Application of a multi-stage neural network approach for time-series landfill gas modeling with missing data imputation

Fault monitoring method of domestic waste incineration slag sorting device based on back propagation neural network

The main monitoring points of traditional sorting equipment fault monitoring methods are usually limited to the inlet and outlet, making it difficult to monitor the internal equipment, which may affect the accuracy of fault monitoring. Therefore, a new fault monitoring method based on back propagation neural network has been studied and designed, which is mainly applied to the sorting device of domestic waste incineration slag. The fault monitoring modeling variables of the domestic waste incineration slag sorting device are selected to determine the operation status of the sorting device. Based on back propagation neural network, a fault monitoring model for the sorting device of municipal solid waste incinerator slag is constructed, and the fault data of the sorting device is trained in the model, so that the fault data of the sorting device can be optimized faster, thus improving the accuracy of fault monitoring. Through comparative experiments with traditional methods, it has been confirmed that this fault monitoring method based on back propagation neural network has significant advantages in detection performance, demonstrating its potential in practical applications.

Municipal solid waste management for low-carbon transition: A systematic review of artificial neural network applications for trend prediction

Improper municipal solid waste (MSW) management contributes to greenhouse gas emissions, necessitating emissions reduction strategies such as waste reduction, recycling, and composting to move towards a more sustainable, low-carbon future. Machine learning models are applied for MSW-related trend prediction to provide insights on future waste generation or carbon emissions trends and assist the formulation of effective low-carbon policies. Yet, the existing machine learning models are diverse and scattered. This inconsistency poses challenges for researchers in the MSW domain who seek to identify and optimize the machine learning techniques and configurations for their applications. This systematic review focuses on MSW-related trend prediction using the most frequently applied machine learning model, artificial neural network (ANN), while addressing potential methodological improvements for reducing prediction uncertainty. Thirty-two papers published from 2013 to 2023 are included in this review, all applying ANN for MSW-related trend prediction. Observing a decrease in the size of data samples used in studies from daily to annual timescales, the summarized statistics suggest that well-performing ANN models can still be developed with approximately 33 annual data samples. This indicates promising opportunities for modeling macroscale greenhouse gas emissions in future works. Existing literature commonly used the grid search (manual) technique for hyperparameter (e.g., learning rate, number of neurons) optimization and should explore more time-efficient automated optimization techniques. Since there are no one-size-fits-all performance indicators, it is crucial to report the model's predictive performance based on more than one performance indicator and examine its uncertainty. The predictive performance of newly-developed integrated models should also be benchmarked to show performance improvement clearly and promote similar applications in future works. The review analyzed the shortcomings, best practices, and prospects of ANNs for MSW-related trend predictions, supporting the realization of practical applications of ANNs to enhance waste management practices and reduce carbon emissions.

Curbing global solid waste emissions toward net-zero warming futures

No global analysis has considered the warming that could be averted through improved solid waste management and how much that could contribute to meeting the Paris Agreement's 1.5° and 2°C pathway goals or the terms of the Global Methane Pledge. With our estimated global solid waste generation of 2.56 to 3.33 billion tonnes by 2050, implementing abrupt technical and behavioral changes could result in a net-zero warming solid waste system relative to 2020, leading to 11 to 27 billion tonnes of carbon dioxide warming-equivalent emissions under the temperature limits. These changes, however, require accelerated adoption within 9 to 17 years (by 2033 to 2041) to align with the Global Methane Pledge. Rapidly reducing methane, carbon dioxide, and nitrous oxide emissions is necessary to maximize the short-term climate benefits and stop the ongoing temperature rise.

An artificial neural network-based data filling approach for smart operation of digital wastewater treatment plants

With the prevalence of digitization, smart operation has become mainstream in future wastewater treatment plants. This requires substantial and complete historical data for model construction. However, the data collected from the front-end sensor contained numerous missing dissolved oxygen (DO) values. Therefore, this study proposed a framework that adaptively adjusted the structure of embedded filling models according to the missing situation. Long short-term memory and gated recurrent units (GRU) were embedded for experiments, and some standard filling methods were selected as benchmarks. The experimental dataset indicated that the K-nearest neighbor could achieve good filling results by traversing the parameters. The effect obtained by the method proposed in this study was slightly better, and GRU was better among the three embedded models. Analysis of the filling results for each DO column revealed that the effect was highly correlated with the dispersion of DO data. The experimental results for the entire dataset demonstrated that the filling effect of the proposed method was significantly better and more stable than the others. The proposed model suffered from the problem of insufficient interpretability and long training time. This study provides an efficient and practical method to solve the intricate missing DO and lays the foundation for the smart operation of wastewater treatment plants.

Optimization of energy production from biogas fuel in a closed landfill using artificial neural networks: A case study of Al Ghabawi Landfill, Jordan

Landfills have high potency as renewable energy sources by producing biogas from organic waste degradation. Landfills biogas (LFG) can be used for power plant purposes instead of allowing it to flare to the atmosphere which contributes to the global warming. The aim of this work was to introduce and examine an optimization model for maximizing the power generation of Al Ghabawi landfill in Amman city, Jordan. The optimization process focused on studying the effect of several operating parameters within the landfill power plant. To achieve this goal, a combustion model had been built and validated against a set of historical real data obtained from the landfill operator. In addition to that, an Artificial Neural Network (ANN) model had been built to perform a multi-objective optimization to obtain the optimal power generation conditions for Al Ghabawi landfill. The combustion model along with the ANN model aim to estimate the best engine operating conditions based on the actual daily data of the landfill. The engine operating parameters includes the intake pressure and temperature, the ignition time and the equivalence ratio. The results of the study indicate that the current operating parameters can be optimized to maximize the gensets power generation. Based on the daily data of the produced LFG, the optimal operating conditions for the landfill are 2.32 bar for the intake pressure, 303 K for the intake temperature, 0.9-1.0 for the equiveillance ratio and for the ignition time it is 13 degrees before the top dead center (BTDC). These optimized operating parameters can maximize the landfill power generation by at least 1 MW for each genset.

Discrete Missing Data Imputation Using Multilayer Perceptron and Momentum Gradient Descent

Data are a strategic resource for industrial production, and an efficient data-mining process will increase productivity. However, there exist many missing values in data collected in real life due to various problems. Because the missing data may reduce productivity, missing value imputation is an important research topic in data mining. At present, most studies mainly focus on imputation methods for continuous missing data, while a few concentrate on discrete missing data. In this paper, a discrete missing value imputation method based on a multilayer perceptron (MLP) is proposed, which employs a momentum gradient descent algorithm, and some prefilling strategies are utilized to improve the convergence speed of the MLP. To verify the effectiveness of the method, experiments are conducted to compare the classification accuracy with eight common imputation methods, such as the mode, random, hot-deck, KNN, autoencoder, and MLP, under different missing mechanisms and missing proportions. Experimental results verify that the improved MLP model (IMLP) can effectively impute discrete missing values in most situations under three missing patterns.

A Method of Pruning and Random Replacing of Known Values for Comparing Missing Data Imputation Models for Incomplete Air Quality Time Series

The data obtained from air quality monitoring stations, which are used to carry out studies using data mining techniques, present the problem of missing values. This paper describes a research work on missing data imputation. Among the most common methods, the method that best imputes values to the available data set is analysed. It uses an algorithm that randomly replaces all known values in a dataset once with imputed values and compares them with the actual known values, forming several subsets. Data from seven stations in the Silesian region (Poland) were analyzed for hourly concentrations of four pollutants: nitrogen dioxide (NO2), nitrogen oxides (NOx), particles of 10 μm or less (PM10) and sulphur dioxide (SO2) for five years. Imputations were performed using linear imputation (LI), predictive mean matching (PMM), random forest (RF), k-nearest neighbours (k-NN) and imputation by Kalman smoothing on structural time series (Kalman) methods and performance evaluations were performed. Once the comparison method was validated, it was determine that, in general, Kalman structural smoothing and the linear imputation methods best fitted the imputed values to the data pattern. It was observed that each imputation method behaves in an analogous way for the different stations The variables with the best results are NO2 and SO2. The UMI method is the worst imputer for missing values in the data sets.

Application of machine learning algorithms in municipal solid waste management: A mini review

Population growth and the acceleration of urbanization have led to a sharp increase in municipal solid waste production, and researchers have sought to use advanced technology to solve this problem. Machine learning (ML) algorithms are good at modeling complex nonlinear processes and have been gradually adopted to promote municipal solid waste management (MSWM) and help the sustainable development of the environment in the past few years. In this study, more than 200 publications published over the last two decades (2000-2020) were reviewed and analyzed. This paper summarizes the application of ML algorithms in the whole process of MSWM, from waste generation to collection and transportation, to final disposal. Through this comprehensive review, the gaps and future directions of ML application in MSWM are discussed, providing theoretical and practical guidance for follow-up related research.

City classification for municipal solid waste prediction in mainland China based on K-means clustering

Cities in mainland China are usually classified according to geographical locations. This traditional city classification system is limited to relative fixed factors, which lives out a gap in terms of the spatial differences of municipal solid waste (MSW). Developing a more comprehensive city classification system is essential for MSW generation prediction and waste management. In this study, six economic, social and climatic indicators that affect MSW generation: population, per capita GDP (PCGDP), environmental sanitation investment (ESI), average temperature, average precipitation, and average humidity, are selected. Weights were calculated for each indicator using a combination of CRITIC weight method and Pearson correlation coefficient prior to cluster analysis. The k-means clustering algorithm was used to classify all cities into four clusters, which differed significantly in the relationships between MSW generation and influencing factors. The results of Kruskal-Wallis test also show that cities in different clusters show different distributions in terms of the indicators selected. The cross-prediction results of the model further validate the reliability of the clustering results from a quantitative perspective. By establishing a city classification system, cities with similar relationships between MSW generation and influencing factors can be placed into one cluster. The model established in one certain city cluster can be used to predict the MSW generation for cities in the same cluster that lack historical data. This may also help to formulate appropriate regional policies according to different relationships between MSW generation and influencing factors, especially for the four city clusters in the mainland China.

The use of a recurrent neural network model with separated time-series and lagged daily inputs for waste disposal rates modeling during COVID-19

A new modeling framework is proposed to estimate mixed waste disposal rates in a Canadian capital city during the pandemic. Different Recurrent Neural Network models were developed using climatic, socioeconomic, and COVID-19 related daily variables with different input lag times and study periods. It is hypothesized that the use of distinct time series and lagged inputs may improve modeling accuracy. Considering the entire 7.5-year period from Jan 2013 to Sept 2020, multi-variate weekday models were sensitive with lag times in the testing stage. It appears that the selection of input variables is more important than waste model complexity. Models applying COVID-19 related inputs generally had better performance, with average MAPE of 10.1%. The optimized lag times are however similar between the periods, with slightly longer average lag for the COVID-19 at 5.3 days. Simpler models with least input variables appear to better simulate waste disposal rates, and both 'Temp-Hum' (Temperature-Humidity) and 'Temp-New Test' (Temperature-COVID new test case) models capture the general disposal trend well, with MAPE of 10.3% and 9.4%, respectively. The benefits of the use of separated time series inputs are more apparent during the COVID-19 period, with noticeable decrease in modeling error.

Demand gap analysis of municipal solid waste landfill in Beijing: Based on the municipal solid waste generation

Achieving accurate prediction of the Municipal Solid Waste (MSW) generation is essential for the sustainable development of the city. This paper selects Beijing as the research object, building a neural network model based on Grey Relational Analysis and Long and Short-Term Memory (GRA-LSTM), and choosing 14 influencing factors of MSW generation as the input indicators, to realize the effective prediction of MSW generation. Then this study obtains the landfill area in Beijing by using the aforementioned prediction results and the calculation formula of the landfill. Firstly, the GRA method is used to sort the influencing factors of the MSW generation for obtain the key influencing indexes. Secondly, the LSTM model is used to learn features of the key influencing indexes. Finally, the area of Beijing landfill is estimated by the calculation formula of landfill area. The results show that, first of all, the MAPE value of the GRA-LSTM combined model established in this paper is 7.3, and the prediction performance of this model is better than the other seven structural methods. Secondly, the area demand for landfills in Beijing shows an upward trend. At last, this paper put forward relevant suggestions to achieve sustainable urban development and deal with the increase in the MSW generation and the demand for landfills.

Modeling of municipal waste disposal rates during COVID-19 using separated waste fraction models

Municipal waste disposal behaviors in Regina, the capital city of Saskatchewan, Canada have significantly changed during the COVID-19 pandemic. About 7.5 year of waste disposal data at the Regina landfill was collected, verified, and consolidated. Four modeling approaches were examined to predict total waste disposal at the Regina landfill during the COVID-19 period, including (i) continuous total (Baseline), (ii) continuous fraction, (iii) truncated total, and (iv) truncated fraction. A single feature input recurrent neural network model was adopted for each approach. It is hypothesized that waste quantity modeling using different waste fractions and separate time series can better capture disposal behaviors of residents during the lockdown. Compared to the baseline approach, the use of waste fractions in modeling improves both result accuracy and precision. In general, the use of continuous time series over-predicted total waste disposal, especially when actual disposal rates were less than 50 t/day. Compared to the baseline approach, mean absolute error (MAE), mean absolute percentage error (MAPE), and mean square error (MSE) were reduced. The R value increased from 0.63 to 0.79. Comparing to the baseline, the truncated total and the truncated fraction approaches better captured the total waste disposal behaviors during the COVID-19 period, probably due to the periodicity of the weeklong data set. For both approaches, MAE and MAPE were lower than 70 and 22%, respectively. The model performance of the truncated fraction appears the best, with an MAPE of 19.8% and R value of 0.92. Results suggest the uses of waste fractions and separated time series are beneficial, especially if the input set is heavily skewed.

Application of periodic parameters and their effects on the ANN landfill gas modeling

To reach a practical landfill gas management system and to diminish the negative environmental impacts from landfills, accurate methane (CH₄) prediction is essential. In this study, the preprocessing steps including minimizing multicollinearity, removal of outliers, and errors with missing data imputation are applied to enhance the data quality. This study is the first at employing periodic parameters in the two-stage non-linear auto-regressive model with exogenous inputs (NARX) with the aim of providing a convenient and precise approach to predict the daily CH₄ collection rate from a municipal landfill in Regina, SK, Canada. Using a stepwise procedure, various volumes of training data were assessed, and concluded that employing the 3-year training data reduced the mean absolute percentage error (MAPE) of the CH₄ prediction model by 26.97% at the testing stage. The favorable artificial neural network model performance was obtained using the day of the year (DOY) as a sole input of the time series model with MAPE of 2.12% showing its acceptable ability in CH₄ prediction. Using an only DOY-based model is especially remarkable because of its simplicity and high accuracy showing a convenient and effective approach in time landfill gas modeling, particularly for the landfills with no reliable climatic data.