Learning by mistakes in memristor networks.

Recent results revived the interest in the implementation of analog devices able to perform brainlike operations. Here we introduce a training algorithm for a memristor network which is inspired by previous work on biological learning. Robust results are obtained from computer simulations of a network of voltage-controlled memristive devices. Its implementation in hardware is straightforward, being scalable and requiring very little peripheral computation overhead.

Benchmarking Soft Sensors for Remote Monitoring of On-Site Wastewater Treatment Plants.

On-site wastewater treatment plants (OSTs) are usually unattended, so failures often remain undetected and lead to prolonged periods of reduced performance. To stabilize the performance of unattended plants, soft sensors could expose faults and failures to the operator. In a previous study, we developed soft sensors and showed that soft sensors with data from unmaintained physical sensors can be as accurate as soft sensors with data from maintained ones. The monitored variables were pH and dissolved oxygen (DO), and soft sensors were used to predict nitrification performance. In the present study, we use synthetic data and monitor three plants to test these soft sensors. We find that a long solids retention time and a moderate aeration rate improve the pH soft-sensor accuracy and that the aeration regime is the main operational parameter affecting the accuracy of the DO soft sensor. We demonstrate that integrated design of monitoring and control is necessary to achieve robustness when extrapolating from one OST to another in the absence of plant-specific fine-tuning. Additionally, we provide a unique labeled dataset for further feature and data-driven soft-sensor development. Our benchmarking results indicate that it is feasible to monitor OSTs with unmaintained sensors and without plant-specific tuning of the developed soft sensors. This is expected to drastically reduce monitoring costs for OST-based sanitation systems.

Modelling quantities and qualities (Q&Q) of faecal sludge in Hanoi, Vietnam and Kampala, Uganda for improved management solutions.

The importance of faecal sludge management is gaining recognition. However, methods are still lacking to reasonably estimate the quantities and qualities (Q&Q) that need to be managed, which makes the planning for and implementing of management solutions quite difficult. The objective of this study was to collect and analyse Q&Q of faecal sludge at a citywide scale, and to evaluate whether "SPA-DET" data (=> spatially analysable - demographic, environmental and technical) could then be used as predictors of Q&Q of faecal sludge. 60 field samples and questionnaires from Hanoi and 180 from Kampala were analysed. Software tools were used in an iterative process to predict total solids (TS) and emptying frequency in both Hanoi, Vietnam and Kampala, Uganda. City-specific data could be predicted with types of "SPA-DET" data as input variables, and model performance was improved by analysing septic tanks and pit latrines separately. Individual models were built for TS concentrations and emptying frequency. In addition, a model was built across both cities for emptying frequency of septic tanks based on number of users and containment volume, indicating predictive models can be relevant for multiple cities. Number of users, containment volume, truck volume and income level were identified as the most common variables for the correction function. Results confirm the high intrinsic variability of faecal sludge characteristics, and illustrate the importance of moving beyond simple reporting of city-wide average values for estimations of Q&Q. The collected data and developed scripts have been made available for replication in future studies.

Characterizing long-term wear and tear of ion-selective pH sensors.

Today, the development and testing of methods for fault detection and identification in wastewater treatment research relies on two important assumptions: (i) that sensor faults appear at distinct times in different sensors and (ii) that any given sensor will function near-perfectly for a significant amount of time following installation. In this work, we show that such assumptions are unrealistic, at least for sensors built around an ion-selective measurement principle. Indeed, long-term exposure of sensors to treated wastewater shows that sensors exhibit fault symptoms that appear simultaneously and with similar intensity. Consequently, this suggests that future research should be reoriented towards methods that do not rely on the assumptions mentioned above. This study also provides the first empirically validated sensor fault model for wastewater treatment simulation, which is useful for effective benchmarking of both fault detection and identification methods and advanced control strategies. Finally, we evaluate the value of redundancy for remote sensor validation in decentralized wastewater treatment systems.

Beyond signal quality: The value of unmaintained pH, dissolved oxygen, and oxidation-reduction potential sensors for remote performance monitoring of on-site sequencing batch reactors.

Sensor maintenance is time-consuming and is a bottleneck for monitoring on-site wastewater treatment systems. Hence, we compare maintained and unmaintained sensors to monitor the biological performance of a small-scale sequencing batch reactor (SBR). The sensor types are ion-selective pH, optical dissolved oxygen (DO), and oxidation-reduction potential (ORP) with platinum electrode. We created soft sensors using engineered features: ammonium valley for pH, oxidation ramp for DO, and nitrite ramp for the ORP. Four soft sensors based on unmaintained pH sensors correctly identified the completion of the ammonium oxidation (89-91 out of 107 cycles), about as many times as soft sensors based on a maintained pH sensor (91 out of 107 cycles). In contrast, the DO soft sensor using data from a maintained sensor showed slightly better (89 out of 96 cycles) detection performance than that using data from two unmaintained sensors (77, respectively 82 out of 96 correct). Furthermore, the DO soft sensor using maintained data is much less sensitive to the optimisation of cut-off frequency and slope tolerance than the soft sensor using unmaintained data. The nitrite ramp provided no useful information on the state of nitrite oxidation, so no comparison of maintained and unmaintained ORP sensors was possible in this case. We identified two hurdles when designing soft sensors for unmaintained sensors: i) Sensors' type- and design-specific deterioration affects performance. ii) Feature engineering for soft sensors is sensor type specific, and the outcome is strongly influenced by operational parameters such as the aeration rate. In summary, the results with the provided soft sensors show that frequent sensor maintenance is not necessarily needed to monitor the performance of SBRs. Without sensor maintenance monitoring small-scale SBRs becomes practicable, which could improve the reliability of unstaffed on-site treatment systems substantially.

Parameter estimation of hydrologic models using a likelihood function for censored and binary observations.

Observations of a hydrologic system response are needed to accurately model system behaviour. Nevertheless, often very few monitoring stations are operated because collecting such reference data adequately and accurately is laborious and costly. It has been recently suggested to use observations not only from dedicated flow meters but also from simpler sensors, such as level or event detectors, which are available more frequently but only provide censored information. Binary observations can be considered as extreme censoring. It is still unclear, however, how to use censored observations most effectively to learn about model parameters. To this end, we suggest a formal likelihood function that incorporates censored observations, while accounting for model structure deficits and uncertainty in input data. Using this likelihood function, the parameter inference is performed within the Bayesian framework. We demonstrate the implementation of our methodology on a case study of an urban catchment, where we estimate the parameters of a hydrodynamic rainfall-runoff model from binary observations of combined sewer overflows. Our results show, first, that censored observations make it possible to learn about model parameters, with an average decrease of 45% in parameter standard deviation from prior to posterior. Second, the inference substantially improves model predictions, providing higher Nash-Sutcliffe efficiency. Third, the gain in information largely depends on the experimental design, i.e. sensor placement. Given the advent of Internet of Things, we foresee that the plethora of censored data promised to be available can be used for parameter estimation within a formal Bayesian framework.

Memristor models for machine learning.

In the quest for alternatives to traditional complementary metal-oxide-semiconductor, it is being suggested that digital computing efficiency and power can be improved by matching the precision to the application. Many applications do not need the high precision that is being used today. In particular, large gains in area and power efficiency could be achieved by dedicated analog realizations of approximate computing engines. In this work we explore the use of memristor networks for analog approximate computation, based on a machine learning framework called reservoir computing. Most experimental investigations on the dynamics of memristors focus on their nonvolatile behavior. Hence, the volatility that is present in the developed technologies is usually unwanted and is not included in simulation models. In contrast, in reservoir computing, volatility is not only desirable but necessary. Therefore, in this work, we propose two different ways to incorporate it into memristor simulation models. The first is an extension of Strukov's model, and the second is an equivalent Wiener model approximation. We analyze and compare the dynamical properties of these models and discuss their implications for the memory and the nonlinear processing capacity of memristor networks. Our results indicate that device variability, increasingly causing problems in traditional computer design, is an asset in the context of reservoir computing. We conclude that although both models could lead to useful memristor-based reservoir computing systems, their computational performance will differ. Therefore, experimental modeling research is required for the development of accurate volatile memristor models.

A computational analysis of motor synergies by dynamic response decomposition.

Analyses of experimental data acquired from humans and other vertebrates have suggested that motor commands may emerge from the combination of a limited set of modules. While many studies have focused on physiological aspects of this modularity, in this paper we propose an investigation of its theoretical foundations. We consider the problem of controlling a planar kinematic chain, and we restrict the admissible actuations to linear combinations of a small set of torque profiles (i.e., motor synergies). This scheme is equivalent to the time-varying synergy model, and it is formalized by means of the dynamic response decomposition (DRD). DRD is a general method to generate open-loop controllers for a dynamical system to solve desired tasks, and it can also be used to synthesize effective motor synergies. We show that a control architecture based on synergies can greatly reduce the dimensionality of the control problem, while keeping a good performance level. Our results suggest that in order to realize an effective and low-dimensional controller, synergies should embed features of both the desired tasks and the system dynamics. These characteristics can be achieved by defining synergies as solutions to a representative set of task instances. The required number of synergies increases with the complexity of the desired tasks. However, a possible strategy to keep the number of synergies low is to construct solutions to complex tasks by concatenating synergy-based actuations associated to simple point-to-point movements, with a limited loss of performance. Ultimately, this work supports the feasibility of controlling a non-linear dynamical systems by linear combinations of basic actuations, and illustrates the fundamental relationship between synergies, desired tasks and system dynamics.

Exploiting the passive dynamics of a compliant leg to develop gait transitions.

In the area of bipedal locomotion, the spring-loaded inverted pendulum model has been proposed as a unified framework to explain the dynamics of a wide variety of gaits. In this paper, we present an analysis of the mathematical model and its dynamical properties. We use the perspective of hybrid dynamical systems to study the dynamics and define concepts such as partial stability and viability. With this approach, on the one hand, we identify stable and unstable regions of locomotion. On the other hand, we find ways to exploit the unstable regions of locomotion to induce gait transitions at a constant energy regime. Additionally, we show that simple nonconstant angle of attack control policies can render the system almost always stable.