Search

The multi-energy system (MES) provides a good environment for the local consumption of renewable energy such as wind and solar power because of its high operational flexibility. In the MES, the hybrid energy storage system (HESS) composed of the battery and thermal storage tank plays an important role in enhancing reliability, economics, and operational flexibility. Hence, determining the optimal size of HESS in the MES is a critical problem but has not received enough attention. In light of this problem, this paper focuses on the optimal HESS planning problem in the community MES (CMES) under diverse uncertainties. Firstly, a two-stage stochastic planning model is proposed for the CMES to coordinate the optimal long-term HESS allocation and the short-term system operation. The thermal inertia in the heating network, space heating demand, and domestic hot water demand is utilized to reduce both the planning and operational cost. Secondly, a deterministic equivalence is proposed for the two-stage planning model to convert it into a mixed-integer linear programming model, which is then solved by off-the-shelf solvers. Finally, simulation results verify the effectiveness of the proposed method. The results reveal that the HESS can enhance the operational flexibility of the CMES but only needs a very few investment costs and prove that the thermal inertia in the CMES can reduce the investment cost of HESS, the fuel, and the operational maintenance cost.

To satisfy the residential cooling demand in tropical large renewable microgrids (MGs), multiple ice-storage air conditionings (IACs) are usually used to supply cold energy. However, these IACs could not directly exchange the cold energy, which seriously limits the combined regulation effect. To address the problem, a novel cooling and power combined supply system (CPCSS) is proposed in this paper, where the trucks are used to distribute the ices among IACs. To evaluate the performance on the multi-time scales, a cooling and power combined supply model is established. On this basis, a two-stage (i.e., day-ahead and real-time stage) scheduling strategy is presented to minimize the MG operation cost. To speed up the solution, this scheduling model is first linearized as a classic mixed-integer second-order cone programming (MISCP) problem, and then is solved by Lagrange solution method. Simulation studies on an IEEE 14-node system indicates that compared with the existing MG operation cost, the annual operational cost of the based-two-truck CPCSS could be reduced by 9868800 yuan (12.6%), while the network loss could be decreased from 5.4 to 3.9 MWh. The results confirm the effectiveness of the proposed strategy.

The integrated energy system can realize the coupling and complementation of various energy sources such as cold, heat and electricity, and plays an important role in the consumption of renewable energy. This paper proposes a bi-level optimal design method for integrated energy system from both economic and carbon emissions aspects. The upper model aims at maximizing the system economy and optimizes the selection and capacity allocation of renewable energy power generation, storage and conversion equipment to meet the demands in the region. The lower model aims at maximizing the environment-protection performance and minimizes the system’s carbon emissions. Since the lower model contains binary variables to characterize trading states and charging and discharging states, the model cannot be transformed into a mathematical program with equilibrium constraints. To effectively handle this problem, the reformulation and decomposition method is adopted. Case studies show that this bi-level model can effectively consider the influence of the objective function in the lower model on the optimal capacity configuration in the upper model, avoid the influence of different objective weights when the multi-objective model is converted to the single-objective model, and obtain the global optimal solution.

In low latitudes, ice storage air conditioners (ISACs) have been widely used to cool while locally responding to the distribution network demand. However, due to the lack of the direct cold energy exchange between ISACs, the cooling load could only be shifted on the time scale instead of the space scale, resulting in an unsatisfactory regulation result. To address the above problem, this article proposes a novel collaborative expansion planning scheme for integrated cooling and power system. Firstly, to increase the regulation flexibility, a novel cold energy supply system is designed, where ice making stations and trucks are used to produce and deliver ices to multiple terminal ISACs. On this basis, taking the capacity of the wind turbines (WTs), large ice makers (LIMs), and trucks as configuration decisions, an optimal expansion planning model is established considering wind generation uncertainties. This model is converted into a classic mixed-integer second-order cone programming (MISOCP) problem using linear techniques, and efficiently solved by the Benders decomposition method. Finally, Shapley value method in economics is used to fairly distribute the revenues between the grid operator (GO) and ISAC owners. Simulation studies on IEEE 14-node distribution network indicate the proposed expansion model is effective and beneficial.

The hybrid neural network model proposed in this paper consists of two main parts: extracting local features of text vectors by convolutional neural network, extracting global features related to text context by BiLSTM, and fusing the features extracted by the two complementary models. In this paper, the pre-processed sentences are put into the hybrid neural network for training. The trained hybrid neural network can automatically classify the sentences. When testing the algorithm proposed in this paper, the training corpus is Word2vec. The test results show that the accuracy rate of text categorization reaches 94.2%, and the number of iterations is 10. The results show that the proposed algorithm has high accuracy and good robustness when the sample size is seriously unbalanced.

The problem of fossil energy shortage and pollution has become increasingly prominent. In China, government-led electric energy substitution projects are being carried out in nationwide, aiming at the end of energy consumption links, using electricity instead of coal-fired and oil-fired energy consumption, in order to optimize the energy structure and achieve the goal of energy conservation and emission reduction. Aiming at the planning and benefit sharing of electric energy substitution project, this research first designs the planning mechanism of electric energy substitution for productive users, analyses distribution network planning, transformation subsidy and so on, and puts forward three schemes. Secondly, based on the cooperative game theory, the benefit allocation model of electric energy substitution project is established. Finally, we take the flue-cured tobacco project about replacing coal with electricity in three areas of Yunnan province as an example to analyze, optimize the game strategy, and put forward three indicators to measure the benefits of the participants in the project. The pollution reduction situation that can not be converted into economic benefits is counted. The modeling results show that the density of users geographically will directly affect the cost of the distribution network transformation. The coal flue-cured tobacco heat loss and serious pollution, after optimization, power grid enterprises can quickly recover investment.

The rise of blockchain technology has injected new vitality into the energy market. At present, blockchain projects for distributed photovoltaic power generation and carbon trading are rapidly emerging, but due to the barrier of two markets and the independent operation of such projects in two chains, the two market values cannot be two-way circulation, which is not conducive to the expansion of power trading business and the scale of carbon market. This paper proposed a joint operation mechanism of cross-chain trading, combined distributed photovoltaic power generation market and the carbon market by the blockchain technology. The novelty is to construct two chains which includes the mainchain and sidechain that enables the two markets to share data and circulate value, and to design a two-way anchoring method that achieves equating between carbon trading and electricity trading by cryptocurrency. The simulation studies took the improved IEEE 33 system as an example to simulate transactions on the Ethereum platform, and technically verify the feasibility of cross-chain transactions in these two markets, and use Ether as the settlement currency.

The cross-bonding is the most common shield/sheath-bonding system now in use on single-conductor shielded power cables for long distance power transmission. But the complex bonding brings great challenges to the assessment of cable ageing conditions. This paper proposes a non-intrusive ageing assessment method using the transient disturbances originating from the system. The cable relative permittivities, instead of the traditional dielectric loss angle, were extracted from the responses of the transient disturbances to character the ageing state more sensitively. The decoupling analysis of the cross-bonded cable and the corresponding relative permittivity identification model were investigated. Since the transient disturbances appear naturally and frequently, a continuous tracking of the cable ageing conditions can be established without interferences to the power system. Moreover, only the magnitudes rather than the phase angles of the cable currents/voltages are required in this method, which effectively simplifies the functions of the signal acquisition devices. The proposed method provides a new idea to assess the ageing conditions of the cross-bonded cables, and has the potential to be applied in industry.

Deep trackers have proven success in visual tracking. Typically, these trackers employ optimally pre-trained deep networks to represent all diverse objects with multi-channel features from some fixed layers. The deep networks employed are usually trained to extract rich knowledge from massive data used in object classification and so they are capable to represent generic objects very well. However, these networks are too complex to represent a specific moving object, leading to poor generalization as well as high computational and memory costs. This paper presents a novel and general framework termed channel distillation to facilitate deep trackers. To validate the effectiveness of channel distillation, we take discriminative correlation filter (DCF) and ECO for example. We demonstrate that an integrated formulation can turn feature compression, response map generation, and model update into a unified energy minimization problem to adaptively select informative feature channels that improve the efficacy of tracking moving objects on the fly. Channel distillation can accurately extract good channels, alleviating the influence of noisy channels and generally reducing the number of channels, as well as adaptively generalizing to different channels and networks. The resulting deep tracker is accurate, fast, and has low memory requirements. Extensive experimental evaluations on popular benchmarks clearly demonstrate the effectiveness and generalizability of our framework.

The multi-energy system (MES) is believed to have bright prospects in the future for its advantages of high energy efficiency, flexible operation condition, and environmentally friendly. The heating network and the heat load play an important role in the MES and have great potentials in improving the system performance. In this paper, an optimal planning method is proposed for the MES that considers the thermal storage capacity of the heating network and the heat load. The objective includes the investment cost, the fuel cost, the grid cost, the maintenance cost, and the environmental cost. The constraints of the cogeneration, the heating network, and the heat load are all taken into consideration. The simulation based on the typical working conditions in annual operation is performed to verify the effectiveness of the proposed planning method. Four cases are set in order to comprehensively investigate the effect of the thermal storage capacity of the heating network and the heat load on the MES planning. The results indicate that the proposed method can decrease the capacity of the devices and reduce the fuel cost.

The microgrid is considered to have bright prospects in the future for its advantages of high economic benefits and environmentally friendly and alleviate the antinomy for energy’s supply and demand. Due to the introduction of renewable energy such as wind power into the microgrid, the intermittent, random, and unpredictability of its output makes it difficult to control the frequency when the microgrid operates in the isolated islands. In order to reduce the frequency fluctuation, this paper proposes a secondary frequency control strategy for isolated microgrid based on the linear active disturbance rejection control (LADRC) technology. Extended state observer (ESO) is used to estimate the extended state online and compensate the disturbance estimation actively so that the disturbance can be eliminated quickly and the frequency stability can be maintained. In addition, considering the demand response, the influence of different demand response coefficients on frequency control is analyzed. By a series of trial declarations, the LADRC control strategy can effectively suppress the frequency oscillation, and after adding the demand responses, the response speed can be faster than the traditional frequency modulation, reflecting its better robustness and stability. The simulation based on the frequency response model of AC microgrid results shows the effectiveness of the proposed method.

When combining very different images which often contain complex objects and backgrounds, producing consistent compositions is a challenging problem requiring seamless image editing. In this paper, we propose a general approach, called object-aware image editing, to obtain consistency in structure, color, and texture in a unified way. Our approach improves upon previous gradient-domain composition in three ways. Firstly, we introduce an iterative optimization algorithm to minimize mismatches on the boundaries when the target region contains multiple objects of interest. Secondly, we propose a mixed-domain consistency metric for measuring gradients and colors, and formulate composition as a unified minimization problem that can be solved with a sparse linear system. In particular, we encode texture consistency using a patch-based approach without searching and matching. Thirdly, we adopt an object-aware approach to separately manipulate the guidance gradient fields for objects of interest and backgrounds of interest, which facilitates a variety of seamless image editing applications. Our unified method outperforms previous state-of-the-art methods in preserving global texture consistency in addition to local structure continuity.

Combined cooling, heating and power (CCHP) microgrid has the advantage of high energy utilization efficiency. The fluctuation of renewable energy sources and multiple load demands challenges the economic operation of CCHP microgrid. In this paper, we propose a novel two-stage coordinated control approach for CCHP microgrid energy management, which consists of two stages: the economic dispatching stage (EDS) and the real-time adjusting stage (RTAS). In EDS, it utilizes a model predictive control incorporating piecewise linear efficiency curves to schedule the operation based on the forecast information. In RTAS, the schedule obtained in EDS is adjusted based on the real-time information to tackle the power fluctuations. A typical-structure CCHP microgrid is analyzed in the case study and simulation results are presented to demonstrate the performance of the proposed two-stage coordinated control approach.

Energy management is facing new challenges due to the increasing supply and demand uncertainties, which is caused by the integration of variable generation resources, inaccurate load forecasts and non-linear efficiency curves. To meet these challenges, a robust optimization method incorporating piecewise linear thermal and electrical efficiency curve is proposed to accommodate the uncertainties of cooling, thermal and electrical load, as well as photovoltaic (PV) output power. Case study results demonstrate that the robust optimization model performs better than the deterministic optimization model in terms of the expected operation cost. The fluctuation of net electrical load has greater effect on the dispatching results of the combined cooling, heating and power (CCHP) microgrid than the fluctuation of the cooling and thermal load. The day-ahead schedule is greatly affected by the uncertainty budget of the load demand. The economy of the optimal decision could be achieved by adjusting different uncertainty budget levels according to control the conservatism of the model.

Dissolved gas analysis (DGA) is the most important tool for fault diagnosis in electric power transformers. To improve accuracy of diagnosis, this paper proposed a new model (SDAE-LSTM) to identify the dissolved gases in the insulating oil of power transformers and perform parameter analysis. The performance evaluation is attained by the case studies in terms of recognition accuracy, precision ratio, and recall ratio. Experiment results show that the SDAE-LSTM model performs better than other models under different input conditions. As evidenced from the analyses, the proposed model achieves considerable results of recognition accuracy (95.86%), precision ratio (95.79%), and recall ratio (97.51%). It can be confirmed that the SDAE-LSTM model using the dissolved gas in the power transformer for fault diagnosis and analysis has great research prospect.

In order to reduce the impact of the randomness and volatility of renewable energy on the economic operation of AC/DC hybrid microgrids, a multi-time-scale rolling optimization strategy is proposed for the grid-connected AC/DC hybrid microgrids. It considers the source-load uncertainty declined with time scale reduction, and the scheduling cooperation problem of different units on different time scales. In this paper, we propose a three-time-scale optimal strategy of the day-ahead, intraday and real-time dispatching stage and a two-level rolling optimal strategy of the intraday and real-time stage, aiming at minimizing the operating cost. We added the power penalty cost in the rolling optimization model to limit the energy state of the energy storage system in the constraint, and improve the power correction and tracking effect of the rolling optimization. A typical-structure AC/DC hybrid microgrid is analyzed in this paper and the simulation results are shown to demonstrate the feasibility and effectiveness of the proposed multi-time-scale rolling optimal dispatch.

In order to attract more flexible resource to take part in integrated demand response (IDR), this can be realized by introducing load aggregator-based framework. In this paper, based on residential smart energy hubs (S.E. Hubs), a two-level IDR framework is proposed, in which S.E. Hub operators play the role of load aggregators. The framework includes day-ahead bidding and real-time scheduling. In day-ahead bidding, S.E. Hub operators have to compete dispatching amount for maximal profit; hence, noncooperative game approach is formulated to describe the competition behavior among operators. In real-time scheduling, the dispatching model is formulated to minimize the error between real-time scheduling amount and bidding amount. Moreover, in order to reduce the influence of IDR on residential users, 4 categories of users’ flexible loads are modeled according to load consumption characteristic, and then these models are considered as the constraints in real-time scheduling. A case study is designed to validate the effectiveness of the proposed two-level IDR framework. And simulation results confirm that smart grid, S.E. Hub operators, and residential users can benefit simultaneously.

The pumped storage power station has the characteristics of frequency-phase modulation, energy saving, and economy, and has great development prospects and application value. In order to cope with the large-scale integration and intermittency of renewable energy and improve the ability of pumped storage units to participate in power grid frequency modulation, this paper proposed a load frequency control (LFC) strategy for pumped storage units based on linear active disturbance rejection technology. Firstly, based on the operating characteristics of the pumped storage power station, the LFC model of the two-area reheat steam turbine under nonlinear conditions such as governor dead zone and generation rate constrains was established. Secondly, a second-order linear active disturbance rejection control (LADRC) was designed. The feasibility and control performance of the proposed LFC system were quantitatively analyzed through simulation. The results show that the LADRC has better control effect and stronger robustness than fractional-order proportion integration differentiation (FOPID) and traditional proportion integration differentiation (PID) controller. Finally, the pumped storage power station was added, and it was found that it has better correction performance under both generating and pumping operations, which greatly improved the dynamic response of secondary frequency modulation.

With renewable generation resources and multiple load demands increasing, the combined cooling, heating, and power (CCHP) microgrid energy management system has attracted much attention due to its high efficiency and low emissions. In order to realize the integration of substation resources and solve the problems of inaccurate, random, volatile and intermittent load forecasting, we propose a three-stage coordinated optimization scheduling strategy for a CCHP microgrid. The strategy contains three stages: a day-ahead economic scheduling stage, an intraday rolling optimization stage, and a real-time adjustment stage. Forecasting data with different accuracy at different time scales were used to carry out multilevel coordination and gradually improve the scheduling plan. A case study was used to verify that the proposed scheduling strategy can mitigate and eliminate the load forecasting error of renewable energy (for power balance and scheduling economy).

Recent deep neural networks have achieved impressive performance in image classification. However, these networks are sensitive to the attack of adversarial examples, leading to a sharp drop in accuracy. To address this issue, this paper proposes a learning approach to improve the robustness by distilling deep neural networks with soft decision trees. This approach learns a decision tree in a softening manner by jointly using data and the predictions of a well-trained deep neural network. In this way, the resulting soft decision tree can distil the knowledge from deep neural network when preserving the efficiency of decision tree. Experimental results show that the proposed approach has better robustness again adversarial examples than deep neural networks and decision trees.

Recent deep trackers have achieved impressive performance in visual tracking. Typically, these trackers apply complex deep networks with massive parameters for representing objects, which makes their deployment on resource-limited devices very challenging due to two reasons: (1) high computation complexity, and (2) high storage footprint. To address these two issues, this paper proposes a lightweight deep tracker to facilitate efficient tracking by using a small deep convolutional neural network. This tracker adaptively extracts features of interest from different deep layers for representing different objects, and then integrates into discriminative correlation filter formulation. Due to the usage of small deep networks and selection of deep features, the drop on tracking accuracy could be effectively alleviated, while the costs in computation and storage could be greatly reduced. This tracker can run at a very fast speed of 55 fps when only taking 4.8M parameters. Experimental results on the public OTB2013 and OTB100 benchmarks demonstrate the effectiveness and efficiency of the proposed tracker.

The deployments of deep neural network models on mobile or embedded devices have been challenged due to two main reasons: 1) the large model size for storage, and 2) the large memory bandwidth for inference. To address these issues, this paper develops a deep neural network compression framework to reduce the resource usage for efficient visual inference. By reviewing the trained deep model, we propose a hybrid model compression algorithm via four major modules. Approximation module reduces the number of weights in each fully connected layer with low rank approximation. Then, quantization module analyzes weight distribution in each layer and represents them with low precision fixed point, which reduces the bits for storing each weight. After that, pruning module suppresses small weights to further reduce the number of parameters. Finally, coding module joint optimizes the representation and encoding of the sparse structure of the pruned weights with relative index by Huffman coding. Beyond the compression of model size, we propose an adaptive fixed point memory allocation algorithm to reduce memory footprint in inference. The proposed framework, along with the model compression and memory allocation algorithms, can provide 20–30x compression rate with negligible accuracy loss. We conduct an evaluation on two representative models, AlexNet and VGG-16, for object recognition and face verification tasks, which demonstrate the effectiveness of our proposed compression framework.

Detecting faces with occlusions is a challenging task due to two main reasons: 1) the absence of large datasets of masked faces, and 2) the absence of facial cues from the masked regions. To address these two issues, this paper first introduces a dataset, denoted as MAFA, with 30, 811 Internet images and 35, 806 masked faces. Faces in the dataset have various orientations and occlusion degrees, while at least one part of each face is occluded by mask. Based on this dataset, we further propose LLE-CNNs for masked face detection, which consist of three major modules. The Proposal module first combines two pre-trained CNNs to extract candidate facial regions from the input image and represent them with high dimensional descriptors. After that, the Embedding module is incorporated to turn such descriptors into a similarity-based descriptor by using locally linear embedding (LLE) algorithm and the dictionaries trained on a large pool of synthesized normal faces, masked faces and non-faces. In this manner, many missing facial cues can be largely recovered and the influences of noisy cues introduced by diversified masks can be greatly alleviated. Finally, the Verification module is incorporated to identify candidate facial regions and refine their positions by jointly performing the classification and regression tasks within a unified CNN. Experimental results on the MAFA dataset show that the proposed approach remarkably outperforms 6 state-of-the-arts by at least 15.6%.

Combined cooling, heating, and power (CCHP) systems have been widely applied in various kinds of buildings. Most operation strategies for CCHP microgrids are designed based on day-ahead profiles. However, prediction error for renewable energy resources (RES) and load leads to suboptimal operation in dispatch scheduling. In this paper, we propose an online optimal operation approach for CCHP microgrids based on model predictive control with feedback correction to compensate for prediction error. This approach includes two hierarchies: 1) rolling optimization; and 2) feedback correction. In the rolling part, a hybrid algorithm based on integrating time series analysis and Kalman filters is used to forecast the power for RES and load. A rolling optimization model is established to schedule operation according to the latest forecast information. The rolling dispatch scheduling is then adjusted based on ultra-short-term error prediction. The feedback correction model is applied to minimize the adjustments and to compensate for prediction error. A case study demonstrates the effectiveness of the proposed approach with better matching between demand and supply.

The deployments of deep neural network models on mobile or embedded devices have been hindered due to their large number of weights. In this work, we develop a deep neural network (DNN) model compression service termed MicroBrain to reduce the resource usage for energy-efficient visual inference. By automatically analyzing the trained DNN models, we propose a high-performance DNN model compression approach to perform resource control via four modules. The proposed service, along with the compression approach, can provide 20–30x compression rate with negligible accuracy loss to condense DNN models, which facilitates their deployments on mobile devices for energy-efficient visual inference. We conduct an evaluation on two representative models, AlexNet and VGG-16, for object recognition and face verification tasks, which demonstrate the effectiveness of our proposed approach.

The deployments of current object trackers on mobile or embedded devices are hindered due to high computation complexity and high resource footprint. To address these issues, this paper proposes an ultra-fast and energyefficient object tracker, to perform accurate and robust object tracking at a speed of more than 56fps with a memory cost of less than 40KB and a clock rate of 480MHz via a unified feature representation and matching. First, we propose a compact feature representation for describing the object, which utilizes the response distribution to provide sparse representation. In this manner, the object can be represented with a low-dimensional and fixed-point vector. Further, we propose a fast feature matching method to jointly perform object verification and regression. Experiments on the public OTB benchmark and our collected weak object tracking (WTB) dataset demonstrate the effectiveness of our proposed tracker.

The performance of face detection in heavy occlusion conditions (e.g., occluded by masks) will drop dramatically since facial cue is missing or polluted by occluders. To address this issue, we propose a novel method by cascaded outrageous proposal generation and deep convolutional neural network (DCNN) via two major modules. First, the proposal generation module assembles many fast face detectors with low threshold to generate massive candidate proposals in a high recall rate. Then, the verification module, which is trained with a ResNet network, jointly performs face classification and regression tasks to identifies a proposal to be real face or not and refine its location respectively. Experiments on OFD-UCI and our collected masked face dataset MAFA demonstrate that our approach effectively outperforms the three representative face detectors by at least 3.3% and 11.6% in detecting crowd occluded faces and masked faces respectively.

From the Internet of Energy point of view, adjacent building combined cooling, heating and power with microgrids (BCCHP microgrids) should be interconnected, which will generate better benefits, both economically and environmentally, from scarce energy resources through intelligent coordination. The interconnected-BCCHP (IBCCHP) microgrids will not only support energy with each other but also bring better benefits compared to the non-interconnected-BCCHP (NBCCHP) microgrids. A mixed-integer linear programming (MILP) economic dispatching approach incorporating piecewise linear efficiency curves model is proposed to compares the performance of IBCCHP microgrids and NBCCHP microgrids. Three buildings with CCHP microgrids in Shanghai are studied as an example, and simulation results are presented to demonstrate that the primary energy consumption (PEC), the carbon dioxide emissions (CDE) and the operation cost of the proposed IBCCHP microgrids are better than the NBCCHP microgrids, and the overall lifetime of the complete battery is improved. Furthermore, a sensitivity analysis is presented to provide various policy recommendations to promote the interconnection among buildings with the change of electricity prices and gas prices.

The increasing penetration of distributed generators (DGs) makes further interconnection of multiple energy carriers possible. In this paper, an integration model comprising IEEE-33 bus distribution network and a correspondingly designed water system is proposed with DG coupling of thermal and electrical energy. The integrated energy system incorporates the constraints of the distribution network, water system, and DGs and ensures that the electrical and thermal demands of the integrated energy system are satisfied simultaneously by energy flow. An optimal procedure including optimal sizing, positioning, and operation of DGs and gas boilers in a given system was implemented while minimizing the total costs of investment and operation. The model constitutes a mixed integer nonlinear problem, as it takes the power flow calculation and the nonlinear characteristics of DG efficiency into consideration. Therefore, the particle swarm optimization algorithm is employed to simulate the proposed model. The simulation results in conjunction with numerical studies reveal the impact of the water system on the optimal configuration of DGs and the advantages of the integrated energy system on economic considerations, voltage level, and net loss.