Search

CN-121998497-A - Emulsified asphalt quality control method and system

CN121998497ACN 121998497 ACN121998497 ACN 121998497ACN-121998497-A

Abstract

The application discloses a quality control method and a system for emulsified asphalt, which relate to the technical field of quality control of a road material production process and comprise the following steps of collecting production process parameters and raw material characteristic parameters on line to form a multi-dimensional real-time parameter set in the emulsified asphalt production process, inputting the multi-dimensional real-time parameter set into a soft measurement model to obtain a predicted value of off-line detection key quality index data, comparing and judging whether the predicted value of the off-line detection key quality index data is in an out-of-tolerance state, and generating a targeted deviation rectifying instruction based on the multi-dimensional real-time parameter set if the predicted value is in the out-of-tolerance state, so that the problem of control lag that the off-line detection cannot be timely rectified when the off-line detection of the key quality index of emulsified asphalt production is mostly failed in the prior art is effectively solved by the design of real-time deviation rectifying in the production process.

Inventors

  • NI XINMING
  • FU LI
  • YIN QING

Assignees

  • 新疆心路科技有限公司
  • 乌鲁木齐魁道路面材料科技有限公司

Dates

Publication Date
20260508
Application Date
20260122

Claims (10)

  1. 1. The emulsified asphalt quality control method is characterized by comprising the following steps: step 1, in the emulsified asphalt production process, acquiring production process parameters and raw material characteristic parameters on line to form a multi-dimensional real-time parameter set; Step 2, acquiring qualified batch data and unqualified batch data of emulsified asphalt historical production, and constructing a model training sample library; according to the model training sample library, training by adopting an integrated learning algorithm to obtain a soft measurement model, wherein the soft measurement model is used for outputting a predicted value of off-line detection key quality index data; step 3, acquiring a qualified range of the off-line detection key quality index data, and judging that the predicted value of the off-line detection key quality index data is in an out-of-tolerance state if the predicted value of the off-line detection key quality index data exceeds the qualified range of the off-line detection key quality index data; And 4, if the predicted value of the off-line detection key quality index data is in an out-of-tolerance state, generating a targeted deviation correcting instruction based on the multi-dimensional real-time parameter set.
  2. 2. The emulsified asphalt quality control method as set forth in claim 1, further comprising the steps of: Preprocessing the raw material characteristic parameters, removing abnormal data in the raw material characteristic parameters by using a Laida criterion, and supplementing missing data in the raw material characteristic parameters by using a median filling method and a linear interpolation method.
  3. 3. The emulsified asphalt quality control method as set forth in claim 1, wherein in step 2, qualified batch data and unqualified batch data of the emulsified asphalt for historical production are obtained, and a model training sample library is constructed as follows: acquiring qualified batch data and unqualified batch data of the emulsified asphalt in historical production, extracting historical production process parameters, historical raw material characteristic parameters and corresponding off-line detection key quality index data from the historical production process parameters and the historical raw material characteristic parameters, and constructing a model training sample library; the off-line detection key quality index data comprise particle size distribution, residual quantity, screen residue, viscosity and sedimentation stability.
  4. 4. The emulsified asphalt quality control method as set forth in claim 3, wherein in step 2, according to a model training sample library, a soft measurement model is obtained by training with an ensemble learning algorithm, and the soft measurement model is used for outputting predicted values of offline detection key quality index data, as follows: Carrying out normalization processing on data in a model training sample library, screening out core parameters related to off-line detection key quality index data through feature importance analysis, and constructing a core parameter set; based on the core parameter set and the corresponding off-line detection key quality index data, an integrated learning algorithm is adopted for training to obtain a soft measurement model.
  5. 5. The emulsified asphalt quality control method as set forth in claim 4, wherein in step 2, data in the model training sample library is normalized, core parameters associated with the off-line detection key quality index data are screened out through feature importance analysis, and a core parameter set is constructed as follows: Normalizing the historical production process parameters and the historical raw material characteristic parameters in the model training sample library; And carrying out feature importance analysis by adopting a random forest algorithm: Acquiring importance scores of the input features of each random forest algorithm on the output labels through the random forest model, quantifying the information gain of the input features in the random forest algorithm decision tree splitting process, screening out the input features with the importance scores higher than the feature importance threshold as core parameters, and acquiring all the core parameters to form a core parameter set; And obtaining importance scores of input features of each random forest algorithm on the output labels through the random forest model to form a correlation mapping table of the historical production process parameters, the historical raw material characteristic parameters and the corresponding offline detection key quality index data.
  6. 6. The emulsified asphalt quality control method according to claim 5, wherein in the step 2, the feature importance threshold is obtained by obtaining importance scores of input features of all random forest algorithms on output tags to form an importance score set, obtaining an average value of importance scores in the importance score set, and taking the average value of the importance scores in the importance score set as the feature importance threshold.
  7. 7. The emulsified asphalt quality control method according to claim 5, wherein in step 2, based on the core parameter set and the corresponding offline detection key quality index data, a soft measurement model is obtained by training with an ensemble learning algorithm, as follows: Taking the core parameter set as an input characteristic of the integrated learning algorithm, and taking off-line detection key quality index data corresponding to the core parameter set one by one as an output label of the integrated learning algorithm; based on the input features of the integrated learning algorithm and the output label of the integrated learning algorithm, performing model training by adopting the integrated learning algorithm, and learning the mapping relation between the core parameters and the key quality indexes by adopting the integrated learning algorithm to obtain an initial model; Performing parameter optimization on the initial model by adopting a cross validation method, removing abnormal parameters related to over-fitting and under-fitting of the model to obtain an optimal model parameter combination; the input of the soft measurement model is a core parameter, and the output is a predicted value of off-line detection key quality index data corresponding to the core parameter.
  8. 8. The emulsified asphalt quality control method as set forth in claim 7, wherein in step 4, if it is determined that the predicted value of the off-line detection key quality index data is in the out-of-tolerance state, a targeted deviation correcting instruction is generated based on the multi-dimensional real-time parameter set as follows: when the predicted value of the off-line detection key quality index data is judged to be out of tolerance, extracting target parameters related to the off-line detection key quality index data in the multi-dimensional real-time parameter set; according to the historical production process parameters, the historical raw material characteristic parameters and the corresponding offline detection key quality index data, an association rule mining algorithm is adopted to extract association relations among the historical production process parameters, the historical raw material characteristic parameters and the offline detection key quality index data, so as to form parameter-quality association rules, all parameter-quality association rules are obtained, and a parameter-quality association rule base is formed; Based on the parameter-quality association rule library, generating a quantitative association relation between the deviation degree of the target parameter and the out-of-tolerance off-line detection key quality index, wherein the quantitative association relation is as follows: Firstly, calling parameter-quality association rules corresponding to current out-of-tolerance off-line detection key quality index data and target parameters from a parameter-quality association rule library; Step two, according to qualified batch data of the emulsified asphalt historical production, acquiring a qualified batch target parameter mean value and a qualified batch target parameter standard deviation corresponding to the target parameter; Obtaining a current target parameter value of a target parameter, and obtaining a deviation degree quantized value through (the current target parameter value-a qualified batch target parameter mean value)/a qualified batch target parameter standard deviation; thirdly, taking the predicted value of the off-line detection key quality index data as the out-of-tolerance amplitude of the off-line detection key quality index data if the predicted value exceeds the out-of-tolerance value of the off-line detection key quality index data; Based on the called parameter-quality association rule, establishing a mapping relation between the deviation degree quantized value and the out-of-tolerance amplitude of the off-line detection key quality index data; And determining the adjustment direction and the adjustment amplitude of the target parameter according to the mapping relation between the deviation degree quantized value and the out-of-tolerance amplitude of the off-line detection key quality index data, and generating a preliminary deviation correcting instruction.
  9. 9. The method according to claim 8, wherein in step 4, when it is determined that the predicted value of the key quality index exceeds the tolerance, the target parameters associated with the exceeding index in the multi-dimensional real-time parameter set are extracted as follows: When the predicted value of a certain off-line detection key quality index data is out of tolerance, a correlation mapping table of the historical production process parameters, the historical raw material characteristic parameters and the corresponding off-line detection key quality index data is called, the historical production process parameters and/or the historical raw material characteristic parameters with the importance scores higher than the characteristic importance threshold corresponding to the off-line detection key quality index data are screened out, and then the corresponding production process parameters and/or the raw material characteristic parameters are extracted from the multi-dimensional real-time parameter set to serve as target parameters correlated with the off-line detection key quality index data.
  10. 10. A emulsified asphalt quality control system for use in a method of controlling the quality of emulsified asphalt according to any one of claims 1 to 9, comprising: The data acquisition module is used for acquiring production process parameters and raw material characteristic parameters on line to form a multi-dimensional real-time parameter set in the emulsified asphalt production process; the key quality index data prediction module is used for obtaining qualified batch data and unqualified batch data of the emulsified asphalt in historical production and constructing a model training sample library, obtaining a soft measurement model by training an integrated learning algorithm according to the model training sample library, and outputting a predicted value of the key quality index data in an offline detection mode; The out-of-tolerance state judging module is used for acquiring the qualified range of the off-line detection key quality index data, and judging that the predicted value of the off-line detection key quality index data is in the out-of-tolerance state if the predicted value of the off-line detection key quality index data exceeds the qualified range of the off-line detection key quality index data; And the deviation rectifying instruction generation module is used for generating a targeted deviation rectifying instruction based on the multi-dimensional real-time parameter set if the predicted value of the off-line detection key quality index data is in an out-of-tolerance state.

Description

Emulsified asphalt quality control method and system Technical Field The invention relates to the technical field of quality control in the production process of road materials, in particular to a method and a system for controlling the quality of emulsified asphalt. Background The emulsified asphalt is used as a key material in the road construction and maintenance fields, and the quality of the emulsified asphalt directly influences the service life and the driving safety of road engineering, so the quality control in the production process is very important. In the existing emulsified asphalt production technology, quality control is mainly realized by an off-line sampling inspection mode, namely, after production is finished, finished products are sampled, key quality indexes such as particle size distribution, residual quantity, screen residue, viscosity, sedimentation stability and the like are detected, and whether batches are qualified or not is judged according to detection results. However, the control mode has the following problems that on one hand, the detection result of the key quality index has obvious hysteresis, when the detection finds that the index is unqualified, the emulsified asphalt of the corresponding batch is finished in production, deviation correction can not be timely carried out on the production process, the emulsified asphalt can only be processed in modes of reworking, scrapping and the like, so that a large amount of waste of raw materials and energy sources is caused, the production period is prolonged, the downtime is increased, on the other hand, the off-line selective inspection is difficult to cover the whole production process, the influence of factors such as raw material fluctuation, equipment running state change and the like on the quality index in the production process can not be captured in real time, the occurrence rate of out-of-tolerance batches is higher, and the production cost is further increased. Therefore, how to realize the real-time prediction and judgment of key quality indexes in the emulsified asphalt production process, avoid the hysteresis of offline sampling inspection, correct deviation in time to reduce the occurrence rate of out-of-tolerance batches, and become the technical problem to be solved urgently in the current emulsified asphalt quality control field. Disclosure of Invention In order to solve the technical problems in the background technology, the invention provides a method and a system for controlling the quality of emulsified asphalt. The invention provides a quality control method of emulsified asphalt, which comprises the following steps: step 1, in the emulsified asphalt production process, acquiring production process parameters and raw material characteristic parameters on line to form a multi-dimensional real-time parameter set; Step 2, acquiring qualified batch data and unqualified batch data of emulsified asphalt historical production, and constructing a model training sample library; according to the model training sample library, training by adopting an integrated learning algorithm to obtain a soft measurement model, wherein the soft measurement model is used for outputting a predicted value of off-line detection key quality index data; step 3, acquiring a qualified range of the off-line detection key quality index data, and judging that the predicted value of the off-line detection key quality index data is in an out-of-tolerance state if the predicted value of the off-line detection key quality index data exceeds the qualified range of the off-line detection key quality index data; And 4, if the predicted value of the off-line detection key quality index data is in an out-of-tolerance state, generating a targeted deviation correcting instruction based on the multi-dimensional real-time parameter set. Preferably, in step 1, the method further comprises the following steps: Preprocessing the raw material characteristic parameters, removing abnormal data in the raw material characteristic parameters by using a Laida criterion, and supplementing missing data in the raw material characteristic parameters by using a median filling method and a linear interpolation method. Preferably, in step 2, qualified batch data and unqualified batch data of emulsified asphalt in historical production are obtained, and a model training sample library is constructed as follows: acquiring qualified batch data and unqualified batch data of the emulsified asphalt in historical production, extracting historical production process parameters, historical raw material characteristic parameters and corresponding off-line detection key quality index data from the historical production process parameters and the historical raw material characteristic parameters, and constructing a model training sample library; the off-line detection key quality index data comprise particle size distribution, residual quantity, screen residue, viscosity and sed