A vehicle axle load measurement method based on load state reconstruction

Abstract

The application discloses a vehicle axle load measurement method based on load state reconstruction, solves the problem that the axle load estimation method of the existing vehicle axle group scale is easy to produce systematic deviation and low precision, and comprises the following steps: acquiring a weighing signal, detecting a vehicle motion event based on the weighing signal, and generating a vehicle motion event sequence; dividing continuous and channel signals into a plurality of stable load states according to the vehicle motion event sequence, and constructing a stable load state matrix H, each state corresponding to a static load combination under the joint action of several vehicle axles; based on the stable load state matrix, a state space mapping model between vehicle axle load parameters and stable load state observations is established; a state confidence matrix is constructed to weight the state space mapping model, the weighted state space mapping model is solved, an optimal solution with minimum observation error is acquired, and vehicle axle loads are obtained. The weighted optimization estimation based on the state confidence is adopted, and the vehicle axle load estimation precision is improved.

Description

Technical Field

[0001] This invention relates to the field of electronic weighing technology, and in particular to a method for measuring vehicle axle load based on load state reconstruction. Background Technology

[0002] Vehicle axle load measurement is a core technical means to ensure the safety of road and bridge structures, prevent traffic accidents, regulate law enforcement and toll collection, and optimize traffic management. Currently, vehicle axle load measurement simply uses an axle load cell to sense wheel pressure through sensors, measure the load on a single axle or axle group, and sum the results to obtain the total vehicle weight. However, this method uses a limited number of sampling points, leading to a large variance in the estimated value and reduced accuracy. Furthermore, when the wheelbase between two axles is close to the length of the weighing platform, the difference in steady-state weight before and after the axle enters the scale does not equal the axle's weight, resulting in an incorrect estimation of that axle's weight.

[0003] Patent CN120702577A describes a horizontal calibration device and weighing method for axle load scales used in vehicle wheel load measurement. The axle load scale includes a weighing platform, weighing sensors, and a control terminal. An electronic gyroscope is installed in mounting slots at each of the four corners of the weighing platform. The electronic gyroscopes are connected to the control terminal via signal circuits. This simplifies weighing and platform leveling operations, increases measurement efficiency, eliminates manual leveling steps, and reduces measurement time from 5 minutes to 1 minute. The weighing accuracy is high, with a measured error ≤ ±1.2%, meeting national standards. However, this solution only improves measurement accuracy by leveling the weighing platform; it does not improve the processing method of the measurement data collected by the weighing sensors themselves. It still suffers from problems such as decreased accuracy due to a small number of measurement points and insufficient measurement accuracy in scenarios where the axle distance is close to the length of the weighing platform. Summary of the Invention

[0004] The purpose of this invention is to solve the problems of systematic bias and low accuracy in the existing axle load estimation methods of vehicle axle scales. It provides a vehicle axle load measurement method based on load state reconstruction, which adopts weighted optimization estimation based on state confidence, thereby improving the accuracy of vehicle axle load estimation.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: A method for measuring vehicle axle load based on load state reconstruction includes the following steps: S1: Acquire the weighing signal, detect vehicle motion events based on the weighing signal, and generate a vehicle motion event sequence E; S2: Based on the vehicle motion event sequence, the continuous channel signal is divided into multiple stable load states, and a stable load state matrix H is constructed. Each state corresponds to the static load combination under the combined action of several axles of the vehicle. S3: Based on the stable load state matrix, establish a state-space mapping model between vehicle axle load parameters and stable load state observations; S4: Construct a state confidence matrix to weight the state-space mapping model, solve the weighted state-space mapping model, obtain the optimal solution with the minimum observation error, and obtain the vehicle axle load.

[0006] The vehicle axle load measurement method based on load state reconstruction provided by this invention utilizes the detected weighing / unweighing events to divide the dynamic weighing process into multiple stable load states, each state corresponding to a different vehicle axle combination, and establishes a state space mapping model; constructs a confidence matrix to perform weighted optimization estimation on the state space mapping model to obtain the optimal axle load, thereby reducing axle load measurement error and improving axle load measurement accuracy.

[0007] Preferably, S2 includes: The event sequence E is reordered in ascending order of the start time of the vehicle motion events to obtain the sorted vehicle motion event sequence Esort. Starting from the first event in the sorted vehicle motion event sequence Esort, a stable load state value is assigned to the time interval between every two adjacent events. Construct a stable load state matrix H. The number of rows in the stable load state matrix is ​​the total number of vehicle motion events, and the number of columns is the total number of axles of the vehicle under test. The elements of the stable load state matrix H are whether the axle is on the scale in state i.

[0008] Preferably, the state-space mapping model is as follows: the observed average weight vector Y on the scale is equal to the sum of the product of the stable load state matrix H and the axle load vector X and the measurement error noise vector N; the average weight vector Y on the scale is a single-column matrix, and the elements y in the average weight vector Y on the scale... i The axle load vector X represents the average weight on the scale under stable load state i; the axle load vector X is a single-column matrix, and the elements x in the axle load vector X are... i The weight of the i-th axle is represented; the measurement error noise vector N is a single-column matrix, and the elements N in the measurement error noise vector N are... i The noise of the observed weight on the scale represents the stable load state i.

[0009] Preferably, the state confidence matrix is ​​a diagonal matrix, where the number of rows and columns of the state confidence matrix is ​​the total number of vehicle motion events, and the elements on its diagonal represent the confidence level of the measurement data of the stable load state.

[0010] As a preferred option, constraints are constructed to verify the optimal solution. The constraints include: all axle loads are positive numbers and all axle loads are less than or equal to a preset axle load threshold. If all components of the optimal solution satisfy the constraints, the optimal solution is a feasible solution; otherwise, the optimal solution is rejected, and a new round of weighing / unweighing event identification and analysis is triggered.

[0011] Preferably, the weighing signal includes the output of a pressure sensor, the output s1(t) of the weighing sensor pair, the output s2(t) of the weighing sensor pair, the output s3(t) of the weighing counter, the output s4(t) of the weighing counter, and the output ss(t) of the weighing platform sensor and the channel, wherein the output ss(t) is the sum of the output s1(t) of the weighing sensor pair and the output s2(t) of the weighing sensor pair.

[0012] Preferably, vehicle motion events are detected from the outputs s3(t) of the weighing input counter and s4(t) of the weighing output counter, including axle weighing input events and axle weighing output events; each vehicle motion event includes an event type, an event start time, an event end time, and an event intensity, wherein the event intensity is a rough estimate of the axle weight of the axle corresponding to the event.

[0013] Preferably, the number of elements in the axle load vector X is equal to the total number of axles of the vehicle under test, and the number of elements in the average weight vector Y on the scale and the measurement error noise vector N are equal to the total number of vehicle motion events.

[0014] Preferably, for each stable load state i, the duration of the stable load state, the mean weight on the scale under the stable load state, and the sample standard deviation are calculated; the confidence level of the measurement data of the stable load state is calculated based on the duration, the mean weight on the scale under the stable load state, and the sample standard deviation.

[0015] Preferably, the duration of the current stable load state is the time from the end of the current weighing event to the start of the next vehicle movement event; if the duration is 0, a valid measurement sample of the stable load state cannot be obtained.

[0016] Therefore, the present invention has the following beneficial effects: it transforms single-event measurement into multi-state joint estimation, can accurately calculate axle load for various wheelbase models, and makes full use of all available sampling points to achieve high-precision axle load estimation, achieving an accuracy of <0.5% in application scenarios such as ladle scales. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating the overall steps of the vehicle axle load measurement method based on load state reconstruction in this invention.

[0018] Figure 2This is a schematic diagram of the existing shaft group scale system structure.

[0019] Figure 3 This is a schematic diagram of the ss(t) signal and vehicle movement event when the vehicle crosses the scale in Example 2. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this application will be further described in detail below through embodiments and in conjunction with the accompanying drawings. It should be understood that the specific embodiments described herein are merely one preferred embodiment of this application and are only used to explain this application. They do not limit the scope of protection of this application. All other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention. The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments.

[0021] Example 1: This embodiment provides a vehicle axle load measurement method based on load state reconstruction, such as... Figure 1 As shown, the operation process is as follows: Step 1, acquire the weighing signal, detect vehicle motion events based on the weighing signal, and generate a vehicle motion event sequence E; Step 2, divide the continuous channel signal into multiple stable load states according to the vehicle motion event sequence, construct a stable load state matrix H, and each state corresponds to a combination of static loads under the combined action of several axles of the vehicle; Step 3, based on the stable load state matrix, establish a state-space mapping model between vehicle axle load parameters and stable load state observations; Step 4, construct a state confidence matrix to weight the state-space mapping model, solve the weighted state-space mapping model, obtain the optimal solution with the minimum observation error, and obtain the vehicle axle weight.

[0022] The current mainstream axle scales in the industry consist of a vehicle separator, two front and rear axle counters, and a weighing platform (including four sensors). Figure 2 The weighing system consists of A1, B1, A2, and B2. Two axle counters are installed at the inlet and outlet of the weighing platform, respectively, to count the number of axles entering / leaving the platform. Four main sensors are installed at the four corners of the platform to measure the weight of the vehicles.

[0023] Specifically, the system structure of mainstream shaft group scales on the market is as follows: Figure 2As shown, the weighing instrument directly calculates the number of axles on and off the scale using the infeed sensor pair (infeed sensor A1, infeed sensor B1) and the outfeed sensor pair (infeed sensor A2, infeed sensor B2). It obtains the weight of each axle by the steady-state weight difference on the scale platform before and after it enters the scale, and then sums the axle weights to obtain the total vehicle weight. The data from the infeed and outfeed sensor pairs is transmitted to the weighing instrument through a junction box, where the weighing instrument processes the data to obtain the axle weight.

[0024] However, the traditional axle load estimation method for axle scales has the following drawbacks: Pain Point 1: When measuring axle weight, the steady-state value of the sum of the signals from the four main sensors on the weighing platform before and after the axle enters the scale is used. The number of sampling points used is relatively small, resulting in a large variance in the estimated quantity and a decrease in accuracy. For some scenarios that require high weighing accuracy, such as steel ladle scales in steel mills, it is difficult to meet the accuracy requirements.

[0025] Pain Point 2: For some vehicle models, when the wheelbase of two axles is close to the length of the weighing platform, the entry time of one axle overlaps with the exit time of the other axle. This causes the difference in steady-state weight before and after the entry time of the axle to be equal to the weight of the axle itself. Instead, it is the difference between the weight of the axle entering the weighing platform and the weight of the axle exiting the weighing platform at the same time, which leads to an incorrect estimation of the weight of the axle.

[0026] Pain Point 3: To avoid the estimation error in Pain Point 2, the weight of the axle can be estimated by the change of the edge of the weighing channel when measuring the axle weight. However, since the signal model of the weighing channel is not an ideal step signal, there may be a ramp effect on both sides of the edge, which will produce systematic deviation and seriously reduce the weighing accuracy.

[0027] In summary, the vehicle axle load measurement method based on load state reconstruction provided in this embodiment is mainly used to solve the above-mentioned pain points in traditional methods. It can accurately calculate the axle load for vehicle models with various wheelbases, and makes full use of all available sampling points to achieve high-precision axle load estimation. In application scenarios such as ladle scales, it achieves an accuracy of less than 0.5%.

[0028] It should be noted that the measurement method in this embodiment also adopts... Figure 2 The axle weighing system shown in the diagram performs axle load measurement, but processes the obtained measurement data to obtain a more accurate measurement value.

[0029] The following examples and specific application scenarios further illustrate the technical solution and effects of the present invention. The following examples are explanations of the present invention, but the present invention is not limited to the following examples.

[0030] A method for measuring vehicle axle load based on load state reconstruction specifically includes the following steps: Step 1: Obtain the weighing signal.

[0031] The weighing signal includes the outputs of multiple pressure sensors, specifically the feed sensor pair (i.e., ... Figure 2 The outputs s1(t) of the infeed sensor A1 and the infeed sensor B1, and the output sensor pair (i.e. Figure 2 The outputs s2(t) of the weighing sensor A2 and weighing sensor B2, the outputs s3(t) of the weighing input axis, the outputs s4(t) of the weighing output axis, and the outputs ss(t) of the weighing platform sensor and the channel are all included. The outputs ss(t) of the weighing platform sensor and the channel are the sum of the outputs s1(t) of the weighing input sensor pair and the outputs s2(t) of the weighing output sensor pair.

[0032] Step 2: Detect vehicle motion events based on weighing signals and generate vehicle motion event sequence E.

[0033] Vehicle motion events, including axle entry and exit events, are detected from the outputs of the infeed counter s3(t) and the outfeed counter s4(t), forming a vehicle motion event sequence E: E={v1, v2, ..., v i , ..., v n}

[0034] It should be noted that the detection of vehicle motion events from the input scale counter s3(t) and the output scale counter s4(t) is only applicable to... Figure 2 The axle scale with a multi-axle unit is just one implementation. In other embodiments, vehicle motion events can be detected directly from s1(t) and s2(t). That is, the solution of the present invention can also be used in axle scales with numerous axle units to realize axle load measurement.

[0035] Where n represents the total number of vehicle movement events, v i This represents the i-th vehicle movement event, i.e., a vehicle entering or leaving the scale.

[0036] Each vehicle motion event includes the following events: { type, start time t_start, end time t_stop, event intensity amp}.

[0037] Where, type represents the event type, 0 represents entering the scale and 1 represents exiting the scale; t_start represents the start time of the vehicle movement event; t_stop represents the end time of the vehicle movement event; amp represents the intensity of the vehicle movement event, which is a rough estimate of the axle weight corresponding to this vehicle movement event.

[0038] Step 3: Divide the continuous sum channel signal ss(t) into multiple stable load states according to the event sequence. Each stable load state corresponds to the static load combination under the combined action of several axles of the vehicle.

[0039] Specifically, the following steps are included: Step (3.1): First, reorder the vehicle motion event sequence E in ascending order of the vehicle running event start time t_start to obtain the sorted vehicle motion event sequence E. sort E sort ={v s1 v s2 , ..., v si , ..., v sn}

[0040] Among them, v si This represents the i-th vehicle movement event after sorting.

[0041] Step (3.2): Then, starting from the first event, assign a stable load state value {state1, state2, ..., statei, ..., staten-1, staten} to the time interval between every two adjacent events, where staten represents the time interval between the last vehicle motion event and the end of the sampling signal.

[0042] The state here refers to which axles are currently on the scale. Specifically, the state can be represented by a Boolean vector [a1, a2, ..., am], where ai=1 indicates that the axle is on the scale, and ai=0 indicates that the axle is not on the scale.

[0043] Step (3.3): Define the stable load state matrix H. The dimension of the stable load state matrix H is n*m, where m represents the total number of axles of the vehicle.

[0044] In this embodiment, the stable load state matrix H consists of n state vectors h i Composed of each state vector h i It also includes m elements, that is, the stable load state matrix H is an n×m matrix.

[0045] Specifically, the state vector h i =[a i1 a i2 ... a ij ... a im ], where a ij The value is {0, 1}, which means that in state i, axle j is on the scale; 0 means that in state i, axle j is not on the scale.

[0046] In this embodiment, the stable load state matrix H can be automatically generated based on the analysis of vehicle motion events.

[0047] Step (3.4): For each stable load state i, define the following property: The duration of this stable load state is w(i) = max(0, t_start(i+1)-t_stop(i)), which means that the duration of the current stable load state i is the maximum value between the end time of the current weighing event and the start time of the next vehicle movement time and 0.

[0048] Where w(i) equals 0, it means that the weighing entry event and the weighing exit event are completely in conflict, and it is impossible to obtain a valid measurement sample of the stable load state i.

[0049] The average weight y on the scale under this stable load state i i This is the ratio of the sum of the current weighing event's end time t_stop(i) to the next vehicle movement event's start time t_start(i+1) and the channel signal ss(t), to the current weighing event's end time minus the next vehicle movement event's start time plus 1 (i.e., t_start(i+1) - t_stop(i) + 1). In other words, it's the average of the channel signal ss(i) within the range [t_stop(i), t_start(i+1)].

[0050] The sample standard deviation z of the weight on the scale under this stable load condition i i Let y be the sum of the average weight on the scale and the channel signal ss(t) within the time interval from the end time t_stop(i) of the current weighing event to the start time t_start(i+1) of the next vehicle movement event. i The square of the difference is the ratio of the difference between the end time of the current weighing event and the start time of the next vehicle movement event. This is also the sample standard deviation of the channel signal ss(i) within the range [t_stop(i), t_start(i+1)].

[0051] Step 4: Establish a state-space mapping model between vehicle axle load parameters and stable load state observations, i.e., the observation model under stable load state of the vehicle.

[0052] Specifically, the steps include the following: Step (4.1): First, define the axle weight vector X. In this embodiment, the axle weight vector x is a single-column matrix with m elements, each element x i This represents the weight of the i-th axle.

[0053] Step (4.2): Define the observed average weight vector Y on the scale. In this embodiment, the average vector Y is a single-column matrix with n elements, each element y i This represents the average weight on the scale under stable load state i.

[0054] Step (4.3): Define the measurement error noise vector N. In this embodiment, the measurement error noise vector N is a single-column matrix with n elements, each element N i The observed noise of the weight on the scale represents the steady load state i (i.e., the i-th steady state). The observation error noise vector N is a random vector, a nondeterministic vector.

[0055] Step (4.4): Establish the observation model under stable load conditions of the vehicle as follows: the mean vector (observation vector) Y is equal to the product of the stable load state matrix H and the axle load vector X plus the measurement error noise vector N, that is, Y=HX+N.

[0056] The observation model is a noisy linear observation equation.

[0057] Step 5: Construct the state confidence matrix C.

[0058] In this embodiment, the confidence matrix C is a diagonal matrix that assigns different confidence weights to each stable load state equation. The element c(i) on its diagonal represents the confidence of the measurement data of stable load state i.

[0059] The confidence level c(i) is based on the duration w(i) output in the third step and the average weight y on the scale under stable load state i. i And the sample standard deviation z of the weight on the scale under stable load condition i. i To estimate.

[0060] The confidence level c(i) can be calculated in different ways. Here are a few examples: Example 1: Confidence level c(i) equals duration w(i).

[0061] This confidence level calculation method only uses the number of samples, making it simple and intuitive.

[0062] Example 2: The confidence level c(i) is equal to the sample standard deviation z with base e. i With the mean y i The negative of the absolute value of the ratio is the value of the exponential function of the independent variable.

[0063] This calculation method only uses the normalized standard deviation of the sample, which is intuitive, but the indicator becomes unreliable when the sample size is small; a better way is to use a fusion method to calculate the confidence score.

[0064] Example 3: The confidence score c(i) equals exp(-abs(z(i)) / (y(i) * sqrt(w(i)))), where sqrt represents the square root function. This calculation method combines the normalized variance of the samples and the sample size, making it more adaptable to different scenarios.

[0065] Construct an optimization problem: Given the confidence matrix C (i.e., the weighting matrix), the stable load state matrix H, and the observation vector Y, find the optimal solution X' that minimizes the overall weighted observation error.

[0066] The optimal solution X' is one that makes (HX-Y)... T C(HX-Y) represents the minimum value of X. (HX-Y) is the product of the stable load state matrix H and the axle load vector X, minus the transpose of the mean vector Y. It represents the observation error vector, (HX-Y). T (HX-Y) represents the 2-norm of the observation error noise vector. T C(HX-Y) is the 2-norm of the weighted observation error noise vector; that is, the objective is to minimize the weighted 2-norm of the measurement error noise vector N. The measurement error noise vector N is a random vector, and for equations with random vectors, it is necessary to solve them based on statistical characteristics.

[0067] This embodiment uses quadratic optimization theory to solve it, and the optimal solution is (H). T CH) + H T CY is the product of the transpose of the stable load state matrix H and the confidence matrix C, plus the product of the transpose of the stable load state matrix H and the confidence matrix C, plus the product of the mean vector Y.

[0068] Step 6: Construct the state normalization error vector Q.

[0069] The state normalization error vector Q is the difference between the product of the stable load state matrix H and the optimal solution X' of the axle load vector, and the observed vector Y. The ratio of this difference to |Y| is the state normalization error vector. |Y| represents taking the absolute value point by point.

[0070] Each element Q(i) of the state normalization error vector Q represents the normalization estimation error of state i.

[0071] Step 7: Construct constraints and verify the optimal solution X'.

[0072] This embodiment includes two constraints: The first constraint is: x(i) > 0, i = 1, ..., m; this constraint means that all axle loads must be positive.

[0073] The second constraint is: Q(i) < threshold TH; this constraint means that all axle loads must be below the preset threshold.

[0074] If all components of the optimal solution satisfy both sets of constraints, the optimal solution is feasible; otherwise, the optimal solution should be rejected. It is very likely that the event detection stage in the second step of this round of analysis has failed, and there has been a missed event or a false alarm. This can be used as a basis to trigger a new round of weighing / unloading event identification analysis.

[0075] It should be noted that in some scenarios, due to sensor failure, signal acquisition link failure, or errors in the detection of vehicle motion events, solutions that do not meet the constraints may occur.

[0076] The vehicle axle load measurement method based on load state reconstruction provided in this embodiment has the following beneficial effects: (1) This addresses the problem that traditional axle load estimation relies on the difference between the edges of a single entry and exit points of the weighbridge. When the axle spacing is close to the weighbridge length, the load changes of multiple axles are superimposed and cannot be independently separated. In addition, it does not make full use of all effective weighing samples, which reduces the weighing accuracy. By using the detected entry / exit events of the weighbridge, the dynamic weighing process is divided into multiple stable load states. Each state corresponds to a different combination of vehicle axles. A load state reconstruction model based on vehicle motion events is established, which transforms "single event measurement" into "multi-state joint estimation". This fully utilizes the data of the entire vehicle passage process and improves the accuracy of axle load estimation.

[0077] (2) A weighted optimization estimation based on state confidence is adopted to solve the problems of different steady-state vinegar time, different noise, and different reliability. By constructing a confidence matrix, the state space mapping model is weighted and optimized using the confidence matrix to reduce the impact of abnormal states, short-term states, and noisy states on axle weight estimation and further improve measurement accuracy.

[0078] Example 2: This embodiment provides a vehicle axle load measurement method based on load state reconstruction. Based on Embodiment 1, specific data is introduced to illustrate the vehicle axle load measurement method based on load state reconstruction in Embodiment 1.

[0079] This example illustrates a scenario where a 4-axle vehicle is being weighed.

[0080] like Figure 3 The image shows the SS(t) signal of a 4-axle vehicle passing through an axle scale and the sequence of vehicle motion events detected by the axle counter. The blue curve represents the SS(t) signal, the red string Ai indicates the start point of the i-th vehicle motion event, and the black string Bi indicates the end point of the i-th vehicle motion event. Red / black dashed lines below indicate the precise time of each event. In other words, this vehicle passage generated a total of 8 events: 4 entry events {v1, v3, v5, v6} and 4 exit events {v2, v4, v7, v8}.

[0081] Sorted vehicle motion event sequence E sort ={v1, v2, v3, v4, v5, v6, v7, v8}.

[0082] In this embodiment, the stable load state matrix H is a 5 (number of events) × 4 (number of axles) matrix. The 5 row vectors hi of the stable load state matrix H correspond to the stable load states between vehicle motion events v1 / v2, v3 / v4, v5 / v6, v6 / v7, and v7 / v8, respectively.

[0083] Specifically, the elements in the first row and first column, the second row and second column, the third row and third column, the fourth row and third column, the fourth row and fourth column, and the fifth row and fourth column are all 1, and the remaining elements are all 0.

[0084] In this embodiment, the stable load state matrix H, as described in Embodiment 1, should be an 8×4 matrix. However, in the actual implementation, the time periods between three vehicle motion events were discarded. The v2 / v3 time period was discarded because the events v2 and v3 partially overlapped in time, resulting in poor stability of the ss(t) signal. The v4 / v5 and v8-end time periods were discarded because they did not contain any axle load, thus resulting in a 5×4 matrix.

[0085] Using the confidence level c(i) as equal to the duration w(i), the resulting confidence matrix is ​​a 5×5 diagonal matrix, where the diagonal elements are 2073 (first row, first column), 1615 (second row, second column), 247 (third row, third column), 786 (fourth row, fourth column) and 281 (fifth row, fifth column).

[0086] The average weight vector Y on the scale is a 5×1 single-column matrix, with each row containing the elements 71116, 120890, 72394, 198710, and 95919.

[0087] Solving the state-space mapping model yields an optimal solution that is a 4×1 single-column matrix. The elements in the first to fourth rows are 71116, 120890, 86253, and 108100, respectively; that is, the weight of the first axle is 71116, the weight of the second axle is 120890, the weight of the third axle is 86253, and the weight of the fourth axle is 108100.

[0088] Using this invention, a measurement error of < 0.5% can be achieved.

[0089] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any way. Other variations and modifications are possible without departing from the technical solutions described in the claims.

Claims

1. A method for measuring vehicle axle load based on load state reconstruction, characterized in that, include: S1: Acquire the weighing signal, detect vehicle motion events based on the weighing signal, and generate a vehicle motion event sequence; S2: Based on the vehicle motion event sequence, the continuous channel signal is divided into multiple stable load states, and a stable load state matrix H is constructed; S3: Based on the stable load state matrix, establish a state-space mapping model between vehicle axle load parameters and stable load state observations; S4: Construct a state confidence matrix to weight the state-space mapping model, solve the weighted state-space mapping model, obtain the optimal solution with the minimum observation error, and obtain the vehicle axle load.

2. The vehicle axle load measurement method based on load state reconstruction according to claim 1, characterized in that, S2 includes: The event sequence is reordered in ascending order of the start time of the vehicle motion events to obtain the sorted vehicle motion event sequence. Starting from the first event in the sorted sequence of vehicle motion events, assign a stable load state value to the time interval between every two adjacent events; Construct a stable load state matrix H. The number of rows in the stable load state matrix is ​​the total number of vehicle motion events, and the number of columns is the total number of axles of the vehicle under test. The elements of the stable load state matrix H are whether the axle is on the scale in state i.

3. A vehicle axle load measurement method based on load state reconstruction according to claim 1 or 2, characterized in that, The state-space mapping model is as follows: the observed average weight vector Y on the scale is equal to the sum of the product of the stable load state matrix H and the axle load vector X and the measurement error noise vector N; the average weight vector Y on the scale is a single-column matrix, and the elements y in the average weight vector Y on the scale... i The axle load vector X represents the average weight on the scale under stable load state i; the axle load vector X is a single-column matrix, and the elements x in the axle load vector X are... i The weight of the i-th axle is represented; the measurement error noise vector N is a single-column matrix, and the elements N in the measurement error noise vector N are... i The noise of the observed weight on the scale represents the stable load state i.

4. The vehicle axle load measurement method based on load state reconstruction according to claim 1, characterized in that, The state confidence matrix is ​​a diagonal matrix, with the number of rows and columns being the total number of vehicle motion events. The elements on the diagonal represent the confidence level of the measurement data for the stable load state.

5. A method for measuring vehicle axle load based on load state reconstruction according to claim 1, 2, or 4, characterized in that, Construct constraints and verify the optimal solution. The constraints include: all axle loads are positive numbers and all axle loads are less than or equal to a preset axle load threshold. If all components of the optimal solution satisfy the constraints, the optimal solution is a feasible solution; otherwise, the optimal solution is rejected, and a new round of weighing / unweighing event identification and analysis is triggered.

6. A method for measuring vehicle axle load based on load state reconstruction according to claim 1, 2, or 4, characterized in that, The weighing signal includes the output of the pressure sensor, the output s1(t) of the feed sensor pair, the output s2(t) of the output sensor pair, the output s3(t) of the feed counter axis, the output s4(t) of the output counter axis, and the output ss(t) of the platform sensor and channel. The output of the channel is the sum of the output s1(t) of the feed sensor pair and the output s2(t) of the output sensor pair.

7. The vehicle axle load measurement method based on load state reconstruction according to claim 6, characterized in that, Vehicle motion events are detected from the outputs s3(t) of the weighing input counter and s4(t) of the weighing output counter. These events include axle weighing input events and axle weighing output events. Each vehicle motion event includes an event type, an event start time, an event end time, and an event intensity, where the event intensity is a rough estimate of the axle weight corresponding to the event.

8. The vehicle axle load measurement method based on load state reconstruction according to claim 3, characterized in that, The number of elements in the axle load vector X is equal to the total number of axles of the vehicle under test, and the number of elements in the average weight vector Y on the scale and the measurement error noise vector N are equal to the total number of vehicle motion events.

9. The vehicle axle load measurement method based on load state reconstruction according to claim 4, characterized in that, For each stable load state i, calculate the duration of the stable load state, the mean weight on the scale under the stable load state, and the sample standard deviation; calculate the confidence level of the measurement data for the stable load state based on the duration, the mean weight on the scale under the stable load state, and the sample standard deviation.

10. The vehicle axle load measurement method based on load state reconstruction according to claim 9, characterized in that, The duration of the current stable load state is the time from the end of the current weighing event to the start of the next vehicle movement event; if the duration is 0, a valid measurement sample of the stable load state cannot be obtained.

Images