Multi-scene weight detection and optimization method and system for electronic scale

Abstract

The invention provides a multi-scene weight detection and optimization method and system for an electronic scale, and the method comprises the following steps: S1, detecting the weight of a target object through the electronic scale, generating initial weight data, and monitoring the environment in a preset range to generate environment data; s2, based on data analysis, inputting the environment data into a scene recognition model to recognize a current weighing scene, and outputting a scene recognition result; s3, an optimization algorithm matched with the scene recognition result is called from a preset algorithm library according to the scene recognition result, the optimization algorithm is executed, an optimization result is output, and the optimization result at least comprises optimized weight data; s4, displaying the optimized weight data at a preset position of an electronic scale screen; and S5, storing an optimization result in a database so as to carry out incremental training on the scene recognition model.

Description

Technical Field

[0001] The present invention belongs to the field of electronic scales, and in particular relates to a multi-scenario weight detection and optimization method and system for electronic scales. Background Art

[0002] Electronic scales are widely used in logistics, retail, industrial production and other fields. In actual use, the diversity and complexity of different scenarios bring different challenges to the weighing system, resulting in some limitations in weighing accuracy and stability of existing technologies, which are mainly reflected in the following aspects: In different environments, factors such as temperature, humidity and tilt angle will have a significant impact on the weighing results of electronic scales. The current electronic scale system cannot automatically adjust the measurement parameters according to changes in the external environment; existing technologies are usually unable to maintain data accuracy and stability in high-frequency weighing environments, which may cause data lags, frequent fluctuations or weighing errors; existing systems usually rely on preset fixed algorithms or calibration methods, and it is difficult to dynamically adjust the weighing algorithm in a variety of application scenarios. Summary of the invention The purpose of this application is to provide a multi-scenario weight detection and optimization method and system for an electronic scale to improve the adaptability and weighing accuracy of the system and meet the diverse needs of multiple scenarios.

[0003] In order to achieve the above-mentioned objectives, one aspect of the present application provides a multi-scenario weight detection and optimization method for an electronic scale, comprising the following steps: S1: using an electronic scale to detect the weight of a target item to generate initial weight data, and monitoring the environment within a preset range to generate environmental data; S2: based on data analysis, inputting the environmental data into a scene recognition model to identify the current weighing scene, and outputting the scene recognition result; S3: based on the scene recognition result, calling an optimization algorithm matching the scene recognition result from a preset algorithm library, executing the optimization algorithm, and outputting the optimization result, wherein the optimization result at least includes the optimized weight data; S4: displaying the optimized weight data at a preset position on the electronic scale screen; S5: storing the optimization result in a database to perform incremental training on the scene recognition model.

[0004] Preferably, based on data analysis including: the environmental data including temperature, humidity and tilt angle; and the weighing frequency of the electronic scale.

[0005] Preferably, S2 includes: S201: determining whether the current weighing scene can be identified through environmental data; if not, executing S202: using weight data as an auxiliary judgment, identifying the current weighing scene through the environmental data and weight data; if yes, executing S3.

[0006] Preferably, the categories of multiple scenarios include high environmental fluctuation scenarios, high frequency weighing scenarios and high precision requirement scenarios. Preferably, the algorithm library includes an environmental fluctuation optimization algorithm, a high-frequency weighing optimization algorithm and a high-precision optimization algorithm.

[0007] Preferably, the environmental fluctuation optimization algorithm calculates the corrected weight data using a compensation model according to the environmental data, and the calculation formula is: ; Among them, W i Indicates the initial weight data, K T Indicates the temperature compensation coefficient, T indicates the current temperature value, T0 indicates the reference temperature value, K H Indicates the humidity compensation coefficient, H indicates the current humidity value, and H0 indicates the reference humidity value; The high-frequency weighing optimization algorithm reduces short-term fluctuations caused by frequent measurements through a smoothing algorithm, and its calculation formula is: ; Among them, W i represents the weight data of the i-th weighing, and n represents the window size of data smoothing; The high-precision optimization algorithm removes noise and micro-fluctuations through a filtering algorithm, and its calculation formula is: ; Among them, W g represents the filtering result of the previous moment, and K represents the Kalman gain.

[0008] Preferably, S2 includes: setting priorities for different scene categories, and when more than two scene categories are detected simultaneously according to the environmental data, outputting a scene recognition result of a scene category with a higher priority.

[0009] Preferably, S3 includes a scene fusion method: S301: setting importance weights for different scene categories, and forming a weight matrix according to the combination of scene fusion, the calculation formula of which is: ; Among them, W q represents the optimized weight after fusion, W e represents the weight of the high environmental fluctuation scenario, W r Indicates high-frequency weighing scenarios, W p Indicates the scene with high precision requirement, α, β and γ represent the weighting coefficients of the scene fusion type, and α+β+γ=1; S302: According to the fused scene category weights, an optimization algorithm is selected for dynamic adjustment, and its calculation formula is: ; Here, P represents a set of adjustable parameters used in the optimization algorithm.

[0010] On the other hand, the present application provides a multi-scenario weight detection and optimization system for an electronic scale, using the above method, including: a weight detection module, the weight detection module is configured to detect the weight of a target item and generate initial weight data; an environment monitoring module, the environment monitoring module is configured to monitor the environment within a preset range to generate environment data; a scene recognition module, the scene recognition module is configured to identify the current weighing scene according to the environment data and output the scene recognition result; an algorithm library calling module, the algorithm library calling module is configured to call a matching optimization algorithm according to the scene recognition result and output the optimization result; a display module, the display module is configured to display the optimized weight data at a preset position on the electronic scale screen; a storage module, the storage module is configured to store the optimization result in a database.

[0011] Preferably, the environmental monitoring module includes: a temperature sensor, which is configured to monitor temperature changes within a preset range; a humidity sensor, which is configured to monitor humidity changes within a preset range; and a tilt sensor, which is configured to detect the tilt angle of the electronic scale.

[0012] The technical solution provided by this application can achieve the following beneficial effects: 1. This solution designs an algorithm library and a multi-scenario recognition mechanism. The scenario recognition module automatically identifies the type of weighing scenario (such as high environmental fluctuations, high-frequency weighing, and high-precision requirement scenarios), and calls the corresponding optimization algorithm to ensure that when the scenario changes, the system can make adaptive adjustments according to the specific scenario requirements, improve the weighing accuracy and stability, and expand the scope of application of electronic scales in different application scenarios; 2. This solution integrates processing strategies for multiple scene types and can dynamically adjust optimization methods when multiple scene types are detected, especially scene fusion methods, so that the system can smoothly switch between multiple scene requirements, ensuring the consistency and accuracy of weighing data during scene changes, and avoiding performance degradation due to scene complexity. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are merely embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying any creative work.

[0014] Figure 1It is a flow chart of a multi-scenario weight detection and optimization method provided in an embodiment of the present application; Figure 2 is a flow chart of a scene recognition method provided by an embodiment of the present application; Figure 3 is a flow chart of the scene fusion method provided in an embodiment of the present application; Figure 4 is a flow chart of the scene recognition model training method provided in an embodiment of the present application; Figure 5 It is a structural diagram of a multi-scenario weight detection and optimization system provided in an embodiment of the present application; Figure 6 is a structural diagram of an environment detection module provided in an embodiment of the present application; Among them: a weight detection module 10, an environment monitoring module 20, a temperature sensor 201, a humidity sensor 202, an inclination sensor 203, a scene recognition module 30, an algorithm library calling module 40, a display module 50 and a storage module 60. DETAILED DESCRIPTION

[0015] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present invention, and should not be construed as limiting the present invention.

[0016] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.

[0017] like Figure 1 As shown, this embodiment provides a multi-scenario weight detection and optimization method for an electronic scale, comprising the following steps S1-S5: S1: Use an electronic scale to detect the weight of the target object to generate initial weight data, and monitor the environment within a preset range to generate environmental data.

[0018] Specifically, the target object is placed on an electronic scale, and the system detects the weight of the object through the weighing sensor of the electronic scale to generate initial weight data.

[0019] The system includes a control unit, which processes various data and stores initial weight data in a system database. The initial weight data provides a reference for the final weighing result.

[0020] During the detection process, the system samples the weight data multiple times and calculates its average value to reduce the impact of instantaneous interference and obtain more stable initial weight data. The electronic scale is provided with a temperature sensor and a humidity sensor within a preset range to collect current ambient temperature and humidity data. The temperature sensor and the humidity sensor collect data at regular intervals, for example, once every 5 seconds, so as to reflect the environmental conditions in real time. The collected temperature and humidity data are transmitted to the control unit and stored together with the initial weight data to provide a reference for subsequent optimization algorithms.

[0021] To ensure the accuracy of the weighing data, the system also uses tilt sensors such as accelerometers or gyroscopes to detect the tilt angle of the electronic scale platform. The tilt sensor collects data at regular intervals, for example, every 5 seconds, to ensure the stability of the electronic scale platform during the weighing process. When it is detected that the tilt angle exceeds the preset safety range, the system will prompt to adjust the position of the electronic scale, or perform tilt compensation when recording the initial weight data. The tilt angle data is also transmitted to the control unit and used together with other environmental data for subsequent optimization steps.

[0022] It can be imagined that the control unit integrates the initial weight data and environmental data to form a complete data set, which includes multi-dimensional information such as initial weight value, temperature, humidity, tilt angle, etc., and in the data integration process, the system verifies all data to eliminate abnormal values. For example, if a certain data collection moment has a large fluctuation (such as abnormal temperature jump), the system will discard the data and re-sample. After completing the data integration and preprocessing, the system stores the data set in the database. In order to facilitate distinction, it can also be marked as "initial detection data."

[0023] S2: Based on data analysis, the environmental data is input into a scene recognition model to identify the current weighing scene, and the scene recognition result is output.

[0024] Specifically, the environmental data collected by S1 is used as the input data of the scene recognition model. The system pre-trains the scene recognition model according to the scene recognition requirements. The scene recognition model selects at least one of the algorithm models such as decision tree, support vector machine and neural network, and stores the scene recognition model in the algorithm library. When new environmental data is input into the scene recognition model, the scene recognition model analyzes and judges the current weighing environment according to the characteristics of the input data. The scene recognition model outputs a scene label to distinguish the current weighing scene, such as "high environmental fluctuation scene", "high frequency weighing scene" or "high precision requirement scene".

[0025] It can be imagined that the high environmental fluctuation scenario refers to the scenario where the temperature and humidity change frequently in a short period of time. This scenario may occur in unstable environments such as outdoors and factory workshops. For this, the temperature and humidity change rate or amplitude can be monitored through real-time data of temperature and humidity; when the temperature and humidity change frequency is high or the change amplitude is large, it can be determined as a high environmental fluctuation scenario. For example, if the temperature change fluctuates by more than 3°C within 10 minutes, or the humidity fluctuates by more than 10%, it can be marked as a high environmental fluctuation scenario. The high-frequency weighing scenario is judged by the frequency of weighing operations and the frequency of weighing operations is recorded. Assuming that the system detects that the weighing frequency is higher than a certain threshold in a unit time, the system automatically identifies it as a high-frequency weighing scenario. The high-precision requirement scenario usually occurs when there are high precision requirements for the weighing results, such as laboratory environments or fine measurement scenarios. This is related to the stability of temperature, humidity and tilt angle. For example, when the temperature, humidity and tilt angle data are kept in a small range of fluctuations, such as temperature changes less than 1°C, humidity changes less than 5%, and tilt angles less than 1°, it can be regarded as a high-precision requirement scenario.

[0026] During the recognition process, the scene recognition model determines the fluctuation of the current environment by analyzing environmental data such as temperature, humidity and tilt angle. It determines whether the frequency of weighing requirements meets the standard of high-frequency weighing by identifying the weighing frequency characteristics. Based on the specific temperature and humidity range and the stability of the tilt angle, the model can determine whether the current scene requires high precision.

[0027] One possible scenario is that when multiple scene categories are detected, the current weighing scene is determined by a priority strategy. Specifically, priorities are defined for different scene categories. For example, the priorities are from high to low as "high environmental fluctuation scene" > "high precision requirement scene" > "high frequency weighing scene". In this way, when the system detects both high environmental fluctuation scene and high frequency weighing scene, the algorithm that adapts to the high environmental fluctuation scene is adopted first. In one implementation, in a logistics scene, if the temperature and humidity change dramatically and the weighing frequency is higher than the threshold, the system will preferentially identify it as a "high environmental fluctuation scene".

[0028] like Figure 2As shown, S2 includes: S201: determine whether the current weighing scene can be identified through environmental data; if not, execute S202: use weight data as auxiliary judgment, and identify the current weighing scene through the environmental data and weight data; if so, execute S3.

[0029] Specifically, in the scene recognition process, weight data can be used as an auxiliary factor, especially when environmental data is not sufficient to identify the scene alone. For example, if the weight data fluctuates frequently or jumps abnormally during a weighing process, and the environmental data does not change significantly, it may be because the system is disturbed by micro-vibration or other unstable factors. In one implementation, in a logistics scenario, if the weight data continues to fluctuate up and down, it may be due to large external vibrations or unstable placement of the goods. In this case, the abnormal weight data can assist in identifying unstable scenes, and the system can enhance its anti-interference ability through algorithms.

[0030] The output result of the scene recognition model is the recognition result of the current weighing scene, and the output is the scene label. The system stores the scene label in the database and records it together with the environmental data and initial weight data for subsequent traceability analysis and algorithm verification.

[0031] In one implementation, a classification model based on a support vector machine is used as a scene recognition model. The model is suitable for processing multi-dimensional feature data (such as temperature, humidity, tilt angle, and weighing frequency). It is assumed that the scene recognition model is trained using an SVM model to obtain the following decision boundary interval: the decision value difference for high-frequency weighing scenarios is -0.5 ≤ f(X')<0, the decision value difference for high environmental fluctuation scenarios is 0 ≤ f(X')<0.5, and the decision value difference for high-precision requirement scenarios is f(X') ≥0.5.

[0032] When the system detects weight data, it automatically collects environmental data and weighing frequency. For example, in a weighing task, the following data is obtained: temperature T = 35 ° C, humidity H = 80%, tilt angle θ = 3.5 °, weighing frequency f = 15 times / minute. The collected data is combined into a feature vector X = [T, H, θ, f] = [35, 80, 3.5, 15] and the feature vector is standardized. The standardization formula is as follows: ; Among them, X' is the new vector obtained after standardization, μ is the mean of each feature, and σ is the standard deviation.

[0033] For the SVM classification model, the decision function can be expressed as: ; Among them, w=[w1, w2, w3, w4] represents the weight vector learned by the model during training. Each w i(i=1, 2, 3, 4) represents the weight of the corresponding feature, b represents the bias term, which is a constant learned by the model during training. The feature vector X is normalized to obtain X'=[0.5, 1.2, 0.3, -0.8], and the weight and bias parameters of the model are as follows: weight vector w=[0.4, -0.3, 0.5, 0.1], bias term b=-0.2, and the calculation process is as follows: Dot product part: w×X'=0.4×0.5+(−0.3)×1.2+0.5×0.3+0.1×(−0.8)=0.2−0.36+0.15−0.08=−0.09; Decision value calculation: f(X')=−0.09+(−0.2)=−0.29; According to the above interval setting, the calculated decision value is f(X′)=−0.29, which belongs to the interval -0.5 ≤ f(X')<0, so the system recognizes the input data as a "high-frequency weighing scenario".

[0034] S3: calling an optimization algorithm matching the scene recognition result from a preset algorithm library according to the scene recognition result, executing the optimization algorithm, and outputting an optimization result, wherein the optimization result at least includes optimized weight data.

[0035] Specifically, the system builds an algorithm library, which includes multiple optimization algorithms. Each algorithm in the algorithm library is designed for different scene characteristics. In this embodiment, the optimization algorithms in the algorithm library are divided into three categories, corresponding to the identified scene categories, namely, environmental fluctuation optimization algorithm, high-frequency weighing optimization algorithm and high-precision optimization algorithm. The algorithm library supports updating and expansion to add new optimization algorithms in future applications, and each algorithm has a corresponding identifier in the algorithm library to facilitate automatic calling according to the scene recognition results.

[0036] When the scene recognition model outputs a scene label, the system selects the corresponding optimization algorithm from the algorithm library based on the label. For example, when the scene label is a high environmental fluctuation scene, the environmental fluctuation optimization algorithm is selected. The algorithm analyzes data such as temperature and humidity, performs dynamic compensation, and the corrected weight data The calculation formula is as follows: ; Among them, W i represents the initial weight data, ∆T represents the temperature offset, that is, the difference between the current temperature and the reference temperature, ∆H represents the humidity offset, that is, the difference between the current humidity and the reference temperature, K T Indicates the temperature compensation coefficient, K H Indicates the humidity compensation coefficient, K T and K HCalibration is performed through historical data to reflect the specific impact of temperature and humidity on weighing accuracy. It can be imagined that ∆T=T-T0, ∆H=H-H0, T represents the current temperature value, T0 represents the reference temperature value, H represents the current humidity value, and H0 represents the reference humidity value.

[0037] In one implementation, assuming that goods are weighed in a logistics warehouse, the ambient temperature and humidity may affect the weighing results. For example, when the temperature rises, the material of the scale may expand, thereby affecting the weighing accuracy. The data obtained is as follows: W i =100kg, T=30℃, H=80%, T0=25℃, H0=60%,K T =0.05, K H =0.02, it should be noted that K T It means that when the temperature changes by 1℃, the weight will be affected by 0.05kg. H It means that when humidity changes by 1%, the impact on weight is 0.02kg. The calculation process is as follows: Temperature deviation: ∆T = T-T0 = 30℃-25℃ = 5℃; Humidity deviation: ∆H=H-H0=80%-60%=20%; Corrected weight data: W o =W i +K T ×∆T+K H ×ΔH=100kg+0.05×5+0.02×20=100.65kg; Output result: optimized weight data W o =100.65kg.

[0038] When the scene label is a high-frequency weighing scene, select the high-frequency weighing optimization algorithm. This algorithm uses a smoothing algorithm to reduce short-term fluctuations caused by frequent measurements and improve data stability. It is based on the moving average method and the corrected weight data The calculation formula is as follows: ; Among them, W i represents the weight data of the i-th weighing, n represents the window size for data smoothing, that is, the number of continuous weighing data required to calculate the moving average. The window size is determined according to the weighing frequency and real-time requirements, and is usually optimized through experiments.

[0039] In one implementation, suppose that on a production line, an electronic scale needs to perform weighing operations frequently, which may cause fluctuations due to noise or vibration of the sensor. In order to smooth these fluctuations, a moving average algorithm is used to process the high-frequency weighing data, and the data obtained are as follows: once per second, a total of 5 times, W1=100.5kg, W2=100.7kg, W3=100.6kg, W4=100.8kg, W5=100.9kg, smoothing window size n=3, that is, the latest 3 weighing data are taken to calculate the moving average, and the calculation process is as follows: First calculation (windows W1, W2, W3): W a1 =(100.5+100.7+100.6) / 3=100.6kg; Second calculation (windows W2, W3, W4): W a2 =(100.7+100.6+100.8) / 3=100.7kg; The third calculation (windows W3, W4, W5): W a3 =(100.6+100.8+100.9) / 3=100.77kg; Output results: The optimized weight data are 100.6kg, 100.7kg and 100.77kg respectively.

[0040] When the scene label is a high-precision requirement scene, select the high-precision optimization algorithm, which removes noise and small fluctuations through the filtering algorithm to obtain more accurate weight data. It is based on Kalman filtering, and the corrected weight data The calculation formula is as follows: ; Among them, W g It represents the filtering result of the previous moment, that is, the weight data after the last optimization. K represents the Kalman gain, which is dynamically adjusted according to the measurement noise and prediction error. It is usually determined by experiments or adaptive algorithms to balance the response speed and data smoothness.

[0041] In one implementation, such as laboratory weighing, the weighing result of the electronic scale may be affected by small vibrations and noise, so Kalman filtering is used to optimize the result, and the data obtained is as follows: W i =100.3kg, W g =100.25kg, K=0.1, the calculation process is as follows: W o =W g + K×(W i -W g )=100.25+0.1×(100.3-100.25)=100.255kg; Output result: optimized weight data W o =100.255kg.

[0042] like Figure 3 As shown, in some complex scenarios, scene fusion methods can also be considered, including: S301: setting importance weights for different scene categories, and forming a weight matrix according to the combination of scene fusion, the calculation formula of which is: ; Among them, W q represents the optimized weight after fusion, W e represents the weight of the high environmental fluctuation scenario, W r Indicates high-frequency weighing scenarios, W p represents the scene with high precision requirement, α, β and γ represent the weighting coefficients of the above three different scene fusion types, and α+β+γ=1; S302: According to the fused scene category weights, an optimization algorithm is selected to dynamically adjust the weight data. The optimized weight data The calculation formula is: ; Here, P represents a set of adjustable parameters used in the optimization algorithm.

[0043] Wo is the weight data after correction of three independent scenes, and Wf is the weight data result of fusion scene optimization.

[0044] Specifically, assuming that the scene weighting coefficients are: α=0.4, β=0.35, γ=0.25, if the system obtains W based on scene data analysis e =0.8,W r =0.6, W p =0.4, then the fusion weight W q =0.4×0.8+0.35×0.6+0.25×0.4=0.32+0.21+0.1=0.63.

[0045] W f The calculation formula can be expanded as: ; The data obtained are as follows: Initial weight data W i =10kg, fusion weight W q =0.63, adjust the parameter P=0.05, and calculate W f =10.0+(0.63×0.05)=10.0+0.0315=10.0315 kg, the final optimized weight data W f=10.0315 kg. This result is a dynamically optimized weight based on the current multi-scenario fusion weights and adjustment parameters. It can more accurately reflect the weighing results in high-frequency weighing, environmental fluctuations, or high-precision scenarios. This example integrates the weight influencing factors of different scenarios and adds real-time adjustment parameters to ensure that the measurement results of the electronic scale have high accuracy and consistency under a variety of complex environmental conditions.

[0046] S4: Display the optimized weight data at a preset position on the electronic scale screen.

[0047] In order to ensure that operators can obtain accurate weighing information in real time, the display position, display mode and display accuracy need to be adjusted and optimized according to different scenarios. The operating interface of the electronic scale screen includes weight display area, status display area and environmental data area, etc. Accordingly, different areas of the electronic scale screen display different data, and the preset position is designed according to the interface.

[0048] The electronic scale screen needs to display different contents according to the data type. For example, the optimized weight data is displayed digitally in the following format: f =xx.xx kg. The number of digits after the decimal point can be set as needed to ensure data accuracy. Units can also be converted freely.

[0049] It is understandable that in high-frequency weighing scenarios, the update frequency can be set to once per second to update the screen display content in real time; in low-frequency weighing scenarios, the update frequency is reduced, for example, updating every few seconds or after each weighing is completed.

[0050] To optimize the display effect, control the screen brightness and / or font size according to the ambient light and usage scenario to ensure that the screen content is clearly readable under different lighting conditions. For example, when the ambient light becomes brighter, increase the screen brightness and / or enlarge the font.

[0051] S5: Storing the optimization results in a database to perform incremental training on the scene recognition model.

[0052] Specifically, the data storage structure includes four main fields, namely initial weight data, environmental data, weighing frequency and optimized weight data. Each data record forms an independent database entry. After each optimization algorithm is executed, the optimization results (including initial weight data, environmental data, weighing frequency and optimized weight data) will be stored in the database.

[0053] The stored data can be used to perform incremental training on the scene recognition model. During the training process, the stored optimized data is used to update the model weights to improve the model's recognition ability for different scenes.

[0054] like Figure 4 As shown, incremental training includes the following steps: S501: regularly reading optimization result data from a database according to a preset training cycle; S502: Using an incremental learning algorithm, inputting data into a scene recognition model for training; S503: Save the updated scene recognition model.

[0055] This embodiment provides a multi-scenario weight detection and optimization system for an electronic scale, using the multi-scenario weight detection and optimization method described above, such as Figure 5 As shown, including: A weight detection module 10, wherein the weight detection module 10 is configured to detect the weight of a target item and generate initial weight data; an environment monitoring module 20, wherein the environment monitoring module 20 is configured to monitor an environment within a preset range to generate environment data; a scene recognition module 30, wherein the scene recognition module 30 is configured to recognize a current weighing scene according to the environment data and output a scene recognition result; an algorithm library calling module 40, wherein the algorithm library calling module 40 is configured to call a matching optimization algorithm according to the scene recognition result and output an optimization result; a display module 50, wherein the display module 50 is configured to display the optimized weight data at a preset position on the electronic scale screen; and a storage module 60, wherein the storage module 60 is configured to store the optimization result in a database.

[0056] Among them, Figure 6 As shown, the environmental monitoring module 20 includes: a temperature sensor 201, which is configured to monitor temperature changes within a preset range; a humidity sensor 202, which is configured to monitor humidity changes within a preset range; and a tilt sensor 203, which is configured to detect the tilt angle of the electronic scale.

[0057] Specifically, the system of this embodiment includes six main modules: a weight detection module 10, an environment monitoring module 20, a scene recognition module 30, an algorithm library calling module 40, a display module 50 and a storage module 60. When the target object is placed on the electronic scale, the weight detection module 10 can transmit the measured weight data to the control unit to generate initial weight data; the environment monitoring module 20 is responsible for monitoring the environmental data within a preset range, including temperature, humidity and the tilt angle of the electronic scale; the scene recognition module 30 inputs the temperature, humidity and tilt data received from the environment monitoring module 20 into the scene recognition model, and the scene recognition model judges whether the current weighing scene belongs to a high-frequency weighing scene, a high environmental fluctuation scene, or a high-precision requirement scene according to a predetermined algorithm, combined with historical data and environmental characteristics, and outputs a recognition result, indicating the current weighing scene category; the algorithm library calling module 40 calls the corresponding optimization algorithm according to the scene recognition result, and the optimization algorithm adjusts the initial weight data; after receiving the optimization result, the display module 50 displays the optimized weight data on the screen of the electronic scale; the storage module 60 stores the optimized weight data, environmental data, weighing frequency and other related information in the database.

[0058] Although the embodiments of the present invention have been shown and described above, it is to be understood that the above embodiments are exemplary and are not to be construed as limitations of the present invention. A person skilled in the art may change, modify, replace and vary the above embodiments within the scope of the present invention.

Claims

1. A multi-scenario weight detection and optimization method for an electronic scale, characterized in that: The following steps are involved: S1: Use an electronic scale to detect the weight of the target object to generate initial weight data, and monitor the environment within a preset range to generate environmental data; S2: Based on data analysis, the environmental data is input into a scene recognition model to identify the current weighing scene, and a scene recognition result is output; S3: calling an optimization algorithm matching the scene recognition result from a preset algorithm library according to the scene recognition result, executing the optimization algorithm, and outputting an optimization result, wherein the optimization result at least includes optimized weight data; S4: Displaying the optimized weight data at a preset position on the electronic scale screen; S5: Storing the optimization results in a database to perform incremental training on the scene recognition model.

2. The multi-scenario weight detection and optimization method according to claim 1 is characterized in that: Based on data analysis including: The environmental data include temperature, humidity and tilt angle; and The weighing frequency of the electronic scale.

3. The multi-scenario weight detection and optimization method according to claim 1, characterized in that S2 include: S201: Determine whether the current weighing scene can be identified through environmental data; If not, executing S202: using the weight data as an auxiliary judgment, identifying the current weighing scene through the environment data and the weight data; If yes, execute S3.

4. The multi-scenario weight detection and optimization method according to claim 1 is characterized in that: The categories of multiple scenarios include high environmental fluctuation scenarios, high-frequency weighing scenarios, and high-precision requirement scenarios.

5. The multi-scenario weight detection and optimization method according to claim 4 is characterized in that: The algorithm library includes an environmental fluctuation optimization algorithm, a high-frequency weighing optimization algorithm and a high-precision optimization algorithm.

6. The multi-scenario weight detection and optimization method according to claim 5 is characterized in that: The environmental fluctuation optimization algorithm calculates the corrected weight data based on the environmental data using the compensation model , and its calculation formula is: ; Among them, W i Indicates the initial weight data, K T Indicates the temperature compensation coefficient, T indicates the current temperature value, T0 indicates the reference temperature value, K H Indicates the humidity compensation coefficient, H indicates the current humidity value, and H0 indicates the reference humidity value; The high-frequency weighing optimization algorithm reduces short-term fluctuations caused by frequent measurements through a smoothing algorithm, and the corrected weight data The calculation formula is: ； Among them, W i represents the weight data of the i-th weighing, and n represents the window size of data smoothing; The high-precision optimization algorithm removes noise and micro-fluctuations through filtering algorithms, and the corrected weight data The calculation formula is: ； Among them, W g represents the filtering result of the previous moment, and K represents the Kalman gain.

7. The multi-scenario weight detection and optimization method according to claim 4, characterized in that S2 include: Priorities are set for different scene categories. When more than two scene categories are detected simultaneously based on environmental data, the scene recognition result of a scene category with a higher priority is output.

8. The multi-scenario weight detection and optimization method according to claim 1, characterized in that: S3 includes scene fusion methods: S301: setting importance weights for different scene categories, and forming a weight matrix according to the combination of scene fusion, the calculation formula of which is: ； Among them, W q represents the optimized weight after fusion, W e represents the weight of the high environmental fluctuation scenario, W r Indicates high-frequency weighing scenarios, W p represents the scene with high precision requirement, α, β and γ represent the weighting coefficients of the above three different scene fusion types, and α+β+γ=1; S302: According to the fused scene category weights, an optimization algorithm is selected for dynamic adjustment. The optimized weight data The calculation formula is: ; Here, P represents a set of adjustable parameters used in the optimization algorithm.

9. A multi-scenario weight detection and optimization system for electronic scales, characterized in that: The multi-scenario weight detection and optimization method according to any one of claims 1 to 8 comprises: A weight detection module, wherein the weight detection module is configured to detect the weight of the target object and generate initial weight data; An environment monitoring module, wherein the environment monitoring module is configured to monitor an environment within a preset range to generate environment data; A scene recognition module, wherein the scene recognition module is configured to recognize a current weighing scene according to the environmental data and output a scene recognition result; An algorithm library calling module, wherein the algorithm library calling module is configured to call a matching optimization algorithm according to the scene recognition result and output an optimization result; A display module, wherein the display module is configured to display the optimized weight data at a preset position on the electronic scale screen; A storage module is configured to store the optimization result in a database.

10. The multi-scenario weight detection and optimization system according to claim 9, characterized in that: The environmental monitoring module comprises: A temperature sensor, wherein the temperature sensor is configured to monitor temperature changes within a preset range; a humidity sensor configured to monitor humidity changes within a preset range; The inclination sensor is configured to detect an inclination angle of the electronic scale.

Images