Intelligent production management system and method for high-concentration polyacrylamide fracturing fluid

Abstract

The invention relates to an intelligent production management system and method for high-concentration polyacrylamide fracturing fluid, and the method comprises the steps: obtaining multi-source heterogeneous data through designing a unique algorithm model and an innovative data processing flow: an innovative data collection and fusion strategy, and designing a comprehensive and high-precision data collection system; based on intelligent control algorithm innovation of a fuzzy neural network and MPC, the fuzzy neural network and model prediction control are combined to carry out production process control, a robust optimization method is introduced to enhance the robustness of a control strategy, and meanwhile, a rule-based emergency processing mechanism is established; intelligent mechanism innovation of quality monitoring and optimization is realized, and real-time monitoring and analysis of product quality are realized by establishing a quality prediction model based on combination of support vector machine regression (SVR) and a particle swarm optimization (PSO) algorithm.

Description

Technical Field

[0001] The present invention relates to the technical field of highly concentrated polyacrylamide fracturing fluid production, and in particular to an intelligent production management system and method for highly concentrated polyacrylamide fracturing fluid, which aims to improve the efficiency, quality and stability of highly concentrated polyacrylamide fracturing fluid production through innovative algorithm models and intelligent management and control means, effectively cope with complex and changeable production conditions and strict quality requirements, and promote the efficient development of oil and gas field production increase operations. Background Art

[0002] Highly concentrated polyacrylamide fracturing fluid plays a critical role in oil and gas field exploitation. It can effectively increase the production and recovery rate of oil and gas wells. However, its production process faces many challenges. Traditional production management of highly concentrated polyacrylamide fracturing fluid mainly relies on manual operation and experience judgment, which has many disadvantages.

[0003] In the raw material configuration link, it is difficult to accurately control the ratio accuracy of various raw materials (such as polyacrylamide dry powder, cross-linking agent, additives, etc.), which can easily lead to unstable performance of fracturing fluid due to human errors, affecting the subsequent oil and gas production effect. During the reaction process, the regulation of key process parameters such as temperature, pressure, and stirring speed mainly relies on the experience of operators and simple instrument monitoring, and cannot achieve accurate dynamic optimization, making it difficult for the reaction process to always be in the best state, and the product quality fluctuates greatly. For example, indicators such as viscosity and cross-linking degree are difficult to stabilize within the ideal range. Moreover, the traditional production management method does not monitor the operating status of production equipment comprehensively and timely, and often repairs are carried out only after the equipment fails, resulting in production interruptions and increasing production costs and time costs. In addition, there is a lack of deep mining and analysis capabilities for large amounts of production data, and it is impossible to summarize effective production optimization strategies from historical data, making it difficult to adapt to changing market demands and technological development requirements.

[0004] With the development of intelligent technology, although some companies have tried to introduce some automation and information technology into production management, there are still many problems. The existing intelligent production management system has deficiencies in dealing with the strong nonlinearity, multivariable coupling and high real-time requirements in the production process of highly concentrated polyacrylamide fracturing fluid, resulting in limited accuracy and reliability of the control model. Most studies only focus on the intelligent improvement of a single production link, and fail to optimize and integrate from a systematic perspective of the entire production process, making it difficult to achieve a significant improvement in overall production efficiency and quality. At the same time, when dealing with uncertain factors such as fluctuations in raw material quality and changes in the production environment, the current production management method lacks sufficient robustness and adaptability, and cannot adjust the production strategy in a timely and effective manner to ensure the continuity of production and the stability of product quality. Therefore, there is an urgent need for an innovative intelligent production management system and method for highly concentrated polyacrylamide fracturing fluid to break through the bottleneck of existing technologies and meet the urgent needs of the oil and gas field mining industry for efficient and high-quality fracturing fluid production. Summary of the invention

[0005] The present invention provides an intelligent production management system and method for highly concentrated polyacrylamide fracturing fluid, which effectively solves the problems in the prior art by designing a unique algorithm model and an innovative data processing flow.

[0006] An intelligent production management system for highly concentrated polyacrylamide fracturing fluid, comprising:

[0007] The data acquisition and integration module is used to deploy high-precision sensor arrays at various key locations on the high-concentration polyacrylamide fracturing fluid production line to comprehensively collect data during the production process, and transmit the collected data to the data processing center in real time via high-speed industrial Ethernet. The data processing center uses ontology-based data fusion technology to integrate multi-source heterogeneous data. The data acquisition and integration module includes:

[0008] The high-precision mass flowmeter, material level sensor and humidity sensor installed in the raw material storage and transportation system are used to accurately measure the flow rate of polyacrylamide dry powder, cross-linking agent and additive raw materials, the material level height in the storage tank and the humidity of the raw materials; the high-sensitivity temperature sensor, pressure sensor, viscosity sensor and torque sensor equipped in the reactor are used to monitor the temperature, pressure, viscosity changes of the reaction system and the torque of the agitator in real time during the reaction process; the density sensor, conductivity sensor and flow sensor installed in the finished product storage and canning process are used to detect the quality indicators such as density and conductivity of the finished fracturing fluid as well as the canning flow.

[0009] Furthermore, in the data collection and integration module, the data collection frequency is dynamically adjusted according to the characteristics of the production process. In the intense reaction stage, such as the initial stage of polymerization reaction, the collection frequency of temperature, pressure and other data is increased to 2 times per second. In the relatively stable reaction stage, the collection frequency is reduced to 1 time per second. The raw material data is intensively collected according to the batch supply situation when each batch is supplied, and the collection interval is 1 minute.

[0010] Furthermore, it also includes an intelligent production control module, which uses an algorithm combining fuzzy neural network and model predictive control to control the production process based on the integrated data, including: using fuzzy neural network to model and approximate the nonlinear relationship in the production process of highly concentrated polyacrylamide fracturing fluid, and automatically constructing a complex nonlinear mapping relationship between input and output through learning a large amount of historical production data, wherein the input includes raw material ratio, reaction temperature, and pressure, and the output includes fracturing fluid viscosity and cross-linking quality indicators; embedding the fuzzy neural network model into the model predictive control (MPC) framework, taking the quality indicators of fracturing fluid as the control target, comprehensively considering various constraints in the production process, and solving the future control sequence through rolling optimization according to the system output in the future time predicted by the fuzzy neural network; introducing the interval description of uncertain parameters in the MPC model, and obtaining a control strategy that can still ensure product quality and production stability within a certain uncertainty range by solving the robust optimization problem.

[0011] Furthermore, the intelligent production control module also includes a rule-based emergency response mechanism, which is used to quickly take corresponding emergency measures according to preset rules when sudden abnormal situations such as equipment failure and raw material supply interruption are detected, so as to ensure the safety and continuity of the production process.

[0012] On the other hand, the present invention also provides an intelligent production management system for highly concentrated polyacrylamide fracturing fluid, which also includes a quality monitoring and optimization module. The quality monitoring and optimization module uses online quality detection equipment and data analysis technology to monitor the quality of highly concentrated polyacrylamide fracturing fluid in real time, and uses a quality prediction model based on a combination of support vector machine regression (SVR) and particle swarm optimization algorithm (PSO) to predict and analyze the quality of the fracturing fluid, including: the online quality detection equipment samples and detects the finished fracturing fluid at regular intervals to obtain key quality indicator data, and the online quality detection equipment includes a high-precision rheometer, a laser particle size analyzer, and a chemical analyzer; the process parameters in the production process and the quality indicators of the raw materials are used as input variables, and the key quality indicators of the fracturing fluid are used as output variables, and a mapping relationship between the input and the output is established through SVR, and the PSO algorithm is used to optimize the model parameters of the SVR to improve the prediction accuracy of the model.

[0013] The quality monitoring and optimization module optimizes and adjusts the production process according to the quality monitoring and prediction results. When it is found that the quality indicators deviate from the target range, the corresponding process parameters are adjusted through the intelligent production control module. At the same time, the quality data feedback is used to continuously optimize the quality prediction model and production control strategy.

[0014] On the other hand, the present invention also provides a highly concentrated polyacrylamide fracturing fluid intelligent production management method, including data collection and preprocessing steps, wherein the data collection and preprocessing steps comprehensively collect multi-source heterogeneous data from the raw material supply, reaction process, and finished product processing links of the production of highly concentrated polyacrylamide fracturing fluid, and adopt ontology-based data fusion technology for preprocessing, including: in the raw material supply link, when each batch of raw materials is supplied, the mass flow meter, the material level sensor, and the humidity sensor collect data at intervals of 1 minute; in the reaction process, the temperature, pressure and other data are collected twice per second at the initial stage of the reaction, and once per second after the reaction is stable; in the finished product processing link, sampling and testing are performed every 10 minutes using online quality detection equipment, and relevant data are collected; a highly concentrated polyacrylamide fracturing fluid production ontology model is constructed, various entities and their attributes and relationships are defined, and data collected by different sensors are semantically mapped and converted according to the ontology model; in the data cleaning process, an outlier detection method based on statistical analysis and data trend judgment is adopted to ensure the accuracy and reliability of the data.

[0015] On the other hand, the present invention also provides a method for intelligent production management of highly concentrated polyacrylamide fracturing fluid, including intelligent production control modeling and solving steps, wherein the intelligent production control modeling and solving steps use fuzzy neural networks to model the nonlinear relationship of the production process of highly concentrated polyacrylamide fracturing fluid, and embed the fuzzy neural network model into a model predictive control (MPC) framework, and also include establishing a rule-based emergency response mechanism, specifically including: the fuzzy neural network adopts a five-layer structure, trains the fuzzy neural network through a large amount of historical production data, optimizes the network weights and fuzzy rule parameters, so that it can accurately approximate the nonlinear relationship in the production process; according to the production process requirements and equipment performance, It can determine the control objectives and constraints of MPC. In each control cycle, MPC obtains the optimal control sequence in the future by solving the quadratic programming problem based on the current production status and the system output in the future predicted by the fuzzy neural network. The uncertain parameters of the raw material quality fluctuation range and the equipment performance parameter variation interval are expressed as interval numbers. By solving the robust optimization problem, a control strategy that can still ensure product quality and production stability within the uncertain parameter fluctuation range is obtained. Emergency handling rules corresponding to various abnormal situations are pre-set. When an abnormal situation is detected, the system quickly takes corresponding emergency measures according to the preset rules to ensure the safety and continuity of the production process.

[0016] On the other hand, the present invention also provides a highly concentrated polyacrylamide fracturing fluid intelligent production management method, including quality monitoring and optimization steps, wherein the quality monitoring and optimization step uses an online quality detection device to sample and detect the highly concentrated polyacrylamide fracturing fluid every 10 minutes, obtains key quality indicator data such as viscosity, particle size distribution, and chemical composition content of the fracturing fluid, and performs comprehensive analysis in combination with production process data, and also uses a quality prediction model based on support vector machine regression (SVR) combined with a particle swarm optimization algorithm (PSO) to perform quality prediction, and optimizes and adjusts the production process according to the quality monitoring and prediction results, specifically including:

[0017] The process parameters in the production process and the quality indicators of raw materials are used as input variables, and the key quality indicators of fracturing fluid are used as output variables to build the SVR model. The PSO algorithm is used to optimize the model parameters of the SVR and improve the prediction accuracy of the model.

[0018] When it is found that the quality indicators deviate from the target range, the corresponding process parameters are adjusted through the intelligent production control module. In the feedback process, the fuzzy neural network and SVR model are retrained according to the adjusted actual production data and quality inspection data, and the model parameters and control rules are updated to improve the adaptability and accuracy of the production management system.

[0019] Beneficial technical effects of the invention:

[0020] Innovative data collection and fusion strategies, acquire multi-source heterogeneous data by designing a comprehensive and high-precision data collection system, and adopt ontology-based data fusion technology to solve the semantic heterogeneity problem of multi-source heterogeneous data; innovation of intelligent control algorithms based on fuzzy neural networks and MPC, combine fuzzy neural networks with model predictive control for production process control, introduce robust optimization methods to enhance the robustness of control strategies, and establish rule-based emergency response mechanisms; innovation of intelligent mechanisms for quality monitoring and optimization, achieve real-time monitoring and analysis of product quality by establishing a quality prediction model based on support vector machine regression (SVR) and particle swarm optimization algorithm (PSO), and use quality data feedback to continuously optimize quality prediction models and production control strategies. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 , schematic diagram of the working process of the invention. DETAILED DESCRIPTION

[0022] Example 1

[0023] 1. System Architecture

[0024] 1. Data collection and integration module:

[0025] 1. Deploy high-precision sensor arrays at key locations on the production line of highly concentrated polyacrylamide fracturing fluid to comprehensively collect all types of data during the production process. Install high-precision mass flow meters (accuracy of ±0.05kg / h), material level sensors (accuracy of ±0.02m) and humidity sensors (accuracy of ±

[0026] 0.05%), which are used to accurately measure the flow of raw materials such as polyacrylamide dry powder, cross-linking agent, additives, the material level height in the storage tank and the humidity of the raw materials to ensure the accuracy and stability of the raw material supply; in the reactor, it is equipped with a high-sensitivity temperature sensor (accuracy ±0.05℃), a pressure sensor (accuracy ±0.005MPa), a viscosity sensor (accuracy ±0.5mPa·s) and a torque sensor (accuracy ±0.1N·m) to monitor the temperature, pressure, viscosity changes of the reaction system and the torque of the agitator in real time during the reaction process, so as to timely grasp the reaction progress and status; in the finished product storage and canning process, a density sensor (accuracy ±0.002g / cm 3 ), conductivity sensor (accuracy ±0.1μS / cm) and flow sensor (accuracy ±0.03m 3 / h), which is used to detect the density, conductivity and other quality indicators of the finished fracturing fluid as well as the filling flow rate, to ensure that the quality of the finished product meets the standards and achieves accurate filling.

[0027] 2. The frequency of data collection is dynamically adjusted according to the characteristics of the production process. In the intense reaction stage, such as the initial stage of polymerization, the frequency of data collection such as temperature and pressure is increased to 2 times per second; in the relatively stable reaction stage,

[0028] The collection frequency can be reduced to once per second. Raw material data is collected intensively according to the batch supply situation, with a collection interval of 1 minute. The collected data is transmitted to the data processing center in real time via high-speed industrial Ethernet, where ontology-based data fusion technology is used to integrate multi-source heterogeneous data. An ontology model for the production of highly concentrated polyacrylamide fracturing fluid is constructed, and the data collected by different sensors are semantically associated and mapped according to the ontology model. For example, the temperature data in the reactor is associated with the corresponding reaction stage, raw material information, etc., to solve the problem of semantic heterogeneity of the data and provide a unified data basis for subsequent data analysis and decision-making.

[0029] 2. Intelligent production control module:

[0030] 1. Based on the integrated data, an innovative algorithm combining fuzzy neural network and model predictive control is used to control the production process. First, the fuzzy neural network is used to model and approximate the nonlinear relationship in the production process of highly concentrated polyacrylamide fracturing fluid. Fuzzy neural network combines the fuzziness of fuzzy logic with the self-learning ability of neural network. By learning a large amount of historical production data, it automatically constructs a complex nonlinear mapping relationship between input (such as raw material ratio, reaction temperature, pressure, etc.) and output (such as fracturing fluid viscosity, cross-linking degree and other quality indicators). For example, by learning the influence of different raw material ratios and reaction temperatures on the viscosity of fracturing fluid through fuzzy neural network, when the raw material ratio or temperature changes, the change trend of fracturing fluid viscosity can be predicted quickly and accurately.

[0031] 2. Then, the fuzzy neural network model is embedded in the model predictive control (MPC) framework to achieve dynamic optimization control of the production process. The MPC algorithm takes the quality indicators of the fracturing fluid (such as viscosity, cross-linking degree, etc.) as the control target, comprehensively considers various constraints in the production process (such as equipment capacity limitations, upper and lower limits of process parameters, etc.), and solves the future control sequence through rolling optimization based on the system output in the future period predicted by the fuzzy neural network. For example, in each control cycle, MPC calculates the optimal adjustment strategy of control variables such as reaction temperature and raw material addition amount in the next 3 minutes based on the current production status and the changes in the viscosity of the fracturing fluid in the next 10 minutes predicted by the fuzzy neural network, so as to ensure that the viscosity of the fracturing fluid is stable within the target range. At the same time, in order to improve the robustness of the control strategy, the interval description of uncertain parameters is introduced in the MPC model, such as the fluctuation range of raw material quality, the variation range of equipment performance parameters, etc. By solving the robust optimization problem, a control strategy that can still guarantee product quality and production stability within a certain uncertainty range is obtained.

[0032] 3. In addition, in order to deal with sudden abnormal situations in the production process, such as equipment failure, interruption of raw material supply, etc., a rule-based emergency response mechanism is established. When an abnormal situation is detected, the system quickly takes corresponding emergency measures according to the preset rules, such as switching to backup equipment, adjusting production process parameters to adapt to temporary substitution of raw materials, etc., to ensure the safety and continuity of the production process.

[0033] 3.Quality monitoring and optimization module:

[0034] 1. Use online quality inspection equipment and data analysis technology to monitor the quality of highly concentrated polyacrylamide fracturing fluid in real time. Online quality inspection equipment includes high-precision rheometer (which can accurately measure rheological parameters such as viscosity and elastic modulus of fracturing fluid with an accuracy of ±0.2mPa·s), laser particle size analyzer (used to detect the particle size distribution of polymer particles in fracturing fluid with an accuracy of ±1μm) and chemical analyzer (used to analyze the chemical composition content in fracturing fluid with an accuracy of ±0.01%), etc. The finished fracturing fluid is sampled and tested at regular intervals (such as 10 minutes) to obtain key quality indicator data.

[0035] 2. A quality prediction model based on support vector machine regression (SVR) combined with particle swarm optimization (PSO) is used to predict and analyze the quality of fracturing fluid. The process parameters in the production process (such as reaction temperature, pressure, raw material ratio, etc.) and the quality indicators of raw materials (such as the molecular weight of polyacrylamide powder, the activity of cross-linking agent, etc.) are used as input variables, and the key quality indicators of fracturing fluid (such as viscosity, cross-linking degree, stability, etc.) are used as output variables. The mapping relationship between input and output is established through SVR. The PSO algorithm is used to optimize the model parameters of SVR (such as kernel function parameters, penalty factors, etc.) to improve the prediction accuracy of the model. For example, using a large amount of historical production data and test data, the SVR model optimized by PSO can accurately predict the viscosity change trend of fracturing fluid based on the current production process parameters and raw material information, with a prediction error within ±5%.

[0036] 3. Optimize and adjust the production process based on the quality monitoring and prediction results. When it is found that the quality indicators deviate from the target range, the corresponding process parameters are adjusted through the intelligent production control module, such as optimizing the raw material ratio, adjusting the reaction temperature and time, etc., to improve the stability of product quality. At the same time, the quality data feedback is used to continuously optimize the quality prediction model and production control strategy. For example, if the viscosity of the fracturing fluid is lower than the target value in actual production, it may be due to insufficient addition of polyacrylamide dry powder or low reaction temperature. The system will automatically adjust the amount of dry powder added or increase the reaction temperature, and feed back the adjustment experience to the quality prediction model and production control strategy, update the model parameters and control rules, and improve the adaptability and accuracy of the production management system.

[0037] 2. Methodology

[0038] 1. Data collection and preprocessing steps:

[0039] 1. Comprehensively collect multi-source heterogeneous data from the raw material supply, reaction process, and finished product processing of highly concentrated polyacrylamide fracturing fluid production. In the raw material supply link, when each batch of raw materials is supplied, the mass flow meter, material level sensor, and humidity sensor collect data at intervals of 1 minute; in the reaction process, the temperature, pressure and other data are collected twice per second at the initial stage of the reaction, and once per second after the reaction is stable; in the finished product processing link, online quality inspection equipment is used for sampling and inspection every 10 minutes, and relevant data is collected.

[0040] 2. After the collected data is transmitted to the data processing center, it is preprocessed using ontology-based data fusion technology. Construct a highly concentrated polyacrylamide fracturing fluid production ontology model to define various entities (such as raw materials, equipment, process parameters, quality indicators, etc.) and their attributes and relationships (such as the usage relationship between raw materials and equipment, the influence relationship between process parameters and quality indicators, etc.). The data collected by different sensors are semantically mapped and converted according to the ontology model. For example, the data collected by the temperature sensor is associated with the reactor entity, and its corresponding reaction stage attributes are marked. In the data cleaning process, an outlier detection method based on statistical analysis and data trend judgment is adopted. For example, for temperature data, if it deviates from the range of 3 times the standard deviation of the historical data average, and the data trend does not match the temperature change trend during the normal reaction process, it is judged as an outlier and replaced with the average value of adjacent normal data points to ensure the accuracy and reliability of the data.

[0041] 2. Intelligent production control modeling and solution steps:

[0042] 1. Use fuzzy neural network to model the nonlinear relationship in the production process of highly concentrated polyacrylamide fracturing fluid. The fuzzy neural network adopts a five-layer structure, including input layer, fuzzification layer, fuzzy rule layer, defuzzification layer and output layer. The input layer receives production process data such as raw material ratio, reaction temperature, pressure, etc. The fuzzification layer fuzzifies the input data and divides it into different fuzzy subsets (such as low, medium, and high). The fuzzy rule layer constructs fuzzy rules based on expert experience and historical data. The defuzzification layer defuzzifies the output of the fuzzy rules and obtains accurate prediction values ​​to output to the output layer (such as fracturing fluid viscosity, cross-linking degree and other quality indicators). The fuzzy neural network is trained through a large amount of historical production data to optimize the network weights and fuzzy rule parameters so that it can accurately approximate the nonlinear relationship in the production process.

[0043] 2. Embed the fuzzy neural network model into the model predictive control (MPC) framework. First, the control objectives (such as the target value and allowable deviation range of the fracturing fluid viscosity) and constraints (such as the upper and lower limits of the reaction temperature, the limit on the amount of raw material added, etc.) of the MPC are determined according to the production process requirements and equipment performance. Then, in each control cycle, the MPC obtains the optimal control sequence for a period of time in the future (such as 3 minutes) by solving the quadratic programming problem based on the current production status and the system output predicted by the fuzzy neural network for a period of time in the future (such as 10 minutes), that is, the adjustment strategy of the control variables such as reaction temperature and raw material addition. In the solution process, in order to improve the robustness of the control strategy, the uncertain parameters such as the raw material quality fluctuation range and the equipment performance parameter variation range are expressed as interval numbers. For example, the molecular weight fluctuation range of the raw material polyacrylamide dry powder is expressed as [M-ΔM, M+ΔM]. By solving the robust optimization problem, a control strategy that can still ensure product quality and production stability within the fluctuation range of uncertain parameters is obtained.

[0044] 3. Establish a rule-based emergency response mechanism. Pre-set emergency response rules for various abnormal situations (such as equipment failure type, reasons for interruption of raw material supply, etc.). For example, when the stirring motor of the reactor fails, the rule stipulates that it should immediately switch to the spare stirring motor and adjust the control strategy of the reaction temperature and stirring speed to maintain the continuation of the reaction. When an abnormal situation is detected, the system will quickly take corresponding emergency measures according to the preset rules to ensure the safety and continuity of the production process.

[0045] 3. Quality monitoring and optimization steps:

[0046] 1. Use online quality inspection equipment to sample and inspect highly concentrated polyacrylamide fracturing fluid every 10 minutes to obtain key quality indicator data such as viscosity, particle size distribution, and chemical composition content of the fracturing fluid, and conduct a comprehensive analysis in combination with production process data.

[0047] 2. Use a quality prediction model based on support vector machine regression (SVR) combined with particle swarm optimization (PSO) to predict quality. The process parameters in the production process (such as reaction temperature, pressure, raw material ratio, etc.)

[0048] The quality indicators of raw materials (such as the molecular weight of polyacrylamide powder, the activity of the cross-linking agent, etc.) are used as input variables, and the key quality indicators of fracturing fluid (such as viscosity, cross-linking degree, stability, etc.) are used as output variables to construct the SVR model.

[0049] The PSO algorithm is used to optimize the model parameters of the SVR. By setting the size of the particle swarm (such as 30 particles), the maximum number of iterations (such as 100 times), the learning factor (such as c1 = 1.5, c2 = 1.8) and other parameters, the optimal SVR model parameter combination is searched in the solution space to improve the prediction accuracy of the model. For example, the SVR model optimized by PSO can accurately predict the viscosity change trend of the fracturing fluid based on the current production process parameters and raw material information, with a prediction error within ±5%.

[0050] 3. Optimize and adjust the production process based on the quality monitoring and prediction results. When it is found that the quality indicators deviate from the target range, the corresponding process parameters are adjusted through the intelligent production control module. For example, if it is predicted that the cross-linking degree of the fracturing fluid will be too high, it may be due to excessive addition of cross-linking agent or too long reaction time. The system will automatically reduce the amount of cross-linking agent or shorten the reaction time, and feed back the adjustment experience to the quality prediction model and production control strategy. In the feedback process, the fuzzy neural network and SVR model are retrained according to the actual production data and quality inspection data after adjustment, and the model parameters and control rules are updated to improve the adaptability and accuracy of the production management system.

[0051] 3. Innovation

[0052] 1. Innovative data collection and fusion strategy:

[0053] 1. A comprehensive and high-precision data acquisition system is designed, which can obtain multi-source heterogeneous data from all aspects of the production of highly concentrated polyacrylamide fracturing fluid, including raw material data, reaction process data, and finished product quality data, etc., providing a rich data foundation for intelligent production management. Compared with traditional data acquisition methods, the data acquisition range of the present invention is wider and the data accuracy is higher, which can more comprehensively and accurately reflect the actual situation of the production process.

[0054] 2. The use of ontology-based data fusion technology effectively solves the semantic heterogeneity problem of multi-source heterogeneous data. By constructing an ontology model for the production of highly concentrated polyacrylamide fracturing fluid, the semantic association and sharing of different sensor data are realized, and the availability and comprehensibility of the data are improved. This data fusion method can better explore the intrinsic connections between data and provide more accurate and comprehensive information support for intelligent production control and quality prediction.

[0055] 2. Innovation of intelligent control algorithm based on fuzzy neural network and MPC:

[0056] 1. An algorithm combining fuzzy neural network and model predictive control (MPC) is proposed for the production process control of highly concentrated polyacrylamide fracturing fluid. Fuzzy neural network can effectively handle the nonlinear relationship in the production process and establish an accurate input-output mapping relationship by learning a large amount of historical data; MPC can realize dynamic optimization control of the production process based on the prediction results of fuzzy neural network and comprehensive consideration of various constraints. This hybrid algorithm gives full play to the advantages of both methods, overcomes the limitations of traditional production control methods in dealing with nonlinear and multivariable coupling problems, and improves the accuracy and flexibility of production.

[0057] 2. The robust optimization method is introduced to enhance the robustness of the control strategy, taking into account the influence of uncertain factors such as raw material quality fluctuations and equipment performance changes in the production process of highly concentrated polyacrylamide fracturing fluid. By solving the robust optimization problem, a control strategy that can still ensure product quality and production stability within a certain uncertainty range is obtained, which effectively reduces the interference of uncertain factors on the production process and improves the stability and reliability of production.

[0058] 3. Establish a rule-based emergency response mechanism, which can quickly take corresponding emergency measures when abnormal situations occur during the production process to ensure the safety and continuity of the production process. This emergency response mechanism effectively improves the risk resistance of the production system and reduces production losses caused by abnormal situations.

[0059] 3. Innovation of intelligent mechanism for quality monitoring and optimization:

[0060] 1. A quality prediction model based on support vector machine regression (SVR) combined with particle swarm optimization (PSO) was established, which can accurately predict the quality change trend of highly concentrated polyacrylamide fracturing fluid based on production process data and discover potential quality problems in advance. Through the combination of online quality detection equipment and quality prediction models, real-time monitoring and analysis of product quality is achieved, providing a timely and accurate basis for the optimization and adjustment of production processes.

[0061] 2. Use quality data feedback to continuously optimize quality prediction models and production control strategies. This feedback optimization mechanism enables the production management system to continuously learn and improve, adapt to various changes and challenges in the production process, and improve the adaptability and intelligence of the production management system.

[0062] Example 2

[0063] (I) Data acquisition module

[0064] 1. Data collection in production process

[0065] Sensors are installed on key equipment such as dissolving tanks and mixing kettles to collect data such as temperature, pressure, stirring speed, material flow, etc., to reflect the status of the production process in real time.

[0066] 2. Raw material data collection

[0067] Detect parameters of raw materials such as polyacrylamide, various additives, such as molecular weight, purity, particle size, active ingredient content, as well as batch and source information of raw materials.

[0068] 3. Product quality data collection

[0069] By using online detection instruments, the quality index data of fracturing fluid such as viscosity, density, pH value, and filtration loss can be obtained in real time.

[0070] (II) Data preprocessing module

[0071] 1. Data cleaning and calibration

[0072] Remove outliers caused by sensor failure or detection errors, and calibrate data based on sensor calibration parameters and detection standards to ensure data accuracy and reliability.

[0073] 2. Data normalization and feature extraction

[0074] Data of different magnitudes are normalized, and features related to fracturing fluid quality and production efficiency are extracted through data analysis and professional knowledge, such as the relationship between temperature and viscosity changes, and the effect of additive dosage on fluid loss.

[0075] (III) Intelligent control and optimization module

[0076] 1. Control model construction

[0077] An intelligent control model is constructed based on production characteristics and requirements, with the goal of ensuring product quality, improving production efficiency and reducing costs, taking into account constraints such as raw material changes and equipment status.

[0078] 2. New algorithm model driven control and optimization strategy generation

[0079] The following three unreported algorithm models are used to generate control and optimization strategies, calculate control instructions based on real-time collected and preprocessed data, and adjust production parameters.

[0080] (IV) Execution module

[0081] 1. Equipment control execution

[0082] Send control instructions to the production equipment control system to adjust parameters such as dissolution temperature, stirring motor speed, material addition amount, etc. to achieve precise production control.

[0083] 2. Feedback and Adjustment

[0084] Monitor equipment operating status and product quality changes in real time, and transmit feedback information to the intelligent control module to adjust the control strategy in time.

[0085] 4. Control model based on hybrid algorithm of quantum neural network and simulated annealing

[0086] 1. Model Principle

[0087] Quantum neural network (QNN) uses the characteristics of quantum bits to increase the network's information processing and representation capabilities, and can better handle complex nonlinear relationships. The simulated annealing algorithm (SA) has the ability to jump out of the local optimal solution. Combining the two, SA is used to optimize the parameters of QNN so that the network can find a better global solution during training to adapt to the complex chemical and physical processes in the production of highly concentrated polyacrylamide fracturing fluid.

[0088] 2. Algorithm Implementation

[0089] Construct the quantum neural network structure, determine the input layer (production data and raw material data), hidden layer and output layer (equipment control parameters). Use the simulated annealing algorithm to initialize and optimize the QNN parameters such as weights and biases. In the simulated annealing process, define the energy function (such as based on the error between the predicted quality and the target quality, production cost, etc.), and gradually optimize the QNN parameters through the simulated annealing cooling process and probability acceptance criteria. The trained model is used to generate production control instructions based on real-time data.

[0090] 3. Advantages

[0091] This model combines the powerful nonlinear processing capability of QNN and the global optimization capability of SA, which can effectively avoid falling into local optimality, improve the ability to respond to complex changes in the production process, ensure stable product quality, and optimize production efficiency.

[0092] (II) Production parameter optimization model integrating variational autoencoder and particle swarm optimization

[0093] 1. Model Principle

[0094] Variational Autoencoder (VAE) can learn the potential distribution of data, reduce the dimension and extract features of production data, and mine key information hidden in complex production data. Particle Swarm Optimization (PSO) algorithm is used to search for the optimal combination of production parameters in the potential space extracted by VAE, and find the parameter value that optimizes product quality and minimizes cost through particle position update and information sharing.

[0095] 2. Algorithm Implementation

[0096] First, the variational autoencoder is trained using historical data from the production process to map high-dimensional production data (such as temperature, pressure, material flow, etc.) to a low-dimensional latent space. In the latent space, each point represents a potential representation of a production state. Then, the particle swarm optimization algorithm is applied to the latent space, and the position of the particle corresponds to the combination of production parameters in the latent space. According to the defined fitness function (such as a comprehensive evaluation of product quality indicators and production costs), the particles search for the optimal position in the latent space. Finally, the optimal latent space parameters are mapped back to the actual production parameters to guide production.

[0097] 3. Advantages

[0098] This fusion model can effectively utilize the feature extraction ability of VAE and the optimization ability of PSO. In the production of highly concentrated polyacrylamide fracturing fluid, it can process complex production data relationships, quickly find the optimal production parameters, improve product quality, and reduce production costs. It is especially suitable for processing multivariable and nonlinear production optimization problems.

[0099] 3. Quality control model based on deep reinforcement learning and fuzzy reasoning system

[0100] 1. Model Principle

[0101] Deep reinforcement learning (DRL) finds the optimal action strategy through continuous attempts and learning by the agent in the environment. In production management, the production environment is regarded as a reinforcement learning environment, the action of the agent is to adjust the production parameters, and the reward is the improvement of product quality and the reduction of costs. Fuzzy reasoning system (FIS) is used to process fuzzy information in the production process, such as fuzzy concepts such as "slightly poor quality" and "high viscosity", as well as the fuzzy relationship between production parameters and product quality, to assist deep reinforcement learning in making more reasonable decisions.

[0102] 2. Algorithm Implementation

[0103] A deep reinforcement learning model is constructed to define the state space (including production data, raw material data, and current product quality information), action space (adjustable range of production parameters), and reward function (calculated based on product quality indicators, production efficiency, and cost). The agent continuously explores during the production process and adjusts its action strategy based on reward feedback. At the same time, a fuzzy reasoning system is established to convert fuzzy data and expert experience in the production process into fuzzy rules, which are combined with the decision-making of deep reinforcement learning. For example, when the fuzzy reasoning system determines that the viscosity tends to be high, deep reinforcement learning adjusts the control parameters based on this information, such as reducing the addition rate of polyacrylamide.

[0104] 3. Advantages

[0105] This model combines the autonomous learning and optimization capabilities of deep reinforcement learning with the advantages of fuzzy reasoning systems in processing fuzzy information. In the production of highly concentrated polyacrylamide fracturing fluid, it can cope with complex production environments and fuzzy quality feedback, achieve more precise quality control, and improve product qualification rate and production stability.

[0106] The present invention has been described in detail above in conjunction with the embodiments. However, those skilled in the art will appreciate that, without departing from the spirit of the present invention, the specific parameters in the above embodiments may be changed to form a plurality of specific embodiments, which are all within the common variation range of the present invention and will not be described in detail herein.

Claims

1. An intelligent production management system for highly concentrated polyacrylamide fracturing fluid, characterized in that: include: The data acquisition and integration module is used to deploy high-precision sensor arrays at various key locations on the high-concentration polyacrylamide fracturing fluid production line to comprehensively collect data in the production process, and transmit the collected data to the data processing center in real time via high-speed industrial Ethernet. The data processing center uses ontology-based data fusion technology to integrate multi-source heterogeneous data. The data acquisition and integration module includes: high-precision mass flow meters, level sensors, and humidity sensors installed in the raw material storage and transportation system, which are used to accurately measure the flow of polyacrylamide dry powder, cross-linking agent, and additive raw materials, the level height in the storage tank, and the humidity of the raw materials.

2. The intelligent production management system for highly concentrated polyacrylamide fracturing fluid according to claim 1, characterized in that: In the data collection and integration module, the data collection frequency is dynamically adjusted according to the characteristics of the production process.

3. An intelligent production management system for highly concentrated polyacrylamide fracturing fluid, characterized in that: It also includes an intelligent production control module, which uses an algorithm combining fuzzy neural network and model predictive control to control the production process based on the integrated data, including: using fuzzy neural network to model and approximate the nonlinear relationship in the production process of highly concentrated polyacrylamide fracturing fluid, and automatically constructing a complex nonlinear mapping relationship between input and output through learning from historical production data, wherein the input includes raw material ratio, reaction temperature, and pressure, and the output includes fracturing fluid viscosity and cross-linking quality indicators; embedding the fuzzy neural network model into the model predictive control (MPC) framework, taking the quality indicators of fracturing fluid as the control target, comprehensively considering various constraints in the production process, and solving the future control sequence through rolling optimization based on the system output in the future time predicted by the fuzzy neural network; introducing the interval description of uncertain parameters in the MPC model, and obtaining a control strategy that can still ensure product quality and production stability within a certain uncertainty range by solving the robust optimization problem.

4. The intelligent production management system for highly concentrated polyacrylamide fracturing fluid according to claim 3, characterized in that: The intelligent production control module also includes a rule-based emergency response mechanism, which is used to quickly take corresponding emergency measures according to preset rules when sudden abnormal situations such as equipment failure and raw material supply interruption are detected, so as to ensure the safety and continuity of the production process.

5. An intelligent production management system for highly concentrated polyacrylamide fracturing fluid, characterized in that: It also includes a quality monitoring and optimization module, which uses online quality detection equipment and data analysis technology to monitor the quality of highly concentrated polyacrylamide fracturing fluid in real time, and uses a quality prediction model based on a combination of support vector machine regression (SVR) and particle swarm optimization algorithm (PSO) to predict and analyze the quality of the fracturing fluid.

6. The intelligent production management system for highly concentrated polyacrylamide fracturing fluid according to claim 5, characterized in that: The quality monitoring and optimization module optimizes and adjusts the production process according to the quality monitoring and prediction results. When it is found that the quality indicators deviate from the target range, the corresponding process parameters are adjusted through the intelligent production control module. At the same time, the quality data feedback is used to continuously optimize the quality prediction model and production control strategy.

7. A highly concentrated polyacrylamide fracturing fluid intelligent production management method, characterized in that: The method comprises data collection and preprocessing steps, wherein the data collection and preprocessing steps comprehensively collect multi-source heterogeneous data from raw material supply, reaction process and finished product processing links of high-concentrated polyacrylamide fracturing fluid production, and perform preprocessing using ontology-based data fusion technology.

8. A highly concentrated polyacrylamide fracturing fluid intelligent production management method, characterized in that: The method comprises the steps of intelligent production control modeling and solving, wherein the steps utilize fuzzy neural network to model the nonlinear relationship of the production process of highly concentrated polyacrylamide fracturing fluid, embed the fuzzy neural network model into the model predictive control (MPC) framework, and also include establishing a rule-based emergency handling mechanism.

9. A highly concentrated polyacrylamide fracturing fluid intelligent production management method, characterized in that: The method includes a quality monitoring and optimization step, wherein the method uses an online quality detection device to sample and detect the highly concentrated polyacrylamide fracturing fluid every 10 minutes, obtains key quality indicator data of the fracturing fluid such as viscosity, particle size distribution, and chemical composition content, and performs a comprehensive analysis in combination with production process data. The method also uses a quality prediction model based on a combination of support vector machine regression (SVR) and a particle swarm optimization algorithm (PSO) to predict quality, and optimizes and adjusts the production process based on the quality monitoring and prediction results, specifically including: The process parameters in the production process and the quality indicators of raw materials are used as input variables, and the key quality indicators of fracturing fluid are used as output variables to build the SVR model. The PSO algorithm is used to optimize the model parameters of the SVR and improve the prediction accuracy of the model. When it is found that the quality indicators deviate from the target range, the corresponding process parameters are adjusted through the intelligent production control module. During the feedback process, the fuzzy neural network and SVR model are retrained according to the adjusted actual production data and quality inspection data, and the model parameters and control rules are updated to improve the adaptability and accuracy of the production management system.

Images