Remote automatic quality control and calibration method and system for pH meter

Abstract

The invention discloses a remote automatic quality control and calibration method and system for a pH meter, and the method comprises the following steps: S1, carrying out the initial temperature compensation correction of a pH value, and generating a corrected pH value; s2, environment parameters collected by an environment sensor and the corrected pH value are combined and formatted; s3, cleaning the standardized data set at an edge computing node, and generating an optimized measurement data set; s4, performing confidence evaluation based on a committee query algorithm, and triggering a calibration process for a detected low-confidence sample; s5, generating an enhanced calibration curve by using a double-neural network regression model, and optimizing measurement data in real time by combining particle filtering; s6, displaying the final data in real time through a visual tool, and uploading the final data to a cloud end; and S7, dynamically adjusting the calibration period, the acquisition frequency and the communication strategy of the sensor through the edge computing node. According to the method, double neural network regression, particle filtering, cloud edge cooperation technologies and the like are utilized, and accurate measurement and remote intelligent calibration of the pH meter are achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of remote calibration, and in particular to a method and system for remote automatic quality control and calibration of a pH meter. Background Art

[0002] As an important measuring instrument widely used in industries such as industry, agriculture, and environmental monitoring, pH meters are used to accurately measure the pH value of liquids and provide key data for production process control, water quality testing, and laboratory research. Traditional pH meters rely on manual regular calibration and on-site operation. Although certain technical progress has been made, they still face many limitations in practical applications. Especially in complex or remote environments, existing technologies have obvious deficiencies in data reliability, calibration efficiency, and degree of automation.

[0003] Traditional pH meters are mainly based on glass electrodes, composite electrodes or solid-state electrodes for measurement, and the pH value is calculated by measuring the potential difference generated by the activity of hydrogen ions in the liquid. However, due to the complexity of the measurement environment, the performance of the pH meter will be significantly affected by external environmental factors such as temperature, humidity, and ion concentration. Therefore, in order to improve the accuracy of the measurement, the existing technology usually relies on a temperature compensation module to correct the pH value to reduce the interference of temperature changes on the measurement results. These compensation mechanisms often require manual setting of fixed parameters and cannot respond to dynamic changes in the environment in real time, resulting in errors in the measurement results in extreme environments or changing conditions.

[0004] On the other hand, the existing pH meter calibration method mainly relies on manual operation, and usually requires regular multi-point calibration using standard buffer solutions. Although this calibration mode can guarantee measurement accuracy to a certain extent, its limitations are obvious: first, the calibration frequency is fixed and cannot be flexibly adjusted according to actual needs, which can easily lead to waste of calibration resources or insufficient calibration; second, the calibration process relies on manual intervention, which is difficult to implement in remote or unattended scenarios; third, the storage and use of standard solutions require strict environmental control, otherwise it is easy to lead to inaccurate calibration results. Therefore, in application scenarios that require long-term continuous monitoring, the limitations of existing calibration methods have become a significant bottleneck.

[0005] In addition, existing technologies still have shortcomings in data anomaly detection and sensor failure warning. Although some systems have introduced simple statistical methods, such as fixed threshold detection based on upper and lower limits, in actual environments, due to the nonlinear characteristics of the measurement data, these methods are often not accurate enough in identifying outliers. Especially in complex dynamic environments, it is impossible to effectively distinguish between normal measurement fluctuations and real anomalies, resulting in frequent false alarms and missed alarms. At the same time, when the sensor drifts or fails, the existing system lacks a timely response mechanism and cannot actively trigger self-calibration or repair actions. This lag not only affects the reliability of the data, but may also have a serious impact on key monitoring tasks.

[0006] In multi-sensor collaborative application scenarios, existing technologies also show certain limitations. For example, when multiple pH sensors work together in a monitoring network, due to differences in measurement accuracy, drift characteristics, and environmental adaptability, data consistency between different sensors is difficult to guarantee. Some current data fusion technologies mainly rely on simple weighted average or fixed rule fusion methods, which cannot dynamically adjust the weights of different sensors, resulting in insufficient accuracy of data fusion. In addition, existing sensor communication strategies usually adopt fixed data transmission intervals and modes, and cannot be dynamically adjusted according to actual network load or data urgency. This communication strategy not only increases the energy consumption of the system, but also easily causes transmission delays of important data.

[0007] In recent years, with the rapid development of cloud computing, edge computing and artificial intelligence technologies, remote automated quality control and calibration systems have gradually become a trend. For example, cloud analysis platforms can generate optimization suggestions through the accumulation and analysis of historical data, and edge computing nodes can achieve real-time localized adjustments. However, the architectural design of existing systems often lacks sufficient dynamic adaptability and global optimization capabilities. For example, the optimization suggestions generated by the cloud are usually static and cannot respond to sensor operating status and environmental changes in real time; the processing power of edge nodes is also relatively limited, and it is impossible to efficiently perform real-time reasoning and adjustments on complex models.

[0008] Therefore, how to provide a remote automatic quality control and calibration method and system for a pH meter is a problem that those skilled in the art need to solve urgently. Summary of the invention

[0009] One purpose of the present invention is to propose a remote automatic quality control and calibration method and system for a pH meter. The present invention adopts dual neural network regression, particle filtering and dynamic model committee algorithm, and combines the cloud-edge collaborative architecture to achieve accurate measurement, intelligent calibration and remote automatic quality control of the pH meter. Through dynamic environmental adaptation and real-time optimization, the present invention can effectively improve the accuracy and reliability of data measurement in a complex and changeable environment, while reducing the need for manual intervention, and is suitable for unmanned remote monitoring scenarios. Compared with traditional methods, the present invention has the significant advantages of high precision, low energy consumption, high degree of automation and strong environmental adaptability, and can be widely used in industrial monitoring, environmental protection, water quality management and other fields.

[0010] The pH meter remote automatic quality control and calibration method according to an embodiment of the present invention comprises the following steps: S1. Collect the real-time pH value of the liquid to be tested through the pH sensor, and use the temperature compensation module to obtain real-time temperature data, perform preliminary temperature compensation correction on the pH value, and generate a corrected pH value; S2, combining and formatting the environmental parameters collected by the environmental sensor and the corrected pH value to generate a standardized data set, which is transmitted to the edge computing node; S3, cleaning the standardized data set at the edge computing node, removing outliers and noise data, and generating an optimized measurement data set; S4. Use the committee query algorithm to perform confidence assessment on the optimized measurement data set, screen low-confidence samples through the inconsistency of the model committee, and trigger the calibration process for the detected low-confidence samples; S5. In the calibration process, the dual neural network regression model is used to generate an enhanced calibration curve, and the particle filter is combined to optimize the measurement data in real time; S6. Display the final measurement output value, equipment operation status and calibration records in real time through visualization tools, and upload the equipment operation status data and historical measurement data to the cloud; S7. Based on the optimization suggestions and calibration frequency adjustment schemes generated in the cloud, the calibration cycle, acquisition frequency and communication strategy of the sensor are dynamically adjusted through the edge computing nodes.

[0011] Optionally, the S1 specifically includes: S11. The real-time pH value of the liquid to be tested is collected through the pH sensor, and the original pH value signal is obtained after analog-to-digital conversion. , the pH sensor uses a glass electrode, a composite electrode or a solid-state electrode; S12, using a temperature compensation module to obtain real-time liquid temperature data, wherein the temperature compensation module includes at least one temperature sensor to detect the real-time temperature of the liquid to be measured ; S13, according to the real-time temperature of the liquid to be tested , calculate the effect of temperature on pH value , and obtain the compensation factor; S14. Using compensation factors to correct the original pH signal Temperature compensation is performed to convert the original pH signal The effect of temperature on pH value Perform comprehensive calculations to obtain the pH value after temperature compensation: ; in, represents the corrected pH value, represents the reference temperature, , , and represents the temperature compensation coefficient; S15. Corrected pH value Perform anomaly detection if If the preset standard pH value range is exceeded, an alarm is triggered and the abnormal situation is recorded.

[0012] Optionally, the S4 specifically includes: S41. Perform preliminary processing on the optimized measurement data set to generate a data set to be evaluated , and for each data point defining a feature vector, wherein the feature vector includes an original pH value, a compensated temperature, an environmental parameter, and a timestamp; S42. Construct a dynamic model committee, wherein the dynamic model committee comprises Independent models , and assign dynamic weights to each model: ; in, Representation Model At data point The dynamic weight on represents the exponential function, Representation Model For data points The standard deviation of the prediction results, Representation Model For data points The performance indicator function is represents the weight sensitivity parameter, represents the performance bias coefficient, Representation Model For data points The standard deviation of the prediction results, Representation Model For data points Performance indicator function; S43. Data set to be evaluated Each data point in , by each model in the model committee Calculate the expected value and generate a set of weighted prediction results: ; in, Represents data points The weighted prediction result set of Representation Model For data points The predicted value of S44. Calculate each data point The overall confidence level of: ; in, Represents data points The overall confidence level, represents the standard deviation of the weighted prediction result set, represents the mean of the weighted prediction result set, represents the nonlinear response adjustment factor, represents the smoothing factor, Represents data points The environmental parameter vector, Indicates a stable environment; S45, the comprehensive confidence is lower than the set threshold The data points are screened out from the data set to be evaluated to generate a low confidence data set : ; S46, trigger the calibration process, for low confidence data set Further processing including data optimization, model updating and calibration logging is performed while dynamic weight adjustments and changes in environmental parameters are recorded.

[0013] Optionally, the S5 specifically includes: S51, low confidence data set Each data point in The feature vector is input into a dual neural network regression model, which includes a main network and auxiliary network The main network Responsible for learning the global relationship between pH value and environmental parameters, the auxiliary network Learning local correction relations based on environment parameters: ; ; in, represents the predicted value of the main network, represents the prediction value of the auxiliary network, represents the set of model parameters of the main network, represents the set of model parameters of the auxiliary network, Represents data points The characteristic vector of Represents data points The environmental parameter vector; The final preliminary calibration prediction value is generated by weighted fusion of the main network and the auxiliary network: ; in, represents the initial calibration prediction value, represents the network fusion weight; S52, based on the initial calibration prediction value , combined with real-time temperature and data points The environmental parameter vector , generate the enhanced calibration curve: ; in, Indicates the pH value after calibration, represents the normalized environmental parameters, Indicates the minimum value of the environmental parameter, Indicates the maximum value of the environmental parameter. , , , , , , , , and represents the optimization coefficient; S53, initialize the particle filter algorithm and define the state vector: ; in, Indicates Particles at time Status; Define the state transition equation: ; in, represents the state transition function, Indicates Particles at time status, Indicates the dynamic adjustment value of the current calibration parameter. represents process noise; Define the measurement update equation: ; in, Indicates Particles at time The measured value of represents the measurement function, represents the measurement noise; S54. Resample particle distribution and update particle status according to weight: ; in, Indicates Particles at time The weight of Indicates Particles at time The weight of represents the exponential function, represents the actual observed value, represents the measurement noise covariance; S55, calculate the final calibration value, and take the weighted average of the particle states as the calibration result: ; in, represents the final calibrated pH value, Represents the total number of particles.

[0014] Optionally, the S7 specifically includes: S71, transmitting the optimization suggestions and calibration frequency adjustment scheme generated in the cloud to the edge computing node, and the edge computing node parses the optimization suggestions after receiving the data, including adjustment parameters of the calibration period, collection frequency and communication strategy; S72. Based on the analysis results, the edge computing node combines the current equipment operation status data, including pH value, temperature and environmental parameters collected by the sensor in real time, to make local adjustments to the cloud optimization suggestions. S73, generating a dynamic calibration plan in the edge computing node, the dynamic calibration plan including the trigger time of the next calibration, the calibration duration, and the estimated amount of calibration resources required, and sending the dynamic calibration plan to the sensor; S74, generating a dynamic collection strategy including a collection interval and a collection range according to the rate of change of environmental parameters, the quality assessment result of real-time collection data, and the equipment operation load, and synchronizing the dynamic collection strategy to the sensor; S75, formulate a communication optimization strategy, adjust the data transmission interval, data compression level and priority of different data types according to the current network status and the urgency of data upload, and send the communication optimization strategy to the sensor; S76. The sensor adjusts its own working mode according to the dynamic calibration plan, dynamic collection strategy and communication optimization strategy issued by the edge computing node, including dynamically adjusting the calibration cycle, collection frequency and communication frequency.

[0015] The remote automatic quality control and calibration system for a pH meter according to an embodiment of the present invention includes the following modules: A data acquisition module is used to collect real-time pH value, temperature data and environmental parameters through a pH sensor to generate a corrected pH value; Data processing module, used to format, clean and optimize data, remove outliers and noise data, and generate standardized measurement data sets; A model analysis module is used to perform confidence assessment on the optimized measurement data set based on the committee query algorithm to filter out low-confidence samples; Calibration optimization module, which is used to generate enhanced calibration curves using dual neural network regression models and optimize measurement data in real time in combination with particle filtering; Display and upload module, used to display measurement results, equipment status and calibration records in real time, and upload data to the cloud; The dynamic adjustment module is used to dynamically adjust the sensor calibration cycle, collection frequency and communication strategy according to the cloud optimization suggestions.

[0016] The beneficial effects of the present invention are: First, the present invention adopts a calibration model based on dual neural network regression, which can achieve high-precision automatic calibration under a wider range of environmental conditions compared to the traditional multi-point calibration mode that relies on manual adjustment. The main network is responsible for learning the global relationship between pH value and environmental parameters, and the auxiliary network performs local optimization based on real-time environmental parameters. The two generate more accurate calibration results through weighted fusion. Combined with the real-time optimization capability of particle filtering, the present invention can dynamically correct the measurement results when the sensor drifts or the data fluctuates, thereby significantly improving the stability and reliability of the data.

[0017] Secondly, the present invention can intelligently screen low-confidence samples and trigger the calibration process through the confidence assessment mechanism of the dynamic model committee, effectively avoiding the waste of resources or insufficient calibration problems that may be caused by the traditional fixed calibration cycle. The model committee achieves accurate analysis of complex data distribution by dynamically weighting the performance of multiple prediction models. The introduction of piecewise nonlinear functions enables the confidence assessment mechanism of the present invention to adapt to a variety of environmental conditions, thereby improving the ability to detect abnormal data and the accuracy of calibration triggering.

[0018] In addition, the present invention fully combines the advantages of cloud computing and edge computing in the design of system architecture. The cloud platform generates optimization suggestions through historical data analysis, and the edge nodes process and locally adjust the real-time data, realizing an efficient cloud-edge collaborative working mode. This architecture not only greatly reduces the dependence on cloud resources and reduces the response delay of the system, but also improves the system's adaptability to dynamic environments. The sensor can dynamically adjust the calibration cycle, acquisition frequency, and communication strategy according to the optimization strategy issued by the edge node, thereby achieving a dynamic balance between energy consumption, accuracy, and communication efficiency.

[0019] The present invention also pays special attention to the intelligent design of anomaly detection and fault warning. Through the collaborative work of the model committee query algorithm and the neural network regression model, it can promptly identify problems when sensor drift, drastic environmental changes or hardware failures occur, and actively trigger calibration or repair operations, avoiding the reduction in measurement accuracy and operating efficiency caused by delayed response in traditional systems. In addition, the system's multi-sensor data fusion function can dynamically adjust the weight of each sensor, significantly improving data consistency and overall monitoring accuracy in multi-device collaborative scenarios.

[0020] Through dynamic calibration and data optimization, the present invention can also effectively reduce the need for manual intervention, so that the pH meter can operate stably in unattended or remote monitoring scenarios. This automated quality control capability greatly reduces the cost of equipment maintenance and is particularly suitable for application scenarios that require long-term continuous monitoring, such as industrial process control, environmental monitoring, food processing, and agricultural water quality management. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings: Figure 1 This is an overall flow chart of the remote automatic quality control and calibration method for a pH meter proposed by the present invention; Figure 2 This is a schematic diagram of the structure of the remote automatic quality control and calibration system for a pH meter proposed by the present invention. DETAILED DESCRIPTION

[0022] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, which only illustrate the basic structure of the present invention in a schematic manner, and therefore only show the components related to the present invention.

[0023] refer to Figure 1 , a pH meter remote automatic quality control and calibration method, comprising the following steps: S1. Collect the real-time pH value of the liquid to be tested through the pH sensor, and use the temperature compensation module to obtain real-time temperature data, perform preliminary temperature compensation correction on the pH value, and generate a corrected pH value; S2, combining and formatting the environmental parameters collected by the environmental sensor and the corrected pH value to generate a standardized data set, which is transmitted to the edge computing node; S3, cleaning the standardized data set at the edge computing node, removing outliers and noise data, and generating an optimized measurement data set; S4. Use the committee query algorithm to perform confidence assessment on the optimized measurement data set, screen low-confidence samples through the inconsistency of the model committee, and trigger the calibration process for the detected low-confidence samples; S5. In the calibration process, the dual neural network regression model is used to generate an enhanced calibration curve, and the particle filter is combined to optimize the measurement data in real time; S6. Display the final measurement output value, equipment operation status and calibration records in real time through visualization tools, and upload the equipment operation status data and historical measurement data to the cloud; S7. Based on the optimization suggestions and calibration frequency adjustment schemes generated in the cloud, the calibration cycle, acquisition frequency and communication strategy of the sensor are dynamically adjusted through the edge computing nodes.

[0024] In this implementation manner, the S1 specifically includes: S11. The real-time pH value of the liquid to be tested is collected through the pH sensor, and the original pH value signal is obtained after analog-to-digital conversion. , the pH sensor uses a glass electrode, a composite electrode or a solid-state electrode; S12, using a temperature compensation module to obtain real-time liquid temperature data, wherein the temperature compensation module includes at least one temperature sensor to detect the real-time temperature of the liquid to be measured ; S13, according to the real-time temperature of the liquid to be tested , calculate the effect of temperature on pH value , and obtain the compensation factor; S14. Using compensation factors to correct the original pH signal Temperature compensation is performed to convert the original pH signal The effect of temperature on pH value Perform comprehensive calculations to obtain the pH value after temperature compensation: ; in, represents the corrected pH value, represents the reference temperature, , , and represents the temperature compensation coefficient; S15. Corrected pH value Perform anomaly detection if If the preset standard pH value range is exceeded, an alarm is triggered and the abnormal situation is recorded.

[0025] In this implementation manner, the S4 specifically includes: S41. Perform preliminary processing on the optimized measurement data set to generate a data set to be evaluated , and for each data point defining a feature vector, wherein the feature vector includes an original pH value, a compensated temperature, an environmental parameter, and a timestamp; S42. Construct a dynamic model committee, wherein the dynamic model committee comprises Independent models , and assign dynamic weights to each model: ; in, Representation Model At data point The dynamic weight on represents the exponential function, Representation Model For data points The standard deviation of the prediction results, Representation Model For data points The performance indicator function is represents the weight sensitivity parameter, represents the performance bias coefficient, Representation Model For data points The standard deviation of the prediction results, Representation Model For data points Performance indicator function; S43. Data set to be evaluated Each data point in , by each model in the model committee Calculate the expected value and generate a set of weighted prediction results: ; in, Represents data points The weighted prediction result set of Representation Model For data points The predicted value of S44. Calculate each data point The overall confidence level of: ; in, Represents data points The overall confidence level, represents the standard deviation of the weighted prediction result set, represents the mean of the weighted prediction result set, represents the nonlinear response adjustment factor, represents the smoothing factor, Represents data points The environmental parameter vector, Indicates a stable environment; S45, the comprehensive confidence is lower than the set threshold The data points are screened out from the data set to be evaluated to generate a low confidence data set : ; S46, trigger the calibration process, for low confidence data set Further processing including data optimization, model updating and calibration logging is performed while dynamic weight adjustments and changes in environmental parameters are recorded.

[0026] In this implementation manner, S5 specifically includes: S51, low confidence data set Each data point in The feature vector is input into a dual neural network regression model, which includes a main network and auxiliary network The main network Responsible for learning the global relationship between pH value and environmental parameters, the auxiliary network Learning local correction relations based on environment parameters: ; ; in, represents the predicted value of the main network, represents the prediction value of the auxiliary network, Represents the model parameter set of the main network, represents the set of model parameters of the auxiliary network, Represents data points The characteristic vector of Represents data points The environmental parameter vector; The final preliminary calibration prediction value is generated by weighted fusion of the main network and the auxiliary network: ; in, represents the initial calibration prediction value, represents the network fusion weight; S52, based on the initial calibration prediction value , combined with real-time temperature and data points The environmental parameter vector , generate the enhanced calibration curve: ; in, Indicates the pH value after calibration, represents the normalized environmental parameters, Indicates the minimum value of the environmental parameter, Indicates the maximum value of the environmental parameter. , , , , , , , , and represents the optimization coefficient; S53, initialize the particle filter algorithm and define the state vector: ; in, Indicates Particles at time Status; Define the state transition equation: ; in, represents the state transition function, Indicates Particles at time status, Indicates the dynamic adjustment value of the current calibration parameter. represents process noise; Define the measurement update equation: ; in, Indicates Particles at time The measured value of represents the measurement function, represents the measurement noise; S54. Resample particle distribution and update particle status according to weight: ; in, Indicates Particles at time The weight of Indicates Particles at time The weight of represents the exponential function, represents the actual observed value, represents the measurement noise covariance; S55, calculate the final calibration value, and take the weighted average of the particle states as the calibration result: ; in, represents the final calibrated pH value, Represents the total number of particles.

[0027] In this implementation manner, the S7 specifically includes: S71, transmitting the optimization suggestions and calibration frequency adjustment scheme generated in the cloud to the edge computing node, and the edge computing node parses the optimization suggestions after receiving the data, including adjustment parameters of the calibration period, collection frequency and communication strategy; S72. Based on the analysis results, the edge computing node combines the current equipment operation status data, including pH value, temperature and environmental parameters collected by the sensor in real time, to make local adjustments to the cloud optimization suggestions. S73, generating a dynamic calibration plan in the edge computing node, the dynamic calibration plan including the trigger time of the next calibration, the calibration duration, and the estimated amount of calibration resources required, and sending the dynamic calibration plan to the sensor; S74, generating a dynamic collection strategy including a collection interval and a collection range according to the rate of change of environmental parameters, the quality assessment result of real-time collection data, and the equipment operation load, and synchronizing the dynamic collection strategy to the sensor; S75, formulate a communication optimization strategy, adjust the data transmission interval, data compression level and priority of different data types according to the current network status and the urgency of data upload, and send the communication optimization strategy to the sensor; S76. The sensor adjusts its own working mode according to the dynamic calibration plan, dynamic collection strategy and communication optimization strategy issued by the edge computing node, including dynamically adjusting the calibration cycle, collection frequency and communication frequency.

[0028] refer to Figure 2 , pH meter remote automatic quality control and calibration system, including the following modules: A data acquisition module is used to collect real-time pH value, temperature data and environmental parameters through a pH sensor to generate a corrected pH value; Data processing module, used to format, clean and optimize data, remove outliers and noise data, and generate standardized measurement data sets; A model analysis module is used to perform confidence assessment on the optimized measurement data set based on the committee query algorithm to filter out low-confidence samples; Calibration optimization module, which is used to generate enhanced calibration curves using dual neural network regression models and optimize measurement data in real time in combination with particle filtering; Display and upload module, used to display measurement results, equipment status and calibration records in real time, and upload data to the cloud; The dynamic adjustment module is used to dynamically adjust the sensor calibration cycle, collection frequency and communication strategy according to the cloud optimization suggestions.

[0029] Embodiment 1:

[0030] In order to verify the feasibility of the present invention in implementation, the present invention is applied to a food processing plant in a certain industrial park. The plant needs to perform real-time monitoring and calibration of the pH value of wastewater in the production process every day to ensure that the discharged water quality meets the national environmental protection standards. However, due to the complex chemical composition of the wastewater, frequent temperature changes, and scattered monitoring points, traditional pH meters have obvious problems in data reliability and calibration efficiency. For example, the pH value of certain monitoring points often has measurement errors due to temperature fluctuations and sensor drift, resulting in unstable wastewater treatment process. In addition, calibration needs to be completed manually on a regular basis, which is not only time-consuming and labor-intensive, but also difficult to adapt to the needs of dynamic changes on site.

[0031] To solve the above problems, this embodiment applies a pH meter method based on a remote automatic quality control and calibration system. The entire system includes a data acquisition module, a data processing module, a model analysis module, a calibration optimization module, a display and upload module, and a dynamic adjustment module. pH sensors, temperature sensors, and environmental sensors are installed at the wastewater outlet to collect the pH value, temperature, and chemical environmental parameters of the water sample in real time. These data are transmitted to the edge computing node via a wireless network for preliminary processing, and then uploaded to the cloud for further analysis.

[0032] In actual applications, the test data of a certain day showed that the initial measurement data had large deviations due to environmental fluctuations and sensor aging. For example, the standard value of pH should be between 6.5 and 7.5, but the initial measurement values ​​of some points were 6.1 or 7.9. Through data cleaning and dynamic model committee analysis, low-confidence sample points were identified and the calibration process was automatically triggered. The dual neural network regression model combined with environmental parameters generated an enhanced calibration curve, and the particle filter algorithm further optimized the measurement value, and finally calibrated the data to a reasonable range.

[0033] During one month of continuous monitoring, the system significantly improved monitoring efficiency and data accuracy. Compared with traditional manual calibration, the system reduced the number of calibrations per week from 8 to 2, the time required for a single calibration from 30 minutes to 10 minutes, and the overall data error was reduced by 70%. The following are the specific data: Table 1 pH monitoring data during wastewater treatment

[0034] Table 2 Calibration efficiency and error analysis table

[0035] Table 3 System working status record

[0036] From the data in Table 1 above, it can be seen that the pH values ​​initially measured have deviations at multiple measuring points. The deviations of some measuring points, such as A1 and A5, reached -0.4 and -0.5, far from the standard range (6.5-7.5). After the calibration optimization of this system, the pH values ​​of each measuring point were adjusted to a reasonable range, and the deviations were significantly reduced. For example, the final deviations of A2 and A4 were adjusted from +0.4 and +0.3 to -0.1 and 0, respectively. In addition, the results after calibration are closer to the mean, showing higher data stability, proving that the system can dynamically correct drift and improve measurement accuracy in complex environments.

[0037] It can be found from Table 2 that the automatic calibration system of the present invention shows obvious advantages in calibration efficiency and data accuracy. Traditional manual calibration needs to be performed 8 times a week, each time takes 30 minutes, the overall error reaches 15.8%, the work efficiency is low, and manual intervention is frequent. However, this system only needs 2 automatic calibrations per week, each time takes 10 minutes, and the overall error is reduced to 4.7%. This significant efficiency improvement and error reduction show that the system can greatly reduce the manual burden through the intelligent calibration mechanism, while improving data reliability and stability.

[0038] Table 3 records the dynamic adjustment capability of the system in actual operation. It can be seen from the data that the system can flexibly adjust the calibration trigger times, acquisition frequency, and communication optimization strategy according to the changes in ambient temperature and real-time data. For example, when the ambient temperature is high, the calibration trigger times are 1 and 3 respectively, and the acquisition frequency and communication optimization strategy are also adjusted accordingly with the environmental fluctuations, which maximizes the smooth operation of the system under high load conditions. This dynamic adaptability shows that the system can perceive environmental changes in real time and adjust the working mode, optimizing resource utilization efficiency while ensuring measurement accuracy.

[0039] From the comprehensive analysis of the data in the three tables, it can be concluded that the system of the present invention solves the problems of fixed calibration frequency, high manual dependence, and poor data reliability of traditional pH meters by combining intelligent calibration methods and cloud-edge collaborative architecture. Specifically, by screening low-confidence samples through a dynamic model committee and triggering the calibration process, combined with dual neural network regression and particle filter optimization, the system can effectively improve data accuracy in complex environments. At the same time, the mechanism of dynamically adjusting the calibration cycle and acquisition frequency not only optimizes the energy consumption of the system, but also significantly reduces the cost of manual operation. Compared with traditional methods, the present invention has demonstrated significant advantages in dynamically changing scenarios such as wastewater treatment, providing strong support for unattended monitoring and remote management.

[0040] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A method for remote automatic quality control and calibration of a pH meter, characterized in that: The steps include: S1. Collect the real-time pH value of the liquid to be tested through the pH sensor, and use the temperature compensation module to obtain real-time temperature data, perform preliminary temperature compensation correction on the pH value, and generate a corrected pH value; S2, combining and formatting the environmental parameters collected by the environmental sensor and the corrected pH value to generate a standardized data set, which is transmitted to the edge computing node; S3, cleaning the standardized data set at the edge computing node, removing outliers and noise data, and generating an optimized measurement data set; S4. Use the committee query algorithm to perform confidence assessment on the optimized measurement data set, screen low-confidence samples through the inconsistency of the model committee, and trigger the calibration process for the detected low-confidence samples; S5. In the calibration process, the dual neural network regression model is used to generate an enhanced calibration curve, and the particle filter is combined to optimize the measurement data in real time; S6. Display the final measurement output value, equipment operation status and calibration records in real time through visualization tools, and upload the equipment operation status data and historical measurement data to the cloud; S7. Based on the optimization suggestions and calibration frequency adjustment schemes generated in the cloud, the calibration cycle, acquisition frequency and communication strategy of the sensor are dynamically adjusted through the edge computing nodes.

2. The pH meter remote automatic quality control and calibration method according to claim 1, characterized in that: The S1 specifically includes: S11. The real-time pH value of the liquid to be tested is collected through the pH sensor, and the original pH value signal is obtained after analog-to-digital conversion. , the pH sensor uses a glass electrode, a composite electrode or a solid-state electrode; S12, using a temperature compensation module to obtain real-time liquid temperature data, wherein the temperature compensation module includes at least one temperature sensor to detect the real-time temperature of the liquid to be measured ; S13, according to the real-time temperature of the liquid to be tested , calculate the effect of temperature on pH value , and obtain the compensation factor; S14. Using compensation factor to adjust the original pH signal Temperature compensation is performed to convert the original pH signal The effect of temperature on pH value Perform comprehensive calculations to obtain the pH value after temperature compensation: ； in, represents the corrected pH value, represents the reference temperature, , , and It represents the temperature compensation coefficient; S15. Corrected pH value Perform anomaly detection if If the preset standard pH value range is exceeded, an alarm is triggered and the abnormal situation is recorded.

3. The pH meter remote automatic quality control and calibration method according to claim 1, characterized in that: The S4 specifically includes: S41. Perform preliminary processing on the optimized measurement data set to generate a data set to be evaluated , and for each data point defining a feature vector, wherein the feature vector includes an original pH value, a compensated temperature, an environmental parameter, and a timestamp; S42. Construct a dynamic model committee, wherein the dynamic model committee comprises Independent models , and assign dynamic weights to each model: ； in, Representation Model At data point The dynamic weight on represents the exponential function, Representation Model For data points The standard deviation of the prediction results, Representation Model For data points The performance indicator function is represents the weight sensitivity parameter, represents the performance bias coefficient, Representation Model For data points The standard deviation of the prediction results, Representation Model For data points Performance indicator function; S43. Data set to be evaluated Each data point in , by each model in the model committee Calculate the expected value and generate a set of weighted prediction results: ； in, Represents data points The weighted prediction result set of Representation Model For data points The predicted value of S44. Calculate each data point The overall confidence level of: ； in, Represents data points The overall confidence level, represents the standard deviation of the weighted prediction result set, represents the mean of the weighted prediction result set, represents the nonlinear response adjustment factor, represents the smoothing factor, Represents data points The environmental parameter vector, Indicates a stable environment; S45, the comprehensive confidence is lower than the set threshold The data points are screened out from the data set to be evaluated to generate a low confidence data set : ； S46, trigger the calibration process, for low confidence data set Further processing including data optimization, model updating and calibration logging is performed while dynamic weight adjustments and changes in environmental parameters are recorded.

4. The pH meter remote automatic quality control and calibration method according to claim 1, characterized in that: The S5 specifically includes: S51, low confidence data set Each data point in The feature vector is input into a dual neural network regression model, which includes a main network and auxiliary network The main network Responsible for learning the global relationship between pH value and environmental parameters, the auxiliary network Learning local correction relations based on environment parameters: ； ； in, represents the predicted value of the main network, represents the prediction value of the auxiliary network, represents the set of model parameters of the main network, represents the set of model parameters of the auxiliary network, Represents data points The characteristic vector of Represents data points The environmental parameter vector; The final preliminary calibration prediction value is generated by weighted fusion of the main network and the auxiliary network: ; in, represents the initial calibration prediction, represents the network fusion weight; S52, based on the initial calibration prediction value , combined with real-time temperature and data points The environmental parameter vector , generate the enhanced calibration curve: ； in, Indicates the pH value after calibration, represents the normalized environmental parameters, Indicates the minimum value of the environmental parameter, Indicates the maximum value of the environmental parameter. , , , , , , , , and represents the optimization coefficient; S53, initialize the particle filter algorithm and define the state vector: ； in, Indicates Particles at time Status; Define the state transition equation: ； in, represents the state transition function, Indicates Particles at time status, Indicates the dynamic adjustment value of the current calibration parameter. represents process noise; Define the measurement update equation: ； in, Indicates Particles at time The measured value of represents the measurement function, represents the measurement noise; S54. Resample particle distribution and update particle status according to weight: ； in, Indicates Particles at time The weight of Indicates Particles at time The weight of represents the exponential function, represents the actual observed value, represents the measurement noise covariance; S55, calculate the final calibration value, and take the weighted average of the particle states as the calibration result: ； in, represents the final calibrated pH value, Represents the total number of particles.

5. The pH meter remote automatic quality control and calibration method according to claim 1, characterized in that: The S7 specifically includes: S71, transmitting the optimization suggestions and calibration frequency adjustment scheme generated in the cloud to the edge computing node, and the edge computing node parses the optimization suggestions after receiving the data, including adjustment parameters of the calibration period, collection frequency and communication strategy; S72. Based on the analysis results, the edge computing node combines the current equipment operation status data, including pH value, temperature and environmental parameters collected by the sensor in real time, to make local adjustments to the cloud optimization suggestions. S73, generating a dynamic calibration plan in the edge computing node, the dynamic calibration plan including the trigger time of the next calibration, the calibration duration, and the estimated amount of calibration resources required, and sending the dynamic calibration plan to the sensor; S74, generating a dynamic collection strategy including a collection interval and a collection range according to the rate of change of environmental parameters, the quality assessment result of real-time collection data, and the equipment operation load, and synchronizing the dynamic collection strategy to the sensor; S75, formulate a communication optimization strategy, adjust the data transmission interval, data compression level and priority of different data types according to the current network status and the urgency of data upload, and send the communication optimization strategy to the sensor; S76. The sensor adjusts its own working mode according to the dynamic calibration plan, dynamic collection strategy and communication optimization strategy issued by the edge computing node, including dynamically adjusting the calibration cycle, collection frequency and communication frequency.

6. A pH meter remote automatic quality control and calibration system, which implements the pH meter remote automatic quality control and calibration method according to any one of claims 1 to 5, characterized in that: Includes the following modules: A data acquisition module is used to collect real-time pH value, temperature data and environmental parameters through a pH sensor to generate a corrected pH value; Data processing module, used to format, clean and optimize data, remove outliers and noise data, and generate standardized measurement data sets; A model analysis module is used to perform confidence evaluation on the optimized measurement data set based on the committee query algorithm to filter out low-confidence samples; Calibration optimization module, which is used to generate enhanced calibration curves using dual neural network regression models and optimize measurement data in real time in combination with particle filtering; Display and upload module, used to display measurement results, equipment status and calibration records in real time, and upload data to the cloud; The dynamic adjustment module is used to dynamically adjust the sensor calibration cycle, collection frequency and communication strategy according to the cloud optimization suggestions.

Images