A method for determining abnormalities in hazardous waste management data

By analyzing and recording the business processes of hazardous waste management, identifying the logical dependencies between processing links and performing key feature evaluations, and combining historical data for prediction and abnormal detection, the problem that traditional methods cannot integrate business processes is solved, and data management is improved Accuracy and security reduce environmental and security risks.

CN118585944BActive Publication Date: 2024-11-15SHENZHEN DEEP STATE ENVIRONMENTAL TECH CO LTD +1
View PDF 4 Cites 0 Cited by

Patent Information

Application Number
CN202411067284.6
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-08-06
Publication Date
2024-11-15
Estimated Expiration
2044-08-06

AI Technical Summary

Technical Problem

Traditional data anomaly judgment methods cannot effectively integrate multiple links of business processes in the field of hazardous waste management, resulting in the data analysis results being unable to fully reflect the complexity of actual operation, lack of immediate feedback, affecting the accuracy of the prediction model and the flexibility of the processing strategy, increasing The risk of environmental pollution and safety accidents is evident.

Method used

By analyzing the business processes of hazardous waste management, collecting and recording the type, quantity, time and location data of waste treatment, generating waste process analysis data, identifying logical dependencies between processing links, calculating the influence scores of key characteristics, and combining historical data Perform weight allocation and prediction, set abnormality detection standards, and perform data quality analysis to identify abnormal data points.

Benefits of technology

It improves the accuracy and consistency of data, enhances the ability to identify data abnormalities, optimizes the forecast of waste treatment demand and resource allocation, reduces the risks of environmental pollution and safety accidents, and improves the overall efficiency and safety of data management.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118585944B_ABST
    Figure CN118585944B_ABST
Patent Text Reader

Abstract

The present invention relates to the technical field of data quality inspection, and specifically to a method for judging abnormalities in hazardous waste management data, comprising the following steps: based on waste management process data, analyzing the business process of hazardous waste management, collecting and recording the type, quantity, time and location data of waste treatment, and generating waste process analysis data. In the present invention, by analyzing the business process of waste treatment and recording key data, the accuracy and consistency of data are enhanced, the logical dependencies between processing links are identified, the key connections in data flow and processing are evaluated, the ability to discover and resolve data anomalies is improved, the identification and weight allocation of key features are improved, the accuracy of waste treatment demand forecasting is improved, and by setting anomaly detection standards and performing data quality analysis, the ability to identify abnormal data points is enhanced, environmental pollution and safety accidents are avoided, the overall efficiency and safety of data management are improved, and the reliability of data analysis results is ensured.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of data quality inspection, and in particular to a method for determining abnormalities in hazardous waste management data. Background Art

[0002] The field of data quality inspection technology ensures the accuracy, consistency, completeness and availability of data by using a variety of methods and tools. It is a key branch of data management and is applied in multiple fields such as data science, big data analysis, and information system management. It aims to identify and correct data errors, fill in data gaps, standardize data formats, and verify the logical consistency of data, improve the credibility of data, and provide a solid data foundation for decision support. It plays a key role in business intelligence, risk management, and customer relationship management.

[0003] Among them, the method for judging anomalies in hazardous waste management data aims to use data analysis technology to identify anomalies and inconsistencies in the data set and make corrections and adjustments during the hazardous waste management process, to ensure that the data accurately reflects the actual situation in the process of handling, transferring and storing hazardous wastes, optimize the compliance and safety of risk waste management, prevent environmental pollution and safety accidents, improve the safety and efficiency of hazardous waste treatment, and reduce economic and legal risks caused by data problems.

[0004] Traditional data anomaly judgment methods cannot effectively integrate multiple links of the business process in the field of hazardous waste management, resulting in data analysis results that cannot fully reflect the complexity of actual operations. This decentralized data processing method leads to the neglect of key features, affecting the accuracy and practicality of the prediction model. The lack of immediate feedback on data quality analysis results leads to delayed response measures when data anomalies occur, increasing the risk of environmental pollution and safety accidents. The lack of effective data difference analysis and dependency identification leads to inflexible processing strategies that cannot adapt to rapidly changing environmental conditions, affecting waste treatment efficiency and compliance. Summary of the invention

[0005] The purpose of the present invention is to solve the shortcomings of the prior art and to propose a method for determining abnormalities in hazardous waste management data.

[0006] In order to achieve the above object, the present invention adopts the following technical scheme, a method for determining abnormality in hazardous waste management data, comprising the following steps:

[0007] S1: Based on the waste management process data, analyze the business process of hazardous waste management, collect and record the type, quantity, time and location data of waste treatment, and generate waste process analysis data;

[0008] S2: Based on the waste process analysis data, perform data difference analysis, compare the processing volume and processing time of multiple links, identify the logical dependency between the processing links, and generate a dependency analysis network;

[0009] S3: Based on the dependency analysis network, identifying key features of waste treatment types and links in multiple waste treatment links, and calculating influence scores of multiple features to generate key feature identification information;

[0010] S4: Based on the key feature identification information and in combination with historical data, weights are assigned to key features of multiple data processing links, waste treatment demand is predicted, and a treatment scale prediction model is generated;

[0011] S5: Based on the processing scale prediction model, the processing demand prediction values ​​of various types of waste are calculated, and by comparing and analyzing with the data of the same period in the past, the model parameters are adjusted to optimize the prediction accuracy, and the prediction parameter adjustment results are generated;

[0012] S6: Based on the prediction parameter adjustment results, set anomaly detection standards and perform data quality analysis, combine historical data patterns, identify and record abnormal data points, and generate abnormal data judgment results.

[0013] As a further solution of the present invention, the waste process analysis data includes waste classification information, waste treatment time information, and waste treatment location data; the dependency analysis network includes data flow information between processing links, logical dependency indicators and correlation analysis results, and comparative analysis data of processing volume and processing time; the key feature identification information includes key waste treatment types, feature influence scores, and correlation scores; the processing scale prediction model includes weight distribution ratios of key features, waste type treatment demand prediction algorithms, and historical data comparison results; the prediction parameter adjustment results include feature weight adjustment results, prediction algorithm parameter optimization information, and prediction accuracy improvement data; the abnormal data judgment results include abnormal data point identification results, abnormal detection standard setting results, and historical pattern comparison abnormal data records.

[0014] As a further solution of the present invention, based on the waste management process data, the business process of hazardous waste management is analyzed, and the type, quantity, time and location data of waste treatment are collected and recorded. The steps of generating waste process analysis data are specifically as follows:

[0015] S101: Based on the waste management process data, collect hazardous waste management process data, record the waste types and quantities in multiple processing links, archive these waste process records, and generate quantity type analysis results;

[0016] S102: Based on the quantitative type analysis result, analyzing and recording the processing time and location of multiple processing links, and optimizing the consistency and accuracy of the data to obtain time and location record data;

[0017] S103: Analyze the business process of waste treatment by using the time and location record data, identify and record the treatment links required for various types of waste, and generate waste process analysis data.

[0018] As a further solution of the present invention, based on the waste process analysis data, data difference analysis is performed to compare the processing volume and processing time of multiple links, identify the logical dependencies between the processing links, and generate the dependency analysis network in the following steps:

[0019] S201: using the waste process analysis data, calculating the statistical data of waste treatment volume and treatment time in multiple links, analyzing the distribution characteristics of time and quantity, and generating a treatment volume and time statistics table;

[0020] S202: Based on the processing volume and time statistics table, comparative analysis is performed on the processing volume and processing time of multiple links, and processing differences between multiple links are identified through numerical comparison, so as to generate link processing difference data;

[0021] S203: Utilize the process difference data of the links to identify the logical dependencies between the processing links, build a dependency model based on the correlation between the processing time and the amount, and generate a dependency analysis network.

[0022] As a further solution of the present invention, based on the dependency analysis network, the key features of waste treatment types and links in multiple waste treatment links are identified, and the influence scores of multiple features are calculated. The steps of generating key feature identification information are specifically as follows:

[0023] S301: Based on the dependency analysis network, identifying characteristic data in multiple waste treatment links, including treatment type, time and efficiency, analyzing the occurrence frequency and correlation of target characteristics, and generating characteristic correlation data;

[0024] S302: Based on the feature correlation data, calculate the influence scores of multiple features in the waste treatment process, evaluate the importance of multiple features, and obtain influence score data;

[0025] S303: Based on the influence scoring data, identify key features that affect the waste treatment process, analyze the actual impact of the target features on the treatment process, and generate key feature identification information.

[0026] As a further solution of the present invention, based on the key feature identification information and in combination with historical data, the key features of multiple data processing links are weighted and the waste treatment demand is predicted. The specific steps of generating a treatment scale prediction model are as follows:

[0027] S401: Based on the key feature identification information and in combination with historical data, identifying historical performance data of multiple key features, including processing volume and time changes, and generating historical feature performance data;

[0028] S402: Based on the historical feature performance data, according to the historical impact of multiple features on processing efficiency, weights are assigned to key features, the predicted impact of the features is optimized, and feature weight adjustment parameters are obtained;

[0029] S403: Based on the feature weight adjustment parameters, a support vector machine regression algorithm is used to construct a waste treatment demand prediction model, calculate and predict the demand for multiple waste treatment links, and generate a treatment scale prediction model.

[0030] As a further solution of the present invention, the support vector machine regression is according to the formula:

[0031] ;

[0032] Calculate the forecast of waste treatment demand, where: For the predicted waste disposal demand, is the weight vector, For environmental factors, is a seasonal variable, For waste urgency, For historical processing efficiency, is the weight coefficient of environmental factors, is the seasonal variable weight coefficient, is the waste urgency weight coefficient, is the historical processing efficiency weight coefficient, is the intercept.

[0033] As a further solution of the present invention, based on the processing scale prediction model, the processing demand prediction values ​​of various types of waste are calculated, and by comparing and analyzing with the data of the same period in the past, the model parameters are adjusted to optimize the prediction accuracy. The steps of generating the prediction parameter adjustment results are specifically as follows:

[0034] S501: Based on the processing scale prediction model, the processing demand prediction values ​​of various types of waste are calculated, and the demand prediction calculation results are obtained by combining the historical processing volume and the environmental changes in the real-time period;

[0035] S502: Based on the demand forecast calculation result, by comparing with historical similar processing data, identifying the difference between the predicted value and the actual processing volume, and generating difference comparison analysis data;

[0036] S503: Based on the difference comparison and analysis data, adjust and process demand forecasting model parameters, including weights and deviation values, optimize the matching degree and consistency between the forecasted value and the actual data, and generate forecast parameter adjustment results.

[0037] As a further solution of the present invention, based on the prediction parameter adjustment result, the abnormality detection standard is set and data quality analysis is performed, and the abnormal data points are identified and recorded in combination with the historical data pattern, and the steps of generating the abnormal data judgment result are specifically as follows:

[0038] S601: Based on the prediction parameter adjustment result, according to the deviation rate between the predicted value and the actual value, analyze and calculate the anomaly detection standard, set the threshold of data quality analysis, and generate the detection standard matching result;

[0039] S602: Based on the detection standard matching result, a Z-score algorithm is used to perform data quality analysis, multiple data points are analyzed and checked according to the anomaly detection standard, and abnormal data points are identified and recorded using a set threshold to obtain an abnormal identification record list;

[0040] S603: Based on the abnormal identification record list, combined with historical data patterns and current business trends, the occurrence frequency, type and impact scope of multiple abnormal data are analyzed to generate abnormal data judgment results.

[0041] As a further solution of the present invention, the Z score algorithm is according to the formula:

[0042] ;

[0043] Calculates abnormal data points, where Z is the Z score used to identify the degree of abnormality of the data point. is the value of a single data point, is the mean of the data set, is the weight of the data point, is the standard deviation of the data set, is a constant term.

[0044] Compared with the prior art, the advantages and positive effects of the present invention are:

[0045] In the present invention, by analyzing the business process of waste treatment and recording key data, the accuracy and consistency of data are enhanced, difference analysis is performed and a dependency network is generated, the logical dependencies between processing links are identified, the key connections in data flow and processing are evaluated, the ability to discover and resolve data anomalies is improved, the identification and weight allocation of key features are improved, the accuracy of waste treatment demand forecasts is improved, resource allocation and risk management are optimized, and the ability to identify abnormal data points is enhanced by setting anomaly detection standards and performing data quality analysis, environmental pollution and safety accidents are avoided, the overall efficiency and security of data management are improved, and the reliability of data analysis results is ensured. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 It is a schematic diagram of the main steps of the present invention;

[0047] Figure 2 It is a schematic diagram of the refinement of S1 of the present invention;

[0048] Figure 3 It is a schematic diagram of the refinement of S2 of the present invention;

[0049] Figure 4 It is a schematic diagram of the refinement of S3 of the present invention;

[0050] Figure 5 It is a schematic diagram of the refinement of S4 of the present invention;

[0051] Figure 6 It is a detailed schematic diagram of S5 of the present invention;

[0052] Figure 7 It is a detailed schematic diagram of S6 of the present invention. DETAILED DESCRIPTION

[0053] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0054] In the description of the present invention, it should be understood that the terms "length", "width", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, in the description of the present invention, "multiple" means two or more, unless otherwise clearly and specifically defined.

[0055] Example 1: Please refer to Figure 1 The present invention provides a technical solution, a method for determining abnormalities in hazardous waste management data, comprising the following steps:

[0056] S1: Based on the waste management process data, analyze the business process of hazardous waste management, collect and record the type, quantity, time and location data of waste treatment, and generate waste process analysis data;

[0057] S2: Based on the waste process analysis data, perform data difference analysis, compare the processing volume and processing time of multiple links, identify the logical dependencies between the processing links, and generate a dependency analysis network;

[0058] S3: Based on the dependency analysis network, identify the key features of waste treatment types and links in multiple waste treatment links, calculate the influence scores of multiple features, and generate key feature identification information;

[0059] S4: Based on the key feature identification information and combined with historical data, the key features of multiple data processing links are weighted, the demand for waste treatment is predicted, and a treatment scale prediction model is generated;

[0060] S5: Based on the treatment scale prediction model, calculate the treatment demand forecast values ​​of various waste types, compare and analyze with the data of the same period in the past, adjust the model parameters to optimize the prediction accuracy, and generate the prediction parameter adjustment results;

[0061] S6: Based on the prediction parameter adjustment results, set the anomaly detection standard and perform data quality analysis, combine the historical data patterns, identify and record abnormal data points, and generate abnormal data judgment results.

[0062] Waste process analysis data include waste classification information, waste treatment time information, and waste treatment location data. The dependency analysis network includes data flow information between treatment links, logical dependency indicators and correlation analysis results, and comparative analysis data of treatment volume and treatment time. Key feature identification information includes key waste treatment types, feature influence scores, and correlation scores. The treatment scale prediction model includes the weight distribution ratio of key features, waste type treatment demand prediction algorithm, and historical data comparison results. The prediction parameter adjustment results include feature weight adjustment results, prediction algorithm parameter optimization information, and prediction accuracy improvement data. The abnormal data judgment results include abnormal data point identification results, abnormal detection standard setting results, and historical pattern comparison abnormal data records.

[0063] See also Figure 2 , based on the waste management process data, analyze the business process of hazardous waste management, collect and record the type, quantity, time and location data of waste treatment, and the steps to generate waste process analysis data are as follows:

[0064] S101: Based on the waste management process data, collect hazardous waste management process data, record the waste types and quantities in multiple processing links, archive these waste process records, and generate the quantity type analysis results. The specific process is as follows;

[0065] In sub-step S101, based on the waste management process data, the hazardous waste management process data is collected, the waste types and quantities of multiple processing links are recorded, and the data classification algorithm is used to perform statistical analysis on the types and quantities of wastes. The collected waste types and quantity data are grouped according to the preset classification standards, including classification according to the hazard level and physical state. The classification statistical algorithm is used to accumulate the quantity of each type of waste to obtain the total amount of each type of waste. The formula is: ,in, represents the total quantity of waste of category i, represents the quantity of waste of type i in the jth treatment stage, It represents the total number of processing links, identifies the distribution of various wastes through calculation, and generates quantitative type analysis results.

[0066] S102: Based on the results of the quantity type analysis, the processing time and location of multiple processing links are analyzed and recorded, and the consistency and accuracy of the data are optimized. The specific process of obtaining the time location record data is as follows;

[0067] In sub-step S102, based on the results of the quantity type analysis, the processing time and location of multiple processing links are analyzed and recorded, the consistency and accuracy of the data are optimized, and data cleaning and correction methods are used to ensure the quality of the recorded data. The integrity and logic of the data records are checked, including the rationality of the processing time and the accuracy of the location information. The erroneous data are corrected by querying the original records or re-collecting the data. The linear regression model is used to predict the impact of the processing time on the waste treatment efficiency. The formula is: ,in, Represents processing time, is the intercept term, The serial number representing the treatment process, i.e. the number of multiple stages in the waste treatment process. Represents the amount of waste processed in the target processing link, used to measure how much waste is processed in each link and to evaluate the workload and processing capacity of each link. is the processing step number in the linear regression model Processing time The coefficient of influence, is the amount of waste treated in the linear regression model Processing time The coefficient of influence, As the error term, the data processing process is optimized to improve the consistency and accuracy of the data and obtain the time and location record data.

[0068] S103: Using the time location record data, analyzing the business process of waste treatment, identifying and recording the treatment links required for various types of waste, and generating waste process analysis data, the specific process is as follows;

[0069] In sub-step S103, the time and location record data are used to analyze the business process of waste treatment. Through the process analysis method, the treatment links required for various waste types are identified and recorded. For each type of waste, the treatment links required are determined based on the physical and chemical characteristics, including pretreatment, chemical treatment, and safe landfill. The process analysis model is used to combine the target links with the time and location data to conduct efficiency and risk assessment. The formula is: ,in, Represents processing efficiency, is the adjustment factor, represents the weight of the kth processing link, To identify the key links in the treatment of various waste types and generate waste process analysis data for the completion of target links.

[0070] See also Figure 3 ,Based on the waste process analysis data, perform data difference analysis, compare the processing volume and processing time of multiple links, identify the logical dependencies between the processing links, and generate the dependency analysis network in the following steps:

[0071] S201: using waste process analysis data, calculating the statistical data of waste treatment volume and treatment time in multiple links, analyzing the distribution characteristics of time and quantity, and generating a treatment volume and time statistical table;

[0072] In sub-step S201, based on the waste process analysis data, statistical analysis is performed on the waste treatment volume and treatment time in multiple treatment links, and the mean and variance of the waste treatment volume in multiple links are calculated. At the same time, similar statistical processing is performed on the treatment time, and the distribution characteristics of the target data points are calculated using the statistical analysis formula. The formula is: , ,in, represents the average processing time, represents the variance of processing time, is the processing time of the kth stage, The total number of links is generated by analyzing the distribution characteristics of processing time and quantity to generate a processing volume time statistics table.

[0073] S202: Based on the processing volume and time statistics table, a comparative analysis of the processing volume and processing time between multiple links is performed, and the processing differences between multiple links are identified through numerical comparison. The specific process of generating link processing difference data is as follows;

[0074] In sub-step S202, based on the processing volume and time statistics table, a comparative analysis of the processing volume and processing time between multiple links is performed. By constructing a difference analysis formula, the processing volume and time of multiple links are compared to identify the links with differences. The formula is: ,in, represents the ratio of the treatment differences between the two links, and They represent the processing volume or time of the i-th and j-th links respectively, revealing the relative differences in processing volume or time between differentiated links and generating link processing difference data.

[0075] S203: Using the process difference data of each link, identifying the logical dependency between the processing links, building a dependency model according to the correlation between the processing time and the amount, and generating a dependency analysis network, the specific process is as follows;

[0076] In sub-step S203, the process difference data is used to identify the logical dependencies between the processing links, and a dependency model is constructed. The dependencies between multiple links are determined by analyzing the correlation between processing time and quantity. The model is established using the correlation analysis formula. The formula is: ,in, is the correlation coefficient, which indicates the correlation between the processing time and quantity of multiple links. represents the processing time of the i-th link, represents the processing volume of the i-th link, represents the average processing time, It represents the average value of the processing amount, constructs a network of logical dependencies, and generates a dependency analysis network.

[0077] See also Figure 4 ,Based on the dependency analysis network, the key features of waste treatment types and links in multiple waste treatment links are identified, and the influence scores of multiple features are calculated.,The steps for generating key feature identification information are as follows:

[0078] S301: Based on the dependency analysis network, identifying characteristic data in multiple waste treatment links, including treatment type, time and efficiency, analyzing the frequency and correlation of target characteristics, and generating characteristic correlation data in detail;

[0079] In sub-step S301, based on the dependency analysis network, the characteristic data in the waste treatment process is analyzed. By counting the treatment type, time and efficiency of each process, and calculating the occurrence frequency and correlation of the target characteristics, the frequency analysis and correlation measurement formula are used to quantify the impact of multiple treatment characteristics. The formula is: , ,in, Representation characteristics The frequency of occurrence, It is a feature The number of occurrences, is the total number of observations, It is a feature and The correlation between and Represents the features in the kth observation and The value of and It is the average value of the corresponding feature, generating feature association data.

[0080] S302: Based on the feature correlation data, the influence scores of multiple features in the waste treatment process are calculated, the importance of multiple features is evaluated, and the process of obtaining the influence score data is as follows;

[0081] In sub-step S302, based on the feature correlation data, the influence scoring model is used to evaluate the importance of multiple features in the waste treatment process and calculate the influence score. The formula is: ,in, Representative features The influence score of It is a feature The weight factor is assigned according to its importance in the process. Features and The correlation between features is evaluated, the importance of multiple features is evaluated, and the influence score data is obtained.

[0082] S303: Based on the impact score data, identify the key features that affect the waste treatment process, analyze the actual impact of the target features on the treatment process, and generate the key feature identification information.

[0083] In sub-step S303, based on the impact score data, the key features that affect the waste treatment process are analyzed. The impact scores of multiple features and the actual operation of the treatment process are considered to identify the key features that have the greatest impact on the link. The key features are prioritized using a criticality analysis. The formula is: ,in, Representation characteristics The critical rating of To adjust the coefficient, adjust it according to the influence of the feature in actual operation. Features The influence score of the ,generated key feature identification information.

[0084] See also Figure 5 Based on the key feature identification information and combined with historical data, the key features of multiple data processing links are weighted and the waste treatment demand is predicted. The specific steps of generating the treatment scale prediction model are as follows:

[0085] S401: Based on the key feature identification information and in combination with historical data, the historical performance data of multiple key features are identified, including the processing volume and time changes. The specific process of generating the historical feature performance data is as follows;

[0086] In sub-step S401, based on the key feature identification information and combined with historical data, the historical performance analysis of multiple key features is performed. By integrating the past processing volume and time data of multiple features, the change trend and volatility of each feature over time are calculated. Using time series analysis, the historical performance of each feature is quantitatively analyzed. The formula is: ,in, represents the performance value of the feature at time t, is the performance value of the previous period, is the smoothing coefficient, Generate historical characteristic performance data for random fluctuations.

[0087] S402: Based on historical feature performance data, according to the historical impact of multiple features on processing efficiency, weights are assigned to key features, the predicted impact of features is optimized, and the process of obtaining feature weight adjustment parameters is specifically as follows;

[0088] In sub-step S402, based on the historical feature performance data, the impact of multiple features on historical processing efficiency is analyzed, and a linear weighted model is applied to assign corresponding weights to each key feature to optimize the prediction of future processing efficiency. The formula is: ,in, represents the weight of the i-th feature, Score the impact of the target feature, is the attenuation coefficient, which controls the sensitivity of the impact score. is the total number of features, Represents the base of the natural logarithm, generating the feature weight adjustment parameter.

[0089] S403: Based on the feature weight adjustment parameters, a support vector machine regression algorithm is used to construct a waste treatment demand prediction model, calculate and predict the demand for multiple waste treatment links, and generate a treatment scale prediction model. The specific process is as follows;

[0090] In sub-step S403, based on the characteristic weight adjustment parameters, the data sets in multiple waste treatment links are analyzed, the data are cleaned and preprocessed, including normalization and missing value processing, to ensure data quality, and weights are assigned to the characteristic data of each waste treatment link. The weight assignment is based on the correlation analysis between multiple characteristics and demand in historical data, including calculating the correlation coefficients between multiple characteristics and target variables, adjusting weight parameters according to the size of the correlation coefficients, integrating the characteristics with adjusted weights, building a waste treatment demand forecasting model, forecasting waste treatment demand, optimizing the allocation and scheduling of waste treatment resources, and generating a treatment scale forecasting model.

[0091] Support vector machine regression, according to the formula:

[0092] ;

[0093] Calculate the forecast of waste treatment demand, where: is the predicted waste treatment demand, which represents the predicted value of waste treatment output by the model. is a weight vector used to measure the influence of each input feature on the predicted output. Environmental factors reflect the impact of current environmental conditions on waste treatment needs. is a seasonal variable, considering the impact of seasonal changes on waste treatment demand. The waste urgency level reflects the urgency of waste treatment. The historical treatment efficiency reflects the efficiency level of treating similar wastes in history. is the environmental factor weight coefficient, which is used to quantify the impact of environmental variables on waste treatment demand. is the seasonal variable weight coefficient, which is used to quantify the impact of seasonal changes on waste treatment demand. is the waste urgency weight coefficient, which is used to quantify the impact of waste treatment urgency on the treatment demand forecast. is the historical treatment efficiency weight coefficient, which is used to quantify the impact of the efficiency data of previous treatment of similar waste on the current treatment demand forecast. is the intercept, which is used to adjust the output of the model.

[0094] The specific implementation process of the formula is as follows: collect environmental factor data, seasonal data, waste urgency and historical treatment efficiency data related to waste treatment, use historical data sets to determine the weight coefficient of the target parameter through cross-validation method, including using statistical analysis and machine learning techniques to analyze the correlation between multiple parameters and waste treatment demand, substitute the weight parameters into the support vector machine regression model, train the model to fit the actual data, and adjust the model's weight vector and bias term, and combine the current input data to predict the waste treatment demand.

[0095] See also Figure 6 Based on the treatment scale prediction model, the treatment demand forecast values ​​of various types of waste are calculated. By comparing and analyzing the data of the same period in the past, the model parameters are adjusted to optimize the prediction accuracy. The specific steps for generating the prediction parameter adjustment results are as follows:

[0096] S501: Based on the treatment scale prediction model, the treatment demand prediction values ​​of various types of waste are calculated, and the process of obtaining the demand prediction calculation results by combining the historical treatment volume and the environmental changes of the real-time cycle is specifically as follows;

[0097] In sub-step S501, based on the treatment scale prediction model, the treatment demand forecast values ​​of various waste types are calculated. The historical treatment volume and current environmental change factors, including seasonal changes, policy impacts and market demand, are combined to use a multivariate linear regression model to predict demand. The formula is: ,in, represents the predicted processing demand, is the intercept of the model, Represents the historical processing volume, represents the environmental change parameter, is the regression coefficient in the multivariate linear regression model, which indicates the impact of historical processing volume on the predicted demand. is the regression coefficient in the multiple linear regression model, which indicates the impact of environmental change parameters on the predicted demand. It represents the error term, and the processing demand is predicted based on the input historical data and environmental variables to obtain the demand forecast calculation result.

[0098] S502: Based on the demand forecast calculation result, by comparing with the historical similar processing data, identifying the difference between the forecast value and the actual processing volume, and generating the difference comparison analysis data, the specific process is as follows;

[0099] In sub-step S502, based on the demand forecast calculation results, by comparing with historical similar processing data, the difference between the predicted value and the actual processing volume is identified, and the difference analysis method is used to calculate the deviation between the predicted value and the actual value, and statistical analysis is performed. The formula is: ,in, represents the total deviation, and They represent the predicted value and actual processing volume of the i-th data point respectively. In this way, the accuracy and reliability of the prediction model are quantified and difference comparison analysis data are generated.

[0100] S503: Based on the difference comparison analysis data, the demand forecast model parameters are adjusted, including weights and deviation values, to optimize the matching degree and consistency between the forecast value and the actual data, and the process of generating the forecast parameter adjustment result is as follows;

[0101] In sub-step S503, based on the difference comparison analysis data, the demand forecasting model parameters, including weights and bias values, are adjusted and the model parameters are optimized using the gradient descent method to reduce the forecast error and improve the forecast accuracy. The formula is: ,in, represents the model parameters, is the value of the parameter before updating, is the new value obtained after applying the gradient descent method, is the learning rate, Represents the loss function About parameters Through iterative training, the model parameters are adjusted to match the actual data, the matching degree and consistency between the predicted value and the actual data are optimized, and the prediction parameter adjustment results are generated.

[0102] See also Figure 7 , based on the prediction parameter adjustment results, set the anomaly detection standard and perform data quality analysis, combine the historical data pattern, identify and record abnormal data points, and generate the abnormal data judgment results in the following steps:

[0103] S601: Based on the prediction parameter adjustment result, according to the deviation rate between the predicted value and the actual value, the abnormal detection standard is analyzed and calculated, the threshold of the data quality analysis is set, and the specific process of generating the detection standard matching result is as follows;

[0104] In sub-step S601, based on the prediction parameter adjustment results, according to the deviation rate between the predicted value and the actual value, the anomaly detection standard is analyzed and calculated, and the deviation rate of each data point is calculated, which is defined as the ratio of the difference between the predicted value and the actual value to the actual value, to evaluate whether the data point is within the normal range. The formula is ,in, is the deviation rate, is the predicted value of the ith data point, It is the corresponding actual value. According to the statistical distribution of the deviation rate, the threshold of data quality analysis is set, including setting the upper limit of the deviation rate to 5%. Data points exceeding the threshold are recorded as abnormalities, the acceptable deviation level is determined, and the detection standard matching result is generated.

[0105] S602: Based on the detection standard matching result, the Z score algorithm is used to perform data quality analysis, multiple data points are analyzed and checked according to the anomaly detection standard, and the abnormal data points are identified and recorded using the set threshold to obtain the abnormal identification record list. The specific process is as follows;

[0106] In sub-step S602, based on the detection standard matching results, the data set is normalized to the same scale, and the Z-score algorithm is used to perform statistical analysis on the data points collected in the processing link, including calculating the Z-score of each data point, that is, the deviation of each data point from the mean divided by the standard deviation, and identifying data points with abnormally large deviations by comparing the Z-score of each data point with a preset threshold. The set threshold is obtained through the analysis of the characteristics of the data distribution and historical abnormal data, and a list of abnormal identification records is generated.

[0107] Z score algorithm, according to the formula:

[0108] ;

[0109] Calculates abnormal data points, where Z is the Z score used to identify the degree of abnormality of the data point. is the value of a single data point, representing the data measurement value at the target time point and location. is the mean of the data set, which represents the average level of all data point values. is the weight of the data point, which is adjusted according to the timeliness or business importance of the data point. is the standard deviation of the data set, which indicates the dispersion or variability of the data point values. is a constant term used to adjust the scale of the denominator and avoid divisors that are too small.

[0110] The specific implementation process of the formula is as follows: Assign a weight to each data point based on its time proximity and business importance , and assign higher weights to data points that are closer to the current data and related to key business events in time. Consider the weights and constant terms of the data points, use the Z-score formula to calculate the Z score of each data point, compare the Z of each data point with the preset anomaly threshold, mark the data points that exceed the threshold as abnormal, record and summarize the data points identified as abnormal, build a list of abnormal data judgment results, and perform data quality management and processing optimization.

[0111] S603: Based on the abnormal identification record list, combined with the historical data pattern and the current business trend, the occurrence frequency, type and impact scope of multiple abnormal data are analyzed to generate the abnormal data judgment result. The specific process is as follows;

[0112] In sub-step S603, based on the abnormal identification record list, combined with historical data patterns and current business trends, the occurrence frequency, type and impact range of multiple abnormal data are analyzed, the abnormal data are classified through cluster analysis, and the common characteristics and frequent conditions of each type of abnormal data are analyzed. The K-means clustering algorithm is used to divide the abnormal data into multiple groups, and the characteristics of each group are analyzed. The formula is ,in, is the number of groups, is the data point in the ith population, is the center point of the ith group, Represents the sum of squared errors within a group, analyzes the pattern and business impact of abnormal data, and generates abnormal data judgment results.

[0113] The above are only preferred embodiments of the present invention and are not intended to limit the present invention in other forms. Any technician familiar with the profession may use the technical contents disclosed above to change or modify them into equivalent embodiments with equivalent changes and apply them to other fields. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. A method for determining abnormalities in hazardous waste management data, characterized in that: The following steps are involved: Based on the waste management process data, analyze the business process of hazardous waste management, collect and record the type, quantity, time and location data of waste treatment, and generate waste process analysis data; Based on the waste process analysis data, perform data difference analysis, compare the processing volume and processing time of multiple links, identify the logical dependency between the processing links, and generate a dependency analysis network; Based on the dependency analysis network, identifying key features of waste treatment types and links in multiple waste treatment links, and calculating influence scores of multiple features to generate key feature identification information; Based on the key feature identification information and in combination with historical data, weights are assigned to key features of multiple data processing links, waste treatment demand is predicted, and a treatment scale prediction model is generated; Based on the processing scale prediction model, the processing demand prediction values ​​of various types of waste are calculated, and by comparing and analyzing with the data of the same period in the past, the model parameters are adjusted to optimize the prediction accuracy, and the prediction parameter adjustment results are generated; Based on the prediction parameter adjustment results, set anomaly detection standards and perform data quality analysis, combine historical data patterns, identify and record abnormal data points, and generate abnormal data judgment results; Based on the waste management process data, the business process of hazardous waste management is analyzed, and the type, quantity, time and location data of waste treatment are collected and recorded. The specific steps for generating waste process analysis data are as follows: Based on the waste management process data, collect hazardous waste management process data, record the waste types and quantities in multiple processing links, archive these waste process records, and generate quantity type analysis results; Based on the quantitative type analysis results, the processing time and location of multiple processing links are analyzed and recorded, and the consistency and accuracy of the data are optimized to obtain time and location record data; Utilizing the time location record data, analyzing the business process of waste treatment, identifying and recording the treatment links required for various types of waste, and generating waste process analysis data; Based on the waste process analysis data, data difference analysis is performed to compare the processing volume and processing time of multiple links, identify the logical dependencies between the processing links, and generate the dependency analysis network in the following steps: Using the waste process analysis data, calculate the statistical data of waste treatment volume and treatment time in multiple links, analyze the distribution characteristics of time and quantity, and generate a treatment volume and time statistical table; Based on the processing volume and time statistics table, comparative analysis of the processing volume and processing time between multiple links is performed, and processing differences between multiple links are identified through numerical comparison to generate link processing difference data; Using the process difference data of the links, the logical dependency relationship between the processing links is identified, and according to the correlation between the processing time and quantity, a dependency model is constructed to generate a dependency analysis network; Based on the dependency analysis network, the key features of waste treatment types and links in multiple waste treatment links are identified, and the influence scores of multiple features are calculated. The steps of generating key feature identification information are specifically as follows: Based on the dependency analysis network, identifying characteristic data in multiple waste treatment links, including treatment type, time and efficiency, analyzing the occurrence frequency and correlation of target characteristics, and generating characteristic correlation data; Based on the feature correlation data, the influence scores of multiple features in the waste treatment process are calculated, the importance of multiple features is evaluated, and influence score data is obtained; Based on the impact score data, identify key features that affect the waste treatment process, analyze the actual impact of the target features on the treatment process, and generate key feature identification information; Based on the key feature identification information, combined with historical data, weights are assigned to the key features of multiple data processing links, and the waste treatment demand is predicted. The specific steps of generating a treatment scale prediction model are as follows: Based on the key feature identification information, combined with historical data, identifying historical performance data of multiple key features, including processing volume and time changes, and generating historical feature performance data; Based on the historical feature performance data, weights are assigned to key features according to the historical impact of multiple features on processing efficiency, and the predicted impact of the features is optimized to obtain feature weight adjustment parameters; Based on the feature weight adjustment parameters, a support vector machine regression algorithm is used to construct a waste treatment demand prediction model, calculate and predict the demand for multiple waste treatment links, and generate a treatment scale prediction model; Based on the processing scale prediction model, the processing demand prediction values ​​of various types of waste are calculated, and by comparing and analyzing with the data of the same period in the past, the model parameters are adjusted to optimize the prediction accuracy. The specific steps of generating the prediction parameter adjustment results are as follows: Based on the processing scale prediction model, the processing demand prediction values ​​of various types of waste are calculated, and the demand prediction calculation results are obtained by combining the historical processing volume with the environmental changes in the real-time cycle; Based on the demand forecast calculation results, by comparing with historical similar processing data, identifying the difference between the predicted value and the actual processing volume, and generating difference comparison analysis data; Based on the difference comparison and analysis data, the demand forecasting model parameters are adjusted, including weights and deviation values, to optimize the matching degree and consistency between the forecasted value and the actual data, and generate forecast parameter adjustment results; Based on the prediction parameter adjustment results, the abnormal detection standards are set and data quality analysis is performed. In combination with the historical data patterns, abnormal data points are identified and recorded, and the steps of generating abnormal data judgment results are specifically as follows: Based on the prediction parameter adjustment result, according to the deviation rate between the predicted value and the actual value, the anomaly detection standard is analyzed and calculated, the threshold of the data quality analysis is set, and the detection standard matching result is generated; Based on the detection standard matching results, a Z-score algorithm is used to perform data quality analysis, multiple data points are analyzed and checked according to the anomaly detection standard, and abnormal data points are identified and recorded using a set threshold to obtain an abnormal identification record list; Based on the abnormality identification record list, combined with historical data patterns and current business trends, the occurrence frequency, type and impact scope of multiple abnormal data are analyzed to generate abnormal data judgment results.

2. The method for determining abnormality in hazardous waste management data according to claim 1, characterized in that: The waste process analysis data includes waste classification information, waste treatment time information, and waste treatment location data; the dependency analysis network includes data flow information between treatment links, logical dependency indicators and correlation analysis results, and comparative analysis data of treatment volume and treatment time; the key feature identification information includes key waste treatment types, feature influence scores, and correlation scores; the treatment scale prediction model includes weight distribution ratios of key features, waste type treatment demand prediction algorithms, and historical data comparison results; the prediction parameter adjustment results include feature weight adjustment results, prediction algorithm parameter optimization information, and prediction accuracy improvement data; the abnormal data judgment results include abnormal data point identification results, abnormal detection standard setting results, and historical pattern comparison abnormal data records.

3. The method for determining abnormality in hazardous waste management data according to claim 1, characterized in that: The support vector machine regression is based on the formula: y=ω·(x1·f1+x2·f2+x3·f3+x4·f4)+b Calculate the predicted value of waste treatment demand, where y is the predicted waste treatment demand, ω is the weight vector, x1 is the environmental factor, x2 is the seasonal variable, x3 is the waste urgency, x4 is the historical treatment efficiency, f1 is the environmental factor weight coefficient, f2 is the seasonal variable weight coefficient, f3 is the waste urgency weight coefficient, f4 is the historical treatment efficiency weight coefficient, and b is the intercept.

4. The method for determining abnormality in hazardous waste management data according to claim 3, characterized in that: The Z-score algorithm follows the formula: Calculate the abnormal data points, where Z is the Z score used to identify the degree of abnormality of the data point, X is the value of a single data point, μ is the mean of the data set, and w is the weight of the data point. is the standard deviation of the data set, and C is a constant term.

Citation Information

Patent Citations

  • Visual intelligent emergency command platform for medical waste disposal in epidemic situation

    CN113822394A

  • Hazardous waste supervision method, system and device based on artificial intelligence

    CN116109142A

  • Dangerous waste quantity change detection system and method based on video analysis

    CN117078024A

  • Prediction model training method and hazardous waste monitoring method

    CN117591887A