CN-122018333-A - Automatic control system for coastal wind power environment-friendly lightweight concrete preparation

CN122018333ACN 122018333 ACN122018333 ACN 122018333ACN-122018333-A

Abstract

The invention relates to the field of industrial automation control and building material preparation, in particular to an automatic control system for coastal wind power environment-friendly lightweight concrete preparation, which comprises a multi-source heterogeneous data acquisition unit, a state evaluation and risk quantification unit, a dynamic risk pair impulse strategy unit and a system confidence coefficient restoration and closed loop unit, wherein the system combines sensor data flow and coastal environment disturbance characteristics, calculates state evaluation entropy by using a neural network, predicts a safe time window from a pouring cold joint failure boundary, and is characterized in that the water adding amount and stirring rate are regulated by dynamically switching a nonlinear optimal or suboptimal robust control law based on the comparison of the window and a dangerous threshold value, and the closed loop restoration of the system is realized by recalculating the evaluation entropy.

Inventors

LI RUIZHI
ZHANG XUEYAN
WANG HAO
Feng Zhaoteng

Assignees

山东世纪鑫源建筑科技有限公司

Dates

Publication Date: 20260512
Application Date: 20260410

Claims (10)

1. Automatic control system of coastal wind-powered electricity generation environmental protection lightweight concrete preparation is applied to concrete preparation platform that contains sensor group and on-the-spot actuating mechanism, its characterized in that includes: the multi-source heterogeneous data acquisition unit is used for acquiring multi-source sensor data streams containing the water content of the light solid waste raw materials, the running state of the stirrer and the temperature and humidity of the coastal environment through the sensor group; the state evaluation and risk quantification unit is used for constructing a control state evaluation model based on a neural network based on the multisource sensor data flow and the coastal environment disturbance characteristics, calculating a state evaluation entropy according to the control state evaluation model according to time sequence, and predicting a safety time window of a current control system from a preset pouring cold joint failure boundary based on the state evaluation entropy; The dynamic risk pair flushing decision unit is used for comparing the safety time window with a preset dangerous threshold, activating a nonlinear optimal control law to generate a first control instruction for adjusting the water adding amount or stirring rate if the safety time window is larger than or equal to the dangerous threshold, triggering a control degradation mode if the safety time window is smaller than the dangerous threshold, and activating a preset suboptimal robust control law to generate a second control instruction for adjusting the water adding amount or stirring rate.
2. The automatic control system for coastal wind power environment-friendly lightweight concrete preparation according to claim 1, further comprising a system confidence repair and closed loop unit, wherein the system confidence repair and closed loop unit is used for sending the first control command or the second control command to the on-site execution mechanism, collecting feedback states of the on-site execution mechanism and recalculating the state evaluation entropy, restoring the nonlinear optimal control law if the recalculated state evaluation entropy converges to a preset safety interval, and maintaining the suboptimal robust control law and updating the control state evaluation model if the recalculated state evaluation entropy does not converge to the safety interval.
3. The automated control system for coastal wind power environment-friendly lightweight concrete production according to claim 1, wherein the means for extracting coastal environmental disturbance features in the multi-source sensor data stream comprises: identifying temperature and humidity abrupt change signals in the multi-source sensor data stream; Acquiring heterogeneous sensor space-time dislocation data in the multi-source sensor data stream; fusing the temperature and humidity mutation signal with the space-time dislocation data of the heterogeneous sensor to generate the coastal environment disturbance characteristic; The coastal environment disturbance characteristic is physical attribute jump characteristic caused by coastal sea fog microclimate reversal.
4. An automated coastal wind power environment-friendly lightweight concrete production control system according to claim 3, wherein the means for calculating a state evaluation entropy by the control state evaluation model comprises: mapping the multi-source sensor data stream to a unified dimensionless feature space to generate a feature vector, acquiring a preset reference consistency vector, and calculating a feature vector value deviation rate in the dimensionless feature space based on a difference value between the feature vector and the reference consistency vector; When the deviation rate of the feature vector values is smaller than a preset deviation threshold value, judging that the data are consistent, converting each feature value in the feature vector into specific gravity distribution, calculating the information entropy of the feature vector in the dimensionless feature space as a basic entropy value, and setting a data consistency abnormal factor to be zero; When the characteristic vector value deviation rate is greater than or equal to the deviation threshold, judging that data are in contradiction, calculating information entropy of the characteristic vector in the dimensionless characteristic space as a basic entropy value, and extracting the ratio of the characteristic vector value deviation rate to the deviation threshold as a data consistency anomaly factor; And carrying out linear weighted summation on the basic entropy value, the data consistency anomaly factor and a quantized value obtained by quantizing the coastal environment disturbance characteristic to obtain the state evaluation entropy, wherein the state evaluation entropy is used for representing the degree of anomaly of the instruction confidence coefficient of the control system.
5. The automatic control system for coastal wind power environment-friendly lightweight concrete production according to claim 4, wherein the means for predicting a safety time window of a current control system from a preset pouring cold joint failure boundary based on the state evaluation entropy comprises: acquiring a system operation history log; Extracting a critical entropy value time sequence corresponding to the pouring cold joint failure boundary based on the system operation history log; calculating the difference value between the state evaluation entropy at the current moment and the state evaluation entropy at the last moment recorded in the system operation history log, dividing the difference value by the sampling time interval corresponding to the multi-source sensor data stream to obtain an entropy value deterioration rate; When the entropy value deterioration rate is larger than a preset tiny fluctuation threshold value, dividing the difference value between the corresponding critical entropy value in the critical entropy value time sequence and the state evaluation entropy at the current moment by the entropy value deterioration rate, and calculating to obtain the remaining time approaching the pouring cold joint failure boundary as the safety time window.
6. The automatic control system for coastal wind power environment protection lightweight concrete production according to claim 5, wherein the means for activating the nonlinear optimal control law to generate the first control command for adjusting the water addition amount or the stirring rate comprises: loading a preset multi-agent deep reinforcement learning model; Inputting the multi-source sensor data stream into the multi-agent deep reinforcement learning model; Based on the multi-agent deep reinforcement learning model, generating a nonlinear optimal solution by taking minimized water content fluctuation compensation delay time and minimized energy consumption of the on-site execution mechanism as combined optimization targets, wherein the water content fluctuation compensation delay time is from the time when the multi-source sensor data stream monitors that the water content of the light solid waste raw material is suddenly changed to the time when the on-site execution mechanism outputs and adjusts, so that data items reflecting the water content of the light solid waste raw material in the multi-source sensor data stream are returned to a steady state threshold zone calibrated in advance based on historical steady operation data; and converting the nonlinear optimal solution into the first control instruction.
7. The automatic control system for coastal wind power environment-friendly lightweight concrete production according to claim 6, wherein the means for triggering the control degradation mode and activating the preset suboptimal robust control law to generate the second control command for adjusting the water addition amount or the stirring rate comprises: freezing the adaptive weights of the multi-agent deep reinforcement learning model; extracting a conservative dead zone control parameter corresponding to the control degradation mode from a preset system configuration library; Based on the conservative dead zone control parameters, discarding the joint optimization objective, and constructing a suboptimal robust control law for minimizing the state evaluation entropy as a single objective; and executing the suboptimal robust control law to generate the second control instruction, wherein the second control instruction reduces the response frequency to the fluctuation of the multi-source sensor data stream.
8. The automated coastal wind power environment-friendly lightweight concrete production control system of claim 7, wherein the means for collecting feedback status of the on-site actuators and recalculating the status assessment entropy comprises: monitoring whether a manual override signal is generated in the field execution mechanism; When the manual override signal is not monitored, marking the manual intervention frequency as zero, and directly collecting mechanical execution feedback data generated by the on-site execution mechanism as the feedback state; When the manual override signal is monitored, recording manual intervention frequency, extracting a control quantity contained in the manual override signal, and carrying out weighted splicing on the control quantity and mechanical execution feedback data generated by the on-site execution mechanism to generate the feedback state; and inputting the feedback state into the control state evaluation model, and recalculating to obtain updated state evaluation entropy.
9. The automatic control system for preparing coastal wind power environment-friendly lightweight concrete according to claim 8, wherein if the recalculated state evaluation entropy converges to a preset safe interval, the mode of restoring the nonlinear optimal control law comprises: comparing the updated state evaluation entropy with the upper limit value and the lower limit value of the preset safety interval; if the updated state evaluation entropy is larger than the lower limit value of the preset safety interval and smaller than the upper limit value of the preset safety interval, judging that the system confidence coefficient is recovered; Thawing the adaptive weights of the multi-agent deep reinforcement learning model; and reactivating the joint optimization target, and recovering and outputting the first control instruction.
10. The automatic control system for coastal wind power environment-friendly lightweight concrete production according to claim 9, wherein if the recalculated state evaluation entropy does not converge to the safe interval, the manner of maintaining the suboptimal robust control law and updating the control state evaluation model comprises: if the updated state evaluation entropy is greater than or equal to the upper limit value of the preset safety interval or less than or equal to the lower limit value of the preset safety interval, the abnormal state of the system confidence coefficient is judged to be continuous; Maintaining to output the second control instruction; Taking the product of the manual intervention frequency and the data consistency abnormal factor as a punishment item to construct a punishment function; And combining a loss function with a prediction error between the state evaluation entropy output by the control state evaluation model and an actual risk label generated by dynamic conversion of a posterior calibration rule, and correcting the evaluation weight of the control state evaluation model by using back propagation of the penalty function to finish updating of the control state evaluation model.

Description

Automatic control system for coastal wind power environment-friendly lightweight concrete preparation Technical Field The invention relates to the field of industrial automation control and building material preparation, in particular to an automatic control system for coastal wind power environment-friendly lightweight concrete preparation. Background Because the site is in a high-salt-fog and high-humidity environment and the pouring process is not interruptible, the traditional control system relies on single parameter static optimization, and lacks fusion evaluation of multi-source heterogeneous sensor data under microclimate mutation, the control instability is easy to cause and pouring cold joints are caused; In the prior art, although a part of systems try conventional closed-loop adjustment, the problems of inaccurate control state confidence evaluation and insufficient intelligence for coping with data conflict generally exist, meanwhile, the existing logic ignores the space-time dislocation of a sensor caused by environmental mutation, so that the control is easy to fail under complex disturbance, and in addition, when the failure critical or manual intervention is faced, a dynamic degradation and closed-loop repair mechanism is lacking, so that the system is difficult to adapt to the robust operation requirement of continuous pouring under the high disturbance condition. Therefore, how to provide an automatic control system for coastal wind power environment-friendly lightweight concrete preparation is a problem to be solved by those skilled in the art. Disclosure of Invention In order to solve the technical problems, the invention provides an automatic control system for preparing coastal wind power environment-friendly lightweight concrete, which is applied to a concrete preparation platform comprising a sensor group and an on-site execution mechanism, and specifically, the technical scheme of the invention comprises the following steps: the multi-source heterogeneous data acquisition unit is used for acquiring multi-source sensor data streams containing the water content of the light solid waste raw materials, the running state of the stirrer and the temperature and humidity of the coastal environment through the sensor group; the state evaluation and risk quantification unit is used for constructing a control state evaluation model based on a neural network based on the multisource sensor data flow and the coastal environment disturbance characteristics, calculating a state evaluation entropy according to the control state evaluation model according to time sequence, and predicting a safety time window of a current control system from a preset pouring cold joint failure boundary based on the state evaluation entropy; The dynamic risk pairing strategy unit is used for comparing the safety time window with a preset dangerous threshold value, activating a nonlinear optimal control law to generate a first control instruction for adjusting the water adding amount or stirring rate if the safety time window is larger than or equal to the dangerous threshold value, triggering a control degradation mode if the safety time window is smaller than the dangerous threshold value, and activating a preset suboptimal robust control law to generate a second control instruction for adjusting the water adding amount or stirring rate; The system confidence repairing and closed loop unit is used for sending the first control instruction or the second control instruction to the on-site execution mechanism, collecting the feedback state of the on-site execution mechanism, recalculating the state evaluation entropy, restoring the nonlinear optimal control law if the recalculated state evaluation entropy is converged to a preset safety interval, and maintaining the suboptimal robust control law and updating the control state evaluation model if the recalculated state evaluation entropy is not converged to the safety interval. Preferably, the method for extracting the coastal environment disturbance feature in the multi-source sensor data stream includes: identifying temperature and humidity abrupt change signals in the multi-source sensor data stream; Acquiring heterogeneous sensor space-time dislocation data in the multi-source sensor data stream; and fusing the temperature and humidity mutation signal with the space-time dislocation data of the heterogeneous sensor to generate the coastal environment disturbance characteristic. Preferably, the method for calculating the state evaluation entropy through the control state evaluation model includes: Mapping the multi-source sensor data stream to a unified dimensionless feature space, generating a feature vector, acquiring a preset reference consistency vector, and calculating a feature vector value deviation rate in the dimensionless feature space based on a difference value between the feature vector and the reference consistency vector; When the deviation rate of the