Fault self-adaptive method of vehicle dynamic weighing system based on multi-mode fusion

Abstract

The application relates to a fault self-adaptive method of a vehicle dynamic weighing system based on multi-mode fusion. The method mainly comprises the following steps: collecting output signals of weighing sensors on each weighing platform, and converting the output signals into corresponding weight data; based on the weight data, vehicle total weight values in various weighing modes are calculated according to preset weighing modes; when it is detected that a weighing sensor fault exists in a weighing platform, a weighing mode depending on the fault weighing platform is marked as an invalid mode, and a weighing mode not depending on the fault weighing platform is marked as a valid mode; the total weight values calculated by the valid modes are weighted and fused to obtain a final vehicle measurement total weight value and output. The application can automatically exclude abnormal modes and fuse weighing data of the remaining valid modes when a sensor fault occurs, so that the weighing precision is maintained without interrupting the lane passage.

Description

Technical Field

[0001] This application relates to the field of vehicle weighing technology, and in particular to a fault adaptive method for a vehicle dynamic weighing system based on multi-mode fusion. Background Technology

[0002] In the field of highway traffic management, dynamic vehicle weighing equipment (such as axle weighing scales and axle group scales) is widely used for vehicle weighing at highway entrances.

[0003] Existing axle weighing scales or axle group scales typically operate using a single weighing mode, such as a single-platform axle group scale or a dual-platform axle group scale (one large and one small). During actual operation, the load cells on the weighing platform may experience data anomalies or malfunctions due to prolonged use, environmental factors, or sudden failures. Since existing equipment usually relies on a single weighing mode, if any load cell on that mode fails, the entire weighing system will be unable to output valid weighing results, forcing the closure of the corresponding lane and severely impacting highway traffic efficiency and equipment availability.

[0004] Based on the above situation, how to solve the technical problem that the vehicle dynamic weighing system cannot output effective results when the weighing sensor fails is a problem to be solved in this technical field. Summary of the Invention

[0005] To address the aforementioned deficiencies or improvement needs of existing technologies, this application provides a fault adaptive method for a vehicle dynamic weighing system based on multi-mode fusion. This method constructs multiple parallel weighing modes based on multiple weighing platforms and employs a sensor fault detection and weighing mode adaptive degradation fusion strategy. When a sensor malfunctions, abnormal modes are automatically eliminated and the weighing data of the remaining valid modes are fused, thereby maintaining weighing accuracy without interrupting lane traffic.

[0006] The embodiments of this application adopt the following technical solutions: This application provides a fault adaptive method for a vehicle dynamic weighing system based on multi-mode fusion, applicable to a weighing system in which multiple weighing platforms are sequentially arranged adjacent to each other along the vehicle's direction of travel, each weighing platform being equipped with an independently connected weighing sensor; the method includes: The output signals of the weighing sensors on each weighing platform are collected and converted into corresponding weight data. Based on the weight data, the total weight of the vehicle in each preset weighing mode is calculated. The weighing platforms used in different weighing modes are not exactly the same. The output signals of each weighing sensor are monitored in real time. Based on the preset fault judgment conditions, it is determined whether there is a weighing sensor fault and the weighing platform where the faulty weighing sensor is located is identified. When a load cell malfunction is detected on a certain weighing platform, the weighing mode that depends on the malfunctioning weighing platform is marked as an invalid mode, and the weighing mode that does not depend on the malfunctioning weighing platform is marked as an effective mode. The total weight values ​​calculated from the effective modes are weighted and fused to obtain the final measured total weight value of the vehicle, which is then output.

[0007] Through the above technical solution, based on the parallel operation of multiple weighing modes, the faulty sensor platform is located by real-time fault detection. The mode that depends on the faulty platform is automatically excluded and only the weighing data of the effective mode is fused. This realizes the adaptive degradation and uninterrupted weighing of the weighing system when the sensor fails, and solves the technical problem of lane closure and equipment unavailability caused by sensor failure in the prior art.

[0008] In some embodiments, the real-time monitoring of the output signals of each weighing sensor and the determination of whether a weighing sensor fault exists based on preset fault determination conditions specifically include: The output signal of each weighing sensor is periodically sampled, and the fluctuation amplitude of the output signal of each weighing sensor within a preset time window is calculated. When the fluctuation amplitude exceeds the preset normal fluctuation threshold range, or when the output signal of the weighing sensor remains at zero for more than a preset duration, the weighing sensor is determined to be faulty.

[0009] By periodically sampling the sensor output signal and setting multiple judgment conditions such as fluctuation threshold and zero value duration, faulty sensors can be identified in a timely and accurate manner, avoiding unnecessary downgrading due to misjudgment or data errors caused by missed judgment.

[0010] In some implementations, the weighted fusion of the total weight values ​​calculated from the effective modes to obtain the final measured total weight value of the vehicle and its output specifically includes: Based on the historical weighing accuracy statistics of each effective mode and the number of weighing platforms used, weighting coefficients are assigned to each effective mode. The final total weight value of the vehicle is obtained by weighting the total weight values ​​of each effective mode based on the weighting coefficients.

[0011] By combining historical accuracy statistics and the number of platforms for differentiated weighting, the more comprehensive and accurate mode contributes more to the final result, thereby improving the accuracy and stability of the fused weighing result.

[0012] In some implementations, the allocation rules for the weighting coefficients include: When all weighing modes are valid, the weighing mode that relies on the most weighing platform has the largest weight coefficient, and the weighing mode that relies on the fewest weighing platform has the smallest weight coefficient. When a weighing mode is marked as invalid, the weight coefficient of the invalid mode is set to zero, and the weight coefficients of the remaining valid modes are normalized.

[0013] By employing the above technical solution, the mode with the most platforms and the most comprehensive information is given the highest weight, and the total weight is kept to 1 through normalization during degradation. This ensures that the fusion result maintains a reasonable accuracy distribution and achieves a smooth transition, regardless of whether it is in a normal or degraded state.

[0014] In some implementations, a consistency verification step is also included: Before weighted fusion of the total weight values ​​calculated by the effective modes, a consistency check is performed on the total weight values ​​calculated by each effective mode. Calculate the deviation between the total weight values ​​of each effective mode. When the deviation between the total weight value of a certain effective mode and the average total weight value of the other effective modes exceeds the preset deviation threshold, the effective mode is marked as an abnormal mode and excluded from the fusion calculation.

[0015] Through the above technical solution, consistency verification can further identify patterns where the data is obviously abnormal even though no fault judgment has been triggered, based on fault detection, thus achieving dual protection and effectively avoiding weighing data deviation caused by soft faults such as sensor output drift.

[0016] In some embodiments, the weighing platform includes a first weighing platform, a second weighing platform, and a third weighing platform, wherein the length of the third weighing platform along the vehicle's travel direction is greater than the lengths of the first and second weighing platforms along the vehicle's travel direction, and the lengths of the first and second weighing platforms along the vehicle's travel direction are the same; the method further includes an axle group identification step: As the vehicle passes over the third weighing platform, the number of axles and axle group composition information of the vehicle are identified based on the time-series change waveform of the output signal of the weighing sensor on the third weighing platform. The number of shafts and shaft group configuration information are used to correct the total weight value calculation under each weighing mode.

[0017] By using the above technical solution, the axle group is identified by utilizing a longer third weighing platform, and the vehicle structure information is introduced into the weighing calculation. This allows the total weight value calculation of each mode to be specifically corrected for different vehicle models, thereby improving the weighing adaptability to vehicles with different axle group configurations.

[0018] In some implementations, a sensor degradation prediction and mode weight pre-adjustment step is also included: Based on the real-time monitoring of the output signals of each weighing sensor, a performance degradation trend model for each weighing sensor is established. The performance degradation trend model calculates the degradation trend index of each weighing sensor based on at least one of the zero drift, sensitivity change rate, and signal-to-noise ratio attenuation trend of each weighing sensor within a preset statistical period. When the degradation trend index of a certain weighing sensor exceeds the preset degradation warning threshold but the fault judgment condition has not yet been triggered, the weighing modes associated with the weighing platform where the weighing sensor is located are marked as warning modes. During weighted fusion, the weight coefficient of the early warning mode is gradually reduced according to the severity of the degradation trend index, while the weight coefficient of the non-early warning mode is increased accordingly, so that the fusion result gradually reduces its dependence on the weighing platform where the deteriorated weighing sensor is located.

[0019] By using the above technical solution, when the sensor has not yet experienced actual failure but its performance has already shown a trend of degradation, the reliance on the degraded sensor data can be reduced in advance by gradually adjusting the mode weights. This avoids abrupt switching between the normal state and the fault-degraded state, and ensures that the weighing results maintain a smooth transition throughout the entire process of sensor degradation. This improves the weighing stability and data continuity of the system in the scenario of gradual sensor degradation.

[0020] In some implementations, a remote monitoring step is also included: The operating status information of each weighing sensor, the calculation results of each weighing mode, the fault detection results, and the mode degradation records are uploaded to the remote monitoring platform through the communication interface. The remote monitoring platform generates equipment operation status reports and maintenance reminders based on the received information.

[0021] By using the above technical solution, the status of each sensor, the calculation results of each mode, and the fault and degradation records are uploaded to the remote monitoring platform in real time, enabling managers to remotely monitor the equipment's operating status and fault conditions in real time, which facilitates the formulation of maintenance plans and improves operation and maintenance efficiency.

[0022] In some implementations, the remote monitoring step further includes: Record the time, faulty weighing sensor identifier, fault type, and valid mode combinations before and after each mode degradation event; Based on the accumulated mode degradation event records, a weighing sensor health assessment report is generated, which is used to guide the preventive maintenance of the equipment.

[0023] By using the above technical solution, and by recording detailed information of each degradation event and generating a sensor health assessment report, it is possible to provide maintenance personnel with sensor degradation trend analysis data, support condition-based preventive maintenance strategies, and reduce the impact of sudden failures on heavy services.

[0024] In some implementations, an accuracy confidence assessment step is also included: The accuracy confidence score of the current weighing result is calculated based on multiple evaluation factors, including the number of currently valid modes, the number of faulty weighing sensors detected, the number of excluded abnormal modes, and the vehicle's passing speed. The accuracy confidence score is used to query the pre-stored accuracy level mapping table to obtain the accuracy confidence level of the current weighing result. The accuracy level mapping table is established based on the measured error statistics under different working conditions during the system calibration stage. When the accuracy confidence level is lower than the preset minimum acceptable level, an accuracy alarm signal is output.

[0025] The above technical solution comprehensively evaluates the current weighing conditions from multiple dimensions, such as the number of effective modes, the number of faulty sensors, the number of abnormal modes eliminated, and the vehicle speed, rather than relying solely on the consistency of results between modes. This makes the accuracy assessment and consistency verification complementary: consistency verification solves the problem of "which modes' data are available", while accuracy confidence assessment solves the problem of "whether the available data under the current conditions is sufficient to support the accuracy requirements". Both ensure the reliability of the weighing results from two different levels: data quality and operating conditions.

[0026] In summary, this application includes at least the following beneficial technical effects: 1. Based on the parallel operation of multiple weighing modes, the faulty sensor platform is located by real-time fault detection. The modes that depend on the faulty platform are automatically excluded and only the weighing data of the effective modes are fused. This realizes the adaptive degradation and uninterrupted weighing of the weighing system when the sensor fails, and solves the technical problem of lane closure and equipment unavailability caused by sensor failure in the existing technology. 2. Through consistency verification, it is possible to further identify patterns where the data is obviously abnormal even though no fault judgment has been triggered, based on fault detection, thus achieving dual protection and effectively avoiding weighing data deviation caused by soft faults such as sensor output drift. 3. The current weighing conditions are comprehensively evaluated from multiple dimensions, such as the number of valid modes, the number of faulty sensors, the number of abnormal modes eliminated, and the vehicle speed, rather than relying solely on the consistency of results between modes. This makes the accuracy assessment and consistency verification complementary: consistency verification solves the problem of "which modes' data are available", while accuracy confidence assessment solves the problem of "whether the available data under the current conditions is sufficient to support the accuracy requirements". Both of them ensure the reliability of the weighing results from two different levels: data quality and operating conditions. Attached Figure Description

[0027] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the embodiments of this application will be briefly described below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0028] Figure 1 A flowchart illustrating a fault adaptive method for a vehicle dynamic weighing system based on multi-mode fusion, provided for embodiments of this application; Figure 2 This is a schematic diagram of the weighing system provided in an embodiment of this application; Figure 3 This is a schematic diagram illustrating the mode downgrade correspondence provided in the embodiments of this application; Figure 4 This is a schematic diagram of weighted fusion calculation provided in an embodiment of this application; Figure 5 A detailed flowchart of the accuracy confidence assessment steps provided for embodiments of this application; Figure 6 A flowchart illustrating the sensor degradation prediction and mode weight pre-adjustment steps provided in this application embodiment.

[0029] In the diagram: 1. First weighing platform; 2. Second weighing platform; 3. Third weighing platform; 4. Control module; 5. Light curtain. Detailed Implementation

[0030] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application. Furthermore, the technical features involved in the various embodiments described below can be combined with each other as long as they do not conflict with each other.

[0031] The present application will now be described in detail with reference to the accompanying drawings and embodiments.

[0032] Example 1: It should be noted that existing vehicle dynamic weighing systems typically operate in a single weighing mode. When the sensors on the weighing platform malfunction, this single mode cannot output valid weighing results, leading to lane closures and impacting highway traffic efficiency.

[0033] To address the aforementioned issues, Embodiment 1 of this application provides a fault adaptive method for a vehicle dynamic weighing system based on multi-mode fusion. This method is applied to a weighing system in which multiple weighing platforms are sequentially arranged adjacent to each other along the vehicle's driving direction, and each weighing platform is equipped with an independently connected weighing sensor.

[0034] refer to Figure 2 As shown, in some embodiments, the weighing system includes a first weighing platform 1, a second weighing platform 2, and a third weighing platform 3, which are arranged adjacent to each other along the vehicle's travel direction. Optionally, the length of the first weighing platform 1 along the vehicle's travel direction is 1 meter, the length of the second weighing platform 2 along the vehicle's travel direction is 1 meter, and the length of the third weighing platform 3 along the vehicle's travel direction is 4 meters, with a total length of 6 meters. The widths of the first weighing platform 1, the second weighing platform 2, and the third weighing platform 3 are the same. Each weighing platform is equipped with multiple independently connected weighing sensors, and each weighing sensor is connected to the weighing instrument, i.e., the control module 4, through an independent data channel. The control module 4 may include a data acquisition submodule, a synchronous calculation submodule, a data fusion submodule, and a data analysis submodule.

[0035] refer to Figure 1 The method of this embodiment specifically includes the following steps: Step S1: Data Acquisition. When the vehicle enters the weighing area, it passes through the first weighing platform 1, the second weighing platform 2, and the third weighing platform 3 in sequence. The data acquisition submodule acquires the output signal (voltage signal) of each weighing sensor on each weighing platform and converts the voltage signal into the corresponding digital format weight data through analog-to-digital conversion. Since each weighing sensor is independently connected, the data acquisition submodule can obtain the independent weight data of each sensor.

[0036] Step S2: Parallel computation step. The synchronous computation submodule receives the weight data output by the data acquisition submodule and simultaneously calculates the total vehicle weight for each of the four preset weighing modes: First Mode (Mode 0): The Mode 0 unit calculates the total weight value W0 using the weight data from the first weighing platform 1, the second weighing platform 2, and the third weighing platform 3. This mode constitutes a three-platform shaft group scale mode, providing the most comprehensive information.

[0037] Second mode (mode 1): The mode 1 unit calculates the total weight value W1 by using the weight data from the first weighing platform 1 and the second weighing platform 2. This mode constitutes a dual-platform axle load cell.

[0038] The third mode (mode 2): The mode 2 unit calculates the total weight value W2 using the weight data from the second weighing platform 2 and the third weighing platform 3. This mode constitutes a large and small axle group scale.

[0039] The fourth mode (mode 3): The total weight value W3 of the fourth mode is calculated by the mode 3 unit using the weight data of a single platform of the third weighing platform 3. This mode constitutes a single-platform shaft group scale mode.

[0040] The above four modes run in parallel and perform real-time synchronous calculations.

[0041] Step S3: Fault Detection Step. The data analysis submodule monitors the output signals of each weighing sensor in real time. Specifically, the data analysis submodule periodically samples the output signal of each weighing sensor and calculates the fluctuation amplitude of each sensor's output signal within a preset time window (e.g., the most recent 10 seconds). When the fluctuation amplitude of a sensor exceeds a preset normal fluctuation threshold range (e.g., the fluctuation amplitude exceeds ±5% of the sensor's full scale, i.e., the normal fluctuation threshold range is set to ±5% of the full scale), the sensor is determined to be faulty. Furthermore, when the output signal of a sensor remains at zero for more than a preset duration (e.g., 3 seconds), the sensor is also determined to be faulty. Based on the above determination conditions, the weighing platform where the faulty sensor is located is identified.

[0042] For example, when a sensor on the first weighing platform 1 is determined to be faulty, the first weighing platform 1 is determined to be a faulty platform.

[0043] Step S4: Mode Degradation Step. When a sensor fault is detected on the first weighing platform 1, the data analysis submodule marks the weighing mode dependent on the first weighing platform 1 as an invalid mode. Specifically: The first mode (mode 0) depends on the first weighing platform 1 and is marked as an invalid mode; The second mode (mode 1) depends on the first weighing platform 1 and is marked as an invalid mode; The third mode (mode 2) does not depend on the first weighing platform 1 and is retained as a valid mode; The fourth mode (mode 3) does not depend on the first weighing platform 1 and is retained as a valid mode.

[0044] At this point, the system automatically downgrades to a state where the third and fourth modes operate in parallel. The data analysis submodule records the time of this downgrade event, the fault sensor identifier (e.g., "first weighing platform - sensor 3"), the fault type (e.g., "signal remains at zero"), and the effective mode combinations before and after the downgrade (e.g., "before downgrade: modes 0 / 1 / 2 / 3 are all effective → after downgrade: modes 2 / 3 are effective").

[0045] Similarly, refer to Figure 3 The diagram illustrates the mode degradation relationship. If a sensor malfunctions on the second weighing platform 2, the first, second, and third modes dependent on the second weighing platform 2 are all marked as invalid, and only the fourth mode is retained. If a sensor malfunctions on the third weighing platform 3, the first, third, and fourth modes dependent on the third weighing platform 3 are marked as invalid, and only the second mode is retained.

[0046] Step S5: Fusion Output Step. The data fusion submodule performs weighted fusion of the total weight values ​​calculated from the valid modes. Under normal conditions where all modes are valid, weight coefficients are assigned to each mode based on historical weighing accuracy statistics and the number of weighing platforms used, and weighted fusion calculations are performed. (Refer to...) Figure 4 The diagram illustrates the weighted fusion calculation. Specifically, the first mode (using three platforms) has the largest weight coefficient α0, for example, α0 = 0.4; the second and third modes (each using two platforms) have weight coefficients α1 = α2 = 0.2; and the fourth mode (using one platform) has a weight coefficient α3 = 0.2. The final total vehicle weight value W = α0 × W0 + α1 × W1 + α2 × W2 + α3 × W3.

[0047] When a mode is marked as invalid, its weight coefficient is set to zero, and the weight coefficients of the remaining valid modes are normalized. For example, when a malfunction of the first weighing platform 1 renders the first and second modes invalid, only the third and fourth modes are valid. The normalized weights are α2'=α2 / (α2+α3)=0.5, α3'=α3 / (α2+α3)=0.5, and finally W=0.5×W2+0.5×W3.

[0048] It should be noted that in multi-mode parallel computing, in addition to sensor hard faults, environmental factors (such as temperature changes, vibration interference, etc.) may cause occasional deviations in the calculation results of a certain mode, even though the sensor itself has not experienced a clear fault. Based on this, a consistency verification step is performed before the fusion output: the deviation value between the total weight values ​​of each valid mode is calculated. When the deviation between the total weight value of a certain mode and the average total weight value of the other valid modes exceeds a preset deviation threshold (e.g., exceeding 3%), the mode is marked as an abnormal mode and excluded from the fusion calculation.

[0049] Specifically, the average weight value (Wavg) of each valid mode is calculated; the deviation ratio (δi) between the total weight value of each mode and the average value is calculated as δi = |Wi - Wavg| / Wavg × 100%. When the deviation ratio (δi) of a mode exceeds a preset deviation threshold (e.g., 3%), the mode is marked as an abnormal mode and excluded from the fusion calculation. After excluding abnormal modes, the weighted fusion result of the remaining valid modes is recalculated.

[0050] For example, under normal conditions, the total weight values ​​of the four modes are W0=30050kg, W1=29980kg, W2=30020kg, and W3=31200kg, with an average Wavg=30312.5kg. The deviation ratio of the fourth mode is δ3=|31200-30312.5| / 30312.5×100%≈2.93%. If the preset deviation threshold is 3%, the fourth mode is not excluded. However, if W3=31500kg, then δ3≈3.92%, exceeding the threshold, and the fourth mode is marked as an abnormal mode and excluded.

[0051] Consistency verification can further identify patterns where data is obviously abnormal but has not triggered a hard fault determination, based on fault detection, thus achieving dual protection and effectively avoiding the impact of occasional data deviations on the final weighing results.

[0052] Step S6: Accuracy Confidence Assessment Step. It should be noted that the consistency verification step described above addresses the issue of "whether there are outliers in the results of each mode," which is a screening step in the data quality dimension. However, even if the results of each mode pass the consistency verification, the overall operating conditions of the current weighing process (such as too few available modes, excessive vehicle speed, etc.) may still result in the weighing accuracy failing to meet requirements. Therefore, refer to... Figure 5 The accuracy confidence assessment step is used to evaluate from the perspective of operating conditions, which complements the consistency verification.

[0053] S6a: Calculate the accuracy confidence score. The data analysis submodule calculates the accuracy confidence score S of the current weighing result based on the following four evaluation factors: (1) Effective pattern quantity factor F1. This reflects the number of currently available independent data sources. The more effective patterns there are, the more sufficient data support the fusion results will be. For example, when the number of effective patterns n is 4, 3, 2, and 1, the F1 values ​​are 1.0, 0.8, 0.5, and 0.2, respectively.

[0054] (2) Fault sensor quantity factor F2. It reflects the current health status of the weighing hardware. Even if some modes are still valid, the more faulty sensors there are, the greater the computing load borne by the remaining sensors, and the lower the data redundancy. Let the total number of sensors be Ntotal, and the number of faulty sensors be Nfault, then: F2=1-Nfault / Ntotal; For example, there are 16 sensors in the system, 2 of which are faulty, then F2=1-2 / 16=0.875.

[0055] (3) Abnormal mode elimination factor F3. It reflects the degree of data quality loss in the consistency check step. After the fault degradation step S4, originally valid modes are eliminated in the consistency check, which indicates that there is data abnormality caused by non-fault reasons (such as temporary interference like uneven road surface, wind load, eccentric load, etc.), and the accuracy confidence should be further reduced in this case. For example, when the number of abnormal modes eliminated in the consistency check is 0, 1, and ≥2 respectively, the values of F3 are 1.0, 0.7, and 0.4 respectively.

[0056] (4) Vehicle passing speed factor F4. The error of dynamic weighing is related to the vehicle speed. The higher the vehicle speed is, the shorter the dynamic action time between the vehicle and the weighing platform is, the fewer the number of signal acquisition points is, and the greater the dynamic impact effect is; however, excessively low vehicle speed may introduce new error sources such as increased unsteady driving state, impaired signal sampling integrity, and amplified influence of external interference. Based on this, let the vehicle passing speed be v (km / h), the nominal upper speed limit designed for the system be Vmax (e.g., 120 km / h), the nominal lower speed limit be Vmin (e.g., 5 km / h), and the optimal accuracy speed range be Vlow to Vhigh (e.g., 20~80 km / h), then: When Vlow≤v≤Vhigh, F4=1.0; When v>Vhigh, F4=1-(v-Vhigh) / (Vmax-Vhigh); for example, when v=100km / h, F4=1-(100-80) / (120-80)=0.5; When v<Vlow, F4=1-(Vlow-v) / (Vlow-Vmin); for example, when v=10km / h, F4=1-(20-10) / (20-5)=0.33.

[0057] The comprehensive accuracy confidence score S is calculated by weighting: S=β1×F1+β2×F2+β3×F3+β4×F4; Where β1, β2, β3, and β4 are the weight coefficients of each factor, satisfying β1+β2+β3+β4=1. In this embodiment, β1=0.4, β2=0.2, β3=0.2, and β4=0.2, meaning that the effective mode quantity factor has the largest weight, because the number of available data sources has the most direct impact on fusion accuracy.

[0058] S6b: Query the accuracy level mapping table. During the system calibration phase, multiple standard load vehicles of known weight are used to pass through the weighing system multiple times under different working conditions (different number of modes, artificially simulated number of sensor failures, different vehicle speed conditions) to statistically analyze the measured weighing errors under various working condition combinations.

[0059] Based on the statistical results, a mapping relationship between the accuracy confidence score S and the accuracy level is established: For example, when the accuracy confidence score S is: S≥0.8; the corresponding measured error statistics (95% confidence interval) is: error ≤3%; the corresponding accuracy confidence level is A; the corresponding meaning is: high reliability, meeting the national standard level 1 accuracy. When the accuracy confidence score S is: 0.6≤S<0.8; the corresponding measured error statistics (95% confidence interval) is: error ≤5%; the corresponding accuracy confidence level is B; the corresponding meaning is: reliable, meeting the national standard level 1 accuracy. When the accuracy confidence score S is: 0.4≤S<0.6; the corresponding measured error statistics (95% confidence interval) is: error 5%~8%; the corresponding accuracy confidence level is C; the corresponding meaning is: accuracy reduced, approaching or exceeding the national standard level 1. When the accuracy confidence score S is: S<0.4; the corresponding measured error statistics (95% confidence interval) is: error >8%; the corresponding accuracy confidence level is D; the corresponding meaning is: unreliable. The permissible error for the vehicle weight corresponding to the Level 1 accuracy of dynamic weighing as specified in the Chinese standard GB / T21296 is ±5%. Levels A and B meet the Level 1 accuracy requirements of the national standard; Level C has insufficient accuracy assurance; and Level D results are unreliable.

[0060] The above correspondence is saved in the form of a precision level mapping table during actual use.

[0061] S6c: Accuracy Alarm Judgment. The minimum acceptable accuracy confidence level is set to B. A or B: Output the total vehicle weight W normally. C: Output W along with a "Reduced Accuracy" flag, indicating a decrease in the reliability of the weighing result. D: Output an accuracy alarm signal, displaying alarm information (including the current values ​​of each evaluation factor, accuracy confidence score S, and level) on the remote monitoring platform, and adding an "Unreliable" flag to the weighing result.

[0062] Complete calculation example: Example 1 (Normal operating condition): All four modes are valid (n=4, F1=1.0), no sensor faults (Nfault=0, F2=1.0), consistency check does not exclude any mode (F3=1.0), vehicle speed v=60km / h (within the optimal range, F4=1.0).

[0063] S = 0.4 × 1.0 + 0.2 × 1.0 + 0.2 × 1.0 + 0.2 × 1.0 = 1.0. Looking up the table, we get grade A, so the output is normal.

[0064] Example 2 (Partial Degradation Condition): A sensor on the second weighing platform 2 malfunctions, causing the first mode to be marked as invalid, and only the second, third, and fourth modes are valid (n=3, F1=0.8); there is one faulty sensor (Nfault=1, F2=1-1 / 16=0.9375); the consistency check did not eliminate the abnormal mode (F3=1.0); the vehicle speed v=70km / h (within the optimal range, F4=1.0).

[0065] S = 0.4 × 0.8 + 0.2 × 0.9375 + 0.2 × 1.0 + 0.2 × 1.0 = 0.32 + 0.1875 + 0.2 + 0.2 = 0.9075. Looking up the table, we get Grade A, so the output is normal.

[0066] Example 3 (Severe Degradation + High-Speed ​​Operation): Two sensor failures result in only the third and fourth modes being effective (n=2, F1=0.5); 2 faulty sensors (Nfault=2, F2=0.875); Consistency check also eliminates 1 abnormal mode (F3=0.7); Vehicle speed v=100km / h (F4=0.5).

[0067] S = 0.4 × 0.5 + 0.2 × 0.875 + 0.2 × 0.7 + 0.2 × 0.5 = 0.2 + 0.175 + 0.14 + 0.1 = 0.615. Looking up the table, we get grade B, which is normal output, but it is close to the boundary of grade C.

[0068] Example 4 (Extreme Degradation Condition): Multiple sensor failures result in only the fourth mode being effective (n=1, F1=0.2); 4 faulty sensors (Nfault=4, F2=0.75); consistency check excludes 2 modes (F3=0.4); vehicle speed v=110km / h (F4=0.25).

[0069] S = 0.4 × 0.2 + 0.2 × 0.75 + 0.2 × 0.4 + 0.2 × 0.25 = 0.08 + 0.15 + 0.08 + 0.05 = 0.36. From the table, we get level D, and output an accuracy alarm signal.

[0070] In combination with the aforementioned steps, the consistency check ensures that "the data used for fusion is clean," and the S6 accuracy confidence assessment step ensures that "the fusion result is reliable under the current conditions."

[0071] Step S7: Remote Monitoring Step. The operating status information of each weighing sensor, the calculation results of each weighing mode, the fault detection results, and the mode degradation records are uploaded to the remote monitoring platform via the communication interface. The remote monitoring platform generates an equipment operating status report and maintenance reminders based on the received information. Simultaneously, based on the accumulated mode degradation event records, a sensor health assessment report is generated to guide preventative equipment maintenance.

[0072] Specifically, the control module 4 (weighing instrument, such as the XKL-3016 model) in the weighing system enables completely independent access to each weighing sensor. The control module 4 uploads the operating status information of each weighing sensor (including real-time output signal values, zero-point drift, sensitivity changes, etc.), the calculation results of each weighing mode (total weight value of each mode, and the fused total weight value), fault detection results (faulty sensor identifier, fault type), and mode degradation records (degradation time, mode combination before and after degradation) to the remote monitoring platform via a communication interface (such as an Ethernet interface or a 4G / 5G communication module).

[0073] The remote monitoring platform generates equipment operation status reports and maintenance reminders based on the received information. Simultaneously, it generates sensor health assessment reports based on accumulated mode degradation event records. For example, if a sensor triggers a fault determination multiple times recently or causes mode degradation, the sensor health assessment report marks the sensor as "replacement recommended" and sends a maintenance reminder to management personnel.

[0074] Managers can remotely view the real-time parameters of each weighing sensor, the overall status of the weighing platform, and various parameters of the weighing instruments through the remote monitoring platform, so as to have a comprehensive understanding of the operating status, vehicle passage conditions, and maintenance plans of the axle weighing system.

[0075] By collecting and displaying equipment operation data in real time through a remote monitoring platform, managers can have a comprehensive understanding of the equipment status without going to the site, enabling them to formulate maintenance plans in a timely manner, achieve condition-based preventive maintenance, improve operation and maintenance efficiency, and reduce operation and maintenance costs.

[0076] Based on the above solution, this application utilizes parallel operation of multiple weighing modes and real-time fault detection. When a sensor on any weighing platform malfunctions, the system can automatically identify the fault and eliminate weighing modes that rely on the faulty platform, fusing only the data from the valid modes to achieve uninterrupted weighing. Even after downgrading, the weighing system still achieves the national standard Level 1 weighing accuracy, without affecting lane traffic, significantly improving the equipment's availability and fault tolerance.

[0077] Example 2: It should be noted that different vehicle models have different numbers of axles and axle group configurations. Using a uniform weighing calculation method to calculate the weight of all vehicle models may result in insufficient weighing accuracy for certain special axle group configurations (such as triple axle groups).

[0078] In this embodiment, the length of the third weighing platform 3 along the vehicle's travel direction is 4 meters, which is much greater than the 1-meter length of the first weighing platform 1 and the second weighing platform 2. As the vehicle passes over the third weighing platform 3, the number of axles and axle group configuration information of the vehicle are identified based on the time-series change waveform of the weighing sensor output signal on the third weighing platform 3.

[0079] Specifically, when a vehicle passes over the third weighing platform 3, each axle's entry and exit from the scale will generate corresponding peaks in the output signal of the weighing sensor. The number of axles is determined by detecting the number of peaks, and the distance between adjacent axles is determined by analyzing the time interval between adjacent peaks, thus identifying the axle group configuration (e.g., single axle, double axle group, triple axle group, etc.). For example, when 1.0 m < wheelbase ≤ 1.8 m: it is determined to be a double axle group; when the adjacent wheelbases of three consecutive axles all satisfy 1.0 m < wheelbase ≤ 1.8 m: it is determined to be a triple axle group; when the wheelbase > 2.5 m: it is considered an independent single axle; when the wheelbase is in the range of 1.8 m to 2.5 m, it is determined to be an independent single axle.

[0080] The identified number of shafts and shaft group composition information are used to correct the total weight calculation in each weighing mode. For example, for double or triple shaft groups, in the second mode (one large and one small shaft group weighing mode) and the fourth mode (single platform shaft group weighing mode), there may be cases where the shafts in the shaft group do not pass through the weighing platform in complete sequence. In this case, based on the identified shaft group composition information, a conventional shaft group overall weighing correction algorithm is used to calculate the weight of the corresponding shaft group as a whole to compensate for the weighing deviation caused by the partial pressure on each shaft in the shaft group.

[0081] The above scheme utilizes a longer third weighing platform for axle group identification and incorporates vehicle structure information into the weighing calculation, enabling the total weight calculation of each mode to be specifically corrected for different vehicle models, thereby improving the weighing adaptability and accuracy for vehicles with different axle group configurations.

[0082] Optional, see reference Figure 2Light curtains 5 are installed on both sides near the intersection of the first weighing platform 1 and the second weighing platform 2, and the light curtains 5 are arranged perpendicular to the vehicle's travel direction. The light curtains 5 are used to constrain the vehicle's travel trajectory and verify whether the axle is completely within the lateral weighing area of ​​the first weighing platform 1 and the second weighing platform 2, so as to avoid weighing errors caused by uneven vehicle loading or oblique passage; at the same time, the timing signal output by the light curtain is synchronized with the weighing sensor data, which is used to help identify the axle group boundary and match the vehicle's travel speed, and to provide a trigger basis for switching the weighing mode and correcting the total weight value using the axle group overall weighing correction algorithm.

[0083] Example 3: It should be noted that in the aforementioned Example 1, mode degradation was triggered only after the sensor was determined to be faulty, which is a reactive response. However, in actual operation, the performance degradation of weighing sensors is usually a gradual process (such as increased zero-point drift and gradual decrease in sensitivity due to long-term use). During the transition period when the sensor performance has significantly deteriorated but has not yet reached the fault determination condition, the data quality of the weighing mode relying on that sensor has already decreased, but the system still uses these data for fusion with normal weights, which may affect the accuracy of the fusion results. Once the sensor finally reaches the fault determination condition and triggers degradation, the mode weights will undergo abrupt adjustments, causing a jump in the weighing results of continuous vehicle passages.

[0084] To address the aforementioned issues, this embodiment 3 adds a sensor degradation prediction and mode weight pre-adjustment step to the fault detection step S3, referencing... Figure 6 Specifically, it includes: S3a: Establish a sensor performance degradation trend model. Based on continuous monitoring data of each weighing sensor, the data analysis submodule establishes a performance degradation trend model for each sensor. Specifically, the data analysis submodule continuously records the following performance parameters of each sensor within a preset statistical period (e.g., every 24 hours as one statistical period): (1) Zero drift: The amount of deviation of the sensor output signal from the calibrated zero point when no vehicle passes by. The zero drift of a normal sensor should fluctuate within a very small range. When the zero drift gradually increases over time, it indicates that the sensor performance is deteriorating.

[0085] (2) Sensitivity change rate: The ratio of the deviation between the sensor output signal amplitude and the calibrated sensitivity under similar load conditions (e.g., when the same type of vehicle passes by). A gradual increase in the sensitivity change rate indicates that the sensor's response characteristics are changing.

[0086] (3) Signal-to-noise ratio decay trend: The ratio of the effective signal amplitude to the noise amplitude of the sensor output signal. A continuous decrease in signal-to-noise ratio indicates that the signal quality of the sensor is deteriorating.

[0087] Based on at least one of the above performance parameters, calculate the degradation trend index D for each sensor. In one embodiment, the degradation trend index D can be calculated using a weighted comprehensive method: D = w1 × (zero drift amount / upper limit of zero drift) + w2 × (sensitivity change rate / upper limit of sensitivity change) + w3 × (1 - current signal-to-noise ratio / calibrated signal-to-noise ratio), where w1, w2, and w3 are the weighting coefficients of each parameter, and the value of D ranges from 0 to 1. The larger the value of D, the more severe the degradation.

[0088] S3b: Degradation warning judgment. Set a degradation warning threshold Dpre (e.g., Dpre=0.5). When the degradation trend index D of a certain sensor exceeds the degradation warning threshold Dpre but the degradation trend index D has not yet reached 1 (i.e., the fault judgment condition has not been triggered), mark each weighing mode associated with the weighing platform where the sensor is located as a warning mode.

[0089] For example, if the degradation trend index D of a certain sensor on the second weighing platform 2 is 0.7, which exceeds the warning threshold Dpre=0.5 but does not reach the fault threshold 1.0, then the weighing modes (first mode, second mode and third mode) that depend on the second weighing platform 2 will be marked as warning modes, and the fourth mode will be a non-warning mode.

[0090] S3c: Mode Weight Pre-adjustment. In the fusion output step S5, the weight coefficient of the early warning mode is gradually reduced according to the severity of the degradation trend index D. Specifically, let the degradation decay factor λ = 1 - (D - Dpre) / (1 - Dpre), then the adjusted weight coefficient of the early warning mode α'i = αi × λ. As D gradually increases from Dpre to 1, λ gradually decreases from 1 to 0, and the weight coefficient of the early warning mode gradually decreases from its original value to zero, achieving a smooth transition.

[0091] Continuing with the example above: if D=0.7 and Dpre=0.5, then: λ = 1 - (0.7 - 0.5) / (1 - 0.5) = 1 - 0.4 = 0.6. Assume that under normal conditions, the weights of the first mode are α0 = 0.4, the second mode α1 = 0.2, the third mode α2 = 0.2, and the fourth mode α3 = 0.2. Since the first, second, and third modes are all warning modes, after adjustment: α'0 = 0.4 × 0.6 = 0.24, α'1 = 0.2 × 0.6 = 0.12, α'2 = 0.2 × 0.6 = 0.12, α'3 = 0.2 (unchanged). After normalization: α''0 = 0.24 / 0.68 ≈ 0.353, α''1 = 0.12 / 0.68 ≈ 0.176, α''2 = 0.12 / 0.68 ≈ 0.176, α''3 = 0.2 / 0.68 ≈ 0.294.

[0092] It can be seen that, compared with the normal state, the weight of the non-warning fourth mode increased from 0.2 to about 0.294, while the weight of the warning mode decreased. The system gradually shifted the fusion focus to the unaffected mode before the sensor completely failed.

[0093] When the degradation trend index D of the sensor finally reaches the fault judgment condition, λ decays to 0, the weight of the early warning mode naturally drops to zero, and the system smoothly transitions to the fault degradation state, avoiding abrupt adjustments of the weight.

[0094] Meanwhile, the data analysis submodule uploads degradation warning information (including warning sensor identifiers, current values ​​of degradation trend indicators, and the remaining time expected to reach the fault threshold) to the remote monitoring platform through remote monitoring steps, enabling managers to arrange sensor replacement or maintenance in advance and avoid sudden sensor failures during operation.

[0095] By establishing a sensor performance degradation trend model and gradually pre-adjusting mode weights during the transition period when the sensor deteriorates but does not fail, a smooth transition from normal to degraded state is achieved. This avoids abrupt changes in weighing results caused by sudden degradation, improving the system's weighing stability and data continuity under progressive sensor degradation scenarios. Simultaneously, remote reporting of degradation early warning information allows managers to perform preventative maintenance in advance, further reducing the risk of sudden sensor failures.

[0096] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A fault adaptive method for a vehicle dynamic weighing system based on multi-mode fusion, characterized in that, A weighing system is applied to a system in which multiple weighing platforms are arranged sequentially and adjacently along the vehicle's direction of travel, each of which is equipped with an independently connected weighing sensor; the method includes: The output signals of the weighing sensors on each weighing platform are collected and converted into corresponding weight data. Based on the weight data, the total weight of the vehicle in each preset weighing mode is calculated. The weighing platforms used in different weighing modes are not exactly the same. The output signals of each weighing sensor are monitored in real time. Based on the preset fault judgment conditions, it is determined whether there is a weighing sensor fault and the weighing platform where the faulty weighing sensor is located is identified. When a load cell malfunction is detected on a certain weighing platform, the weighing mode that depends on the malfunctioning weighing platform will be marked as an invalid mode, and the weighing mode that does not depend on the malfunctioning weighing platform will be marked as an effective mode. The total weight values ​​calculated from the effective modes are weighted and fused to obtain the final measured total weight value of the vehicle, which is then output.

2. The fault adaptive method for a vehicle dynamic weighing system based on multi-mode fusion according to claim 1, characterized in that, The real-time monitoring of the output signals of each weighing sensor and the determination of whether a weighing sensor fault exists based on preset fault judgment conditions specifically include: The output signal of each weighing sensor is periodically sampled, and the fluctuation amplitude of the output signal of each weighing sensor within a preset time window is calculated. When the fluctuation amplitude exceeds the preset normal fluctuation threshold range, or when the output signal of the weighing sensor remains at zero for more than a preset duration, the weighing sensor is determined to be faulty.

3. The fault adaptive method for a vehicle dynamic weighing system based on multi-mode fusion according to claim 1, characterized in that, The process of weighted fusion of the total weight values ​​calculated from the effective modes to obtain and output the final measured total weight value of the vehicle specifically includes: Based on the historical weighing accuracy statistics of each effective mode and the number of weighing platforms used, weighting coefficients are assigned to each effective mode. The final total weight value of the vehicle is obtained by weighting the total weight values ​​of each effective mode based on the weighting coefficients.

4. The fault adaptive method for a vehicle dynamic weighing system based on multi-mode fusion according to claim 3, characterized in that, The rules for allocating the weighting coefficients include: When all weighing modes are valid, the weighing mode that relies on the most weighing platform has the largest weight coefficient, and the weighing mode that relies on the fewest weighing platform has the smallest weight coefficient. When a weighing mode is marked as invalid, the weight coefficient of the invalid mode is set to zero, and the weight coefficients of the remaining valid modes are normalized.

5. The fault adaptive method for a vehicle dynamic weighing system based on multi-mode fusion according to claim 1, characterized in that, It also includes a consistency verification step: Before weighted fusion of the total weight values ​​calculated by the effective modes, a consistency check is performed on the total weight values ​​calculated by each effective mode. Calculate the deviation between the total weight values ​​of each effective mode. When the deviation between the total weight value of a certain effective mode and the average total weight value of the other effective modes exceeds the preset deviation threshold, the effective mode is marked as an abnormal mode and excluded from the fusion calculation.

6. The fault adaptive method for a vehicle dynamic weighing system based on multi-mode fusion according to claim 1, characterized in that, The weighing platform includes a first weighing platform, a second weighing platform, and a third weighing platform. The length of the third weighing platform along the vehicle's travel direction is greater than the lengths of the first and second weighing platforms along the vehicle's travel direction. The lengths of the first and second weighing platforms along the vehicle's travel direction are the same. The method further includes an axle group identification step. As the vehicle passes over the third weighing platform, the number of axles and axle group composition information of the vehicle are identified based on the time-series change waveform of the output signal of the weighing sensor on the third weighing platform. The number of shafts and shaft group configuration information are used to correct the total weight value calculation under each weighing mode.

7. The fault adaptive method for a vehicle dynamic weighing system based on multi-mode fusion according to claim 1, characterized in that, It also includes sensor degradation prediction and mode weight pre-adjustment steps: Based on the real-time monitoring of the output signals of each weighing sensor, a performance degradation trend model for each weighing sensor is established. The performance degradation trend model calculates the degradation trend index of each weighing sensor based on at least one of the zero drift, sensitivity change rate, and signal-to-noise ratio attenuation trend of each weighing sensor within a preset statistical period. When the degradation trend index of a certain weighing sensor exceeds the preset degradation warning threshold but the fault judgment condition has not yet been triggered, the weighing modes associated with the weighing platform where the weighing sensor is located are marked as warning modes. During weighted fusion, the weight coefficient of the early warning mode is gradually reduced according to the severity of the degradation trend index, while the weight coefficient of the non-early warning mode is increased accordingly, so that the fusion result gradually reduces its dependence on the weighing platform where the deteriorated weighing sensor is located.

8. The fault adaptive method for a vehicle dynamic weighing system based on multi-mode fusion according to any one of claims 1-7, characterized in that, It also includes remote monitoring steps: The operating status information of each weighing sensor, the calculation results of each weighing mode, the fault detection results, and the mode degradation records are uploaded to the remote monitoring platform through the communication interface. The remote monitoring platform generates equipment operation status reports and maintenance reminders based on the received information.

9. The fault adaptive method for a vehicle dynamic weighing system based on multi-mode fusion according to claim 8, characterized in that, The remote monitoring steps also include: Record the time, faulty weighing sensor identifier, fault type, and valid mode combinations before and after each mode degradation event; Based on the accumulated mode degradation event records, a weighing sensor health assessment report is generated, which is used to guide the preventive maintenance of the equipment.

10. The fault adaptive method for a vehicle dynamic weighing system based on multi-mode fusion according to claim 5, characterized in that, It also includes a precision confidence assessment step: The accuracy confidence score of the current weighing result is calculated based on multiple evaluation factors, including the number of currently valid modes, the number of faulty weighing sensors detected, the number of excluded abnormal modes, and the vehicle's passing speed. The accuracy confidence score is used to query the pre-stored accuracy level mapping table to obtain the accuracy confidence level of the current weighing result. The accuracy level mapping table is established based on the measured error statistics under different working conditions during the system calibration stage. When the accuracy confidence level is lower than the preset minimum acceptable level, an accuracy alarm signal is output.

Images