 1000; S5: Obtain a large amount of historical sound location training data, analyze and process a large amount of historical results to establish a training results analysis model, and optimize the training parameters based on the training results analysis model to improve the training results; Based on the training results of listening to sounds and identifying positions, the present invention establishes a performance analysis module based on a deep learning model to scientifically guide trainees to improve their training results. The specific operation steps are to process a large number of dynamic training results and fill in the missing data with the mean value, specifically through the formula:

[0013] ;in, The table shows a training performance data set, including: time factor T, number of rings factor R, distance factor D, number of targets N, For the data that needs to be filled, is the jth non-missing data in the training score data set, is the total number of non-missing data in the training performance data set, calculated by this formula It is the data used to fill in missing values; Median filling, using the formula: ; median function means to find the median; if n is an odd number, then , ,arrive The median is the value in the middle after these data are arranged in ascending order; if n is an even number, the average of the two middle numbers is taken as the median; where n represents the total number of data, is the nth data in the data; The outliers are processed using the box plot method to calculate Q1, which is the 25th percentile, meaning the data value in the top 25% after sorting the data from small to large. The calculation method is that if the number of data is n, when n×0.25 is not an integer, the position index is rounded up, and the data at that position is Q1. Calculate Q3, which is the 75th percentile. Similarly, after sorting the data from small to large, the data value in the top 75% position; when n×0.75 is not an integer, round up to get the position index, and the data at that position is Q3; Calculate the interquartile range; use the formula: ; IQR reflects the degree of dispersion of the middle 50% of the data; for the constant value range, the upper limit is: ; The lower limit is: ; Any data less than the lower limit or greater than the upper limit is considered an outlier; The data was standardized using the Z-score standardization formula: ; Where x is the original data, μ is the mean of the data in the training performance data set, and σ is the standard deviation of the data in the training performance data set; Through this formula, the original data x is converted into standardized data x' centered on the mean and scaled by the standard deviation; First calculate the mean μ and standard deviation σ of the data, and then for each data in the data column, calculate according to the above formula to obtain the corresponding standardized data; All training performance data sets are standardized according to the above method, and all standardized feature columns are normalized using the formula: ; Here Xmin is the minimum value in the feature column data, and Xmax is the maximum value; through this formula, the standardized data X is mapped to the range of 0 to 1; The linear regression model is used to establish the performance analysis model, and its formula is: ;in, is the model intercept, , , , is the proportional coefficient; T is the time factor, R is the ring number factor, D is the distance factor, and N is the number of targets. is the error term, which represents the part that cannot be explained by the model; Using the least squares method, the formula is: ; Here D is a matrix composed of the eigenvalues ​​T, R, D, and N in the training performance data set, and each row represents the eigenvalue of a training. is the transposed matrix of D, yes The inverse matrix of are the coefficients of the model; S6: Use a linear regression model for training guidance; illustratively, in a linear regression model, the larger the absolute value of the proportionality coefficient is, the greater the influence of the corresponding characteristic factor on the training performance is. When other characteristic factors remain unchanged, the influence of the characteristic factor on the performance is analyzed to guide the training performance; illustratively, for example If it is equal to -15, then with other characteristic factors unchanged, the score will decrease by 15 points for every 1 meter increase in distance; if the current training results do not match the expected target results, you can , , , The specific value of is used to adjust the non-auditory position identification; The above shows and describes the basic principles, main features and advantages of the present invention. It should be understood by those skilled in the art that the present invention is not limited to the above embodiments. The above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. A mobile voice target system based on multi-dimensional perception technology, characterized in that: include: Shooting end module: uses invisible infrared laser to simulate the shooting process; Target module: receives the invisible infrared laser emitted by the shooting module and forms physical interaction, while simulating training sounds; Shooting training module: Use the shooting end and the target end to build a dual training mode, namely, the sound identification training and the non-sound identification training mode; Score calculation module: Process the data obtained from shooting training, determine the geometric partition of the target board, determine the number of rings, and introduce the area normalization coefficient to calculate the score; The sound identification training results are obtained based on the non-sound identification training results, and the total score is calculated using the accumulation method; Sound identification performance analysis module: processes the sound identification performance and establishes a training performance analysis model; Sound identification training guidance module: Use linear regression model for training guidance.

2. The mobile voice target system based on multi-dimensional perception technology according to claim 1, characterized in that: The training of positioning by sound is specifically as follows: according to the training requirements, targets are set at different distances and orientations, and the targets are adjusted in linkage; after the linkage adjustment, different targets will emit training sounds alternately according to irregular instructions, so as to facilitate shooters to practice positioning by sound.

3. The mobile voice target system based on multi-dimensional perception technology according to claim 1, characterized in that: The non-auditory positioning training is specifically as follows: a suitable shooting distance and number of targets can be set according to training requirements.

4. The mobile voice target system based on multi-dimensional perception technology according to claim 1, characterized in that: The specific process of determining the geometric partition of the target plate surface is as follows: the target surface is partitioned and calculated. Assuming the radius of the semicircles at both ends is R and the distance between the centers is L, the length of the middle rectangle is L and the width of the rectangle is 2R; Assuming the number of rings is N, it is divided into N concentric rings from the center of the target plate outward, and the width of each ring is ; The bull's eye is set at the origin (0,0) and the middle rectangle extends along the x-axis, where the range of x is ; The range of y is The center origins of the radii at both ends are located at and .

5. The mobile voice target system based on multi-dimensional perception technology according to claim 4 is characterized in that: The specific process of determining the number of rings is as follows: for the middle rectangular area ; Determination of vertical distance: As a benchmark, the expression for calculating the number of rings in the middle rectangular area is: ； For semicircular area ; Calculate the shooting point using the Euclidean distance formula The distance d to the nearest semicircle center; where the sign is the same as the sign of x; the expression for calculating the number of rings in the semicircle area is: .

6. The mobile voice target system based on multi-dimensional perception technology according to claim 1, characterized in that: The specific process of introducing the area normalization coefficient to calculate the score is as follows: Assume the area of ​​the kth ring is ; Then the area of ​​the kth ring can be expressed as ,in, k∈(k1,k2);kth ring score The available formula is: ;in, It is the preset reference value for the corresponding number of rings.

7. The mobile voice target system based on multi-dimensional perception technology according to claim 1, characterized in that: The total score is calculated by the accumulation method as follows: for n shots, each single score is recorded as ; The total score is: ;in, Indicates the single score of the nth shooting.

8. The mobile voice target system based on multi-dimensional perception technology according to claim 1, characterized in that: The specific process of obtaining the sound location training results based on the non-sound location training results is: substituting the sound location training data into the formula: ; Calculate its score; where T is the time factor, R is the ring number factor, D is the distance factor, and N is the number of targets; The basic score part reflects the shooting accuracy and difficulty. The farther the distance and the higher the number of rings, the higher the basic score. is the attenuation coefficient of the number of targets; a natural logarithmic function is used to balance multi-target interference, and the score attenuation slows down when the number of targets increases; is a normalization constant, and its value is set to 1000.

9. The mobile voice target system based on multi-dimensional perception technology according to claim 1, characterized in that: The use of the linear regression model for training guidance is specifically as follows: in the linear regression model, the larger the absolute value of the proportionality coefficient is, the greater the influence of the corresponding characteristic factor on the training performance is, and when other characteristic factors remain unchanged, Analyze the impact of characteristic factors on performance to guide training results.

10. A mobile speech target method based on multi-dimensional perception technology, characterized in that: include: S1: Use invisible infrared laser to simulate the shooting process; S2: Receives the invisible infrared laser emitted by the shooting end module and forms physical interaction, while simulating training sounds; S3: Use the shooting end and the target end to build a dual training mode, namely, the sound identification training and the non-sound identification training mode; S4: Processing the data obtained from the shooting training, determining the geometric partitions of the target board, determining the number of rings, and introducing the area normalization coefficient to calculate the score; The sound identification training results are obtained based on the non-sound identification training results, and the total score is calculated using the accumulation method; S5: Process the sound location identification results and establish a training results analysis model; S6: Use linear regression model for training guidance.

Images