Weighing sensor data compensation method, device, equipment and storage medium

Abstract

The present application relates to the field of weighing sensor data processing technology, and discloses a weighing sensor data compensation method, device, equipment, and storage medium, including: when detecting that the weighing sensor starts measuring, obtaining the multi-dimensional environmental characteristics of the weighing sensor in the current measurement environment; extracting the dynamic feature components corresponding to each time dimension of the weighing sensor based on the historical measurement data of the weighing sensor; inputting each dynamic feature component and the multi-dimensional environmental characteristics into different compensation modules to generate multiple compensation parameter components; fusing all compensation parameter components to generate comprehensive compensation parameters, and inputting the comprehensive compensation parameters into the trained compensation operation network to compensate the measurement data of the weighing sensor in the current measurement environment to obtain compensated data. Through a modular compensation method, interference factors such as temperature, humidity, and posture are captured, and the weighing sensor data can be adjusted more finely, improving the accuracy and effectiveness of the compensation.

Description

Technical Field

[0001] The present application relates to the technical field of weighing sensor data processing, and in particular to a weighing sensor data compensation method, device, equipment and storage medium. Background Art

[0002] With the advancement of load cell manufacturing and microelectronics technology, load cells, as core components of truck scales, operate in complex and changing environments, with their performance affected by numerous factors, such as temperature, humidity, and tilt angle. Existing compensation algorithms typically only address the load cell itself, ignoring indirect effects from mounting components (such as the scale platform) and the environment (such as vibration and light). These algorithms struggle to adapt to dynamic changes, such as load cell aging and gradual environmental changes, leading to decreased compensation accuracy after long-term use. Summary of the Invention

[0003] The main purpose of this application is to provide a weighing sensor data compensation method, device, equipment and storage medium, aiming to solve the technical problem that the existing compensation algorithm ignores the indirect influence of installation accessories and usage environment, resulting in a decrease in compensation accuracy after long-term use.

[0004] To achieve the above objectives, the present application proposes a weighing sensor data compensation method, which includes:

[0005] When detecting that the weighing sensor starts measuring, obtaining the multi-dimensional environmental characteristics of the weighing sensor in the current measurement environment;

[0006] Extracting dynamic characteristic components corresponding to each time dimension of the weighing sensor according to historical measurement data of the weighing sensor;

[0007] Inputting each of the dynamic feature components and the multi-dimensional environmental features into different compensation modules respectively to generate multiple compensation parameter components;

[0008] All the compensation parameter components are integrated to generate comprehensive compensation parameters, and the comprehensive compensation parameters are input into the trained compensation operation network to compensate the measurement data of the weighing sensor in the current measurement environment to obtain compensated data.

[0009] Optionally, when detecting that the weighing sensor starts measuring, the step of obtaining the multi-dimensional environmental characteristics of the weighing sensor in the current measurement environment includes:

[0010] Acquire historical measurement data of the weighing sensor, and determine a first compensation factor set for the historical measurement data, where the first compensation factor set is used to characterize different influencing factors of the measurement environment;

[0011] When detecting that the weighing sensor starts measuring, determining a second compensation factor set of the weighing sensor in a current measurement cycle;

[0012] The first compensation factor set and the second compensation factor set are input into the trained environmental feature model to extract the multi-dimensional environmental features of the current measurement environment.

[0013] Optionally, after the step of inputting the first compensation factor set and the second compensation factor set into the trained environmental feature model to extract the multi-dimensional environmental features of the current measurement environment, the method further includes:

[0014] Obtaining benchmark verification data of the load cell within a target observation period;

[0015] adjusting the first set of compensation factors based on the historical measurement data and the benchmark verification data to obtain a third set of compensation factors;

[0016] adjusting the second compensation factor set based on the compensated data and the benchmark verification data to obtain a fourth compensation factor set;

[0017] Determining a measurement deviation value and a compensation deviation value of a weighing sensor according to the historical measurement data, the compensated data, the first compensation factor set, the second compensation factor set, the third compensation factor set, and the fourth compensation factor set;

[0018] If the compensation deviation value exceeds the preset convergence threshold, the historical measurement data is adjusted based on the measurement deviation value of the weighing sensor, and the step of obtaining the benchmark verification data of the weighing sensor within the target observation period is re-executed until the compensation deviation value is less than the preset convergence threshold.

[0019] Optionally, the step of determining a load cell measurement deviation value and a compensation deviation value based on the historical measurement data, the compensated data, the first compensation factor set, the second compensation factor set, the third compensation factor set, and the fourth compensation factor set includes:

[0020] determining a measurement deviation value of a weighing sensor according to a first reconstruction error between the first compensation factor set and the third compensation factor set, and a second reconstruction error between the second compensation factor set and the fourth compensation factor set;

[0021] Performing validity verification on the historical measurement data, the compensated data, the third compensation factor set, and the fourth compensation factor set using a trained data quality evaluator to generate an estimated loss value;

[0022] A compensation deviation value is determined according to the deviation value measured by the weighing sensor and the estimated loss value.

[0023] Optionally, the step of performing validity verification on the historical measurement data, the compensated data, the third compensation factor set, and the fourth compensation factor set by a trained data quality evaluator to generate an estimated loss value includes:

[0024] Obtaining, by a data quality assessor, a first credibility score for the historical measurement data, a second credibility score for the compensated data, a third credibility score for the third compensation factor set, and a fourth credibility score for the fourth compensation factor set;

[0025] Calculating a first deviation value between the second credibility score and the first credibility score, a second deviation value between the third credibility score and a preset measurement score, and a third deviation value between the fourth credibility score and a preset compensation score;

[0026] The first deviation value, the second deviation value, and the third deviation value are weighted and combined to obtain an estimated loss value.

[0027] Optionally, the step of inputting each of the dynamic feature components and the multi-dimensional environmental features into different compensation modules to generate multiple compensation parameter components includes:

[0028] Calculating the exponential function transformation value of the weight coefficient of each sampling point of the weighing sensor;

[0029] Normalize the exponential function transformation value of each sampling point to generate a normalized weight coefficient;

[0030] Assigning a corresponding normalized weight coefficient to each of the dynamic feature components to obtain a target feature component corresponding to each sampling point;

[0031] Each of the target feature components and the multi-dimensional environmental features are input into different compensation modules to generate a plurality of compensation parameter components.

[0032] Optionally, the compensation parameter components include a temperature compensation parameter, a humidity compensation parameter and a posture compensation parameter;

[0033] The step of fusing all the compensation parameter components to generate a comprehensive compensation parameter, inputting the comprehensive compensation parameter into a trained compensation operation network, compensating the measurement data of the weighing sensor in the current measurement environment, and obtaining compensated data includes:

[0034] Performing weighted fusion on the temperature compensation parameter, the humidity compensation parameter, and the attitude compensation parameter through a multi-source compensation parameter fusion algorithm to generate a comprehensive compensation parameter matrix including environmental dynamic characteristics;

[0035] Inputting the comprehensive compensation parameter matrix and the real-time measurement data of the weighing sensor into the trained compensation operation network, and calculating the compensation offset through multi-layer nonlinear mapping;

[0036] The measurement data of the weighing sensor in the current measurement environment is compensated and dynamically corrected according to the compensation offset to generate compensated data.

[0037] In addition, to achieve the above-mentioned purpose, the present application also proposes a weighing sensor data compensation device, which includes:

[0038] A feature extraction module is used to obtain the multi-dimensional environmental features of the weighing sensor in the current measurement environment when it is detected that the weighing sensor starts measuring;

[0039] A component acquisition module, configured to extract dynamic characteristic components corresponding to each time dimension of the weighing sensor based on historical measurement data of the weighing sensor;

[0040] a parameter determination module, configured to input each of the dynamic feature components and the multi-dimensional environmental features into different compensation modules to generate a plurality of compensation parameter components;

[0041] The data compensation module is used to fuse all the compensation parameter components to generate comprehensive compensation parameters, and input the comprehensive compensation parameters into the trained compensation operation network to compensate the measurement data of the weighing sensor in the current measurement environment to obtain compensated data.

[0042] In addition, to achieve the above-mentioned purpose, the present application also proposes a weighing sensor data compensation device, which includes: a memory, a processor, and a computer program stored on the memory and runnable on the processor, and the computer program is configured to implement the steps of the weighing sensor data compensation method as described above.

[0043] In addition, to achieve the above-mentioned purpose, the present application also proposes a storage medium, which is a computer-readable storage medium. A computer program is stored on the storage medium, and when the computer program is executed by the processor, the steps of the weighing sensor data compensation method described above are implemented.

[0044] This application discloses that when a weighing sensor is detected to have started measuring, the multi-dimensional environmental characteristics of the weighing sensor in the current measurement environment are obtained; the dynamic characteristic components corresponding to each time dimension of the weighing sensor are extracted based on the historical measurement data of the weighing sensor; each of the dynamic characteristic components and the multi-dimensional environmental characteristics are input into different compensation modules to generate multiple compensation parameter components; all the compensation parameter components are fused to generate comprehensive compensation parameters, and the comprehensive compensation parameters are input into a trained compensation operation network to compensate the measurement data of the weighing sensor in the current measurement environment to obtain compensated data. The dynamic characteristic components and the multi-dimensional environmental characteristics are input into different compensation modules to generate multiple compensation parameter components, which are then fused into comprehensive compensation parameters and finally input into the trained compensation operation network for compensation. Through a modular and integrated compensation method, interference factors such as temperature, humidity, and posture are fully captured, and the weighing sensor data can be adjusted more finely, improving the accuracy and effectiveness of compensation. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0046] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0047] Figure 1 This is a flow chart of the first embodiment of the weighing sensor data compensation method of the present application;

[0048] Figure 2 This is a flow chart of the second embodiment of the weighing sensor data compensation method of the present application;

[0049] Figure 3 Flowchart of the environmental characteristic model for load cell data compensation in this application;

[0050] Figure 4 This is a flow chart of a third embodiment of the weighing sensor data compensation method of the present application;

[0051] Figure 5 This is a schematic diagram of the module structure of the weighing sensor data compensation device according to an embodiment of the present application;

[0052] Figure 6 Schematic diagram of the device structure of the hardware operating environment involved in the weighing sensor data compensation method in the embodiment of the present application.

[0053] The purpose, features and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0054] It should be understood that the specific embodiments described herein are merely used to explain the technical solutions of the present application and are not intended to limit the present application.

[0055] In order to better understand the technical solution of the present application, a detailed description will be given below in conjunction with the accompanying drawings and specific implementation methods.

[0056] In the field of industrial weighing, load cells serve as the core measurement units of equipment such as truck scales. Their measurement accuracy is directly related to production quality and the reliability of trade settlements. However, load cells are often deployed in complex working conditions such as open air, high temperatures, and dusty environments. Their output signals are not only affected by inherently sensitive factors such as temperature and humidity, but are also susceptible to the coupling effects of mounting components (such as scale platform deformation) and environmental interference (such as mechanical vibration and transient light and heat radiation). Especially for load cells that operate over a long period of time, performance drift caused by material aging and structural creep makes it difficult for traditional single-parameter environmental compensation models to adapt to dynamic changes, resulting in a significant decrease in compensation accuracy over time.

[0057] Existing load cell compensation methods often focus on linear correction of load cell parameters. For example, they use temperature sensors to collect ambient temperature and construct polynomial fitting models of temperature-output characteristics for compensation. While these methods can mitigate static environmental interference, they have significant limitations: 1. The compensation dimension is single, failing to consider the indirect mechanical stress transmitted to the load cell by deformation of mounting accessories, nor does it quantify transient signal distortion caused by vibration and impact. 2. They rely on fixed model parameters and lack the ability to online identify the load cell's inherent performance and time-varying characteristics (such as elastic fatigue and strain gauge creep). This can lead to model mismatch after long-term use. 3. Compensation algorithms often rely on independent variable assumptions and fail to establish nonlinear mapping relationships under the coupled effects of multiple environmental factors. This can easily lead to over- or under-compensation when temperature and humidity fluctuate synchronously or when posture tilt and vibration are superimposed. In recent years, while some studies have attempted to introduce neural networks to construct multi-factor compensation models, limited computing power at the edge of the load cell makes it difficult to balance model complexity with real-time performance. Crucially, existing methods generally ignore the performance degradation trajectory of load cells implicit in historical measurement data and lack a closed-loop verification mechanism for compensation effectiveness, resulting in the algorithm's inability to autonomously iterate and optimize based on long-term monitoring data. Therefore, decoupling multi-dimensional interference factors in complex dynamic environments and constructing a compensation model with long-term adaptive capabilities have become technical bottlenecks in improving the measurement accuracy of industrial load cells.

[0058] Therefore, this application provides an integrated multi-dimensional dynamic compensation and anti-cheating system for weighing sensors for complex measurement scenarios. Through the three core technologies of environmental coupling interference decoupling, online identification of time-varying characteristics, and closed-loop verification and autonomous optimization, it breaks through the limitations of static and fragmented traditional compensation models and achieves high-precision, long-life, and anti-interference industrial weighing protection.

[0059] It should be noted that the execution subject of this embodiment can be a computing service device with data processing, calculation compensation, and program execution functions, such as a data compensation module, or an electronic device capable of implementing the above functions. The following uses a weighing sensor data processing system as an example to illustrate this embodiment and the following embodiments.

[0060] Based on this, the embodiment of the present application provides a weighing sensor data compensation method, referring to Figure 1 , Figure 1 This is a flow chart of the first embodiment of the weighing sensor data compensation method of the present application.

[0061] In this embodiment, the weighing sensor data compensation method includes:

[0062] Step S10 : when it is detected that the weighing sensor starts measuring, obtaining the multi-dimensional environmental characteristics of the weighing sensor in the current measurement environment.

[0063] It should be noted that multidimensional environmental characteristics refer to feature vectors that include multidimensional environmental parameters such as temperature, humidity, vibration, air pressure, and light intensity. They are used to quantify the comprehensive impact of the load cell's current operating environment on the measurement results. For example, temperature changes can cause the elastic modulus of a strain gauge load cell to change, mechanical vibrations can affect the load cell's stability, and in high humidity environments, water vapor can enter the load cell, affecting its normal operation and causing measurement distortion. Multidimensional environmental characteristics can simultaneously include these interference factors.

[0064] It should be understood that the multi-dimensional environmental characteristics of the load cell can be collected in real time by sensing units such as temperature and humidity load cells and triaxial accelerometers installed on the load cell body, such as PT100 temperature probes, MEMS (Micro-Electro-Mechanical System) accelerometers, and capacitive humidity load cells.

[0065] It is understandable that by training time series models such as ARIMA (Autoregressive Integrated Moving Average Model) through historical measurement data, the dynamic response characteristics of the weighing sensor in different time dimensions (such as zero drift trend and temperature and humidity sensitivity) can be extracted for dynamic feature analysis.

[0066] It can be understood that when the weighing sensor starts measuring, the vehicle to be detected has already driven onto the device equipped with the weighing sensor data processing system.

[0067] Step S20 , extracting dynamic characteristic components corresponding to each time dimension of the weighing sensor based on historical measurement data of the weighing sensor.

[0068] It should be noted that historical measurement data includes measurement data continuously collected by the load cell over a period of time, including raw measurement signals (such as AD values), environmental parameters (temperature, humidity, tilt angle, etc.), and corresponding timestamps. The preset weights for each environmental characteristic vary across different time dimensions. In the short term, sudden temperature changes have a significant impact on measurement; in the medium term, zero-point drift may occur due to diurnal temperature differences; and in the long term, load cell materials can age due to humidity, leading to decreased sensitivity. Dynamic characteristic components are characteristic parameters extracted through analysis of historical measurement data that reflect how load cell performance changes over time.

[0069] It is understandable that dynamic characteristic components can be classified according to the environmental factors of the weighing sensor, such as temperature component, humidity component, posture component, etc. In order to reflect the characteristic parameters of the weighing sensor performance that change over time, dynamic characteristic components can also be classified according to the change of characteristics, such as trend component, periodic component, mutation component, etc.

[0070] In one example, the time dimension is divided into short-term (the last hour, with a 5-minute sliding window, to capture transient vibrations and sudden temperature changes); medium-term (the last 24 hours, to analyze zero-point drift caused by diurnal temperature differences); and long-term (the last year, to extract sensor aging trends, such as a 0.1% annual sensitivity drop). After segmenting historical data by time, the trend term, linear regression slope, and period term within each window are calculated. FFT is used to extract the dominant frequency, and the calculated feature parameters are normalized to generate dynamic feature components.

[0071] In step S30 , each of the dynamic feature components and the multi-dimensional environmental feature is input into different compensation modules to generate a plurality of compensation parameter components.

[0072] It should be understood that when different features are input into separate compensation modules for processing, first, the dynamic feature components should be associated with the multidimensional environmental features by type, such as the vibration feature corresponding to the vibration compensation module; second, each compensation module has a built-in specific algorithm, such as the temperature module uses polynomial fitting, and the posture module uses the coordinate transformation matrix.

[0073] It is understood that the input characteristics of each compensation module are not coupled to each other to avoid cross-interference and compensation distortion. Different types of dynamic characteristics must be strictly time-aligned, such as synchronous acquisition of vibration spectrum and weighing instantaneous data, otherwise the compensation parameters may lag or advance.

[0074] It is understandable that after each module independently outputs the compensation parameters, it is necessary to filter out invalid components through a gating mechanism. For example, the vibration compensation module outputs 0 in a static environment to ensure that the multi-source parameters do not conflict.

[0075] In one example, the vibration compensation module for a vehicle scale uses a dynamic feature component to extract a vibration standard deviation of 0.5 kg from the last 10 weighings. Multi-dimensional environmental features capture the current dominant vibration frequency of 25 Hz. A pre-set frequency-amplitude mapping table within the module reveals that 25 Hz vibration corresponds to a historical error of 0.5%. Combined with the current standard deviation, the compensation parameter is calculated to be -0.4%. The final output is "vibration compensation component of -0.4%," which is used to offset inflated measurements caused by vehicle vibration during weighing.

[0076] Step S40: All the compensation parameter components are integrated to generate comprehensive compensation parameters, and the comprehensive compensation parameters are input into the trained compensation operation network to compensate the measurement data of the weighing sensor in the current measurement environment to obtain compensated data.

[0077] It should be noted that the compensation operation network is a pre-trained deep learning model, such as a convolutional neural network, a residual network, etc., which can achieve high-precision data correction by learning the nonlinear relationship between historical error data and compensation parameters.

[0078] It should be understood that when generating comprehensive compensation parameters, the weight distribution of each compensation component must be verified through experiments to avoid over- or under-compensation caused by subjective settings. The training data for the compensation operation network should cover all possible operating conditions encountered by the load cell, such as extreme temperatures and severe vibration. Furthermore, the inference speed of the compensation operation network must meet the load cell sampling rate requirements. For example, a truck scale weighing scale requires a response time of less than 10ms.

[0079] It is understandable that when the comprehensive compensation parameters exceed the preset threshold, the compensation result may deviate significantly from the actual weighing data and the weighing data of the vehicle to be measured. In this case, the compensation operation network can trigger an alarm instruction to remind maintenance personnel to perform maintenance.

[0080] In this embodiment, when a weighing sensor is detected to have started measuring, the multidimensional environmental characteristics of the weighing sensor in the current measurement environment are obtained; the dynamic characteristic components corresponding to each time dimension of the weighing sensor are extracted based on the historical measurement data of the weighing sensor; each of the dynamic characteristic components and the multidimensional environmental characteristics are input into different compensation modules to generate multiple compensation parameter components; all of the compensation parameter components are integrated to generate a comprehensive compensation parameter, which is then input into a trained compensation operation network to compensate the measurement data of the weighing sensor in the current measurement environment to obtain compensated data. Through a modular and integrated compensation approach, interference factors such as temperature, humidity, and posture are fully captured, allowing for more precise adjustments to the weighing sensor data, improving the accuracy and effectiveness of compensation.

[0081] Reference Figure 2 , Figure 2 This is a flow chart of the second embodiment of the weighing sensor data compensation method of the present application. Based on the above-mentioned first embodiment, the second embodiment of the weighing sensor data compensation method of the present application is proposed.

[0082] In the second embodiment, step S10 includes:

[0083] Step S101 : acquiring historical measurement data of a weighing sensor, and determining a first compensation factor set of the historical measurement data, wherein the first compensation factor set is used to characterize different influencing factors of a measurement environment.

[0084] It should be noted that historical measurement data is a collection of raw measurement results and corresponding environmental parameter records accumulated during the load cell's past operation, such as weight, temperature, vibration, and other data that changes over time. The first compensation factor set is a set of parameters obtained through data analysis that quantitatively describes the extent to which different environmental factors (such as temperature and humidity) affect the load cell's measurement error.

[0085] Understandably, historical data must encompass all typical load cell operating conditions, such as extreme temperatures, high humidity, and strong vibration, to prevent model failure under unseen conditions. If the relationship between the environment and the error is nonlinear, a nonlinear model, such as polynomial regression or support vector machines, must be employed. Furthermore, when load cell aging or environmental changes are detected, compensation factors must be regularly updated to maintain accuracy.

[0086] In one example, if the historical measurement value of the pressure load cell is , the true value is , the ambient temperature is T. Then the relative error calculation formula is:

[0087]

[0088] Assume that the current error is linear with temperature , then the regression equation is established, where It indicates the rate of change of error caused by each unit increase in temperature, that is, the temperature compensation factor.

[0089] If the error growth in the high temperature area is accelerated, the quadratic term can be introduced ,at this time Describes the nonlinear intensity of temperature influence. The final set of compensation factors extracted is , used for subsequent compensation calculations.

[0090] Step S102: When it is detected that the weighing sensor starts measuring, a second compensation factor set of the weighing sensor in the current measurement cycle is determined.

[0091] It's important to note that the second compensation factor set is a set of real-time compensation parameters dynamically generated by the load cell during the current measurement cycle. It's used to instantly correct for the effects of sudden environmental changes (such as temperature fluctuations and transient vibrations) or short-term disturbances on measurement. Compared to the first compensation factor set, which is based on historical data, the second compensation factor set is rapidly iteratively updated using real-time data, focusing on the transient characteristics of the current cycle to ensure timely and targeted compensation.

[0092] It's understandable that when the load cell starts measuring, it simultaneously collects current environmental parameters and initial measurements. When generating the second compensation factor, a lightweight algorithm (such as sliding window statistics or fast Fourier transform) should be employed to ensure that the compensation factor generation speed meets real-time requirements. A compensation factor attenuation mechanism can also be implemented to gradually reduce its weight as interference intensity decreases, preventing residual effects from affecting subsequent cycles.

[0093] In one example, let the historical temperature compensation factor be , the current temperature deviation is ; The instantaneous vibration spectrum energy is detected to be concentrated in the frequency band f , the corresponding energy amplitude is A.

[0094] The temperature compensation factor is corrected to ,in is the temperature tolerance parameter, which suppresses overcorrection when the temperature deviates significantly;

[0095] Vibration interference generates additional factors , where α is the sensitivity coefficient, n is the nonlinear index, is the reference vibration amplitude.

[0096] Then the corresponding combined second compensation factor set is , which is used to offset the strain signal distortion caused by temperature offset and vibration interference in real time.

[0097] Step S103: input the first compensation factor set and the second compensation factor set into the trained environment feature model to extract the multi-dimensional environment features of the current measurement environment.

[0098] It should be noted that the environmental feature model is a pre-trained machine learning model, such as a neural network, which is used to map the compensation factor set into an abstract environmental representation.

[0099] In one example, reference Figure 3 , Figure 3 This is a flowchart of the environmental feature model for weighing sensor data compensation in this application. The environmental feature model uses a multimodal time series fusion neural network. Its specific structure includes: input layer: receiving the first compensation factor set (dimension is 24×3) and the second compensation factor set (dimension is 1×3); embedding layer: mapping each compensation factor into a 16-dimensional vector through linear transformation to enhance feature expression capability; bidirectional LSTM (Long Short-Term Memory) Memory (long short-term memory network) layer: performs time series modeling on historical environmental data sequences to capture long-term dependencies. This layer contains 64 memory units, has an output dimension of 24×64, and performs layer normalization at the same time. Attention mechanism layer: calculates the importance weight of each time point in the historical data by scaling the dot product attention, highlighting the impact of key periods (such as high temperature periods), with an output dimension of 24×64. Real-time feature enhancement layer: copies the current environmental data 24 times and splices it with the attention output of the historical data (dimension is 24×67) to enhance the corrective effect of the real-time environment on historical trends. CNN convolution layer: uses three 1×3 convolution kernels to extract local patterns (such as the correlation between sudden temperature changes and increased humidity), with an output dimension of 24×3. The convolution layer is residually connected to the fully connected layer 2. Global average pooling layer: compresses the time series dimension to 1 and generates a fixed-length vector containing spatiotemporal features (dimension is 1×3). Fully connected layers 1 and 2: Two fully connected layers (64 neurons per layer) activated by ReLUs map features into a multidimensional environmental feature space (output dimension 1×10). Output layer: A linear activation function is used to output the final 10-dimensional environmental feature vector, corresponding to abstract features such as temperature dynamic patterns, humidity sensitivity, and tilt angle stability.

[0100] In actual deployment, residual connections can be deleted or the layer order can be adjusted according to computing power, as well as the dimensions of input and output vectors can be changed. If in an actual detection environment, the environmental feature model is a 3-layer fully connected network, the input layer receives , where the first compensation factor set (temperature T, humidity H compensation coefficient); second compensation factor set (Vibration V dynamic compensation coefficient).

[0101] The latent feature vector is output through nonlinear transformation calculation of the hidden layer:

[0102]

[0103]

[0104] in, , ,... are the weight parameters for hidden layer training respectively, and the final layer of the model will be and The output layer generates a two-dimensional environmental feature vector F = [f1, f2], where f1 represents the temperature-humidity coupling effect and f2 represents the intensity of the transient vibration interference, which is used by the downstream compensation network.

[0105] This model mitigates the vanishing gradient problem through residual connections and employs layer normalization to stabilize the training process. During training, benchmark validation data (such as weighing sensor output under a standard laboratory environment) serves as the supervisory signal, using the mean squared error (MSE) as the loss function and the Adam optimizer to iteratively update parameters. Experiments demonstrate that this model effectively extracts complex features such as sudden temperature changes and humidity hysteresis in multiple environmental interference scenarios, reducing the compensated data error to ±0.03%, significantly outperforming traditional linear models.

[0106] In the second embodiment, after step S103, the method further includes:

[0107] Step S104: obtaining benchmark verification data of the weighing sensor within a target observation period.

[0108] It's important to note that benchmark verification data is reference data used to verify the measurement accuracy of load cells. It can be generated by standard metrology equipment or laboratory calibration to provide a true value reference for the load cell's output during a target observation period. For example, in truck scale calibration, benchmark verification data is the measured value of a standard weight of known weight. The target observation period is a pre-defined time period used to verify the effectiveness of load cell compensation. During this period, benchmark verification data (such as the weight of the standard weight) must be acquired simultaneously with the load cell's real-time measurement data.

[0109] Step S105 : adjusting the first compensation factor set based on the historical measurement data and the benchmark verification data to obtain a third compensation factor set.

[0110] It should be noted that the third compensation factor set is a parameter set obtained by dynamically optimizing the first compensation factor set by fusing historical measurement data with benchmark verification data, and is used to correct deviations in the first compensation factor set using the true value constraints of the benchmark data.

[0111] In one embodiment, the first compensation factor is the temperature effect coefficient , humidity influence coefficient , the benchmark data provides a relative error Δ. If high temperature and high humidity lead to nonlinear errors, a model can be established:

[0112]

[0113] in , is the historical mean of temperature and humidity, λ is the synergy coefficient, T is the ambient temperature, and H is the ambient humidity. After optimization, the third compensation factor set is obtained ,λ suppresses the sudden increase of error when the temperature and humidity exceed the standard.

[0114] Step S106: adjusting the second compensation factor set based on the compensated data and the benchmark verification data to obtain a fourth compensation factor set.

[0115] It should be noted that the fourth compensation factor set is a dynamic parameter set that performs feedback optimization on the second compensation factor set by comparing the residuals of the compensated data with the benchmark verification data.

[0116] It is understandable that the baseline data needs to be strictly time-synchronized with the compensated data to avoid residual distortion caused by the response delay of the weighing sensor; during the adjustment process, the adjustment range of the factor needs to be limited to prevent overfitting the current period data and losing the short-term adaptability of the second compensation factor.

[0117] In one example, the second compensation factor set is , corresponding to the high-frequency vibration suppression coefficient and temperature transient compensation coefficient respectively;

[0118] The output after compensation is , where v is the vibration intensity, ΔT is the temperature change rate, is the measured acceleration. The benchmark data gives the true value acceleration , calculate the residual .

[0119] Computing the partial derivatives of the residual with respect to the factors:

[0120]

[0121] Generate the fourth compensation factor set When , it is updated according to the following formula:

[0122]

[0123] in, represents the second compensation factor, represents the fourth compensation factor.

[0124] Step S107 : determining a measurement deviation value and a compensation deviation value of the weighing sensor according to the historical measurement data, the compensated data, the first compensation factor set, the second compensation factor set, the third compensation factor set, and the fourth compensation factor set.

[0125] Furthermore, in order to comprehensively consider the reconstruction error of the compensation factor set and the validity of the data, the deviation of the weighing sensor measurement and compensation can be evaluated more accurately. The step S107 may include:

[0126] determining a measurement deviation value of a weighing sensor according to a first reconstruction error between the first compensation factor set and the third compensation factor set, and a second reconstruction error between the second compensation factor set and the fourth compensation factor set;

[0127] Performing validity verification on the historical measurement data, the compensated data, the third compensation factor set, and the fourth compensation factor set using a trained data quality evaluator to generate an estimated loss value;

[0128] A compensation deviation value is determined according to the deviation value measured by the weighing sensor and the estimated loss value.

[0129] It should be noted that the first reconstruction error (the difference between the first compensation factor set and the optimized third compensation factor set) reflects the adaptability of historical data to the current compensation model. The second reconstruction error (the difference between the second compensation factor set and the optimized fourth compensation factor set) reflects the rationality of real-time environmental parameter adjustments. The data quality evaluator is a model trained with historical data that assesses the credibility of measurement data, compensation data, and compensation factors, outputting an estimated loss value to quantify data validity. The estimated loss value is the deviation of the comprehensive data credibility score from the preset standard, reflecting the potential impact of data quality on compensation effectiveness.

[0130] Understandably, the weights of the first and second reconstruction errors must be set based on the load cell type (e.g., temperature load cells prioritize historical errors, while vibration load cells prioritize dynamic errors) to avoid bias misjudgments caused by balanced weighting. The training set for the data quality evaluator must include abnormal load cell operating conditions (e.g., signal drift and sudden interference); otherwise, the estimated loss may be underestimated. Furthermore, the complexity of the data quality evaluator must be adapted to real-time computing resources to avoid delays in evaluation that affect the timeliness of compensation. Dynamic thresholds must be set for compensation deviations based on the application scenario, triggering alarms when these thresholds are exceeded.

[0131] Specifically, the first reconstruction error and the second reconstruction error can be root mean square errors (RMSs). Their weighted summation yields the load cell measurement deviation value. The data quality assessor scores the credibility of historical measurement data, compensated data, and compensation factors. For historical data, if data fluctuations exceed a normal range, the score is lowered; for compensated data, if the deviation from the benchmark validation data is significant, the score is lowered. The deviation of each score from a preset standard (e.g., the difference between the compensated data score and the full score) is calculated and weightedly combined to yield an estimated loss value. The load cell measurement deviation value is then divided by the estimated loss value to yield the final compensated deviation value.

[0132] It should be understood that the weight coefficients when determining the compensation deviation value need to be dynamically adjusted according to the scenario. For example, the real-time error weight is higher in a high temperature environment to avoid a single factor dominating the deviation calculation.

[0133] Furthermore, in order to dynamically adjust the compensation weight based on the scoring mechanism to avoid interference from low-quality data, the weighted combination of the estimated loss values ​​simultaneously realizes the comprehensive quantification of the compensation effect and supports the objective comparison of the algorithm performance. The step of verifying the validity of the historical measurement data, the compensated data, the third compensation factor set, and the fourth compensation factor set by the trained data quality evaluator to generate the estimated loss value may include:

[0134] Obtaining, by a data quality assessor, a first credibility score for the historical measurement data, a second credibility score for the compensated data, a third credibility score for the third compensation factor set, and a fourth credibility score for the fourth compensation factor set;

[0135] Calculating a first deviation value between the second credibility score and the first credibility score, a second deviation value between the third credibility score and a preset measurement score, and a third deviation value between the fourth credibility score and a preset compensation score;

[0136] The first deviation value, the second deviation value, and the third deviation value are weighted and combined to obtain an estimated loss value.

[0137] It should be noted that the first credibility score can be determined by checking data continuity or the proportion of outliers; the second credibility score can be determined by calculating the deviation from the benchmark verification data; the third credibility score and the fourth credibility score can be determined by evaluating the logical consistency of the compensation factor with historical and real-time environmental parameters.

[0138] In one example, the first compensation factor set , represents the compensation parameters of historical environment (such as temperature, humidity). The third compensation factor set Obtained by optimizing historical data. The second compensation factor set Indicates the compensation parameters collected in real time. The fourth compensation factor set Obtained through real-time data optimization. The regularized Frobenius norm is used to measure the optimization deviation of the compensation factor, the first reconstruction error:

[0139]

[0140] Where, is the Frobenius norm, which calculates the square root of the sum of the squares of the matrix elements. is the L1 regularization coefficient to prevent overfitting. Used to measure the difference before and after the historical compensation factor adjustment, correspondingly, the second reconstruction error:

[0141]

[0142] in, Measure the adjustment of the real-time compensation factor, is the L1 regularization coefficient. The comprehensive measurement deviation value generated by adaptive weighted fusion is:

[0143]

[0144] in, is the historical error weight, which is dynamically adjusted by the sensor usage time. The calculation process is as follows:

[0145]

[0146] in, For usage time, is the design life threshold, is the decay rate. The square root of the real-time error is taken to reduce the impact of sudden interference. The data quality evaluator is a function Q, which outputs the credibility score of each data: historical data score , data score after compensation , optimize historical scoring , optimize real-time scoring The estimated loss value of the evidence is obtained through the compensation effect, the rationality of the measurement factor and the effectiveness of the compensation factor:

[0147]

[0148] Among them, the weight Follow The deviation between them is dynamically adjusted to strengthen the penalty of the main error source.

[0149] Step S108: If the compensation deviation value exceeds the preset convergence threshold, the historical measurement data is adjusted based on the measurement deviation value of the weighing sensor, and the step of obtaining the benchmark verification data of the weighing sensor within the target observation period is re-executed until the compensation deviation value is less than the preset convergence threshold.

[0150] It should be noted that the preset convergence threshold is a set calibration accuracy target value, which represents the maximum allowable compensation deviation.

[0151] It should be understood that when the compensation deviation exceeds the preset convergence threshold, it indicates that the current compensation accuracy cannot meet the preset accuracy requirements. Exceeding the limit may be caused by the following reasons: 1. Load cell aging or environmental drift, causing the statistical characteristics of historical data to deviate from reality; 2. The original benchmark verification data does not cover the current operating conditions (such as extreme temperatures, new interference).

[0152] In one example, when correcting environmental parameters or measurement values ​​in historical data, if there is a systematic deviation in the temperature compensation factor, the historical temperature data can be scaled according to the deviation ratio; if there is a baseline drift in the vibration data, an offset can be added to the historical vibration signal.

[0153] In this embodiment, the historical measurement data of the weighing sensor is obtained, and a first compensation factor set of the historical measurement data is determined, wherein the first compensation factor set is used to characterize different influencing factors of the measurement environment; when it is detected that the weighing sensor starts measuring, a second compensation factor set of the weighing sensor in the current measurement cycle is determined; the first compensation factor set and the second compensation factor set are input into the trained environmental feature model to extract the multidimensional environmental features of the current measurement environment. The multidimensional environmental features of the current measurement environment can be extracted more accurately. The combination of the statistical information of the historical data and the real-time situation of the current environment makes the environmental features more accurate and enhances the robustness of the environmental feature modeling. In addition, the algorithm is forced to optimize by presetting the convergence threshold to avoid the long-term divergence of the compensation deviation, which is particularly suitable for the continuous monitoring needs of industrial scenarios.

[0154] Reference Figure 4 , Figure 4 This is a flow chart of the third embodiment of the weighing sensor data compensation method of the present application. Based on the above second embodiment, the third embodiment of the weighing sensor data compensation method of the present application is proposed.

[0155] In the third embodiment, step S30 includes:

[0156] Step S301: Calculate the exponential function transformation value of the weight coefficient of each sampling point of the weighing sensor.

[0157] It should be noted that the contribution of data from different sampling points of a weighing sensor to the final result varies due to factors such as the measurement environment, measurement accuracy, and data reliability. The sampling point weight coefficient is a numerical value used to measure the importance of the data at each sampling point. It can be set during installation and deployment, and is typically determined based on signal quality (such as signal-to-noise ratio), temporal proximity (for example, more weight for recent data), or environmental stability (for example, more weight for periods of low interference). The exponential function transform maps the weight coefficient to a nonlinear space through exponential calculations, which is used to enhance or suppress the contribution of specific sampling points.

[0158] It can be understood that the sampling points can be different weighing sensors, or different sampling times of a single weighing sensor.

[0159] In one example, according to time decay Assign initial weights to each sampling point .in, is the sampling time series, is the current time. The initial weight is hour, α ∈(0,1), the closer the time is, the higher the weight is. Apply an exponential function to the initial weight to adjust its distribution characteristics:

[0160]

[0161] Where k is the adjustment factor. When k is greater than 0, high weights are amplified, and when k is less than 0, low weights are suppressed. To suppress the influence of early data, k=−1 is taken, and the exponential function transformation value of the transformed weight is obtained:

[0162]

[0163] in, is the transformed exponential function value after transformation.

[0164] Step S302 : normalize the exponential function transformation value of each sampling point to generate a normalized weight coefficient.

[0165] In one example, the exponential function transformation value needs to be performed after the change to ensure that the total weight sums to 1:

[0166] in, is the normalized weight coefficient after normalization.

[0167] Step S303: assigning a corresponding normalized weight coefficient to each of the dynamic feature components to obtain a target feature component corresponding to each sampling point.

[0168] It is understood that for high-sampling-rate weighing sensors, sliding windows or parallel computing can be used to accelerate weight normalization and feature fusion. When assigning a corresponding normalized weight coefficient to each dynamic feature component, the normalized weight can be multiplied by the corresponding dynamic feature component at each sampling point and summed to generate the target feature component.

[0169] Step S304 : Inputting each of the target feature components and the multi-dimensional environmental features into different compensation modules respectively to generate a plurality of compensation parameter components.

[0170] It is understandable that the compensation module may include temperature compensation parameters, the humidity compensation parameters, and the posture compensation parameters, etc., which are not limited in this embodiment.

[0171] In the third embodiment, step S40 includes:

[0172] Step S401 : weightedly fusing the temperature compensation parameter, the humidity compensation parameter, and the attitude compensation parameter using a multi-source compensation parameter fusion algorithm to generate a comprehensive compensation parameter matrix including environmental dynamic characteristics.

[0173] It's important to note that the multi-source compensation parameter fusion algorithm assigns weights to compensation parameters from different compensation modules (such as temperature, humidity, and attitude) based on the dynamic characteristics of the environment and performs a weighted combination. This algorithm aims to eliminate the limitations of a single compensation model and improve the accuracy of comprehensive compensation in complex environments. The comprehensive compensation parameter matrix is ​​composed of the fused compensation parameters.

[0174] In one example, along the sampling sequence of different load cells (sampling points , ,..., ) Generate a comprehensive compensation parameter matrix as follows:

[0175]

[0176] in, is the temperature compensation parameter, is the humidity compensation parameter, is the attitude compensation parameter. This is a comprehensive compensation parameter, obtained by averaging the temperature compensation parameter, humidity compensation parameter, and attitude compensation parameter. Each row in the matrix contains the fused parameter and each independent parameter, supporting subsequent analysis and backtracking.

[0177] Step S402: input the comprehensive compensation parameter matrix and the real-time measurement data of the weighing sensor into the trained compensation operation network, and calculate the compensation offset through multi-layer nonlinear mapping.

[0178] It should be noted that real-time measurement data is the raw measurement value output by the load cell at the current moment. The pre-trained deep learning model of the compensation operation network maps compensation parameters and raw data into a compensation offset through multi-layer nonlinear transformations, addressing nonlinear errors in complex environments. The compensation offset is a correction to the measurement data of the load cell under the current measurement environment.

[0179] In one example, the real-time measurement data H and the current row of the comprehensive compensation parameter matrix are concatenated into an input vector X=[H, Θ(t)].

[0180] The compensation operation network calculates the compensation offset through multi-layer nonlinear mapping :

[0181] Hidden layer 1, activation function ReLU: ;

[0182] Hidden layer 2, activation function Sigmoid: ;

[0183] Output layer, linear activation Where Wi, bi are the weights and biases in the compensation operation network, σ is the activation function, and the compensation offset Superimpose on the original data to obtain compensated data.

[0184] Step S403 : Compensating and dynamically correcting the measurement data of the weighing sensor in the current measurement environment according to the compensation offset to generate compensated data.

[0185] In this embodiment, the exponential function transformation value of the weight coefficient for each sampling point of the load cell is calculated; the exponential function transformation value for each sampling point is normalized to generate a normalized weight coefficient; a corresponding normalized weight coefficient is assigned to each dynamic feature component to obtain a target feature component corresponding to each sampling point; each target feature component and the multidimensional environmental characteristics are input into different compensation modules to generate multiple compensation parameter components. Assigning differentiated weights to the target feature components improves the algorithm's response to critical environmental disturbances (such as transient high temperatures). A comprehensive compensation matrix is ​​constructed through weighted fusion to address complex issues such as temperature-humidity coupling and attitude-temperature cross-interference. This approach can better adapt to complex and changing measurement environments.

[0186] It should be noted that the above examples are only used to understand the present application and do not constitute a limitation on the weighing sensor data compensation method of the present application. More forms of simple transformations based on this technical concept are all within the scope of protection of the present application.

[0187] This application also provides a weighing sensor data compensation device, please refer to Figure 5, the weighing sensor data compensation device includes:

[0188] The feature extraction module 10 is used to obtain the multi-dimensional environmental features of the weighing sensor in the current measurement environment when detecting that the weighing sensor starts measuring;

[0189] A component acquisition module 20 is used to extract the dynamic characteristic components corresponding to each time dimension of the weighing sensor based on the historical measurement data of the weighing sensor;

[0190] a parameter determination module 30 for inputting each of the dynamic feature components and the multi-dimensional environmental features into different compensation modules to generate a plurality of compensation parameter components;

[0191] The data compensation module 40 is used to fuse all the compensation parameter components to generate comprehensive compensation parameters, and input the comprehensive compensation parameters into the trained compensation operation network to compensate the measurement data of the weighing sensor in the current measurement environment to obtain compensated data.

[0192] The load cell data compensation device provided in this application, utilizing the load cell data compensation method described in the aforementioned embodiments, can address the technical issue of existing compensation algorithms neglecting the indirect effects of mounting accessories and the operating environment, leading to decreased compensation accuracy after long-term use. Compared to the prior art, the load cell data compensation device provided in this application achieves the same beneficial effects as the load cell data compensation method described in the aforementioned embodiments. Other technical features of the load cell data compensation device are the same as those disclosed in the aforementioned embodiments and are not further elaborated here.

[0193] The present application provides a weighing sensor data compensation device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the weighing sensor data compensation method in the above-mentioned embodiment 1.

[0194] Reference below Figure 6, which shows a schematic diagram of the structure of a load cell data compensation device suitable for implementing the embodiments of the present application. The load cell data compensation device in the embodiments of the present application can include, but is not limited to, mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Descriptions), PMPs (Portable Media Players), and in-vehicle terminals (e.g., in-vehicle navigation terminals), as well as fixed terminals such as digital TVs and desktop computers. Figure 6 The weighing sensor data compensation device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0195] like Figure 6 As shown, the load cell data compensation device may include a processing device 1001 (e.g., a central processing unit, graphics processing unit, etc.), which can perform various appropriate actions and processes based on programs stored in a read-only memory 1002 or programs loaded from a storage device 1003 into a random access memory 1004. The random access memory 1004 also stores various programs and data required for the operation of the load cell data compensation device. The processing device 1001, the read-only memory 1002, and the random access memory 1004 are interconnected via a bus 1005. An input / output interface 1006 is also connected to the bus. Typically, the following systems can be connected to the input / output interface 1006: an input device 1007 including, for example, a touchscreen, touchpad, keyboard, mouse, image load cell, microphone, accelerometer, gyroscope, etc.; an output device 1008 including, for example, a liquid crystal display (LCD), speaker, vibrator, etc.; a storage device 1003 including, for example, a magnetic tape or hard disk; and a communication device 1009. The communication device 1009 can allow the load cell data compensation device to communicate wirelessly or wired with other devices to exchange data. Although the figure shows a load cell data compensation device with various systems, it should be understood that it is not required to implement or have all of the systems shown. More or fewer systems may be implemented or have alternatively.

[0196] In particular, according to the embodiments disclosed in the present application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, the embodiments disclosed in the present application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program comprising program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network via a communication device, or installed from a storage device 1003, or installed from a read-only memory 1002. When the computer program is executed by the processing device 1001, the above-mentioned functions defined in the method of the embodiment disclosed in the present application are performed.

[0197] The load cell data compensation device provided in this application, utilizing the load cell data compensation method described in the aforementioned embodiment, can address the technical issue of existing compensation algorithms neglecting the indirect effects of mounting accessories and the operating environment, leading to decreased compensation accuracy after long-term use. Compared to the prior art, the load cell data compensation device provided in this application achieves the same beneficial effects as the load cell data compensation method described in the aforementioned embodiment. Other technical features of this load cell data compensation device are the same as those disclosed in the aforementioned embodiment and are not further elaborated here.

[0198] It should be understood that the various parts disclosed in this application can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in any one or more embodiments or examples in a suitable manner.

[0199] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

[0200] The present application provides a computer-readable storage medium having computer-readable program instructions (ie, a computer program) stored thereon, wherein the computer-readable program instructions are used to execute the weighing sensor data compensation method in the above embodiment.

[0201] The computer-readable storage medium provided herein may be, for example, a USB flash drive, but is not limited to electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, systems, or devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to, an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In this embodiment, the computer-readable storage medium may be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, system, or device. The program code contained on the computer-readable storage medium may be transmitted using any suitable medium, including, but not limited to, wires, optical cables, RF (Radio Frequency), etc., or any suitable combination thereof.

[0202] The computer-readable storage medium may be included in the weighing sensor data compensation device; or may exist independently without being assembled into the weighing sensor data compensation device.

[0203] The computer-readable storage medium carries one or more programs. When the one or more programs are executed by the load cell data compensation device, the load cell data compensation device executes the load cell data compensation method described above.

[0204] Computer program code for performing the operations of the present application may be written in one or more programming languages, or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a stand-alone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0205] The flowcharts and block diagrams in the accompanying drawings illustrate the possible implementation architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present application. In this regard, each box in the flowchart or block diagram can represent a module, program segment or part of code, and the module, program segment or part of code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in an order different from that marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flowchart, as well as the combination of boxes in the block diagram and / or flowchart, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or can be implemented using a combination of dedicated hardware and computer instructions.

[0206] The modules described in the embodiments of the present application may be implemented in software or hardware, wherein the name of a module does not necessarily limit the unit itself.

[0207] The computer-readable storage medium provided herein stores computer-readable program instructions (i.e., a computer program) for executing the aforementioned load cell data compensation method. This computer-readable storage medium addresses the technical issue of existing compensation algorithms neglecting the indirect effects of mounting accessories and the operating environment, resulting in decreased compensation accuracy after long-term use. Compared to the prior art, the beneficial effects of the computer-readable storage medium provided herein are similar to those of the load cell data compensation method provided in the aforementioned embodiments and are not further elaborated here.

[0208] The above descriptions are only some embodiments of the present application and do not limit the scope of the present application. All equivalent structural transformations made using the contents of the present application specification and drawings under the technical concept of the present application, or direct / indirect application in other related technical fields are included in the scope of protection of the present application.

Claims

1. A weighing sensor data compensation method, characterized in that: The weighing sensor data compensation method includes: When detecting that the weighing sensor starts measuring, obtaining the multi-dimensional environmental characteristics of the weighing sensor in the current measurement environment; Extracting dynamic characteristic components corresponding to each time dimension of the weighing sensor according to historical measurement data of the weighing sensor; Inputting each of the dynamic feature components and the multi-dimensional environmental features into different compensation modules respectively to generate multiple compensation parameter components; Fusing all the compensation parameter components to generate a comprehensive compensation parameter, and inputting the comprehensive compensation parameter into the trained compensation operation network to compensate the measurement data of the weighing sensor in the current measurement environment to obtain compensated data; The step of obtaining the multi-dimensional environmental characteristics of the weighing sensor in the current measurement environment when detecting that the weighing sensor starts measuring includes: Acquire historical measurement data of the weighing sensor, and determine a first compensation factor set for the historical measurement data, where the first compensation factor set is used to characterize different influencing factors of the measurement environment; When detecting that the weighing sensor starts measuring, determining a second compensation factor set of the weighing sensor in a current measurement cycle; Inputting the first compensation factor set and the second compensation factor set into the trained environmental feature model to extract the multidimensional environmental features of the current measurement environment; The compensation parameter components include temperature compensation parameters, humidity compensation parameters and attitude compensation parameters; The step of fusing all the compensation parameter components to generate a comprehensive compensation parameter, inputting the comprehensive compensation parameter into a trained compensation operation network, compensating the measurement data of the weighing sensor in the current measurement environment, and obtaining compensated data includes: Performing weighted fusion on the temperature compensation parameter, the humidity compensation parameter, and the attitude compensation parameter through a multi-source compensation parameter fusion algorithm to generate a comprehensive compensation parameter matrix including environmental dynamic characteristics; Inputting the comprehensive compensation parameter matrix and the real-time measurement data of the weighing sensor into the trained compensation operation network, and calculating the compensation offset through multi-layer nonlinear mapping; The measurement data of the weighing sensor in the current measurement environment is compensated and dynamically corrected according to the compensation offset to generate compensated data.

2. The weighing sensor data compensation method according to claim 1, characterized in that: After the step of inputting the first compensation factor set and the second compensation factor set into the trained environmental feature model to extract the multi-dimensional environmental features of the current measurement environment, the method further includes: Obtaining benchmark verification data of the load cell within a target observation period; adjusting the first set of compensation factors based on the historical measurement data and the benchmark verification data to obtain a third set of compensation factors; adjusting the second compensation factor set based on the compensated data and the benchmark verification data to obtain a fourth compensation factor set; Determining a measurement deviation value and a compensation deviation value of a weighing sensor according to the historical measurement data, the compensated data, the first compensation factor set, the second compensation factor set, the third compensation factor set, and the fourth compensation factor set; If the compensation deviation value exceeds the preset convergence threshold, the historical measurement data is adjusted based on the measurement deviation value of the weighing sensor, and the step of obtaining the benchmark verification data of the weighing sensor within the target observation period is re-executed until the compensation deviation value is less than the preset convergence threshold.

3. The weighing sensor data compensation method according to claim 2, wherein: The step of determining the load cell measurement deviation value and the compensation deviation value based on the historical measurement data, the compensated data, the first compensation factor set, the second compensation factor set, the third compensation factor set, and the fourth compensation factor set includes: determining a measurement deviation value of a weighing sensor according to a first reconstruction error between the first compensation factor set and the third compensation factor set, and a second reconstruction error between the second compensation factor set and the fourth compensation factor set; Performing validity verification on the historical measurement data, the compensated data, the third compensation factor set, and the fourth compensation factor set using a trained data quality evaluator to generate an estimated loss value; A compensation deviation value is determined according to the deviation value measured by the weighing sensor and the estimated loss value.

4. The weighing sensor data compensation method according to claim 3, wherein: The step of performing validity verification on the historical measurement data, the compensated data, the third compensation factor set, and the fourth compensation factor set by a trained data quality evaluator to generate an estimated loss value includes: Obtaining, by a data quality assessor, a first credibility score for the historical measurement data, a second credibility score for the compensated data, a third credibility score for the third compensation factor set, and a fourth credibility score for the fourth compensation factor set; Calculating a first deviation value between the second credibility score and the first credibility score, a second deviation value between the third credibility score and a preset measurement score, and a third deviation value between the fourth credibility score and a preset compensation score; The first deviation value, the second deviation value, and the third deviation value are weighted and combined to obtain an estimated loss value.

5. The weighing sensor data compensation method according to any one of claims 1 to 4, characterized in that: The step of inputting each of the dynamic feature components and the multi-dimensional environmental features into different compensation modules to generate multiple compensation parameter components includes: Calculating the exponential function transformation value of the weight coefficient of each sampling point of the weighing sensor; Normalize the exponential function transformation value of each sampling point to generate a normalized weight coefficient; Assigning a corresponding normalized weight coefficient to each of the dynamic feature components to obtain a target feature component corresponding to each sampling point; Each of the target feature components and the multi-dimensional environmental features are input into different compensation modules to generate a plurality of compensation parameter components.

6. A weighing sensor data compensation device, characterized in that: The device comprises: A feature extraction module is used to obtain the multi-dimensional environmental features of the weighing sensor in the current measurement environment when it is detected that the weighing sensor starts measuring; A component acquisition module, configured to extract dynamic characteristic components corresponding to each time dimension of the weighing sensor based on historical measurement data of the weighing sensor; a parameter determination module, configured to input each of the dynamic feature components and the multi-dimensional environmental feature into different compensation modules to generate a plurality of compensation parameter components, wherein the compensation parameter components include a temperature compensation parameter, a humidity compensation parameter, and a posture compensation parameter; A data compensation module is used to fuse all the compensation parameter components to generate a comprehensive compensation parameter, and input the comprehensive compensation parameter into the trained compensation operation network to compensate the measurement data of the weighing sensor in the current measurement environment to obtain compensated data; The feature extraction module is further configured to obtain historical measurement data of the weighing sensor, determine a first set of compensation factors for the historical measurement data, the first set of compensation factors being used to characterize different influencing factors of the measurement environment; upon detecting that the weighing sensor has started measuring, determine a second set of compensation factors for the weighing sensor in the current measurement cycle; and input the first set of compensation factors and the second set of compensation factors into a trained environmental feature model to extract multidimensional environmental features of the current measurement environment. The data compensation module is further used to perform weighted fusion of the temperature compensation parameter, the humidity compensation parameter and the posture compensation parameter through a multi-source compensation parameter fusion algorithm to generate a comprehensive compensation parameter matrix containing environmental dynamic characteristics; input the comprehensive compensation parameter matrix and the real-time measurement data of the weighing sensor into the trained compensation operation network, and calculate the compensation offset through multi-layer nonlinear mapping; and dynamically correct the measurement data of the weighing sensor in the current measurement environment according to the compensation offset to generate compensated data.

7. A weighing sensor data compensation device, characterized in that: The device comprises: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program is configured to implement the steps of the weighing sensor data compensation method according to any one of claims 1 to 5.

8. A storage medium, characterized in that: The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, the steps of the weighing sensor data compensation method according to any one of claims 1 to 5 are implemented.

Images