AOA device performance evaluation method and system

Abstract

The present application relates to radio direction finding equipment test and performance evaluation technology, disclose a kind of AOA equipment performance evaluation method, system and computer readable storage medium, solve the problem of repeated reporting and sporadic far point leading to evaluation deviation.The method associates AOA observation with RTK true value trajectory according to time window and carries out five-state adjudication, constructs the logic second excluding RTK long cavity, at most selects a representative observation per second;Respectively establish the assessable, observation and effective logic second set according to distance grade, to sample support quality multiplicative gate cover, hit, long tail angle error and the weighted sum of flow quality, form grade quality coefficient;The farthest actual horizontal distance of best candidate of representative observation in grade is multiplied by grade quality coefficient, and the maximum value is obtained by crossing grade, and quality conversion action distance is obtained.The scheme can reduce the direct influence of message density and sporadic far point on evaluation.

Description

Technical Field

[0001] This invention relates to the fields of radio direction finding equipment testing, low-altitude target surveillance, spatiotemporal data processing, and equipment performance evaluation, and particularly to a method and system for evaluating the performance of AOA (Optical Direction Finding) equipment. Background Technology AOA (Aspect-Oriented Array) devices estimate the azimuth angle of a target relative to a receiving base station by receiving the target's radio signal. When selecting, conducting on-site acceptance tests, or making horizontal comparisons of AOA devices, it is usually necessary to use a high-precision positioning trajectory as the true value, compare the deviation between the azimuth angle output by the AOA device and the true azimuth angle calculated from the position of the receiving base station and the true position of the target, and evaluate the direction-finding accuracy and effective range of the device accordingly.

[0002] In actual testing, the reporting cycle and frequency of different AOA devices may vary, and multiple observations with similar content may appear within the same second. If we directly count each original message, duplicate samples from high-frequency reporting devices will be given a larger counting weight. At the same time, a single or small number of long-distance hits can easily inflate the performance result expressed by the maximum hit distance, but it cannot reflect the sample support, coverage, hit rate, long tail angle error, and flow interruption within the corresponding distance range.

[0003] Furthermore, the target identifier on the AOA device side may be an internal tracking identifier and is not necessarily equivalent to the RTK target identity; directly binding it by identifier will miss candidates that could be identified through time and angle relationships. RTK trajectories themselves may also experience long-term interruptions. If this interruption is directly included in the test duration, it is easy to classify unevaluable periods caused by missing ground truth data as invalid periods for the AOA device under test. For situations such as multiple target time overlaps, missing base station locations, and no RTK points within the time window, it is also difficult to distinguish different sources of anomalies if only a single "hit" or "miss" result is used.

[0004] Therefore, there is a need for an AOA equipment performance evaluation scheme that can control the direct impact of raw message density on statistical weights, eliminate long-term RTK holes, distinguish multiple candidate referee states, and combine the data support quality of different distance levels with the actual effective hit distance. Summary of the Invention

[0005] Therefore, it is necessary to provide an AOA equipment performance evaluation method and system to solve at least one of the above-mentioned technical problems.

[0006] To achieve the above objectives, an AOA equipment performance evaluation method and system includes the following steps: The technical problem to be solved by this invention is: how to use RTK truth values ​​to form an AOA device performance evaluation result that is time-weighted, has range data quality constraints, and allows intermediate quantities to be recalculated, under the circumstances that the AOA observation reporting frequency is inconsistent, the same second may be reported repeatedly, the AOA device side identifier and RTK target identity cannot be strongly bound in advance, and long-distance ranges may only have sparse and occasional hits.

[0007] To address the aforementioned technical problems, this invention provides a method for evaluating the performance of an AOA (Optical Angle of View) device. The method uses the AOA azimuth observation sequence with observation time output by the AOA device under test, the RTK ground truth trajectory with time and target identity, the receiving base station location of the AOA device, and evaluation parameters as physical measurement inputs.

[0008] The AOA azimuth observation sequence and RTK ground truth trajectory are cleaned for validity and sorted by time. Based on the long time intervals in the RTK ground truth trajectory, valid segments are divided. The first and last seconds of the second-level index within each valid segment are concatenated into a logical second sequence, so that long-term RTK gaps between valid segments do not enter the evaluable duration.

[0009] For each cleaned AOA azimuth observation, RTK ground truth points with time differences not exceeding the time matching window are selected as candidates from the RTK ground truth trajectory. The true azimuth from the receiving base station to the candidate, the minimum circular angular error between the AOA azimuth and the true azimuth, and the actual horizontal distance from the receiving base station to the candidate are calculated. Candidates are sorted in ascending order of minimum circular angular error and then in ascending order of time difference. Five judgment results are formed based on the availability of the receiving base station location, the existence of candidates, angular error conditions, and ambiguities between different RTK targets.

[0010] The AOA azimuth observations after the refereeing are mapped to logical seconds. When multiple observations exist for the same logical second, at most one observation is selected to represent that second, in the order of time difference priority, followed by minimum angular error at the circumference, and then state priority. The candidate layer sorting and the second-representative observation sorting serve candidate confirmation and second-level unique weighting, respectively, and their comparison key orders are different.

[0011] For multiple distance ranges, deduplication is performed on logical seconds to construct an evaluable logical second set, an observed logical second set, and a valid logical second set. The evaluable logical second set is archived according to the maximum horizontal distance of all cleaned RTK ground truth points relative to the receiving base station within the same logical second; the observed logical second set is archived according to the best candidate actual horizontal distance of the second-represented observation; the valid logical second set uses the same distance source as the observed logical second set and is formed by the second-represented observations whose adjudication results are valid.

[0012] For each distance range, sample support quality, coverage quality, hit quality, long tail angle error quality, and flow interruption quality are generated. The sample support quality is weighted and multiplicatively gated with respect to the other four quality parameters to obtain the range quality coefficient. Within the data domain where the optimal candidate actual horizontal distance maximum can be determined from observations representing effective state seconds within that distance range, this maximum value is taken as the range's furthest effective hit distance. The furthest effective hit distance is multiplied by the range quality coefficient to obtain the range quality-converted effective distance. The maximum value of the quality-converted effective distance for each distance range generated from effective data is then taken to obtain the quality-converted effective distance evaluation result.

[0013] This invention also provides an AOA (Automatic Optical Array) equipment performance evaluation system, comprising, in sequence, a data acquisition and cleaning module, a logical second construction module, a candidate physical quantity calculation module, an observation status judging module, a second-represented observation selection module, a distance range three-set construction module, a range quality calculation module, and a quality conversion output module. Each module jointly executes the data processing chain described above.

[0014] The present invention also provides a computer-readable storage medium storing a computer program or instructions thereon, wherein when the computer program or instructions are executed by a processor, the above-described AOA equipment performance evaluation method is implemented, and the quality-converted action distance evaluation result is output.

[0015] The beneficial effects of this invention are: (1) Using logical seconds as the evaluation unit and keeping at most one second-representative observation for each logical second can reduce the direct amplification of the sample counting weight by the original message density; this mechanism does not rely on completely eliminating the difference in candidate distribution for the same second.

[0016] (2) Dividing effective segments based on the long time interval of RTK and splicing logical seconds can prevent long-term holes in RTK from entering the evaluable time, while retaining the impact of AOA equipment disconnection on the evaluation results within the effective segments.

[0017] (3) Separating the five states of no receiving base station location, no RTK matching, ambiguity, miss and valid can provide statistical basis for different sources of anomalies.

[0018] (4) Sample support, coverage, hit, long tail angle error and flow interruption quality are generated for the distance range respectively, and multiplicative gating is implemented by the sample support quality, which can reduce the direct contribution of insufficient sample distance range to the evaluation results.

[0019] (5) Using the effective seconds within the range to represent the maximum actual horizontal distance of the best candidate for observation instead of the upper limit of the distance range for conversion can link the output results with the physical measurement samples that are judged to be valid.

[0020] (6) Simultaneously output the cardinality, five quality items, gear quality coefficient and gear quality conversion distance of the three types of logical second sets, so as to facilitate the verification of the contribution of each distance gear to the final result. Attached Figure Description

[0021] Figure 1 A flowchart of the AOA equipment performance evaluation method; Figure 2 Flowchart for selecting candidate referees and second representatives; Figure 3 Flowchart for distance-based quality conversion; Figure 4 This is a block diagram of the AOA equipment performance evaluation system.

[0022] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0023] The technical method of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0024] Furthermore, the accompanying drawings are merely illustrative of the invention and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and therefore repeated descriptions of them will be omitted. Some block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities can be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor methods and / or microcontroller methods.

[0025] It should be understood that although the terms "first," "second," etc., may be used herein to describe various units, these units should not be limited by these terms. These terms are used merely to distinguish one unit from another. For example, without departing from the scope of the exemplary embodiments, a first unit may be referred to as a second unit, and similarly, a second unit may be referred to as a first unit. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0026] To achieve the above objectives, please refer to Figures 1 to 3 A method and system for evaluating the performance of AOA equipment includes the following steps: All specific values ​​involved in this embodiment are exemplary parameters used to clearly illustrate the technical operation process and are not the only limitation of the present invention.

[0027] like Figure 1 As shown, this implementation method sequentially executes S1 to S8: S1 Acquires and cleans the physical measurement input; S2 Divides the effective RTK segments and splices logical seconds; S3 Screens RTK candidates and calculates the true azimuth, minimum circumferential angular error, and horizontal distance; S4 Sorts the candidates and forms a five-state judge; S5 Maps logical seconds and selects seconds to represent observations; S6 Constructs an evaluable set of logical seconds, an observed logical seconds set, and an effective logical seconds set according to distance ranges; S7 Calculates five quality parameters and forms the range quality coefficient by sample-supported quality multiplicative gating; S8 Forms the range quality-converted effective distance based on the actual horizontal distance of the best candidate representing the effective seconds, and outputs the cross-range quality-converted effective distance evaluation result.

[0028] The above steps constitute a continuous data supply relationship. The cleaned RTK truth trajectory is used to construct logical seconds and generate candidates; the candidate physical quantities and the judging results are used to select the second representative observation; the RTK truth points and the second representative observations are used to construct three types of logical second sets; the three types of logical second sets and the second representative observations are used to form the gear quality and the gear's farthest effective hit distance; the gear quality and the gear's farthest effective hit distance together form the final evaluation result.

[0029] Input data and evaluation parameters; a single evaluation receives the following data: (1) The AOA azimuth observation sequence output by the AOA device under test. Each AOA azimuth observation includes at least the observation time and azimuth, and may also include the device-side target identifier, device type, channel and bandwidth. The device-side target identifier is used to retain the semantics reported by the device and is not pre-bound to the RTK target identity.

[0030] (2) One or more RTK ground truth trajectories. Each RTK ground truth point includes at least time, longitude, latitude, altitude, and target identity. Target identity is used to determine whether the best candidate and the second-best candidate belong to different RTK targets.

[0031] (3) The location of the receiving base station of the AOA device, including the longitude and latitude of the receiving base station. The location of the receiving base station serves as the physical origin for calculating the true azimuth and actual horizontal distance.

[0032] (4) Evaluation parameters, including time matching window W, hit angle error threshold H, ambiguity difference threshold A, RTK segmentation threshold, distance range boundary, sample support threshold, coverage threshold, hit threshold, P90 quality threshold, flow interruption threshold and quality weight.

[0033] In one parameter configuration example, the time matching window W is 1500ms, the hit angle error threshold H is 5°, and the ambiguity difference threshold A is 8°; the sample support threshold E_0 is 5 seconds, the coverage threshold C_0 is 0.50, the hit threshold H_0 is 0.50, the P90 quality full value point P_0 is 15°, the P90 quality bad value point P_bad is 35°, and the disconnection threshold M_0 is 20 seconds; the weights w_c, w_h, w_p, and w_l for coverage quality, hit quality, long tail angle error quality, and disconnection quality are 0.35, 0.35, 0.20, and 0.10, respectively. The above values ​​are configurable parameters for this embodiment.

[0034] Acquire and clean physical measurement inputs Obtain the AOA azimuth observation sequence, RTK ground truth trajectory, receiving base station location, and evaluation parameters. Perform validity cleaning on the AOA azimuth observations and RTK ground truth points respectively, and sort them in ascending order according to their respective validity time.

[0035] For AOA azimuth observations, only observations with non-empty and finite azimuth angles and valid observation times are retained. When multiple time fields exist, the observation times used for sorting can be determined according to the preset time field priority.

[0036] For RTK ground truth points, we retain ground truth points that have non-empty and finite latitude and longitude, whose latitude and longitude are within the global legal range and the preset business area, whose latitude and longitude are not simultaneously approximately zero, and which have altitude and valid time. In a deployment instance, the preset business area has a longitude range of 70° to 140° and a latitude range of 3° to 60°. Although altitude is not involved in the subsequent calculation of the two-dimensional true azimuth and Haversine horizontal distance, it serves as a condition for the validity of RTK ground truth points in this implementation.

[0037] The cleaned AOA azimuth observations retain the original observation time and azimuth; the cleaned RTK ground truth points retain the time, location and target identity, thus providing a data foundation for subsequent time window candidate association.

[0038] S2: Divide the RTK valid segments and concatenate them into logical seconds. Let the time matching window be W. The effective segment boundaries are defined as the locations in the RTK ground truth trajectory where the time interval between adjacent RTK ground truth points is greater than the segmentation threshold G. Where: G = max(3000ms, 2W); For the q-th valid segment, let its start time be t. q,start The end time is t q,end Then the number of seconds T of the valid segment. q for: T_q=max(1,ceil((t_q,end-t_q,start) / 1000)); Based on the time sequence of each valid segment, a second-level index is created within each segment. The first and last second-level indices of each segment are then concatenated to form a logical second sequence. The total number of test seconds T is: T = Σ_qT_q; The time of any valid segment is mapped to logical seconds using the second offset within that segment and the segment's starting index in the logical second sequence. RTK long-duration holes exceeding G between adjacent valid segments do not form logical seconds and therefore are not included in the evaluable duration. Valid segments are concatenated end-to-end on their logical second indices for subsequent deduplication and statistical analysis of observations using a unified logical second key.

[0039] S3: Screen RTK candidates and calculate candidate physical quantities For the i-th AOA azimuth observation with observation time t_i and azimuth angle α_i, the RTK ground truth point with time τ_j that satisfies the following formula is determined as an RTK candidate: |t_i-τ_j|≤W; Let the longitude and latitude of the receiving base station be λ_b and φ_b, respectively, and the longitude and latitude of the j-th RTK candidate be λ_j and φ_j, respectively. All longitudes and latitudes are converted to radians during trigonometric function operations. Let: Δλ=λ_j-λ_b, Δφ=φ_j-φ_b; The true azimuth angle β from the receiving base station to the j-th RTK candidate ij for: β_ij=norm_[0°,360°){(180° / π)×atan2(sinΔλ×cosφ_j, cosφ_b×sinφ_j-sinφ_b×cosφ_j×cosΔλ)}; The atan2 function outputs radian values, which are first multiplied by 180° / π to convert to degrees, and then normalized to [0°, 360°). The AOA azimuth α_i and the true azimuth β... ij All are expressed in degrees. Let: d_ij=|norm(α_i)-norm(β_ij)|, Then the minimum angular error of the circle e ij for: e_ij=min(d_ij, 360°-d_ij); This minimum angular error avoids large, non-physical angular differences at the boundaries between 0° and 360°. For example, when the AOA azimuth is 359° and the true azimuth is 1°, the minimum angular error is 2°.

[0040] The actual horizontal distance r from the receiving base station to the j-th RTK candidate Hasersine ij Determined according to the following formula: a_ij=sin²(Δφ / 2)+cosφ_b×cosφ_j×sin²(Δλ / 2), r_ij=2R×atan2(√a_ij,√(1-a_ij)), Where R is the Earth's radius. In one embodiment, R is taken as 6,371,000 m.

[0041] Therefore, a candidate set is formed for each AOA azimuth observation. Each candidate includes at least the time difference between the candidate and the AOA azimuth observation, the minimum circumferential angular error, the actual horizontal distance, and the RTK target identity.

[0042] S4: Candidate sorting and forming a five-state judge. The RTK candidates corresponding to the same AOA azimuth observation are sorted in ascending order of minimum circumferential angular error and then in ascending order of time difference to determine the best candidate and the next best candidates. The candidate judges then distinguish the following states in sequence: (1) When the location of the receiving base station is missing or invalid, and therefore the true azimuth angle cannot be calculated, the judgment result is NO_DEVICE_LOCATION, which indicates that there is no receiving base station location.

[0043] (2) When the location of the receiving base station is valid but there is no RTK candidate within the time matching window, the judgment result is NO_RTK_MATCH, which is the state of no RTK matching.

[0044] (3) When there is a second-best candidate, the minimum circumferential angle error e_i2 of the second-best candidate is not greater than the hit angle error threshold H, the absolute value of the difference between the minimum circumferential angle errors of the second-best candidate and the best candidate |e_i2-e_i1| is not greater than the ambiguity difference threshold A, and the best candidate and the second-best candidate come from different RTK targets, the referee result is an ambiguous state AMBIGUOUS. The ambiguity judgment only checks the sorted best candidate and the second-best candidate.

[0045] (4) When the ambiguity condition is not met and the minimum circumferential angle error e_i1 of the best candidate is greater than H, the referee result is a MISS state.

[0046] (5) When the aforementioned abnormal conditions are not met and the minimum circumferential angular error e_i1 of the best candidate is not greater than H, the judgment result is the valid state VALID.

[0047] The aforementioned rulings caused the receiving base station location to become unavailable, there to be no true value within the time window, different target candidates to be difficult to uniquely identify, and angle miss and valid hit to enter different states.

[0048] S5: Map logical seconds and select seconds to represent observations The time used for logical second mapping is determined from the AOA azimuth observation with the adjudication result. When an optimal candidate exists for the observation, the time of the optimal candidate RTK ground truth point is used; when no RTK candidate exists or a valid receiving base station location is lacking, the observation time of the AOA azimuth observation itself is used. A corresponding logical second is formed only if the time used for mapping falls within a valid RTK segment.

[0049] When multiple AOA azimuth observations exist within the same logical second, one second is selected to represent the observation in ascending order of time difference, ascending order of minimum circumferential angle error, and state priority. State priorities, from highest to lowest, are VALID, MISS, AMBIGUOUS, NO_RTK_MATCH, and NO_DEVICE_LOCATION. Empty time differences are ranked after non-empty time differences, and empty minimum circumferential angle errors are ranked after non-empty minimum circumferential angle errors. Therefore, at most one second is retained to represent the observation for each logical second.

[0050] like Figure 2 As shown, AOA azimuth observations, RTK ground truth trajectories, and receiving base station locations are first incorporated into the time window candidate calculation. After calculation based on the true azimuth, minimum circumferential angular error, and actual horizontal distance, the candidates undergo a first sorting process to form a five-state arbiter. The arbiter results are then mapped to logical seconds, and a second sorting is performed based on time difference, minimum circumferential angular error, and state priority, ultimately resulting in at most one second-representing observation for each logical second. The comparison key order for the two sorting processes is different.

[0051] In a statistical implementation that works in conjunction with second-representation observations, the number of second-representation samples N_s, the number of observation seconds, the number of valid seconds in the VALID state, the number of seconds in the MISS state, the number of seconds in the AMBIGUOUS state, the number of seconds in the NO_RTK_MATCH state, and the number of seconds in the NO_DEVICE_LOCATION state are statistically analyzed based on the concatenated logical second sequence. The longest consecutive valid seconds and the maximum consecutive no valid seconds are also statistically analyzed.

[0052] The second-representation observations with states of VALID, MISS, or AMBIGUOUS and non-empty minimum angular error of the circle constitute the angular error sample set. The N angular error samples are sorted in ascending order, and the samples with indices ceil(0.90N) and ceil(0.95N) are selected as P90 and P95 respectively using the nearest rank method. The mean and median can then be further calculated. For... Calculate Hit@x for x∈{3°,5°,10°,20°}. The denominator of Hit@x is the number of all non-empty angle error samples with the minimum angular error of the circle mentioned above, and the numerator is the number of samples whose judgment result is VALID and whose minimum angular error of the circle is not greater than x, that is: Hit@x=#{s|e_s≠null, status_s=VALID, e_s≤x} / #{s|e_s≠null}; When the denominator is zero, Hit@x is determined to be zero. Let N be the number of AOA azimuth observations cleaned by S1. raw The compression ratio is N. s / N raw When N raw When the value is zero, the compression ratio is determined to be zero.

[0053] S6: Construct three sets of logical seconds based on distance ranges. This implementation method divides the actual horizontal distance into the following six semi-open distance intervals: (1) [0,300) meters; (2) [300, 500) meters; (3) [500, 800) meters; (4) [800, 1000) meters; (5) [1000, 1500) meters; (6) [1500, +∞) meters, where +∞ indicates that the distance range does not have a finite upper bound.

[0054] For the b-th distance range, construct a set of evaluable logical seconds. Observation logic second set and the set of valid logical seconds .

[0055] For the set of evaluable logical seconds Calculate the actual horizontal distance between all cleaned RTK ground truth points and the receiving base station within the same logical second, take the maximum value within that logical second, and add the logical second in which the maximum value falls into the b-th distance range after deduplication by logical second index. .

[0056] For the set of observed logical seconds Distance levels are assigned by using the best candidate actual horizontal distance represented by seconds. Observations falling into the b-th distance level are deduplicated by their logical second index and then added to the next level. .

[0057] For the set of valid logical seconds , adopt and The same best candidate actual horizontal distance source, in The selected second represents the logical second in which the observed referee result is VALID, and is then deduplicated by logical second index to form the result. .

[0058] make: E_b=| |,O_b=| |,V_b=| |; The original coverage ratio C_b and hit ratio H_b are respectively: C_b=O_b / E_b, H_b=V_b / O_b; When E_b is zero, C_b is set to zero; when O_b is zero, H_b is set to zero. Because... Archived based on the maximum actual horizontal distance of the RTK truth points after all cleaning within the same logical second, and and Based on the best candidate actual horizontal distance archived for observations in seconds, the data sources for the three sets are not entirely the same. In the case of multiple targets or cross-architecture within the same second, the original coverage ratio C_b may be greater than 1, and this original ratio is retained; the subsequent coverage quality is limited to [0,1] by a normalization function.

[0059] S7: Calculate the five quality parameters and generate the grade quality coefficient. For the b-th distance setting, it will be according to The distance source is assigned to the distance range, and the second-representing observations with a judgment result of VALID, MISS, or AMBIGUOUS and a non-empty minimum circumferential angle error form the range angle error sample set. The range P90 value P90_b is determined according to the nearest rank method.

[0060] Within the consecutive evaluable logical seconds of this distance range, the accumulated seconds without VALID status represent the number of consecutive observation seconds; when the logical second indices of adjacent evaluable logical seconds are not consecutive, the accumulation is restarted, thus obtaining the maximum interruption M_b of the range.

[0061] Define clip(x,0,1)=max(0,min(x,1)). The sample support quality Q_sup,b, coverage quality Q_cov,b, and hit quality Q_hit,b are respectively: Q_sup,b=clip(E_b / E_0,0,1), Q_cov,b=clip(C_b / C_0,0,1), Q_hit,b=clip(H_b / H_0,0,1).

[0062] The long tail angle error quality Q_p90,b is: Q_p90,b={0, P90_b does not exist; 1, P90_b≤P_0; 0, P90_b≥P_bad; (P_bad-P90_b) / (P_bad-P_0), P_0 <P90_b<P_bad}; The above segmented approach is used when the configurable parameter satisfies P_bad > P_0. In one parameter validity implementation, when the configured P_bad is not greater than P_0, first set P'_bad = max(20°, P_0 + 1°), and then use P'_bad instead of P_bad to calculate the long tail angle error quality, so as to avoid the denominator being zero or negative.

[0063] The interruption quality Q_loss,b is: Q_loss,b={1, M_b≤M_0; M_0 / M_b, M_b>M_0 and M_0>0; 0, M_b>M_0 and M_0≤0}.

[0064] Set non-negative weights w_c, w_h, w_p, and w_l for coverage quality, hit quality, long tail error quality, and disconnection quality, satisfying: w_c+w_h+w_p+w_l=1; The quality coefficient Q_b for the b-th distance gear is: Q_b=Q_sup,b×(w_cQ_cov,b+w_hQ_hit,b+w_pQ_p90,b+w_lQ_loss,b).

[0065] All five quality parameters and Q_b are in the range [0,1]. Among them, the sample support quality Q_sup,b is located outside the parentheses. It is a weighted sum of coverage quality, hit quality, long tail angle error quality, and disconnection quality, and is multiplicatively gated, rather than being added to the other four as a fifth parameter.

[0066] S8: Calculates the effective distance based on the gear's mass and outputs across gears. Within the data domain where the optimal candidate actual horizontal distance can be determined by the second-representation observations with a VALID result within the b-th distance range, the maximum value D_max,b is obtained, and D_max,b is determined as the farthest effective hit distance in the b-th distance range. D_max,b originates from the optimal candidate actual horizontal distance of the VALID second-representation observations within the range; it is not the upper bound of that distance range, nor is it... The maximum distance between the ground truth points in RTK.

[0067] The mass-converted effective distance D_eff,b of the b-th distance range is: D_eff,b = D_max,b × Q_b; The maximum value of the quality-converted effective distance D_eff,b for each distance range formed from valid data is taken to obtain the quality-converted effective distance evaluation result D_s: D_s = max_b(D_eff, b); like Figure 3 As shown, , and The maximum actual horizontal distance of all RTK points washed in the same second, the best candidate actual horizontal distance of observations in seconds, and the other two are respectively represented by the maximum actual horizontal distance of all RTK points washed in the same second. In this context, VALID seconds represent the formation of observations; the cardinality of the three sets further forms five quality parameters: sample support, coverage, hit, long tail angle error, and flow interruption. The sample support quality is multiplicatively gated, and the weighted sum of the other four quality parameters is obtained as Q_b. Q_b is then multiplied by the maximum value of the actual horizontal distance of the best candidate observation, D_max,b, represented by VALID seconds, to obtain D_eff,b. Finally, D_s is formed across the distance range.

[0068] The output range may include intermediate values. , , The base values ​​are E_b, O_b, and V_b; the original coverage ratio is C_b; the hit ratio is H_b; the gear P90 value is P90_b; the gear maximum interruption is M_b; the five quality parameters are: the gear quality coefficient is Q_b; the gear maximum effective hit distance is D_max,b; and the gear quality converted action distance is D_eff,b.

[0069] The candidate layer ranking and the second-representation observation layer ranking follow two different deterministic rules. For the candidate layer, the minimum angular error of the circle is compared first, followed by the time difference; for the second-representation observation layer, the time difference is compared first, followed by the minimum angular error of the circle, and finally the state priority. The comparison order of the two ranking systems must not be interchanged.

[0070] The overall angular error statistics and Hit@x use observations represented by non-empty seconds of the minimum angular error of the circle as the sample domain. This sample domain includes the VALID, MISS, and AMBIGUOUS states. Within this sample domain, the numerator of Hit@x further requires the judging result to be VALID and the minimum angular error of the circle to be no greater than x. The gear hit ratio H_b is V_b / O_b, and these two are statistics at different levels.

[0071] , , Different distance sources are used. The original coverage ratio C_b retains the original value formed by the heterogeneous archiving of the three sets, and the coverage quality Q_cov,b is then normalized by the clip function. Therefore, an original C_b greater than 1 does not necessarily mean that Q_cov,b is greater than 1.

[0072] The domain of D_max,b is the data domain that can form the maximum value from the best candidate actual horizontal distance representing the observation in VALID seconds within the b-th distance range. D_max,b does not use the upper bound of the distance range as a substitute, and the cross-range results are not subject to the condition of continuous achievement from the near distance range to the far distance range.

[0073] When calculating the longest consecutive valid seconds and the maximum consecutive invalid seconds based on the concatenated logical seconds, the continuity of the logical second index is applied. The maximum interruption for a given level is only accumulated within the consecutive evaluable logical seconds of that level, and is re-accumulated when adjacent evaluable logical second indices are not consecutive.

[0074] like Figure 4 As shown, the AOA equipment performance evaluation system includes a data acquisition and cleaning module, a logical second construction module, a candidate physical quantity calculation module, an observation status judgment module, a second-represented observation selection module, a distance range three-set construction module, a range quality calculation module, and a quality conversion output module, which sequentially provide data supply. Figure 4 The eight modules are named according to their functions, without using method / step labels.

[0075] The data acquisition and cleaning module is used to acquire the AOA azimuth observation sequence with observation time, the RTK ground truth trajectory with time and target identity, the receiving base station location of the AOA device, and evaluation parameters. It performs validity cleaning and time sorting on the AOA azimuth observation sequence and RTK ground truth trajectory, and outputs the cleaned AOA azimuth observation sequence, the RTK ground truth trajectory with target identity preserved, the receiving base station location, and evaluation parameters.

[0076] The logical second construction module is used to receive the cleaned RTK truth trajectory, divide the effective segments according to the long time interval between adjacent RTK truth points, establish the intra-segment second-level index according to the duration of each effective segment, and splice the first and last of the intra-segment second-level indexes according to the time order of the effective segments to form a logical second sequence and a mapping relationship from RTK truth points to logical seconds, so that the long-term RTK gaps between adjacent effective segments do not enter the evaluable duration.

[0077] The candidate physical quantity calculation module is used to receive the cleaned AOA azimuth observation sequence, the RTK ground truth trajectory with the target identity preserved, the receiving base station location, and evaluation parameters. For each AOA azimuth observation, it selects RTK ground truth points with a time difference not exceeding the time matching window as candidates. Based on the receiving base station location and each candidate location, it calculates the true azimuth, minimum circumferential angle error, and actual horizontal distance, and outputs a candidate set containing the time difference, minimum circumferential angle error, actual horizontal distance, and target identity.

[0078] The observation state adjudication module is used to receive the candidate set and the availability information of the receiving base station location. It sorts the candidates in ascending order of minimum circumferential angle error and time difference to determine the best candidate. Based on the availability of the receiving base station location, the existence of candidates, the minimum circumferential angle error condition, and the ambiguity between different RTK targets, it forms one of five states: VALID, MISS, AMBIGUOUS, NO_RTK_MATCH, and NO_DEVICE_LOCATION. It outputs the best candidate, the candidate physical quantity, and the adjudication result.

[0079] The second-representation observation selection module receives the logical second sequence, the mapping relationship from the RTK truth point to the logical second, the best candidate, the candidate physical quantity, and the judgment result. It maps the AOA azimuth observation with the judgment result to the logical second, and selects at most one second-representation observation in the same logical second according to the order of time difference priority, followed by minimum circumferential angular error, and then state priority. It outputs the actual horizontal distance, minimum circumferential angular error, and judgment result of the second-representation observation and its best candidate.

[0080] The distance range three-set construction module is used to receive the cleaned RTK ground truth trajectory, logical second sequence, receiving base station location, and second-represented observations, and constructs the module separately for multiple distance ranges by deduplicating logical seconds. , , .in, Archive the maximum actual horizontal distance between the RTK truth points and the receiving base station after all cleaning is completed within the same logical second. Archived data representing the best candidate actual horizontal distance observed in seconds. Adopted and The distances are sourced from the same origin and are formed solely from second-representation observations with a VALID result. This module outputs three sets of data for each distance range, the original coverage ratio, and the hit ratio.

[0081] The gear quality calculation module is used to receive the data for each distance gear. , , The observations are represented by seconds, and the sample support quality, coverage quality, hit quality, long tail angle error quality, and disconnection quality are normalized to [0,1]. The sample support quality is then multiplicatively gated with respect to the non-negative normalized weighted sum of the other four quality items to obtain the gear quality coefficient Q_b.

[0082] The quality conversion output module receives the second-represented observations and the gear quality coefficient Q_b. Within the data domain where the optimal candidate actual horizontal distance maximum value can be determined by the VALID second-represented observations in the b-th distance gear, it obtains D_max,b, calculates D_eff,b = D_max,b × Q_b, and takes the maximum value of D_eff,b for each distance gear formed by valid data. It then obtains and outputs the quality conversion distance evaluation result D_s and the corresponding gear intermediate quantity.

[0083] The system described above can be implemented using a processor, memory, and input / output interfaces. The memory stores computer programs or instructions for executing the functions of each module. The processor calls these computer programs or instructions to perform the aforementioned data processing on the AOA azimuth observations and RTK true trajectory. Each module can be implemented as a software functional unit, or it can be divided according to data processing stages within a program executed by the same processor.

[0084] This embodiment provides a computer-readable storage medium storing a computer program or instructions thereon. When the computer program or instructions are executed by a processor, the processor acquires the AOA azimuth observation sequence, RTK true trajectory, receiving base station location, and evaluation parameters; performs data cleaning, logical second construction, candidate physical quantity calculation, five-state evaluation, second-representative observation selection, distance range three-set construction, range quality calculation, and quality conversion output; and outputs the quality conversion effective distance evaluation result D_s and the range intermediate quantity.

[0085] The computer-readable storage medium may be a storage medium capable of storing programs or instructions and allowing them to be read by a processor. The data processing relationships implemented by the programs or instructions are consistent with the aforementioned method implementation methods.

[0086] The following values ​​are for illustrative purposes only and do not represent actual field test results.

[0087] Assume that a single RTK effective test window, after being segmented by a long hole, forms 20 logical seconds. The AOA ingress data, after being cleaned by S1, retains 30 records, i.e., N_raw=30; after candidate matching, refereeing, and same-second selection, 12 second-representative observations are formed, of which 10 are VALID and 2 are MISS, with the longest consecutive effective seconds being 6.

[0088] The 12 non-empty second-representation observations of the minimum angular error of the circumference are ordered in ascending order of angular error as follows: 1.0°, 1.2°, 1.4°, 1.8°, 2.0°, 2.2°, 2.8°, 3.5°, 4.0°, 4.5°, 6.0°, 12.0°.

[0089] The first 10 samples are VALID, and the last 2 are MISS. Using the nearest rank method, P90 takes the sample with index ceil(12×0.90)=11, which has a 6° error; P95 takes the sample with index ceil(12×0.95)=12, which has a 12° error. The denominator of Hit@3 is all 12 non-empty samples with the smallest circumferential angle error, and the numerator is the 7 VALID samples with an error no greater than 3°, therefore Hit@3=7 / 12; the denominator of Hit@5 is also 12, and the numerator is the 10 VALID samples with an error no greater than 5°, therefore Hit@5=10 / 12. The compression ratio is 12 / 30=0.40.

[0090] Taking the [500, 800) meter distance range as an example, assuming this range forms 4 evaluable logical seconds and 1 observation logical second, with the latter representing a VALID observation, the range P90 is 4°, and the maximum interruption time is 3 seconds, the VALID second represents the actual horizontal distance of the best candidate observation as 550 meters. Therefore: Q_sup=clip(4 / 5,0,1)=0.8; Q_cov=clip((1 / 4) / 0.50,0,1)=0.5; Q_hit=clip(1 / 0.50,0,1)=1; Q_p90=1, Q_loss=1; When w_c=0.35, w_h=0.35, w_p=0.20, w_l=0.10: Q_b=0.8×(0.35×0.5+0.35×1+0.20×1+0.10×1)=0.66; In this example, D_max,b represents the actual horizontal distance of 550 meters to the best candidate observation in VALID seconds, not the upper limit of 800 meters for the [500, 800) meter distance range. Therefore: D_eff,b=550×0.66=363 meters; Assuming that the mass-converted effective distance of other distance ranges already formed from valid data does not exceed 363 meters, then the cross-range result is D_s = 363 meters. This calculation shows that a 550-meter effective hit is not directly equivalent to a 550-meter mass-converted effective distance; sample support, coverage, hit, long tail angle error, and flow interruption mass all participate in the range conversion.

[0091] The above description is merely a specific embodiment of the present invention, enabling those skilled in the art to understand or implement the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the present invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features of the invention herein.

Claims

1. A method for evaluating the performance of AOA equipment, characterized in that, The method uses the AOA azimuth observation sequence with observation time output by the AOA device under test, the RTK ground truth trajectory with time and target identity, the receiving base station location of the AOA device, and evaluation parameters as physical measurement inputs, and includes the following steps: S1, obtain the physical measurement input, perform validity cleaning and time sorting on the AOA azimuth observation sequence and the RTK ground truth trajectory to obtain the cleaned AOA azimuth observation sequence, the RTK ground truth trajectory that retains the target identity, the location of the receiving base station and the evaluation parameters; S2, based on the long time interval between adjacent RTK truth points in the cleaned RTK truth trajectory, divide the effective segments, establish the second-level index within each segment according to the duration of each effective segment, and splice the first and last of the second-level index within each segment according to the time order of the effective segments to form a logical second sequence and a mapping relationship from RTK truth points to logical seconds, so that the long-term RTK gaps between adjacent effective segments do not enter the evaluable duration. S3, for each cleaned AOA azimuth observation, select RTK ground truth points from the RTK ground truth trajectory whose time difference with the AOA azimuth observation does not exceed the time matching window as RTK candidates. Calculate the corresponding true azimuth, the minimum circumferential angle error between the AOA azimuth observation and the true azimuth, and the horizontal distance between the receiving base station and the RTK candidate based on the location of the receiving base station and the location of each RTK candidate, to obtain a candidate set containing time difference, minimum circumferential angle error, horizontal distance, and target identity; S4, sort the RTK candidates in the candidate set in ascending order of minimum circumferential angle error and time difference to determine the best candidate, and based on the availability of the receiving base station location, the existence of RTK candidates, the minimum circumferential angle error condition, and the ambiguous relationship between different RTK targets, form a judgment result for the corresponding AOA azimuth angle observation, including at least the valid state VALID, the miss state MISS, the ambiguous state AMBIGUOUS, the no RTK match state NO_RTK_MATCH, and the no receiving base station location state NO_DEVICE_LOCATION. S5, map the AOA azimuth observation with the judgment result to the logical second sequence. When there are multiple AOA azimuth observations with the judgment result in the same logical second, select at most one second representative observation in the order of time difference priority, minimum circumferential angle error, and state priority, and obtain the second representative observation and its best candidate horizontal distance, minimum circumferential angle error and judgment result. S6, construct an evaluable set of logical seconds by deduplicating multiple distance ranges separately using logical seconds. Observation logic second set and the set of valid logical seconds The set of evaluable logical seconds is defined as the logical seconds in which the maximum horizontal distance between all cleaned RTK truth points and the receiving base station within the same logical second falls into the b-th distance range. The logical seconds representing the best candidate horizontal distance of the observation falling into the b-th distance range constitute the set of observation logical seconds. The set of observation logical seconds The set of valid logical seconds is composed of the logical seconds in which the observed judgment result is VALID. Let E_b, O_b, and V_b be the sets of evaluable logical seconds, respectively. The observation logic second set and the set of valid logical seconds The base number is used to form the original coverage ratio C_b based on the ratio of O_b to E_b, and the hit ratio H_b is formed based on the ratio of V_b to O_b. S7, based on the set of evaluable logical seconds for each distance range. The observation logic second set The set of effective logical seconds The second-represented observations are used to form sample support quality Q_sup,b, coverage quality Q_cov,b, hit quality Q_hit,b, long tail angle error quality Q_p90,b, and disconnection quality Q_loss,b, which are normalized to [0,1]. The quality coefficient Q_b of the b-th distance range is determined according to Q_b=Q_sup,b×(w_cQ_cov,b+w_hQ_hit,b+w_pQ_p90,b+w_lQ_loss,b), where w_c, w_h, w_p, and w_l are non-negative weights and w_c+w_h+w_p+w_l=1, so that the sample support quality Q_sup,b is multiplicatively gated to the weighted sum of the other four quality items. S8. Within the data domain where the best candidate actual horizontal distance maximum value can be determined by the second-represented observation with a VALID result in the b-th distance range, obtain D_max,b. Determine D_max,b as the farthest effective hit distance in the b-th distance range. Determine the quality-converted action distance D_eff,b in the b-th distance range according to D_eff,b=D_max,b×Q_b. Take the maximum value of the quality-converted action distance D_eff,b for each distance range formed by effective data. Obtain and output the quality-converted action distance evaluation result D_s and the corresponding gear intermediate quantity.

2. The AOA equipment performance evaluation method according to claim 1, characterized in that, In step S1, AOA observations with a finite number of azimuth angles and valid observation time are retained and arranged in ascending order of the valid observation time; RTK ground truth points with a finite number of longitudes and latitudes, longitudes and latitudes within a legal range and a preset business area, longitudes and latitudes not simultaneously approximately zero, and possessing altitude and valid time are retained and arranged in ascending order of the valid time; in step S2, assuming the time matching window is W, the positions in the RTK ground truth trajectory where the time interval between adjacent RTK ground truth points is greater than the segmentation threshold G are taken as valid segment boundaries, where G = max(3000ms,2W); For the q-th valid segment, determine the segment's second count T_q=max(1,ceil((t_q,end-t_q,start) / 1000)) based on the segment's start and end times t_q,end. Concatenate the first and last second-level indices of each valid segment according to the segment order to form a logical second sequence, and determine the total test count T=Σ_qT_q, so that long-term RTK gaps between adjacent valid segments are not included in the total test count.

3. The AOA equipment performance evaluation method according to claim 1, characterized in that, In step S3, for the i-th AOA observation with observation time t_i and azimuth angle α_i, the RTK ground truth point with time τ_j and satisfying |t_i-τ_j|≤W is determined as an RTK candidate; the longitude and latitude of the receiving base station are denoted as λ_b and φ_b, respectively, and the longitude and latitude of the j-th RTK candidate are denoted as λ_j and φ_j, respectively. Each longitude and latitude is expressed in radians and Δλ=λ_j-λ_b, Δφ=φ_j-φ_b. The true azimuth angle is determined according to β_ij=norm_[0°,360°){(180° / π)·atan2(sinΔλ·cosφ_j,cosφ_b·sinφ_j-sinφ_b·cosφ_j·cosΔλ)}, where the radian value output by the atan2 function is first multiplied by 180° / π. Converted to angles and then normalized to [0°, 360°), the azimuth angle α_i and the true azimuth angle β_ij are both expressed in degrees. The minimum circumferential angular error is determined according to d_ij=|norm(α_i)-norm(β_ij)| and e_ij=min(d_ij,360°-d_ij). The Haversine horizontal distance from the receiving base station to the RTK candidate is determined according to a_ij=sin²(Δφ / 2)+cosφ_b·cosφ_j·sin²(Δλ / 2) and r_ij=2R·atan2(√a_ij,√(1-a_ij)), where R=6371000m. The RTK candidates are sorted in ascending order of the minimum circumferential angular error and the time difference to determine the best candidate and the second best candidate.

4. The AOA equipment performance evaluation method according to claim 3, characterized in that, In step S4, the adjudication result of each AOA observation is determined to be one of five states: NO_DEVICE_LOCATION, NO_RTK_MATCH, AMBIGUOUS, MISS, and VALID. When the receiving base station location is missing or invalid, it is determined to be in the NO_DEVICE_LOCATION state; when the receiving base station location is valid and no RTK candidate exists, it is determined to be in the NO_RTK_MATCH state; when a suboptimal candidate exists, the minimum circumferential angular error e_i2 of the suboptimal candidate is not greater than... When the hit angle error threshold H, the absolute value of the difference between the minimum circumferential angle error of the second-best candidate and the best candidate (|e_i2-e_i1|) is not greater than the ambiguity difference threshold A, and the best candidate and the second-best candidate come from different RTK targets, the state is determined to be AMBIGUOUS. Ambiguity judgment only checks the sorted best and second-best candidates. When the ambiguity condition is not met and the minimum circumferential angle error e_i1 of the best candidate is greater than H, the state is determined to be MISS. When the aforementioned abnormal condition is not met and e_i1 is not greater than H, the state is determined to be VALID.

5. The AOA equipment performance evaluation method according to claim 1, characterized in that, In step S5, when there is a best candidate for the post-judgment AOA observation, the RTK truth point time of the best candidate is mapped to the logical second. When there is no RTK candidate or no valid receiving base station location, the observation time of the AOA observation itself is mapped to the logical second, and the corresponding logical second is formed only when the time used for mapping falls into a valid RTK segment. The judgment result is further limited to five states: VALID, MISS, AMBIGUOUS, NO_RTK_MATCH, and NO_DEVICE_LOCATION. When there are multiple post-judgment AOA observations in the same logical second, one second is selected to represent the observation in ascending order of time difference, ascending order of minimum circumferential angle error, and state priority. The state priority from high to low is VALID, MISS, AMBIGUOUS, NO_RTK_MATCH, and NO_DEVICE_LOCATION. Empty time difference is ranked after non-empty time difference, and empty minimum circumferential angle error is ranked after non-empty minimum circumferential angle error, so that at most one second is retained to represent the observation for each logical second.

6. The AOA equipment performance evaluation method according to claim 5, characterized in that, Based on the second-represented observations, the number of second-represented samples, the number of observation seconds, the number of valid seconds in the VALID state, the number of seconds in the MISS state, the number of seconds in the AMBIGUOUS state, the number of seconds in the NO_RTK_MATCH state, and the number of seconds in the NO_DEVICE_LOCATION state are statistically analyzed. The longest consecutive valid seconds and the maximum consecutive non-valid seconds are also statistically analyzed based on the concatenated logical second sequence. Second-represented observations in the VALID, MISS, or AMBIGUOUS state with a non-empty minimum angular error are collectively formed into an angular error sample set. The mean, median, P90, and P95 are calculated based on this angular error sample set. The N angular error samples are then sorted in ascending order and... Samples with serial numbers ceil(0.90N) and ceil(0.95N) are selected as P90 and P95 respectively using the nearest rank method. Hit@x is calculated for x∈{3°,5°,10°,20°}. The denominator of Hit@x is the number of samples in the angle error sample set, and the numerator is the number of samples in the angle error sample set that are in the VALID state and whose minimum circumferential angle error is not greater than x. Hit@x is determined to be zero when the denominator is zero. The compression ratio N_s / N_raw is determined as the ratio of the number of second-level representative samples N_s to the number of AOA observations N_raw cleaned in step S1. The compression ratio is determined to be zero when N_raw is zero.

7. The AOA equipment performance evaluation method according to claim 1, characterized in that, The multiple distance ranges are six half-open intervals: [0, 300) meters, [300, 500) meters, [500, 800) meters, [800, 1000) meters, [1000, 1500) meters, and [1500, +∞) meters, where +∞ only indicates that the last half-open interval has no finite upper bound. For the b-th distance range, the maximum horizontal distance between all cleaned RTK ground truth points and the receiving base station within the same logical second falls into the logical second of the b-th distance range. After deduplication, an evaluable logical second set is formed. The logical seconds representing the best candidate horizontal distances of the observations that fall into the b-th distance range are deduplicated to form the set of observation logical seconds. and will The middle second represents the set of valid logical seconds after deduplication of the logical seconds in the observation state of VALID. Let E_b = | |、O_b=| |、V_b=| The original coverage ratio C_b = O_b / E_b and hit ratio H_b = V_b / O_b are determined. When E_b is zero, C_b is set to zero, and when O_b is zero, H_b is set to zero. The original coverage ratio C_b is retained by... and The original ratios formed from different distance sources are not truncated because they are greater than 1.

8. The AOA equipment performance evaluation method according to claim 7, characterized in that, For the b-th range bin, the observation logic second set whose range sources are classified into this range bin, with a state of VALID, MISS or AMBIGUOUS and non-null circular minimum angle error, are grouped into a gear angle error sample set, and the P90 value P90_b of the gear is determined according to the nearest rank method; the number of consecutive seconds with no VALID-state second representative observation is accumulated within the consecutively appearing evaluatable logic seconds of the range bin, and the accumulation is restarted when the logic second indexes of adjacent evaluatable logic seconds are discontinuous, so as to obtain the maximum data interruption M_b of the gear; the sample support quality, coverage quality and hit quality are respectively determined according to Q_sup,b=clip(E_b / E_0,0,1), Q_cov,b=clip(C_b / C_0,0,1) and Q_hit,b=clip(H_b / H_0,0,1), wherein clip is used to limit the calculation result to [0,1]; when P90_b does not exist, P90_b≤P_0, P90_b≥P_bad, and P_0<P90_b<P_bad, the long-tail angle error quality Q_p90,b is determined as 0, 1, 0, and (P_bad-P90_b) / (P_bad-P_0) respectively; when M_b≤M_0, M_b>M_0 and M_0>0, and M_b>M_0 and M_0≤0, the data interruption quality Q_loss,b is determined as 1, M_0 / M_b and 0 respectively; set E_0=5 seconds, C_0=0.50, H_0=0.50, P_0=15°, P_bad=35° and M_0=20 seconds, and set non-negative weights w_c, w_h, w_p, w_l that satisfy w_c+w_h+w_p+w_l=1 and are 0.35, 0.35, 0.20, and 0.10 in sequence; the gear quality coefficient is determined according to Q_b=Q_sup,b×(w_cQ_cov,b+w_hQ_hit,b+w_pQ_p90,b+w_lQ_loss,b), so that the sample support quality implements multiplicative gating on the weighted sum of the remaining four qualities; the maximum value D_max,b is obtained from the actual horizontal distances of the best candidates corresponding to the VALID-state second representative observations in the b-th range bin, D_max,b is determined as the farthest effective hit distance of the range bin instead of the upper bound of the range bin, and D_eff,b=D_max,b×Q_b is calculated, and the quality-converted working distance evaluation result is obtained according to D_s=max_b(D_eff,b) based on D_eff,b of each range bin formed by said valid data.

9. An AOA equipment performance evaluation system, characterized in that, It includes a data acquisition and cleaning module that sequentially supplies data, a logical second construction module, a candidate physical quantity calculation module, an observation status judgment module, a second-represented observation selection module, a distance-gear three-set construction module, a gear mass calculation module, and a mass conversion output module; The data acquisition and cleaning module is used to acquire the AOA azimuth observation sequence with observation time, the RTK ground truth trajectory with time and target identity, the receiving base station location of the AOA device, and evaluation parameters output by the AOA device under test. It performs validity cleaning and time sorting on the AOA azimuth observation sequence and the RTK ground truth trajectory, and outputs the cleaned AOA azimuth observation sequence, the RTK ground truth trajectory retaining the target identity, the receiving base station location, and the evaluation parameters. The logical second construction module is used to receive the cleaned RTK truth trajectory, divide the effective segments according to the long time interval between adjacent RTK truth points, establish a segment-level index based on the duration of each effective segment, and splice the first and last of the segment-level indexes according to the time order of the effective segments to form a logical second sequence and a mapping relationship from RTK truth points to logical seconds, and ensure that long-term RTK gaps between adjacent effective segments do not enter the evaluable duration. The candidate physical quantity calculation module is used to receive the cleaned AOA azimuth observation sequence, the RTK ground truth trajectory with the target identity preserved, the location of the receiving base station, and the evaluation parameters. For each AOA azimuth observation, it selects RTK ground truth points with a time difference not exceeding the time matching window as RTK candidates. Based on the location of the receiving base station and the location of each RTK candidate, it calculates the true azimuth, minimum circumferential angle error, and horizontal distance, and outputs a candidate set containing the time difference, minimum circumferential angle error, horizontal distance, and target identity. The observation state adjudication module is used to receive the candidate set, sort the RTK candidates in ascending order of minimum circumferential angle error and time difference to determine the best candidate, and, based on the availability of the receiving base station location, the existence of RTK candidates, the minimum circumferential angle error condition, and the ambiguous relationship between different RTK targets, form an adjudication result for the corresponding AOA azimuth angle observation, including at least the valid state VALID, the miss state MISS, the ambiguous state AMBIGUOUS, the no RTK match state NO_RTK_MATCH, and the no receiving base station location state NO_DEVICE_LOCATION, and output the best candidate, the candidate physical quantity, and the adjudication result; The second-representation observation selection module is used to receive the logical second sequence, the mapping relationship from the RTK truth point to the logical second, the best candidate, the candidate physical quantity, and the judgment result. It maps the AOA azimuth observation with the judgment result to the logical second sequence, and selects at most one second-representation observation within the same logical second in the order of time difference priority, followed by minimum circumferential angular error, and then state priority. It outputs the horizontal distance, minimum circumferential angular error, and judgment result of the second-representation observation and its best candidate. The distance-range three-set construction module is used to receive the cleaned RTK ground truth trajectory, the logical second sequence, the receiving base station location, and the second-represented observations, and to construct evaluable logical second sets for multiple distance ranges by deduplicating logical seconds. Observation logic second set and the set of valid logical seconds The evaluable logical seconds set The observation logical second set is archived according to the maximum horizontal distance of all cleaned RTK truth points relative to the receiving base station within the same logical second. Archived according to the best candidate horizontal distance of the observation represented by the second, the effective logical second set Using the observed logical second set The distances are from the same source and are formed only by second-represented observations with a VALID result. The three sets, raw coverage ratio, and hit ratio for each distance range are output. The gear quality calculation module is used to receive the evaluable logical second set for each distance gear. The observation logic second set The set of effective logical seconds The observations, represented by seconds, are used to form sample support quality Q_sup,b, coverage quality Q_cov,b, hit quality Q_hit,b, long tail angle error quality Q_p90,b, and disconnection quality Q_loss,b, all normalized to [0,1]. The quality coefficient Q_b is determined according to Q_b=Q_sup,b×(w_cQ_cov,b+w_hQ_hit,b+w_pQ_p90,b+w_lQ_loss,b), where w_c, w_h, w_p, and w_l are non-negative weights and w_c+w_h+w_p+w_l=1. This ensures that the sample support quality Q_sup,b is multiplicatively gated over the weighted sum of the other four quality items. The five quality items and the quality coefficient Q_b are then output. The quality conversion output module is used to receive the second-represented observation and the gear quality coefficient Q_b, obtain D_max,b from the data domain in which the second-represented observation with a VALID adjudication result in the b-th distance gear can determine the maximum value of its best candidate actual horizontal distance, determine D_max,b as the farthest effective hit distance in the b-th distance gear, determine the quality conversion action distance D_eff,b in the b-th distance gear according to D_eff,b=D_max,b×Q_b, and take the maximum value of the quality conversion action distance D_eff,b for each distance gear formed by effective data, obtain and output the quality conversion action distance evaluation result D_s and the corresponding gear intermediate quantity.

10. A computer-readable storage medium having a computer program or instructions stored thereon, wherein when the computer program or instructions are executed by a processor, the AOA equipment performance evaluation method of claim 1 is implemented, and the quality-converted action distance evaluation result is output.

Images