1. bookAHEAD OF PRINT
Informacje o czasopiśmie
License
Format
Czasopismo
eISSN
2444-8656
Pierwsze wydanie
01 Jan 2016
Częstotliwość wydawania
2 razy w roku
Języki
Angielski
Otwarty dostęp

Leakage control of urban water supply network and mathematical analysis and location of leakage points based on machine learning

Data publikacji: 15 Jul 2022
Tom & Zeszyt: AHEAD OF PRINT
Zakres stron: -
Otrzymano: 17 Mar 2022
Przyjęty: 19 May 2022
Informacje o czasopiśmie
License
Format
Czasopismo
eISSN
2444-8656
Pierwsze wydanie
01 Jan 2016
Częstotliwość wydawania
2 razy w roku
Języki
Angielski
Introduction

In recent years, the new generation technology in the 21st century represented by mobile Internet, Internet of things and AI is bringing another information revolution that has a deep impact on people's intelligent life [1]. In terms of urban infrastructure, urban operation and maintenance, intelligent transportation and other aspects, information technology is deeply combined with it, showing a new form in a different way. Among them, the construction of smart water platform is also one of the important components of Smart City mountain construction. In recent years, it has also been highly valued by the state, and its development strategy has also been put on the agenda. The government is also actively promoting the development of smart water, increasing R & D investment and accelerating the construction of water informatization through cooperation with enterprises. The “wisdom” of smart water is reflected in three aspects. From the perception level, the system collects information about water through wireless networks, such as online monitoring equipment for pressure, flow and water quality, and senses the real-time operation status of the whole urban water supply and drainage pipe network system; From the presentation layer, friendly visual interface and organic integration of all water supply and drainage facilities in water affairs for dynamic image display to form a dynamic and intelligent “urban water Internet of things” platform, as shown in Figure 1; In terms of big data, the system timely and effectively analyzes and processes a large amount of water affairs information, and models it in combination with the corresponding mathematical model, so as to provide decision-making help for the better allocation of resources in the city, save energy, reduce the output of redundant labor force, increase the use efficiency of human and material resources, and contribute to urban green environmental protection [2]. Alibaba's main project in recent years is the brain of smart city, realizing the intelligent management of the city, increasing the efficiency of various services in the city and better coordinating various work. Smart water is also a part of it. Smart water system is mainly divided into four parts: SCADA system, GIS geographic information system, revenue system, water quality management, water sample detection and other production management systems. SCADA is referred to as data acquisition and monitoring control system for short. The data acquisition system, which is mainly based on computer technology and integrates communication technology and control technology, can control the instructions issued by the monitoring equipment on the production site and monitor the on-site environment. The administrator can customize the alarm rules, realize the different step alarm mechanism and enhance the monitoring efficiency. The concept of SCADA was formed in the 1960s and took shape in the 1970s. After the 1980s, the rapid development of microcomputer technology also promoted the realization of SCADA system. In fact, the real rapid realization of SCADA system was at the end of the 20th century. With the rapid development of Internet, the demand of enterprises for management automation and monitoring automation promoted the rapid development of SCADA system [3].

Figure 1

Water balance analysis process

He R and others make reasonable planning in advance, promote the implementation by sections, formulate the work objectives in the leakage control stage, explore the optimal management measures and gradually implement them in combination with technical means according to the scale of water supply pipe network and water supply facilities, the asset status and leakage of pipe network and ancillary facilities [4]. Literature shows that Britain is one of the countries with early urban modernization. The construction of centralized water supply facilities has a history of 200 years. In the 1980s, the leakage of water supply system was also very serious. Therefore, they strengthened the frequency and density of leak detection, promoted the renewal and transformation of pipe network, and carried out DMA zoning construction. After 20 years of efforts, the leakage rate of centralized water supply network has been reduced from 29.5% to 18.7%. Yiran L and others found that in the early 1980s, in the practice of pipe network management, in order to monitor the pipe network leakage for a long time, find the macro distribution of pipe network leakage, balance the service pressure of pipe network, and then reduce the leakage rate, the British water management organization put forward the concept of district metered areas (DMA) management [5]. Ulusoy A J and others believe that DMA refers to the construction of an independent area in which the inlet and outlet flow can be recorded in a certain area of the water supply and distribution system by closing some valves and installing flow meters at the inlet and outlet of the independent area [6]. Northwest water has established 3000 DMA management areas in its water supply network; London's water supply network is divided into 16 districts. With the help of DMA, British water company has comprehensively evaluated the water leakage of the pipe network by checking the basic data, leakage level test, “zero pressure” test and other steps. When it is determined that there is water leakage in a water supply area, the location of the water leakage point shall be further determined through appropriate technical means. Although there are many methods to determine the water leakage point of the pipe network, including passive leak detection method, listening leak detection method, regional meter installation method, regional leak detection method, leak detection compound method and pressure adjustment method, none of them is omnipotent, but it is more effective to combine multiple methods to give full play to their respective advantages. The establishment of DMA leak detection area can adopt different modes for different pipe network forms, i.e. tree water supply network and ring water supply network. Due to many factors such as urban planning and water supply planning, the construction of DMA needs to be carried out with the practice of pipe network zoning management and with the means of pipe network modeling. There are many factors affecting DMA leakage rate, such as the accuracy of meter reading, pipeline decompression and flushing, water for pipe network maintenance and emergency repair, etc. Therefore, Xu J and others can effectively carry out regional leakage control with the help of DMA, which requires the organic cooperation of all departments of the water supply enterprise. If a DMA area has reached the leakage control target after evaluation, it can start to prepare the leakage control report of the area, analyze the investment benefit ratio of leakage control, analyze the problems found in the leakage control process, put forward improvement methods, and guide the main direction of leakage control in the future [7]. Developed countries attach great importance to the research and development of pipe network leakage detection equipment. The leak detector developed by the United States, Britain, France, Germany and Japan has been continuously updated with the development of science and technology. Digital leak detector, multi probe correlation instrument and regional leak monitor have been launched one after another, which greatly improves the reliability and accuracy of leak detection.

Method

For most water supply enterprises, the level of non profitable water volume (NRW) is a very important performance indicator”. However, due to institutional and political pressure and lack of knowledge to strictly determine the level of NRW, many water supply enterprises tend to underestimate NRW. Senior managers are eager to see the report of low-level NRW. However, these reports are either based on deliberately provided wrong information or lack of accurate information, which can not help water supply enterprises reduce costs or improve benefits. On the contrary, it will cover up the truth of the problem and affect the operation efficiency of water supply enterprises. Only by quantifying NRW and its constituent elements, calculating reasonable performance indicators, and changing the leakage water to monetary value, can water supply enterprises accurately understand the situation of NRW and take necessary measures. The establishment process of water balance table can help the managers of water supply enterprises understand the size, source and cost of NRW. The International Water Association (IWA) has formulated the elements and technical terms of a standard international water balance (as shown in Table 1) [8].

Water balance recommended by IWA

water system legal water Legal water bill Pay metered water income from buying water

Water usage is not a measure of cost

Legal water is free Free water test no income to irrigate
water outflow Apparent leakage Free and unlimited water illegal use of waterDrowning due to user errormeasurement and data processing errorsopen soil
real leak cover upafter the leak

The total amount of water flowing into the water supply network from the water plant (referred to as “system supply water”) minus the total amount of charged legal industrial and residential water (referred to as “charged legal water”), i.e. non revenue water (NRW), see formula (1).

(NRW)=systemwatersupplyBilledlegalwaterconsumption \left( {{\rm{NRW}}} \right) = {\rm{system}}\,{\rm{water}}\,{\rm{supply}} - {\rm{Billed}}\,{\rm{legal}}\,{\rm{water}}\,{\rm{consumption}}

The assumptions of this equation are as follows:

The system water supply has corrected some known errors;

The charging and metering water consumption recorded by the user's meter reading shall be consistent with the statistical time of the water supplied by the system.

The managers of water supply enterprises should calculate each constituent element according to the water balance table, so as to find out the cause of leakage. This will determine the priority of policy changes and business methods and their implementation plans.

At present, some water departments in China have also begun to try to carry out water balance test, which usually adopts the “table filling” method, that is, fill the corresponding water volume data item by item in the “water balance table”, and calculate the “leakage water volume” and “non revenue water volume” by recursive decline. This case uses wb-easycalcv300 (free software) synchronously recommended by the International Water Association. Z water division was founded in 1953. There are 7 main water supply plants, including 5 surface water plants, which mainly use the water source of South-to-North Water Transfer, the water source of the Yellow River is standby, and 2 groundwater plants use the Yellow River to measure seepage. The daily water supply capacity is 1.77 million m3 / D, the water supply population is 4.5 million, the water supply area is 600 square kilometers, the pipe network above dn75 is 3600 kilometers long, and the registered users are 1.3 million. The statistical cycle of water balance in this case is 181 days in half a year (January 1, 2014 – June 30, 2014). The water balance analysis process is shown in Figure 1 [9].

Step 1: First determine the water supply volume of the water supply system ① The system has 4 water supply systems, 1 pressure make-up water system and 19 pressure make-up wells, all of which have suitable water meters and have been regularly checked;

Step 2: The second step is to count the medicinal water consumption in the calculation cycle ②, including 9 categories such as industrial, living, and wholesale;

Step 3: The third step calculates the unmeasured cost water consumption during the calculation period ③, including prepaid water, construction water, compensation conversion water, etc. The sum of ③ and ② is the water intake of the system;

Step 4: The fourth step is to calculate the measured water consumption free of charge in the calculation cycle ④, such as the water consumption reduced by the policy, the free water consumption agreed in the agreement, etc;

Step 5: The fifth step is to estimate the unpaid and unmeasured water consumption⑤, such as the free water consumption of the contract village, the emptying and filling water consumption of the pipe network construction, the dead end drainage of the pipe network, the temporary water supply of the water supply section, and the household meter conversion. Temporary water supply, long flow of various online instruments in the pipeline network, etc. The sum of ⑤ and ④ is the unpaid legal water consumption;

Step 6: The sixth step is to estimate the apparent leakage ⑥, through a reasonable and feasible sampling detection method, evaluate the illegal water consumption and measurement error, and estimate the apparent leakage. According to the water balance sheet, apparent leakage can be divided into two categories: “illegal water use” and “water loss caused by user measurement errors and data processing errors” are water losses caused by corporate negligence, especially, illegal water pollution in the form of complex of. This article illustrates a methodology for estimating water use in this area using the example of a thorough investigation of water stolen from fire hydrants in violation of sprinkling, sweeping, and dust suppression regulations. The water volume of the sanitary and clean registered water meter is part of the “charged metered water consumption metering”, and its water consumption trend is regularly monitored. Differences can be determined by comparison with the frequency of operation of road cleaning vehicles and municipal administrations; for “water loss due to measurement error,” a sample test can be done by changing water consumption before and after a centralized community replacement of water meters. Similar experimental studies have been carried out by many water conservancy departments in China;

Step 7: Step 7: Calculate the actual leakage ⑦ and calculate or estimate the actual leakage by reasonably feasible methods (night flow analysis, pipeline rupture frequency, flow rate, duration, hydraulic model, etc.). For areas that can be independently measured, the leakage can be pre-analyzed by night current analysis technology and extended to all areas covered by the pipe network; the water loss caused by pipeline rupture can be based on the pipeline rupture repair time recorded by the pipe network management department, crack area and nearby pipe network pressure calculation; according to the leak detection quantity and leak detection period in the statistical period provided by the leak detection department to calculate the hidden leakage amount. For background leakage, the IWA calculation formula for background leakage must be followed;

Step 8: Perform multiple balance checks using data from previous periods.[10].

The city's non-revenue water ratio (production-sales difference rate) is 16%, and the leakage rate is 13.80%. According to the above analysis, first of all, the important factors affecting the non-revenue water ratio (production and sales difference rate) of the water supply system in Z city are 5, 1. Background leakage 2. Various fire hydrants steal water; 3. Light leakage 4. Illegal water use in urban villages; 5. Water loss caused by various measurement errors and data processing errors. Enterprise managers can accurately evaluate the economic benefits of corresponding technical solutions, and decide on investment and management ranking.

In the process of balance analysis, deception in the management stage can be found through reverse guidance. For example, during the calculation period, if it is converted into 36,902 households in Z city for household meter renovation (managed by the household meter renovation department) and 41525 households for new households (managed by the customer service department), the total construction cleaning water volume of the two projects is 7933m' (Operating expenses department management), 0.10m3/household, it can be seen that there is a management loophole in the calculation of construction cleaning water volume. Many components that affect the non-revenue quantity cannot be measured, and different departments of the enterprise need to use certain methods to estimate. Among them, the data reporting department deliberately modified the estimation results based on its own performance interests, resulting in a serious imbalance in the analysis process. Therefore, various statistical methods for estimating water volume must be objective. Perform sample testing or method validation. The water balance table recommended by the International Water Association (IWA) is a horizontal, macro, and system-wide leakage component analysis unit. For water supply enterprises that have achieved classified measurement, they can inherit deeper and different measurement classifications.

Therefore, all kinds of statistical methods for estimating water volume must be objective and verified by sample test or method.

The water balance table recommended by the International Water Association (IWA) is a horizontal, macro and system wide analysis method of water leakage components. For water supply enterprises that have achieved zoning measurement, we can continue to carry out longitudinal water balance analysis for different metering zones. The purpose of constructing the performance evaluation system of water supply industry is to promote the benign objectives of water supply enterprises, such as safe water supply, high-quality service, cost transparency and efficiency improvement, through the vertical comparison of their own business performance and the horizontal comparison of business performance between industries. In view of the relevant performance evaluation indexes of water supply system leakage management, the national industry standard “urban water supply leakage control and evaluation standard” (cjj92-2002) puts forward the basic countermeasures and evaluation standards for strengthening leakage control in China, and determines the benchmark evaluation standards for leakage control, such as “leakage rate” and “unmeasured water rate”. With the improvement of the leakage management level of urban water supply system in the domestic water supply industry, the limitations of indicators such as “leakage rate” and “unmeasured water rate” expressed in percentage in practical application have gradually been exposed, and domestic peers have gradually realized the rationality of the water balance calculation method and water supply service performance indicators recommended by the International Water Association (IWA).

Firstly, HPSO algorithm generates a random particle swarm (random solution). Each particle flies (optimizes) in its multi-dimensional solution space, and the flight speed is dynamically adjusted according to its own and other particle flight conditions and experience. There are N particles in d-dimensional space, and the best position in history is pbest.

The position and velocity vectors of offspring particles are shown in formula (2) (3): x1(t+1)=px1(t)+(1.0p)x1(t) {\vec x_1}\left( {t + 1} \right) = \vec p \cdot {\vec x_1}\left( t \right) + \left( {1.0 - \vec p} \right) \cdot {\vec x_1}\left( t \right) x2(t+1)=px2(t)+(1.0p)x2(t) {\vec x_2}\left( {t + 1} \right) = \vec p \cdot {\vec x_2}\left( t \right) + \left( {1.0 - \vec p} \right) \cdot {\vec x_2}\left( t \right) Where x1(t) {\vec x_1}\left( t \right) and x2(t) {\vec x_2}\left( t \right) are the position vectors of the two parents of D-Dimension respectively; x1(t+1) {\vec x_1}\left( {t + 1} \right) and x2(t+1) {\vec x_2}\left( {t + 1} \right) are two children of d-Dimension respectively, and p \vec p is a random number vector uniformly distributed in d-Dimension. Each component of p \vec p takes value in [0,1], as shown in formula (4) (5). v1(t+1)=v1(t)+v2(t)|v1(t)+v2(t)||v1(t)| {\vec v_1}\left( {t + 1} \right) = {{{{\vec v}_1}\left( t \right) + {{\vec v}_2}\left( t \right)} \over {\left| {{{\vec v}_1}\left( t \right) + {{\vec v}_2}\left( t \right)} \right|}} \cdot \left| {{{\vec v}_1}\left( t \right)} \right| v2(t+1)=v1(t)+v2(t)|v1(t)+v2(t)||v2(t)| {\vec v_2}\left( {t + 1} \right) = {{{{\vec v}_1}\left( t \right) + {{\vec v}_2}\left( t \right)} \over {\left| {{{\vec v}_1}\left( t \right) + {{\vec v}_2}\left( t \right)} \right|}} \cdot \left| {{{\vec v}_2}\left( t \right)} \right| Where v1(t) {\vec v_1}\left( t \right) and v2(t) {\vec v_2}\left( t \right) are the position vectors of the two parents of d-Dimension respectively; v1(t+1) {\vec v_1}\left( {t + 1} \right) and v2(t+1) {\vec v_2}\left( {t + 1} \right) are velocity vectors of two generations of d-Dimension respectively.

Experiment and discussion

Before data modeling, we must first clarify the purpose of modeling and the effect to be achieved [11]. Data modeling is a series of operation sets that do some customization, logicalization, model matching and other work on the data template. The final result of modeling still serves people. How to enable users to clearly get the knowledge they need from the data model is a necessary function of the model.

There are two points to summarize the modeling requirements of this system:

Behavior analysis + intelligent early warning

The purpose of the behavior analysis model is to realize the front-end intelligence, so that the terminal has the ability of self prediction. In case of abnormality, it can wake up from the deep sleep and notify the user. This is mainly based on the practical application scenario of smart water in pipe network monitoring. In pipe network monitoring, the equipment is generally set to sample once in 20 minutes and send it once in four hours. They don't need intensive data collection. The period of manual meter reading is one month. In order to prolong the service life of the equipment to a greater extent, the equipment is in a dormant state before the scheduled sampling, so as to save energy to the greatest extent. However, in the actual modeling and analysis, we should consider the dimension of time. The 20 minute sampling period can only represent the data at a certain time. Therefore, in the process of practical application, we adopt the random discrete sampling distribution algorithm for the equipment to ensure the dispersion of the sampling data, so as to predict the behavior more accurately in the data modeling. However, the device will still keep sampling at the sampling point required by the user. In this way, business requirements and modeling are correct. At the end of the model analysis, the server model middleware issues the simplification strategy and converts it into the instructions that can be recognized by the terminal, which is pushed to the terminal, so that the terminal has some decision-making ability, and sends an alarm in time when the terminal behavior deviates beyond the tolerance range. This will provide timely and important information on leakage detection, control and repair [12].

Production model + auxiliary decision-making

The main purpose of the model is to dynamically adjust the supply according to the actual demand of the water area. First, it can improve user satisfaction and second, it can save energy. For example, at 6 o'clock every day, in a certain area, there is a community with huge water consumption, and the pressure will drop to the normal. In another place, there is less water, and the corresponding pressurizing device should also operate in real time [13]. In view of this situation, we can give auxiliary suggestions according to the pipe network model. It is suggested to add pressurizing devices at the inlet of the community and appropriately reduce the configuration of pressurizing devices at another place. Save resources. The water department can even analyze according to this model, optimize the layout of pipe network and increase the output ratio. However, the model needs to be based on the DMA partition principle.

According to the two requirements, we conduct data statistical analysis on the pressure and flow data. In terms of two types of data statistics, we mainly take the day as the cycle, scatter all sampling data to some time of the day, and form multiple data combinations in divided units. The types of statistical data mainly include: original data, derivative of data relative to time, cumulative value and transient [14]. For the statistical data, we should first remove the noise according to certain rules, and take it out. We will do the positive distribution operation on the data at all times to obtain a more reasonable range. We will dynamically adjust this parameter according to the tolerance range of users. Then take the result of positive Pacific distribution at all times as the a priori probability statistics of Bayesian decision theory, and calculate the a posteriori probability of the next time with time as the reference. However, there are two kinds of algorithms at the moment: one is that all data get a behavior prediction in a day cycle, and the other is multi-dimensional and multi-directional statistics. The second is mainly considering that the water consumption on Monday and Saturday is different, the water consumption on Monday in March and Monday in September may be different, or there may be a factory in this area that needs a lot of water for production on specific days of each month. In view of these different situations, in the subsequent data modeling, we should consider customized selection, distributed statistics and modeling to better meet the practical application scenarios.

After obtaining a reasonable interval distribution, after the user specifies the deviation tolerance of the data, we get the interval range required by the user according to the distribution equation, then screen the data according to the interval range, then predict the trend of the data in the interval, use polynomial to fit the horizontal data behavior of the interval distribution, and display the data analysis results through c# data analysis host computer. In the process of fitting, we divide the data into cumulative quantity and periodic quantity. Later, we will find that the derivative of cumulative quantity is also periodic quantity. In the analysis of the cumulative quantity, after the data distribution operation, the weighted curve is obtained, and then the weighted curve is fitted by the least square method, and the fitting degree value R2 is obtained. In the process of fitting, it is easy to find that the more the order is, the better the fitting degree is, that is, the closer the R2 value is to 1. However, due to the calculation ability, the fitting order cannot be too large. No matter how good the fitting degree is, there will be a certain dispersion between the behavior equation and the actual scatter distribution, because this method is the prediction of the variation, that is, the derivative, and the dispersion within a certain range is allowed. In the actual use of the model, it is also allowed to exist in a certain range of dispersion, which is also the user tolerance range, but how to reflect it offline in the process of model use. This paper designs two schemes, one is using differential mode analysis, the other is using envelope analysis. Behavior model analysis is mainly used in intelligent early warning, which can predict and compare the behavior of the current station in the future to decide whether to give an alarm. This intelligent early warning is mainly caused by pipe explosion in the pipe network, resulting in sudden increase of flow and decrease of pressure. This method can accurately locate the parameters such as pipe explosion section, leakage speed, leakage pipe pressure loss and so on. It is convenient for maintenance personnel to arrive at the site quickly and accurately and reduce the loss. On the other hand, there is abnormal water use. A feedback from the customer is that when the secondary water supplier takes water from the primary water supplier, the measurement is damaged intermittently due to human reasons, so as to reduce the actual statistical flow. This is unacceptable for the primary water supplier. It is precisely because the traditional measurement method can not find this anomaly, which contributes to this bad atmosphere. The secondary water supplier uses intermittent operation, and the primary water supplier is also difficult to find [15]. Using this model analysis, if the secondary water supplier carries out illegal operation at any time, it will trigger the equipment early warning and return to normal. During this period, the primary water supplier can find the abnormality in time and specify that the data in the corresponding time period is illegal data and will not participate in the decision-making. Other data will be used as the source of model data. Timely communication can not only reduce losses, but also reduce this bad atmosphere. The production model mainly analyzes the regional water consumption and pressure of DMA partition, and gives the analysis model. The analysis model will visually display the water supply parameters of each region in combination with GIS pipe network system. The model can point out the water supply suggestions, resource ratio and pipe network optimization suggestions in different periods of time. The water department can conduct secondary transformation evaluation, resource distribution evaluation, zoning optimization and so on according to the analysis model to assist decision-making. This is also the main embodiment of the “wisdom” of the future intelligent water supply system.

Conclusion

The leakage control of urban water supply network is a long-term and arduous task. This paper deeply and systematically studies and analyzes the current situation of non profitable water volume (NRW) of water supply network, and puts forward and gradually implements the non profitable water volume control strategy based on the scientific system of International Water Association (IWA). On this basis, actively explore economic and effective management means to make the leakage level of water supply enterprises adapt to and meet relevant national and industrial standards. This paper systematically studies and discusses the key technical problems of non-profit water volume control commonly existing in the water supply industry, introduces advanced non-profit water volume control technologies and methods, and studies the non-profit water volume management system, main countermeasures and technical scheme of water supply network. The leakage control of water supply network is studied and discussed as a complete system engineering from basic water quantity analysis to management means, from target determination to technical measure selection. Taking the water supply network of KT City, LZ city and Z City as the main objects of analysis and research and the application of specific engineering measures, this paper puts forward a set of relatively complete and flexible pipe network leakage control countermeasures, implements the theoretical research into practice, not only expands and enriches the research content, but also verifies the theoretical research to a certain extent, and determines its feasibility.

Figure 1

Water balance analysis process
Water balance analysis process

Water balance recommended by IWA

water system legal water Legal water bill Pay metered water income from buying water

Water usage is not a measure of cost

Legal water is free Free water test no income to irrigate
water outflow Apparent leakage Free and unlimited water illegal use of waterDrowning due to user errormeasurement and data processing errorsopen soil
real leak cover upafter the leak

Kong Q, Jiang G, Liu Y, et al. Location of the Leakage From a Simulated Water-Cooling Wall Tube Based on Acoustic Method and an Artificial Neural Network[J]. IEEE Transactions on Instrumentation and Measurement, 2021, PP(99):1–1. KongQ JiangG LiuY Location of the Leakage From a Simulated Water-Cooling Wall Tube Based on Acoustic Method and an Artificial Neural Network [J] IEEE Transactions on Instrumentation and Measurement 2021 PP 99 1 1 10.1109/TIM.2020.3048538 Search in Google Scholar

Wang H, Zhao Y, Zhu Y, et al. Prediction of urban water accumulation points and water accumulation process based on machine learning[J]. Earth Science Informatics, 2021, 14(4):2317–2328. WangH ZhaoY ZhuY Prediction of urban water accumulation points and water accumulation process based on machine learning [J] Earth Science Informatics 2021 14 4 2317 2328 10.1007/s12145-021-00700-8 Search in Google Scholar

Gilani, V., Hosseinian, S. M., Ghasedi, M., & Nikookar, M.. (2021). Data-driven urban traffic accident analysis and prediction using logit and machine learning-based pattern recognition models. Mathematical Problems in Engineering, 2021(1), 1–11. GilaniV. HosseinianS. M. GhasediM. NikookarM. 2021 Data-driven urban traffic accident analysis and prediction using logit and machine learning-based pattern recognition models Mathematical Problems in Engineering 2021 1 1 11 10.1155/2021/9974219 Search in Google Scholar

He R, Weizhong M A, Ma X, et al. Modeling and optimizing for operation of CO2-EOR project based on machine learning methods and greedy algorithm[J]. Energy Reports, 2021, 7(342):3664–3677. HeR WeizhongM A MaX Modeling and optimizing for operation of CO2-EOR project based on machine learning methods and greedy algorithm [J] Energy Reports 2021 7 342 3664 3677 10.1016/j.egyr.2021.05.067 Search in Google Scholar

Yiran L. Evaluation of students' IELTS writing ability based on machine learning and neural network algorithm[J]. Journal of Intelligent and Fuzzy Systems, 2020(5):1–11. YiranL Evaluation of students' IELTS writing ability based on machine learning and neural network algorithm [J] Journal of Intelligent and Fuzzy Systems 2020 5 1 11 10.3233/JIFS-189508 Search in Google Scholar

Ulusoy A J, Pecci F, Stoianov I. An MINLP-Based Approach for the Design-for-Control of Resilient Water Supply Systems[J]. IEEE Systems Journal, 2020, PP(99):1–12. UlusoyA J PecciF StoianovI An MINLP-Based Approach for the Design-for-Control of Resilient Water Supply Systems [J] IEEE Systems Journal 2020 PP 99 1 12 10.1109/JSYST.2019.2961104 Search in Google Scholar

Xu J, Liu Q, Zhang Z, et al. Automatic Detection for Battery Leakage Based on DenseNet[J]. Nanoscience and Nanotechnology Letters, 2020, 12(2):230–234. XuJ LiuQ ZhangZ Automatic Detection for Battery Leakage Based on DenseNet [J] Nanoscience and Nanotechnology Letters 2020 12 2 230 234 10.1166/nnl.2020.3087 Search in Google Scholar

Rahul, Kedia P, Sarangi S, et al. Analysis of machine learning models for malware detection[J]. Journal of Discrete Mathematical Sciences and Cryptography, 2020, 23(2):395–407. Rahul KediaP SarangiS Analysis of machine learning models for malware detection [J] Journal of Discrete Mathematical Sciences and Cryptography 2020 23 2 395 407 10.1080/09720529.2020.1721870 Search in Google Scholar

Lee J, Asenov P, Aldegunde M, et al. A Worst-Case Analysis of Trap-Assisted Tunneling Leakage in DRAM using a Machine Learning Approach[J]. IEEE Electron Device Letters, 2020, PP(99):1–1. LeeJ AsenovP AldegundeM A Worst-Case Analysis of Trap-Assisted Tunneling Leakage in DRAM using a Machine Learning Approach [J] IEEE Electron Device Letters 2020 PP 99 1 1 10.1109/LED.2020.3046914 Search in Google Scholar

Zhang L, Mu D, Hu W, et al. Machine-Learning-Based Side-Channel Leakage Detection in Electronic System-Level Synthesis[J]. IEEE Network, 2020, 34(3):44–49. ZhangL MuD HuW Machine-Learning-Based Side-Channel Leakage Detection in Electronic System-Level Synthesis [J] IEEE Network 2020 34 3 44 49 10.1109/MNET.011.1900313 Search in Google Scholar

Taghlabi F, Sour L, Agoumi A. Prelocalization and leak detection in drinking water distribution networks using modeling-based algorithms: a case study for the city of Casablanca (Morocco)[J]. Drinking Water Engineering and Science, 2020, 13(2):29–41. TaghlabiF SourL AgoumiA Prelocalization and leak detection in drinking water distribution networks using modeling-based algorithms: a case study for the city of Casablanca (Morocco) [J] Drinking Water Engineering and Science 2020 13 2 29 41 10.5194/dwes-13-29-2020 Search in Google Scholar

X Qian. Graph processing and machine learning architectures with emerging memory technologies: a survey[J]. Science China Information Sciences, 2021, 64(6):1–25. QianX Graph processing and machine learning architectures with emerging memory technologies: a survey [J] Science China Information Sciences 2021 64 6 1 25 10.1007/s11432-020-3219-6 Search in Google Scholar

Akram A, Mushtaq M, Bhatti M K, et al. Meet the Sherlock Holmes' of Side Channel Leakage: A Survey of Cache SCA Detection Techniques[J]. IEEE Access, 2020, PP(99):1–1. AkramA MushtaqM BhattiM K Meet the Sherlock Holmes' of Side Channel Leakage: A Survey of Cache SCA Detection Techniques [J] IEEE Access 2020 PP 99 1 1 10.1109/ACCESS.2020.2980522 Search in Google Scholar

Ozturk Taha Yasintaha36100@hotmail.comDepartment of Mathematic, Kafkas University, 36100-Kars, Turkey. Some Structures on Neutrosophic Topological Spaces[J]. Applied Mathematics and Nonlinear Sciences, 2021, 6(1):467–478. TahaOzturk Yasintaha36100@hotmail.comDepartment of Mathematic, Kafkas University, 36100-Kars, Turkey. Some Structures on Neutrosophic Topological Spaces [J] Applied Mathematics and Nonlinear Sciences 2021 6 1 467 478 10.2478/amns.2020.2.00069 Search in Google Scholar

Saouli M A. Existence of solution for Mean-field Reflected Discontinuous Backward Doubly Stochastic Differential Equation[J]. Applied Mathematics and Nonlinear Sciences, 2020, 5(2):205–216. SaouliM A Existence of solution for Mean-field Reflected Discontinuous Backward Doubly Stochastic Differential Equation [J] Applied Mathematics and Nonlinear Sciences 2020 5 2 205 216 10.2478/amns.2020.2.00038 Search in Google Scholar

Polecane artykuły z Trend MD

Zaplanuj zdalną konferencję ze Sciendo