Method for identifying state of piezoelectric film sensor of dynamic weighing system

Abstract

The invention discloses a state identification method for a piezoelectric film sensor of a dynamic weighing system, and the method comprises the steps: employing a positioning signal peak value as a symmetric original point, intercepting and transposing a left sample and a right sample, carrying out the quantitative evaluation of the correlation or deviation degree between the transposed sample and the other side, and accurately judging whether the sensor is in a flat installation state or not; and a basis is provided for subsequent data correction or operation and maintenance early warning. According to the method, hardware does not need to be added, the method does not depend on external detection equipment, online, non-inductive and qualitative diagnosis of the installation quality can be achieved only through inherent acquisition signals of the system, the problem of weighing errors caused by the fact that signal distortion caused by installation defects is not perceived is solved, and the reliability and measurement accuracy of the dynamic weighing system are remarkably improved.

Description

Technical Field

[0001] This invention relates to the field of sensor technology, specifically to a method for identifying the state of a piezoelectric thin-film sensor in a dynamic weighing system. Background Technology

[0002] as follows Figure 1 The diagram shows the on-site sensor layout of the piezoelectric film vehicle dynamic weighing system: two sensors spanning the lane are installed within each lane, with a spacing of 3m between them. A coil sensor with a coil size of 2m x 2m is placed in the middle. The system collects the voltage signals (including signal amplitude, pulse width, and time information) generated by the pressure exerted by the vehicle wheels on the piezoelectric film sensors during vehicle movement, and outputs data such as vehicle weight, vehicle speed, and vehicle type.

[0003] A piezoelectric thin-film sensor is a sensor based on the internal piezoelectric effect. When subjected to different load impacts, it generates voltage signals with varying amplitudes and pulse widths for downstream equipment to collect. Therefore, in traditional installation methods, the sensor must be installed flat, with the mounting surface flush with the ground. There should be no protrusions, depressions, or unevenness, otherwise the signal consistency will be affected.

[0004] Traditional vehicle dynamic weighing systems require piezoelectric film sensors to be flush with the road surface and at the same depth during installation to ensure repeatability of signal output under the same load. Traditional methods only specify installation requirements, and the backend data acquisition equipment only integrates and calculates the signal, then combines the integrated signal sum with parameters such as speed for comprehensive calculation, and derives the vehicle weight data based on calibration coefficients. Traditional technology cannot determine whether the installation of the piezoelectric film sensor meets the process requirements by identifying signal characteristics. Summary of the Invention

[0005] The present invention aims to provide a method for identifying the status of piezoelectric thin film sensors in dynamic weighing systems, which solves the problem that traditional methods cannot diagnose the actual installation effect of sensors, and performs correction and compensation to improve measurement accuracy.

[0006] To achieve the above objectives, the technical solution adopted by this invention is: a method for identifying the state of a piezoelectric thin-film sensor in a dynamic weighing system, comprising: S1: Extract the complete bell-shaped voltage signal corresponding to a single axle from the raw voltage signal collected by the dynamic weighing system; S2: In the complete bell-shaped voltage signal, locate the point of maximum amplitude and take the point of maximum amplitude as the origin of symmetry; S3: Taking the symmetrical origin as the center, extract valid data points of equal length on its left and right sides respectively to form left sample and right sample; S4: Perform a transpose operation on the time axis for any sample from the left or right sample to obtain a transposed sample; S5: Based on the correlation or deviation between the transposed sample and the other untransposed sample, calculate the goodness of fit or symmetry error, and determine the installation status of the piezoelectric thin film sensor according to a preset threshold.

[0007] Preferably, in step S5, the correlation is the goodness of fit R² of linear regression; When the goodness of fit is greater than or equal to the preset threshold S, the piezoelectric thin film sensor is determined to be installed normally. When the goodness of fit is less than the preset threshold S, it is determined that there is a defect in the installation of the piezoelectric thin film sensor.

[0008] Preferably, the preset threshold S is determined by statistically analyzing the goodness-of-fit distribution of the bell-shaped voltage signal output by a known properly installed piezoelectric thin-film sensor.

[0009] Preferably, in step S5, the degree of deviation is a symmetrical error, including the sum of squared differences (SSE) or the sum of absolute differences (SAD). When the symmetry error is less than or equal to a preset threshold T, the piezoelectric film sensor is considered to be installed normally. When the symmetry error is greater than the preset threshold T, it is determined that there is a defect in the installation of the piezoelectric thin film sensor.

[0010] Preferably, the preset threshold T is determined by statistically analyzing the symmetry error distribution of the bell-shaped voltage signal output by a piezoelectric thin-film sensor with known installation abnormalities.

[0011] Preferably, in step S4, the transpose operation refers to inverting the time coordinates of the selected sample to achieve a mirror flip.

[0012] Preferably, the method further includes step S6: when step S5 determines that there is a defect in the installation of the piezoelectric thin film sensor, the reconstructed data of the other side is generated by the transpose operation using the relatively complete side of the left sample or the right sample, so as to reconstruct a complete symmetrical bell voltage signal.

[0013] Preferably, in step S6, the reconstructed complete symmetrical bell-shaped voltage signal is used for subsequent vehicle dynamic weighing calculations to compensate for measurement errors introduced by installation defects.

[0014] Preferably, in step S3, the effective data points are continuous sampling points in the bell voltage signal whose amplitude is greater than the peak amplitude by 10% to 30%.

[0015] Preferably, the bell-shaped voltage signal is generated by a piezoelectric thin film sensor laid under the road surface during the application and release of wheel load, and its waveform is Gaussian or Gaussian-like.

[0016] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention uses the peak value of the positioning signal as the origin of symmetry, extracts and transposes left and right samples, and quantitatively evaluates the correlation or deviation between the transposed sample and the other side. This allows for accurate determination of whether the sensor is installed flat, providing a basis for subsequent data correction or maintenance early warning. This method requires no additional hardware and does not rely on external testing equipment. It utilizes only the system's inherent acquisition signals to achieve online, non-intrusive, and qualitative diagnosis of installation quality, solving the problem of weighing errors caused by undetected signal distortion due to installation defects. This significantly improves the reliability and measurement accuracy of dynamic weighing systems. Attached Figure Description

[0017] Figure 1 This is a structural diagram of a piezoelectric thin film weighing system in the prior art; Figure 2 Side view of the sensor mounting of the present invention; Figure 3 This is a diagram showing the signal output during normal installation of the present invention; Figure 4 This is a normal and complete shaft group signal diagram of the present invention; Figure 5 This is a schematic diagram of the installation state of the present invention; Figure 6 This is a signal diagram during abnormal installation of the present invention; Figure 7 This is a flowchart of the piezoelectric thin film sensor state recognition method for the dynamic weighing system of the present invention. Detailed Implementation

[0018] The following description is intended to disclose the invention and enable those skilled in the art to implement it. The preferred embodiments described below are merely examples, and other obvious variations will occur to those skilled in the art.

[0019] This invention proposes a method for identifying the state of a piezoelectric thin film sensor in a dynamic weighing system. By comparing and fitting the signal symmetry of each axle, the method uses the goodness-of-fit value and symmetry error calculation to identify and judge the flatness of the sensor installation.

[0020] Under normal installation conditions, the mounting surface of the piezoelectric thin-film sensor is flush with the road surface, and the bell-shaped signals output by each axle are symmetrical from left to right, centered at their highest point. Utilizing this characteristic, this invention performs symmetrical analysis of the signals from each axle through acquisition and signal comparison at the acquisition end.

[0021] This invention provides two methods for determining and analyzing signal symmetry; either method can be selected or used together: Method 1: Process the output signal of each axle. The original signal is a bell-shaped signal with time on the x-axis and signal amplitude on the y-axis. The symmetrical point of the maximum signal value is determined by sorting, and this point is used as the reference value for the origin time. Data points on both sides of the highest point of the original signal are selected as the analysis sample group. The left (or right) data group is transposed (only one side is transposed). The transposed data is then linearly fitted with the untransposed data on the other side to obtain the goodness of fit between the two sets of data. The goodness of fit is judged as follows: if the goodness of fit ≥ S, it indicates a high correlation between the two sets of data, the original bell-shaped output signal is symmetrical, and the sensor exhibits uniform symmetry during load loading and release, further confirming the installation specifications; if the goodness of fit is lower than S, it indicates asymmetry in the original signal, and unevenness in load loading and release, requiring road surface grinding. The threshold for the S value is determined based on the known error statistics of well-defined symmetrical and asymmetrical signals, or by manually setting a specific value.

[0022] Method 2: Calculate the symmetry error between the two sets of sample data. This can be achieved by calculating the sum of squared differences or the sum of absolute differences. After obtaining the error value SSESSE or SADSAD, compare it with the set acceptable error threshold T. If Error ≤ T, the signal is judged to be symmetrical; if Error > T, the signal is judged to be asymmetrical.

[0023] This invention analyzes the mean and cumulative sum of signals on both sides using goodness-of-fit analysis and symmetry error method to identify the fullness or concavity of the output signals on both sides, thereby identifying the protrusion and concavity features of the sensor mounting plane and providing a basis for construction and maintenance.

[0024] A further optimization of this invention involves first identifying anomalies using metrics such as goodness of fit, signal mean, and cumulative sum, and then correcting and compensating for the abnormal signals. Specifically, the compensation method involves using the signal from the normal side and performing a symmetrical transpose to supplement and replace the signal from the abnormal side, thereby achieving data correction.

[0025] like Figure 7 As shown, this invention proposes a method for identifying the state of a piezoelectric thin-film sensor in a dynamic weighing system, comprising the following steps: S1: Extract the complete bell-shaped voltage signal corresponding to a single axle from the raw voltage signal collected by the dynamic weighing system; S2: In the complete bell-shaped voltage signal, locate the point of maximum amplitude and take the point of maximum amplitude as the origin of symmetry; S3: Taking the symmetrical origin as the center, extract the valid data points of equal length on its left and right sides respectively to form the left sample and the right sample; the valid data points are continuous sampling points in the bell voltage signal whose amplitude is greater than the peak amplitude by 10% to 30%; S4: For any sample in the left or right sample, perform a transpose operation on the time axis to obtain a transposed sample; the transpose operation means inverting the time coordinate of the selected sample to achieve a mirror flip. S5: Based on the correlation or deviation between the transposed sample and the other untransposed sample, calculate the goodness of fit or symmetry error, and determine the installation status of the piezoelectric thin film sensor according to a preset threshold.

[0026] In one embodiment, in step S5, the correlation is the goodness of fit R² of linear regression; when the goodness of fit is ≥ a preset threshold S, the piezoelectric thin film sensor is determined to be installed normally; when the goodness of fit is < the preset threshold S, the piezoelectric thin film sensor is determined to have a defect in installation. The preset threshold S is determined by statistically analyzing the goodness of fit distribution of the bell-shaped voltage signal output by a known normally installed piezoelectric thin film sensor.

[0027] In another embodiment, in step S5, the degree of deviation is a symmetry error, including the sum of squared differences (SSE) or the sum of absolute differences (SAD). When the symmetry error is less than or equal to a preset threshold T, the piezoelectric thin film sensor is determined to be installed normally; when the symmetry error is greater than or equal to the preset threshold T, the piezoelectric thin film sensor is determined to have an installation defect. The preset threshold T is determined by statistically analyzing the symmetry error distribution of the bell-shaped voltage signal output by piezoelectric thin film sensors with known installation abnormalities.

[0028] Furthermore, the above method also includes step S6: when step S5 determines that there is a defect in the installation of the piezoelectric thin film sensor, the reconstructed data of the other side is generated by the transpose operation using the relatively complete side of the left sample or the right sample, so as to reconstruct a complete symmetrical bell voltage signal.

[0029] The reconstructed complete symmetrical bell-shaped voltage signal is used for subsequent vehicle dynamic weighing calculations to compensate for measurement errors introduced by installation defects.

[0030] The bell-shaped voltage signal mentioned above is generated by a piezoelectric thin film sensor laid under the road surface during the application and release of wheel load, and its waveform is Gaussian or Gaussian-like.

[0031] like Figure 2 As shown, when the sensor is installed correctly and flush with the road surface, when the wheel passes through the sensor area, due to the symmetrical mechanical properties, the output signal is bell-shaped (Gaussian shape), with one signal corresponding to each axle. Figure 3 , Figure 4The bell-shaped voltage signals output by the front and rear axles of the two-axle vehicle when passing through the piezoelectric thin film sensor are shown. The output voltage signals of each axle remain symmetrical, which indicates that the mechanical state of the vehicle wheels is normal during the application of load impact and the unloading process, so the output signals exhibit symmetrical characteristics.

[0032] However, when the sensor is installed abnormally, such as when the adhesive surface is not flush with the road surface (e.g.) Figure 5 As shown), due to its uneven contact with the road surface and its asymmetrical physical structure, the elastic deformation process of the wheel during load application and release is asymmetrical, ultimately affecting the sensor's output signal (such as...). Figure 6 As shown in the diagram, there is a significant asymmetry between the left and right signals corresponding to each axle. Therefore, this asymmetry in the output signals can be identified at the acquisition end, and corresponding compensation and correction can be performed in subsequent stages.

[0033] The following are the specific implementation steps and methods for Method 1: Step 1: Extract the output signal for each axle. The original signal is a bell-shaped signal with time as the horizontal axis and signal amplitude as the vertical axis. Find its symmetrical point (i.e., the maximum value of the signal) by sorting, and set this point as the value of the origin time.

[0034] Step 2: Select the data points on both sides of the highest point of the original signal as the analysis sample data group, and determine the length L of the left and right range to be compared. L is taken as the smaller value between the distance from the peak point to the left end and the right end of the array, so as to ensure that the comparison is only performed within the effective range of the array, and only the part with corresponding data points on both the left and right sides is compared.

[0035] Step 3: Transpose the data set on the left (or right), but only on one side. The specific transposition operation is as follows: If the coordinates of the points to the left of the highest point, from near to far, are (x1, y1), (x2, y2)...(xn, yn), and the coordinates of the points to the right of the highest point, from near to far, are (x1, y1), (x2, y2)...(xn, yn), then multiply the x-axis data of the left-side points from near to far by (-1) to obtain the transposed data as (x1*(-1), y1), (x2*(-1), y2)...(xn*(-1), yn). Then, perform a linear fit between the transposed data and the untransposed data on the other side to obtain the goodness of fit between the two sets of data.

[0036] Step 4: Judge the goodness of fit for different values. When the goodness of fit is ≥ S, it indicates high correlation, symmetrical original output bell signal, and uniform symmetry of the sensor during load loading and release, further confirming the installation specifications. If the goodness of fit is lower than S, it indicates that the original signal is asymmetrical, and there is non-uniformity of the sensor during load loading and release, requiring road surface grinding. The threshold setting method for the S value: It can be determined based on the statistical error between known well-symmetrical and asymmetrical signals, or a value can be manually set.

[0037] The following are the specific implementation steps and methods for Method 2: Step 1: Find the peak point Traverse the entire array A to find its maximum value A[ipeak] and its corresponding index ipeak. This peak point is assumed to be the position of the axis of symmetry.

[0038] Step 2: Determine the comparison range or length Determine the left and right ranges L to be compared. L is the smaller of the distances from the peak point to the left and right ends of the array. This ensures that the comparison is only performed within the valid range of the array, and only the portions with corresponding data points on both the left and right sides are expanded.

[0039] Step 3: Define an error function to quantify the degree of asymmetry between the left and right sides. The specific method is to calculate the sum of squared differences or the sum of absolute differences.

[0040] Sum of squared differences:

[0041] Sum of absolute differences: The smaller the error value SSE or SAD, the more symmetrical the signal. For an ideal symmetrical signal, the error is 0.

[0042] Step 4: Set the threshold and make a judgment Set an acceptable error threshold T. Compare the calculated error with this threshold: if Error ≤ T, the signal is considered symmetrical; if Error > T, the signal is considered asymmetrical.

[0043] The threshold T is the core parameter of the algorithm, defining the tolerance range for "sufficient symmetry". The threshold can be set in several ways, including by determining it based on known error statistics of well-symmetric and asymmetric signals, or by setting a specific value manually.

[0044] After determining the asymmetry of the piezoelectric film signal using the above method, this application can further identify symmetry through goodness of fit, signal mean, and summation, and then correct and compensate for the identified anomalies. The compensation method is as follows: using the signal on the normal side, the signal on the abnormal side is replaced by symmetrical transposition to form a complete signal and complete the data correction.

[0045] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claimed invention. The scope of protection claimed by the appended claims and their equivalents is defined.

Claims

1. A method for identifying the state of a piezoelectric thin-film sensor in a dynamic weighing system, characterized in that, include: S1: Extract the complete bell-shaped voltage signal corresponding to a single axle from the raw voltage signal collected by the dynamic weighing system; S2: In the complete bell-shaped voltage signal, locate the point of maximum amplitude and take the point of maximum amplitude as the origin of symmetry; S3: Taking the symmetrical origin as the center, extract valid data points of equal length on its left and right sides respectively to form left sample and right sample; S4: Perform a transpose operation on the time axis for any sample from the left or right sample to obtain a transposed sample; S5: Based on the correlation or deviation between the transposed sample and the other untransposed sample, calculate the goodness of fit or symmetry error, and determine the installation status of the piezoelectric thin film sensor according to a preset threshold.

2. The method for identifying the state of a piezoelectric thin-film sensor in a dynamic weighing system according to claim 1, characterized in that, In step S5, the correlation is the goodness of fit R² of linear regression. When the goodness of fit is greater than or equal to the preset threshold S, the piezoelectric thin film sensor is determined to be installed normally. When the goodness of fit is less than the preset threshold S, it is determined that there is a defect in the installation of the piezoelectric thin film sensor.

3. The method for identifying the state of a piezoelectric thin-film sensor in a dynamic weighing system according to claim 2, characterized in that, The preset threshold S is determined by statistically analyzing the goodness-of-fit distribution of the bell-shaped voltage signals output by a known properly installed piezoelectric thin-film sensor.

4. The method for identifying the state of a piezoelectric thin-film sensor in a dynamic weighing system according to claim 1, characterized in that, In step S5, the degree of deviation is a symmetrical error, including the sum of squared differences (SSE) or the sum of absolute differences (SAD). When the symmetry error is less than or equal to a preset threshold T, the piezoelectric film sensor is considered to be installed normally. When the symmetry error is greater than the preset threshold T, it is determined that there is a defect in the installation of the piezoelectric thin film sensor.

5. The method for identifying the state of a piezoelectric thin-film sensor in a dynamic weighing system according to claim 4, characterized in that, The preset threshold T is determined by statistically analyzing the symmetry error distribution of the bell-shaped voltage signal output by a piezoelectric thin-film sensor with known installation abnormalities.

6. The method for identifying the state of a piezoelectric thin-film sensor in a dynamic weighing system according to claim 1, characterized in that, In step S4, the transpose operation refers to inverting the time coordinates of the selected sample to achieve a mirror flip.

7. The method for identifying the state of a piezoelectric thin-film sensor in a dynamic weighing system according to claim 1, characterized in that, The method further includes step S6: when step S5 determines that there is a defect in the installation of the piezoelectric thin film sensor, the reconstructed data of the other side is generated by the transpose operation using the relatively complete side of the left sample or the right sample, so as to reconstruct a complete symmetrical bell voltage signal.

8. The method for identifying the state of a piezoelectric thin-film sensor in a dynamic weighing system according to claim 7, characterized in that, In step S6, the reconstructed complete symmetrical bell-shaped voltage signal is used for subsequent vehicle dynamic weighing calculations to compensate for measurement errors introduced by installation defects.

9. The method for identifying the state of a piezoelectric thin-film sensor in a dynamic weighing system according to claim 1, characterized in that, In step S3, the effective data points are continuous sampling points in the bell voltage signal whose amplitude is greater than the peak amplitude by 10% to 30%.

10. The method for identifying the state of a piezoelectric thin-film sensor in a dynamic weighing system according to claim 1, characterized in that, The bell-shaped voltage signal is generated by a piezoelectric thin film sensor laid under the road surface during the application and release of wheel load, and its waveform is Gaussian or Gaussian-like.

Images