Method, device and equipment for identifying the operating status of equipment in a sewage-discharging enterprise

Abstract

The present application relates to a method, device and equipment for identifying the operating status of equipment in polluting enterprises. The method generates a real-time electricity consumption data sequence by collecting the electrical parameters of the electrical equipment of the polluting enterprises in real time; performs statistical analysis on the real-time electricity consumption data sequence in the sliding window, and calculates the electricity consumption change rate, power factor and start-stop frequency in the corresponding window; uses a dynamic Bayesian network to build an equipment operating status identification model, and independently builds a sub-model for each process stage to perform parameter learning; inputs multi-dimensional continuous features into the dynamic Bayesian network, and outputs the equipment start-stop status at the current moment; and also includes dynamically adjusting the threshold value used to judge the start-stop status of the equipment. The present application flexibly responds to the dynamic changes in the start-stop status of the equipment, avoids the problem that the fixed threshold value cannot adapt to the fluctuation of the equipment operation, and ensures the accuracy of the equipment start-stop status judgment; avoids the difficulty of selecting the number of clusters, and can better reflect the diversity and complexity of the equipment operation status.

Description

Technical Field

[0001] The present application belongs to the technical field of equipment monitoring and status identification, and specifically relates to a method, device and equipment for identifying the operating status of equipment in a sewage-discharging enterprise. Background Art

[0002] With the continuous advancement of industrialization, environmental regulations have gradually intensified, with the introduction of a series of strict environmental protection policies requiring enterprises to install pollution treatment equipment and discharge pollutants in a scientific manner. To ensure the effective implementation of production suspension and production restriction measures, electricity consumption monitoring technology has emerged. By analyzing a company's electricity consumption data, it is possible to indirectly understand the actual operation of production facilities and pollution control facilities, allowing for the rapid identification and correction of potential pollution violations, significantly improving the efficiency and accuracy of supervision.

[0003] Traditional methods for identifying facility operating status rely primarily on preset power usage thresholds. Once actual power usage data reaches these thresholds, the facility's operating status can be quickly determined. With advances in monitoring technology, clustering algorithms have become widely used in power usage analysis. As an unsupervised learning method, clustering algorithms deeply explore the inherent characteristics of facility power usage data and can automatically classify data with similar power usage patterns into different operating status categories, enabling more accurate and flexible judgment of facility startup and shutdown status.

[0004] While the threshold judgment method is simple and intuitive, its core principle is to use a fixed power usage threshold as a judgment criterion. However, the operating status of a facility is not static but rather changes dynamically with fluctuations in various factors, such as production demand and environmental conditions. Therefore, the threshold judgment method lacks the flexibility to adapt to this dynamic nature, potentially preventing timely and accurate judgments when the actual operating status of a facility deviates from the preset threshold. Furthermore, for certain special conditions, such as when a facility is in standby or low-load operation, power usage data may not be significant, making it difficult to effectively identify using a fixed threshold. Cluster analysis, as a more advanced identification algorithm, has demonstrated certain advantages in dealing with complex and variable power usage data, but it also faces challenges in interpreting the results and determining the right number of clusters. Clustering results are often presented in mathematical form, lacking intuitive interpretation and requiring in-depth interpretation based on practical experience and professional knowledge. Furthermore, determining the optimal number of clusters presents a challenge: too many clusters may lead to information redundancy and difficulty in interpretation, while too few clusters may not accurately reflect the diversity of facility operating conditions. Summary of the Invention

[0005] In view of the above analysis, the embodiments of the present invention aim to provide a method, device and equipment for identifying the operating status of equipment in pollutant discharge enterprises, aiming to solve the challenges faced by the threshold judgment method and cluster analysis method in the existing technology, and provide a more flexible and accurate solution for the dynamic changes of the start and stop status of the equipment and the identification of special operating status, thereby improving the adaptability, accuracy and operability of the system in complex environments.

[0006] In a first aspect of the present application, a method for identifying the operating status of equipment in a pollutant-discharging enterprise is provided, comprising the following steps:

[0007] Real-time collection of electrical parameters of electrical equipment in pollutant-discharging enterprises to generate real-time electricity consumption data series;

[0008] Construct a sliding window, perform statistical analysis on the real-time power consumption data sequence within the sliding window, and calculate the power consumption change rate, power factor, and start-stop frequency within the corresponding window;

[0009] A dynamic Bayesian network is used to build an equipment operation status recognition model. The nodes of the dynamic Bayesian network include the equipment start and stop status at the current moment, current power consumption, power consumption change rate, power factor, start and stop frequency, and the equipment start and stop status at the previous moment;

[0010] The following parameter learning mechanism is used to train the conditional probability distribution of the dynamic Bayesian network: historical data of equipment in pollutant-discharging enterprises is used to identify stages and distinguish them into multiple process stages. A sub-model is independently constructed for each process stage to identify the behavioral characteristics of the equipment in each process stage and perform parameter learning separately.

[0011] Inputting the multi-dimensional continuous features consisting of the real-time power consumption data sequence, the power consumption change rate, the power factor, and the start-stop frequency into a dynamic Bayesian network, performing state inference based on the dynamic Bayesian network, and outputting the start-stop state of the device at the current moment;

[0012] The method further includes dynamically adjusting the threshold used to determine the start and stop status of the device according to changes in the device operating characteristics and identification results.

[0013] Optionally, independently constructing a sub-model for each process stage, identifying behavioral characteristics of the equipment in each process stage, and performing parameter learning separately includes:

[0014] When constructing a sub-model independently in each process stage, the expectation maximization algorithm is used to estimate the conditional probability distribution between the start and stop states of the equipment and other characteristics in the corresponding stage, and the parameters are optimized by expectation maximization.

[0015] Optionally, independently constructing a sub-model for each process stage, identifying behavioral characteristics of the equipment in each process stage, and performing parameter learning separately further includes:

[0016] When performing parameter estimation, obtain labeled device electricity usage sample data and unlabeled device electricity usage sample data;

[0017] The labeled device electricity usage sample data and the unlabeled device electricity usage sample data are input into the device operation status recognition model for semi-supervised learning training.

[0018] Optionally, the stage identification of historical data of equipment of pollutant-discharging enterprises includes:

[0019] A clustering algorithm is used to perform cluster analysis on the historical data to identify different operating stages of the equipment; the operating stages include a working stage, a standby stage, and a cleaning stage.

[0020] Optionally, dynamically adjusting a threshold for determining the start / stop state of a device according to changes in device operation characteristics and recognition results includes:

[0021] Based on the electricity consumption change rate , the following formula is used to dynamically update the threshold used to determine the start and stop status of the device: ;

[0022] in, For the forgetting factor, is the threshold used to judge the start and stop status of the device at the current moment, is the threshold used to judge the start and stop status of the device at the previous moment, is the rate of change of electricity consumption at the current moment;

[0023] Or real-time monitoring of the false alarm rate and missed alarm rate of the equipment operating status of the sewage-discharging enterprises, based on Dynamically adjust the threshold used to determine the start and stop status of the device;

[0024] in, is the threshold used to judge the start and stop status of the device at the current moment, is the threshold used to judge the start and stop status of the device at the previous moment, is the false alarm rate at the current moment, is the target false alarm rate; is the omission rate at the current moment, is the target underreporting rate, , is the adjustment coefficient;

[0025] Or use dynamic Bayesian network to calculate the device is enabled The posterior probability distribution is calculated, and the mean and standard deviation of the posterior probability distribution are calculated. The threshold used to determine the start and stop status of the device is dynamically adjusted as follows: ;

[0026] in, is the threshold used to judge the start and stop status of the device at the current moment, For time series observations arrive Under the condition of , the posterior probability that the device is currently in the enabled state is, is the mean of the posterior probabilities, is the standard deviation of the posterior probability, is the confidence adjustment coefficient.

[0027] Optionally, it also includes:

[0028] Use any of the following strategies to adjust the forgetting factor Make adaptive adjustments:

[0029] Grid search is used within the preset range and combined with cross validation to compare different The recognition accuracy, F1 score or false alarm rate indicators under the value are selected to optimize value;

[0030] Using Bayesian optimization methods, Build a proxy model between the performance indicators and select the best one based on historical recognition performance value;

[0031] Dynamically adjust according to the deviation between the current recognition accuracy and the target accuracy , the update formula is: ;in, is the learning rate, is the current recognition accuracy, is the target accuracy.

[0032] Optionally, it also includes:

[0033] Determining the degree of fluctuation of the electricity consumption data based on an analysis of the real-time electricity consumption data sequence;

[0034] Dynamically adjust the length of the sliding window according to the degree of fluctuation of the electricity consumption data: calculate the degree of fluctuation of the electricity consumption data in the sliding window; when the degree of fluctuation of the electricity consumption data exceeds a set threshold, reduce the sliding window according to a preset step size; when the degree of fluctuation of the electricity consumption data is lower than the set threshold, increase the sliding window according to a preset step size.

[0035] In a second aspect of the present application, a device for identifying the operating status of equipment in a pollutant-discharging enterprise is provided, comprising the following modules:

[0036] The data acquisition module is configured to collect electrical parameters of electrical equipment of pollutant-discharging enterprises in real time and generate a real-time electricity consumption data sequence;

[0037] The statistical calculation module is configured to construct a sliding window, perform statistical analysis on the real-time power consumption data sequence within the sliding window, and calculate the power consumption change rate, power factor, and start-stop frequency within the corresponding window;

[0038] A model building module is configured to build a device operation status recognition model using a dynamic Bayesian network, wherein the nodes of the dynamic Bayesian network include the device start and stop status at the current moment, the current power consumption, the power consumption change rate, the power factor, the start and stop frequency, and the device start and stop status at the previous moment;

[0039] The parameter learning module is configured to train the conditional probability distribution of a dynamic Bayesian network using the following parameter learning mechanism: Historical data of equipment in pollutant-discharging enterprises is used to identify stages and distinguish them into multiple process stages; a sub-model is independently constructed for each process stage to identify the behavioral characteristics of the equipment in each process stage and perform parameter learning on each stage;

[0040] an output module configured to input the multi-dimensional continuous features consisting of the real-time power consumption data sequence, the power consumption change rate, the power factor, and the start-stop frequency into a dynamic Bayesian network, perform state inference based on the dynamic Bayesian network, and output the start-stop state of the device at the current moment;

[0041] The device further includes: a threshold adjustment module configured to dynamically adjust the threshold used to determine the start / stop state of the device according to changes in the device operation characteristics and the recognition results.

[0042] The third aspect of the present application provides a device for identifying the operating status of equipment in a polluting enterprise, comprising a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, it implements any of the above-mentioned methods for identifying the operating status of equipment in a polluting enterprise.

[0043] In a fourth aspect of the present application, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the method for identifying the operating status of equipment in a pollutant-discharging enterprise according to any one of the above-mentioned methods is implemented.

[0044] The method for identifying the operating status of equipment in polluting enterprises provided in this application adopts a dynamic Bayesian network and a dynamic threshold adjustment mechanism, which can automatically adjust the threshold in real time according to changes in the equipment operating characteristics and the identification results. In this way, it is possible to flexibly respond to the dynamic changes in the start and stop status of the equipment, avoid the problem that the fixed threshold cannot adapt to the fluctuations in equipment operation, and ensure the accuracy and flexibility of the equipment start and stop status judgment. By combining characteristics such as the power consumption change rate, the threshold can be sensitively adjusted when the equipment is in standby, low load or other special states, ensuring that these low-fluctuation states can also be accurately identified, thereby improving the comprehensiveness of the identification.

[0045] This application can provide a more intuitive and operational decision-making basis for the identification of the start and stop status of the device by introducing a dynamic Bayesian network and a real-time data feedback mechanism. The application of the dynamic threshold adjustment mechanism makes the identification of the start and stop status of the device not only rely on complex clustering results, but also combines the power consumption data and changes of the actual device to provide a judgment standard that is easier to understand and apply. Unlike traditional cluster analysis methods, this application optimizes the identification strategy of the start and stop status of the device in an adaptive manner, combining real-time data feedback and a dynamic threshold adjustment mechanism, rather than simply relying on a preset number of clusters. In this way, the judgment strategy can be flexibly adjusted according to the actual operating status of the device, avoiding the difficulty of selecting the number of clusters, and better reflecting the diversity and complexity of the device operating status.

[0046] In addition, the present application also provides a device and equipment for identifying the operating status of equipment in a sewage-discharging enterprise having the above-mentioned technical effects. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the embodiments of this specification or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the embodiments of this specification. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0048] Figure 1 A flowchart of a specific implementation method of the method for identifying the operating status of equipment in a pollutant-discharging enterprise provided in this application;

[0049] Figure 2 This is a flowchart of another specific implementation method of the method for identifying the operating status of equipment in a pollutant-discharging enterprise provided in this application;

[0050] Figure 3 This is a schematic diagram of the electricity consumption data sequence and operating status identification of the production facilities of an example enterprise;

[0051] Figure 4 This is a structural block diagram of the device for identifying the operating status of equipment in a pollutant-discharging enterprise provided in this application;

[0052] Figure 5 This is a structural block diagram of the equipment operating status identification device for pollutant discharge enterprises provided in this application. DETAILED DESCRIPTION

[0053] In order to make the purpose, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all of the embodiments. It should be noted that, in the absence of conflict, the embodiments in this disclosure and the features in the embodiments can be combined, separated, interchanged and / or rearranged with each other. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0054] The terms used herein are for the purpose of describing specific embodiments and are not intended to be restrictive. As used herein, unless the context clearly indicates otherwise, the singular forms "one (kind, person)" and "said (the)" are also intended to include plural forms. In addition, when the terms "comprise" and / or "include" and their variations are used in this specification, the features, integral bodies, steps, operations, parts, assemblies and / or their groups stated are explained, but the presence or addition of one or more other features, integral bodies, steps, operations, parts, assemblies and / or their groups is not excluded. It should also be noted that, as used herein, the terms "substantially", "approximately" and other similar terms are used as approximate terms and not as degree terms, so that they are used to explain the inherent deviations of the measured values, calculated values ​​and / or the values ​​provided that will be recognized by those of ordinary skill in the art.

[0055] This embodiment provides a method for identifying the operating status of equipment in pollutant-discharging enterprises, which predicts the start and stop status of equipment and adaptively adjusts the threshold based on real-time electricity consumption data, electricity consumption change rate, and a dynamic Bayesian network (DBN) model. Figure 1 , the process specifically includes the following steps:

[0056] S101: Collect electrical parameters of electrical equipment in pollutant-discharging enterprises in real time to generate a real-time electricity consumption data sequence.

[0057] In this example, smart meters installed on key electrical equipment at pollutant-discharging enterprises first collect real-time electrical parameters, including but not limited to current, voltage, and power. This data is transmitted to a data collection system via Internet of Things (IoT) devices, generating a real-time electricity consumption data series to support subsequent analysis.

[0058] After the real-time collection of electrical parameters of electrical equipment in pollutant-discharging enterprises, the method also includes: cleaning the original data of the electrical parameters, removing null values, duplicate data and abnormal points, obtaining cleaned data, and performing subsequent data processing.

[0059] S102: Construct a sliding window, perform statistical analysis on the real-time power consumption data sequence within the sliding window, and calculate the power consumption change rate, power factor, and start-stop frequency within the corresponding window.

[0060] After data collection, the real-time power consumption data sequence is processed using a sliding window analysis method. The length of the sliding window can be adjusted according to actual needs, usually ranging from tens of seconds to several minutes. Statistical analysis is performed on the power consumption data within the sliding window to calculate the power consumption change rate ( The electricity consumption change rate reflects the fluctuation of equipment electricity consumption, and the calculation formula is:

[0061]

[0062] in, is the current electricity consumption, The power consumption at the last moment. The power consumption change rate can reveal whether the equipment is in a period of start-stop status changes or load fluctuations.

[0063] S103: Build an equipment operation status recognition model using a dynamic Bayesian network, wherein the nodes of the dynamic Bayesian network include the equipment start / stop status at the current moment, current power consumption, power consumption change rate, power factor, start / stop frequency, and the equipment start / stop status at the previous moment.

[0064] Using a dynamic Bayesian network to build a device operating status recognition model includes: determining nodes in the dynamic Bayesian network, each node represents a variable. Specifically, the following nodes are included:

[0065] Device current status node ( ): This node represents the device's current start / stop status. 1 indicates start and 0 indicates stop. The device's start / stop status is the target variable to be predicted.

[0066] Equipment power consumption data node ( ): This node represents the power consumption of the device at the current moment. It is used as an observation variable to reflect the operating intensity and load changes of the device.

[0067] Electricity consumption change rate node ( ): This node represents the power consumption change rate of the device at the current moment. It is used to capture the fluctuation of load changes during device operation and help infer whether the device is in the process of starting and stopping.

[0068] Threshold Node ( ): This node represents the threshold used to determine the start and stop status of the device. It depends on the device's power usage data and change rate, and will be dynamically adjusted as the device's operating status changes.

[0069] Constructing a directed graph model of a dynamic Bayesian network includes the following two types of dependencies:

[0070] Dependencies within a moment: The current start / stop status of the device ( ) and current power consumption ( ) and electricity consumption change rate ( ) have a direct dependency between them. That is, the device's on / off status is not only related to its current power consumption, but also closely linked to its changing state (rate of change). Specifically, when the device is operating at high load, it may be in the on state. However, when the load changes (such as a sudden drop), it may switch to a different state (such as shutting down or switching to standby).

[0071] In the directed graph of DBN, the current state node ( ) will directly point to the electricity consumption data node ( ) and electricity consumption change rate node ( ), forming a dependency relationship within a moment, indicating that the start and stop status of the equipment at the current moment is jointly affected by the power consumption and the power consumption change rate.

[0072] Cross-time dependency: The current start and stop status of the device ( ) depends not only on current electricity consumption data ( ) and electricity consumption change rate ( ), and also with the state node at the previous moment ( ) have dependencies. The current state of a device is often affected by its historical state. Especially during device operation, the previous device start and stop state may affect the current state transition.

[0073] After the network structure is determined, by inputting real-time power consumption data ( ) and electricity consumption change rate ( ) into the dynamic Bayesian network model to perform device start and stop status ( ) inference. During the inference process, the DBN model uses the intra-time dependency and cross-time dependency, combined with the historical state of the device ( ) and the current power consumption data and change rate to predict the start and stop status of the equipment at the current moment.

[0074] Through the forward-backward inference algorithm or Bayesian inference method, the current state of the device (S t ), and continuously update the prediction results of the equipment start and stop status.

[0075] S104: The conditional probability distribution of the dynamic Bayesian network is trained using the following parameter learning mechanism: historical data of the equipment of the pollutant-discharging enterprise is identified as a stage, and divided into multiple different process stages; a sub-model is independently constructed for each process stage to identify the behavioral characteristics of the equipment in each process stage, and parameter learning is performed separately.

[0076] Specifically, a clustering algorithm can be used to perform cluster analysis on the historical data to identify the different operating stages of the equipment; the operating stages include the working stage, the standby stage, and the cleaning stage. This process ensures that the behavior within each stage is independently modeled. In this process, a clustering algorithm (such as K-means or DBSCAN) is used to perform cluster analysis on the equipment's historical data to identify the behavioral characteristics of the equipment at different process stages. For example, based on characteristics such as the equipment's start-stop mode, power consumption, and power factor, different process stages such as working, cleaning, and standby can be identified. To improve accuracy, a time series model (such as the Hidden Markov Model (HMM)) can also be used to classify the equipment status to better capture the cyclical changes of the equipment.

[0077] When constructing a sub-model independently in each process stage, the expectation maximization algorithm is used to estimate the conditional probability distribution between the start and stop states of the equipment and other characteristics in the corresponding stage, and the parameters are optimized by expectation maximization.

[0078] Once the data is divided into distinct process segments, parameter learning can be performed independently within each segment. For example, during the working phase, equipment may require higher power to maintain operation, while during the cleaning phase, power factor and power consumption may drop significantly. The parameters of the Bayesian network can be trained independently within each segment, allowing the model for each process segment to better adapt to the specific equipment behavior.

[0079] When learning parameters independently within each phase, the EM algorithm can help determine the conditional probability distribution at each moment. Specifically, assuming that the device's state (such as on / off status) exhibits different behavioral patterns in different phases, the EM algorithm can be used to estimate the conditional probability distribution within each phase. For example, during the working phase, the model can estimate the conditional probability distribution of power consumption under the current on / off state; while during the cleaning phase, the model can learn different patterns.

[0080] During parameter estimation, labeled and unlabeled device electricity usage sample data are obtained and fed into a device operating status recognition model for semi-supervised training. During parameter learning, incorporating semi-supervised learning methods using pseudo-labeled and unlabeled data can further improve model performance, especially when data is insufficiently labeled.

[0081] S105: Input the multi-dimensional continuous features consisting of the real-time power consumption data sequence, the power consumption change rate, the power factor and the start-stop frequency into a dynamic Bayesian network, perform state inference based on the dynamic Bayesian network, and output the start-stop state of the device at the current moment.

[0082] The real-time collected electricity consumption data series and the calculated electricity consumption change rate ( ), power factor and start-stop frequency, which are composed of multi-dimensional continuous features, are input into the dynamic Bayesian network, and the current state of the equipment is predicted based on the inference algorithm (such as the forward-backward algorithm). During the inference process, historical data and real-time data are combined to output the start and stop status (start or stop) of the device at each time point.

[0083] In this embodiment, input features include the start / stop frequency (indicating the number of state transitions per unit time) and the power factor variation (indicating whether the device is actually loaded). These features help identify behaviors such as idling. The constructed DBN nodes also include a start / stop frequency node and a power factor node.

[0084] After receiving the input features, DBN adopts the following modeling method:

[0085] Conditional Gaussian distribution is used to represent the conditional probability of each continuous variable; or Gaussian mixture model is used to represent the joint probability distribution of states under multi-dimensional input; the learning process maximizes the likelihood through the EM algorithm to train the mean, variance and covariance matrix of the input variables in each state.

[0086] This embodiment constructs multidimensional continuous features (power consumption change rate, power factor, and start / stop frequency) as input observation variables for a dynamic Bayesian network. Based on intra- and inter-temporal dependencies and combined with historical states, state inference is performed to more accurately and dynamically identify the start / stop status of devices. This method offers strong expressiveness, sensitivity, and adaptability, significantly outperforming traditional single-threshold start / stop determination methods. The method also includes dynamically adjusting the threshold used to determine the start / stop status of a device based on changes in the device's operating characteristics and the identification results.

[0087] While inferring the start and stop status of the device, the power consumption change rate of the device is calculated in real time, and the threshold value for judging the start and stop status is determined based on the fluctuation of the power consumption change rate ( ) to make real-time adjustments: determine whether the rate of change of the device power consumption data exceeds a predetermined threshold within a preset time interval. If so, increase the threshold by a preset step size. If the rate of change of the device power consumption data does not detect a start-stop state change within the preset time interval, reduce the threshold by a preset step size.

[0088] When the power consumption change rate is large (i.e. the device starts and stops frequently), increase the threshold ( ), thereby reducing sensitivity to small fluctuations and avoiding false alarms. When the rate of change of power consumption is small (i.e. the equipment runs steadily for a long time), reduce the threshold ( ) to increase sensitivity to small changes and ensure that potential equipment start-up and shutdown changes are captured in a timely manner.

[0089] Specifically, any one or any combination of the following strategies can be used to dynamically adjust the threshold used to determine the start and stop status of the device:

[0090] The first strategy is to make dynamic adjustments based on the rate of change of electricity consumption. , the following formula is used to dynamically update the threshold used to determine the start and stop status of the device: .

[0091] in, For the forgetting factor, is the threshold used to judge the start and stop status of the device at the current moment, is the threshold used to judge the start and stop status of the device at the previous moment, is the rate of change of electricity consumption at the current moment.

[0092] It is understandable that Used to control the weight between historical memory and current feedback. If the value is small, it can respond quickly to current changes; The larger the value, the more dependent on history and the smoother the response. Further, any of the following strategies can be used to adjust the forgetting factor: Make adaptive adjustments:

[0093] (1) Using grid search and cross validation within a preset range (e.g. 0.1~0.9), by comparing different The recognition accuracy, F1 score or false alarm rate indicators under the value are selected to optimize value.

[0094] (2) Using Bayesian optimization methods (such as Gaussian Process), Build a proxy model between the performance indicators and select the best one based on historical recognition performance This method can improve the search efficiency; its advantage is that it is applicable to high-dimensional parameter spaces and non-convex objective functions.

[0095] (3) Dynamically adjust the accuracy based on the deviation between the current recognition accuracy and the target accuracy , the update formula is: ;in, is the learning rate, is the current recognition accuracy, is the target accuracy.

[0096] In a preferred embodiment of the present invention, a linkage control mechanism based on performance feedback is further constructed to dynamically adjust the device start and stop state judgment threshold ( ) and the forgetting factor ( ) to improve the ability to adapt to operational fluctuations and error changes. Specifically, based on the deviation relationship between the current recognition result and the target performance indicators (such as false alarm rate, missed alarm rate, and accuracy), the threshold and value.

[0097] The strategy is as follows: When the current false alarm rate is higher than the set target false alarm rate, it means that the judgment is too sensitive and it is easy to make an erroneous activation judgment. Therefore, the control strategy is to increase the judgment threshold. , and reduce the forgetting factor , in order to respond to new changes more quickly and reduce false alarms. When the current false alarm rate is higher than the target false alarm rate, it means that the judgment is too conservative and the device's activation status cannot be identified in time. At this time, the judgment threshold will be lowered , and also reduce , improve the sensitivity to fluctuation signals. When the current accuracy is stably higher than the set target accuracy, it indicates that the operation is stable and the recognition effect is ideal. In order to avoid over-response to short-term fluctuations, it is appropriate to increase , to enhance the smoothness and stability of the system.

[0098] The second strategy is to dynamically adjust the false alarm rate and missed alarm rate of the real-time monitoring system. When the false alarm rate increases, the threshold is increased to reduce the system sensitivity; when the missed alarm rate increases, the threshold is lowered to increase the system sensitivity. Specifically: real-time monitoring of the false alarm rate and missed alarm rate of the equipment operating status of the pollutant discharge enterprise, based on Dynamically adjust the threshold used to determine the start and stop status of the device.

[0099] in, is the threshold used to judge the start and stop status of the device at the current moment, is the threshold used to judge the start and stop status of the device at the previous moment, is the false alarm rate at the current moment, is the target false alarm rate; is the omission rate at the current moment, is the target underreporting rate, , is the adjustment coefficient.

[0100] The third method is to dynamically adjust the threshold based on the posterior probability distribution inferred by DBN. If the confidence distribution is too concentrated, it means that the model is confident, and the threshold can be increased; if the distribution is more dispersed, it means that the model is uncertain, and the threshold should be lowered to increase sensitivity. Specifically: the device is enabled using the dynamic Bayesian network. The posterior probability distribution is calculated, and the mean and standard deviation of the posterior probability distribution are calculated. The threshold used to determine the start and stop status of the device is dynamically adjusted as follows: .

[0101] in, is the threshold used to judge the start and stop status of the device at the current moment, For time series observations arrive Under the condition of , the posterior probability that the device is currently in the enabled state is, is the mean of the posterior probabilities, is the standard deviation of the posterior probability, is the confidence adjustment coefficient, which can be used to control conservatism: the larger the γ is, the more the threshold tends to be fault-tolerant.

[0102] Threshold adjustment is optimized through a dynamic Bayesian network, which can dynamically adjust the judgment criteria according to the actual operating status of the equipment to ensure the accuracy of status recognition.

[0103] As devices continue to operate, the system continuously adjusts and optimizes thresholds through real-time feedback mechanisms to adapt to long-term changes in the devices. For example, if the system detects that certain devices are frequently starting and stopping, or operating steadily for extended periods, it will adjust the threshold update rate and range through real-time analysis of historical data, further improving the sensitivity and accuracy of detecting device start and stop states.

[0104] Through the above implementation steps, this embodiment provides a method for identifying the operating status of equipment in pollutant-discharging enterprises based on a dynamic Bayesian network and a dynamic threshold adjustment mechanism. This method can flexibly respond to dynamic changes in equipment operation, accurately identify the start and stop status of equipment, and continuously optimize thresholds through a real-time feedback mechanism, thereby improving the efficiency and accuracy of the equipment monitoring system.

[0105] The flowchart of another specific implementation method of the method for identifying the operating status of equipment in a pollutant-discharging enterprise provided in this application is as follows: Figure 2 As shown, the method specifically includes:

[0106] S201: Collecting raw data.

[0107] Install smart meters on the company's key electrical equipment to capture multiple parameters such as current, voltage, power, and electricity in real time.

[0108] S202: Preprocessing the original data.

[0109] Clean the raw data collected in real time to remove null values, duplicate data, and obvious anomalies (such as sudden spikes caused by sensor failure or interference).

[0110] S203: Calculate the instantaneous rate of change.

[0111] Based on historical electricity consumption data, calculate the rate of change for each time segment.

[0112] The instantaneous rate of change is calculated as follows:

[0113]

[0114] Among them, among them, is the power value at time t, It is the time difference between two points in time.

[0115] S204: Perform trend analysis and start / stop status detection.

[0116] (1) Analysis of changing trends

[0117] By dynamically adjusting the size of the sliding window and the window length according to the severity of data fluctuations, the trend within the window is determined by regression.

[0118] Specifically, based on the analysis of the real-time electricity consumption data sequence, the degree of fluctuation of the electricity consumption data is determined; the length of the sliding window is dynamically adjusted according to the degree of fluctuation of the electricity consumption data: the degree of fluctuation of the electricity consumption data within the sliding window is calculated; when the degree of fluctuation of the electricity consumption data exceeds a set threshold, the sliding window is reduced according to a preset step size; when the degree of fluctuation of the electricity consumption data is lower than the set threshold, the sliding window is increased according to a preset step size. The specific implementation of this process can be:

[0119] 1) Initialize the window: define the initial sliding window size W min and W max , representing the minimum and maximum window sizes respectively.

[0120] 2) Calculate the fluctuation characteristics within the window: For each sliding window, calculate the degree of data fluctuation within that window, such as the standard deviation or variance. A larger standard deviation means more severe fluctuations; a smaller standard deviation means more stable data.

[0121] 3) Dynamically adjust the window size: set a fluctuation threshold. When the standard deviation exceeds the threshold, the window is reduced; when the standard deviation is lower than the threshold, the window is expanded.

[0122] 4) For each window, linear regression or polynomial regression can be used to fit a trend line for electricity consumption to see whether the trend is increasing, decreasing, or stable.

[0123] (2) Start-stop status detection

[0124] By setting an initial threshold (which can be determined based on historical device data), the device start / stop status is determined. When the rate of change exceeds the threshold, the device is considered started / stopped.

[0125] S205: Dynamically adjust the threshold.

[0126] As data accumulates, the threshold for change detection is adjusted in real time through algorithms such as EWMA, machine learning, or reinforcement learning.

[0127] The following example uses a dynamic Bayesian network (DBN) as an example. The training process of the model is as follows:

[0128] (1) Data preparation

[0129] Collect historical data of the equipment, including time series power consumption data, start and stop status, and other related features. Extract features from the original data, such as the power consumption change rate, standard deviation, variance, etc. at each moment, and annotate the data.

[0130] (2) Constructing a dynamic Bayesian network

[0131] DBN is a directed graph model that can represent temporal dependencies. Its structure is divided into dependencies between moments (cross-temporal dependencies) and dependencies within moments (same-temporal dependencies). When modeling device power consumption, the basic structure of DBN can be expressed as:

[0132] Device start / stop status: Indicates the start / stop status of the device (S t ), which is the variable we want to predict.

[0133] Equipment power consumption: Indicates the power consumption at each time point (P t ), which is the observed variable.

[0134] Change rate: Equipment power consumption change rate (ΔP t ), are features extracted from historical data.

[0135] Threshold adjustment: Indicates the threshold used to determine the start and stop status of the device (θ t ), which is the parameter you want to adjust dynamically.

[0136] 1) Define DBN nodes

[0137] Current state node ( ): The start / stop status of the device at the current time step. The value can be start / stop (0 or 1).

[0138] Previous state node ( ): The start and stop status of the previous time step.

[0139] Current power consumption ( ): The power consumption of the device in the current time step.

[0140] Rate of Change Node ( ): The rate of change of power consumption of the device in the current time step.

[0141] Threshold Node ( ): Used to determine the change threshold of the device's start and stop status, which depends on the device's historical data.

[0142] 2) DBN modeling structure

[0143] Assume that at time t, the state of the device is From the previous state and current power consumption and the rate of change Determine the rate of change threshold of the device at the same time Adaptively adjust over time.

[0144] 3) Parameter Learning

[0145] DBN parameter learning includes two parts: structure learning and parameter learning.

[0146] Structural learning: Using historical data, algorithms (such as Expectation-Maximization (EM)) are used to determine the dependencies between device start and stop states and features.

[0147] Parameter learning: Determine the parameters of each conditional probability distribution in the Bayesian network based on historical data. t The conditional probability distribution of can be expressed as , the parameters of these conditional probability tables are learned using historical data through maximum likelihood estimation or Bayesian estimation.

[0148] 4) Inference and automatic threshold adjustment

[0149] Inferring the state of the device through the DBN model , and adjust the threshold based on historical data .

[0150] Inferring the device start / stop status: Using the DBN model inference (such as the forward-backward algorithm), predict whether the device is started or stopped at the current moment .

[0151] Adjust threshold: During the inference process, the threshold is automatically adjusted using dynamic adjustment rules Adjustment strategies can be based on the results of Bayesian inference and feedback from historical data:

[0152] In the process of inferring the start and stop status of the equipment, according to the power consumption change rate ( ) and historical data to update the start / stop status judgment threshold in real time ( ) to optimize the sensitivity of device start and stop status recognition. Dynamic threshold updates follow the following rules:

[0153] The dynamic threshold update formula is: .

[0154] in, For the forgetting factor, is the threshold used to judge the start and stop status of the device at the current moment, is the threshold used to judge the start and stop status of the device at the previous moment, is the rate of change of electricity consumption at the current moment.

[0155] Forgetting factor ( ): Controls the degree of influence of historical data on the current threshold adjustment. The value range is between 0 and 1. When it is close to 1, the influence of historical data is greater, and the threshold adjustment is more strongly affected by the historical state; when When it is close to 0, the current electricity consumption change rate has a greater impact, and the threshold adjustment depends more on real-time data.

[0156] Threshold adjustment mechanism: When the device's start and stop status changes frequently, the device's power consumption change rate ( ) is larger, then increase the threshold ( ), reducing the response to small fluctuations; when the equipment starts and stops smoothly and the rate of change of power consumption is small, the threshold is reduced ( ), improve the sensitivity to small changes and ensure timely identification of potential changes in equipment start and stop status.

[0157] By introducing the forgetting factor ( ) and dynamic threshold update formula, the system can adaptively adjust the threshold according to real-time data, avoiding the limitation of traditional fixed threshold method that cannot adapt to changes in equipment start and stop status.

[0158] 5) Model evaluation and optimization

[0159] Using the historical data of the device, we compare the difference between the start and stop status of the device predicted by the DBN model and the actual status, and calculate indicators such as accuracy, recall rate and false alarm rate. According to different device characteristics, we can select different threshold adjustment strategies through cross-validation to optimize parameters. .

[0160] Model evaluation and optimization are performed using historical data, including evaluating the difference between the predicted equipment start and stop status and the actual status, and optimizing the forgetting factor through cross-validation. Ultimately, the optimal forgetting factor value is selected to improve model performance.

[0161] The equipment start and stop status predicted by the model is compared with the actual equipment start and stop status in historical data to calculate evaluation indicators. These evaluation indicators include:

[0162] Accuracy: The ratio of the start and stop states correctly predicted by the model to all predicted states.

[0163] Recall rate: The ratio of the start / stop status predicted by the model to the actual start / stop status, which measures the model's ability to identify changes in the start / stop status of the device.

[0164] False alarm rate: The proportion of start / stop states incorrectly predicted by the model, which measures the degree of false alarm of the model.

[0165] By calculating these evaluation indicators, we can evaluate the performance of the current DBN model in identifying the start and stop status of equipment and provide a basis for the next step of optimization.

[0166] After evaluating the equipment start-stop status prediction model, the forgetting factor is optimized through cross-validation method to improve the performance of the model.

[0167] Cross-validation: K-fold cross-validation is used to divide the device's historical data into k subsets. One subset is selected each time as the validation set, and the remaining subsets are used as the training set. Training and validation are repeated until each subset has been validated once. Cross-validation can prevent model overfitting and provide more accurate performance evaluation.

[0168] Forgetting factor optimization: During the cross-validation process, different forgetting factors are tested Value. Forgetting factor Controls the degree of influence of historical data on the current threshold adjustment, ranging from 0 to 1. When it is close to 1, the threshold update of the model depends more on historical data; when When it is close to 0, the threshold update depends more on the current real-time data.

[0169] Optimization strategy: In each round of cross-validation, calculate the current Evaluation metrics (such as precision, recall, and false alarm rate) under different values.

[0170] By comparing different The evaluation results under the value are used to select the best forgetting factor value, so that the model performs best in identifying the start and stop status of the equipment. After the optimization process is completed, the best output The value is used to update the threshold update mechanism in the dynamic Bayesian network model, thereby improving the model's adaptability to changes in the start and stop states of the equipment.

[0171] After outputting the detected start / stop status of the device in real time, the method may further include a step of updating the log. Figure 3 Shown is a schematic diagram of an example enterprise production facility electricity consumption data sequence and operating status identification.

[0172] Under stationary conditions, the statistical properties of facility electricity usage data (such as the mean) remain constant over time, with only random fluctuations. However, when facilities are started or shut down, the electricity usage data series often experiences structural breaks or changes. These changes introduce non-stationarity, altering the overall distribution of the time series. These points in time, marking the onset of change, are called change points. The greatest advantage of dynamic Bayesian networks (DBNs) lies in their ability to capture temporal data changes and continuously refine the model based on historical data. By inferring implicit states (on / off states), DBNs adjust the rate of change threshold in real time, adapting it to the actual operation of the equipment and dynamically adjusting the sliding window size based on data fluctuations. This mechanism dynamically adjusts the sliding window length by analyzing statistical characteristics such as standard deviation or variance, enabling the system to react quickly to significant changes in the equipment's on / off state while reducing oversensitivity when the equipment's on / off state is stable.

[0173] This application can more accurately identify the start and stop status of equipment by combining sliding windows, trend analysis and dynamic Bayesian networks. Especially during the start and stop process of the equipment, DBN provides higher detection accuracy and reduces misjudgments caused by short-term fluctuations. At the same time, the dynamic threshold adjustment mechanism automatically adjusts the threshold according to the real-time operating status of the equipment to ensure the optimization of detection sensitivity, further reduce the error probability, and better adapt to the complex operating environment of different enterprises and different equipment. This algorithm not only greatly improves the real-time, accuracy and intelligence level of facility operation status identification, but also improves the efficiency of abnormal behavior supervision of polluting enterprises and promotes the continuous progress of environmental protection.

[0174] In some embodiments, the structural block diagram of the device for identifying the operating status of equipment in a pollutant-discharging enterprise provided by this application is as follows: Figure 4 As shown, the device includes the following modules:

[0175] The data acquisition module 401 is configured to collect electrical parameters of electrical equipment of pollutant-discharging enterprises in real time and generate a real-time electricity consumption data sequence.

[0176] The statistical calculation module 402 is configured to construct a sliding window, perform statistical analysis on the real-time power consumption data sequence within the sliding window, and calculate the power consumption change rate, power factor, and start-stop frequency within the corresponding window.

[0177] The model construction module 403 is configured to use a dynamic Bayesian network to construct an equipment operation status recognition model, wherein the nodes of the dynamic Bayesian network include the equipment start and stop status at the current moment, the current power consumption, the power consumption change rate, the power factor, the start and stop frequency and the equipment start and stop status at the previous moment.

[0178] The parameter learning module 404 is configured to train the conditional probability distribution of the dynamic Bayesian network using the following parameter learning mechanism: performing stage identification on the historical data of the equipment of the pollutant discharge enterprise and distinguishing them into multiple different process stages; independently constructing a sub-model for each process stage, identifying the behavioral characteristics of the equipment in each process stage, and performing parameter learning separately.

[0179] The output module 405 is configured to input the multi-dimensional continuous features consisting of the real-time power consumption data sequence, the power consumption change rate, the power factor and the start-stop frequency into a dynamic Bayesian network, perform state inference based on the dynamic Bayesian network, and output the start-stop status of the device at the current moment.

[0180] The device further includes: a threshold adjustment module configured to dynamically adjust the threshold used to determine the start / stop state of the device according to changes in the device operation characteristics and the recognition results.

[0181] The embodiment of the present application also provides a device for identifying the operating status of equipment in a pollutant-discharging enterprise, such as Figure 5 As shown, the device includes a memory 51 and a processor 52, wherein the memory 51 stores a computer program, and when the computer program is executed by the processor 52, the method for identifying the operating status of equipment in a pollutant-discharging enterprise according to any one of the above-mentioned methods is implemented.

[0182] An embodiment of the present application also provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the method for identifying the operating status of equipment in a pollutant-discharging enterprise according to any one of the above-described methods is implemented.

[0183] Computer-readable storage media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. The information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer-readable media does not include transitory computer-readable media (transitory media), such as modulated data signals and carrier waves.

[0184] Professionals should also be further aware that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the above description has generally described the components and steps of each example according to their functions. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0185] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein may be implemented using hardware, a software module executed by a processor, or a combination of the two. The software module may be placed in random access memory (RAM), internal memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, a hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.

[0186] The specific implementation methods described above further illustrate the purpose, technical solutions and beneficial effects of this application. It should be understood that the above description is only the specific implementation methods of this application and is not intended to limit the scope of protection of this application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of this application should be included in the scope of protection of this application.

Claims

1. A method for identifying the operating status of equipment in a pollutant-discharging enterprise, characterized in that: The following steps are involved: Real-time collection of electrical parameters of electrical equipment in pollutant-discharging enterprises to generate real-time electricity consumption data series; Construct a sliding window, perform statistical analysis on the real-time power consumption data sequence within the sliding window, and calculate the power consumption change rate, power factor, and start-stop frequency within the corresponding window; A dynamic Bayesian network is used to build an equipment operation status recognition model. The nodes of the dynamic Bayesian network include the equipment start and stop status at the current moment, current power consumption, power consumption change rate, power factor, start and stop frequency, and the equipment start and stop status at the previous moment; The following parameter learning mechanism is used to train the conditional probability distribution of the dynamic Bayesian network: historical data of equipment in pollutant-discharging enterprises is used to identify stages and distinguish them into multiple process stages. A sub-model is independently constructed for each process stage to identify the behavioral characteristics of the equipment in each process stage and perform parameter learning separately. Inputting the multi-dimensional continuous features consisting of the real-time power consumption data sequence, the power consumption change rate, the power factor, and the start-stop frequency into a dynamic Bayesian network, performing state inference based on the dynamic Bayesian network, and outputting the start-stop state of the device at the current moment; The method further includes dynamically adjusting the threshold used to determine the start and stop status of the device according to changes in the device operating characteristics and identification results.

2. The method for identifying the operating status of equipment in a pollutant-discharging enterprise according to claim 1, characterized in that: The method of independently building a sub-model for each process stage, identifying the behavioral characteristics of the equipment in each process stage, and performing parameter learning separately includes: When constructing a sub-model independently in each process stage, the expectation maximization algorithm is used to estimate the conditional probability distribution between the start and stop states of the equipment and other characteristics in the corresponding stage, and the parameters are optimized by expectation maximization.

3. The method for identifying the operating status of equipment in a pollutant-discharging enterprise according to claim 2, characterized in that: The independent construction of sub-models for each process stage, identification of the behavioral characteristics of the equipment in each process stage, and separate parameter learning also include: When performing parameter estimation, obtain labeled device electricity usage sample data and unlabeled device electricity usage sample data; The labeled device electricity usage sample data and the unlabeled device electricity usage sample data are input into the device operation status recognition model for semi-supervised learning training.

4. The method for identifying the operating status of equipment in a pollutant-discharging enterprise according to claim 2, characterized in that: The stage identification of historical data of equipment of pollutant-discharging enterprises includes: A clustering algorithm is used to perform cluster analysis on the historical data to identify different operating stages of the equipment; the operating stages include a working stage, a standby stage, and a cleaning stage.

5. The method for identifying the operating status of equipment in a pollutant-discharging enterprise according to any one of claims 1 to 4, characterized in that: The dynamically adjusting the threshold for determining the start / stop status of the device according to the changes in the device operation characteristics and the recognition results includes: Based on the electricity consumption change rate , the following formula is used to dynamically update the threshold used to determine the start and stop status of the device: ; in, For the forgetting factor, is the threshold used to judge the start and stop status of the device at the current moment, is the threshold used to judge the start and stop status of the device at the previous moment, is the rate of change of electricity consumption at the current moment; Or real-time monitoring of the false alarm rate and missed alarm rate of the equipment operating status of the sewage-discharging enterprises, based on Dynamically adjust the threshold used to determine the start and stop status of the device; in, is the threshold used to judge the start and stop status of the device at the current moment, is the threshold used to judge the start and stop status of the device at the previous moment, is the false alarm rate at the current moment, is the target false alarm rate; is the omission rate at the current moment, is the target underreporting rate, , is the adjustment coefficient; Or use dynamic Bayesian network to calculate the device is enabled The posterior probability distribution is calculated, and the mean and standard deviation of the posterior probability distribution are calculated. The threshold used to determine the start and stop status of the device is dynamically adjusted as follows: ; in, is the threshold used to judge the start and stop status of the device at the current moment, For time series observations arrive Under the condition of , the posterior probability that the device is currently in the enabled state is, is the mean of the posterior probabilities, is the standard deviation of the posterior probability, is the confidence adjustment coefficient.

6. The method for identifying the operating status of equipment in a pollutant-discharging enterprise according to claim 5, characterized in that: Also includes: Use any of the following strategies to adjust the forgetting factor Make adaptive adjustments: Grid search is used within the preset range and combined with cross validation to compare different The recognition accuracy, F1 score or false alarm rate indicators under the value are selected to optimize value; Using Bayesian optimization methods, Build a proxy model between the performance indicators and select the best one based on historical recognition performance value; Dynamically adjust according to the deviation between the current recognition accuracy and the target accuracy , the update formula is: ;in, is the learning rate, is the current recognition accuracy, is the target accuracy.

7. The method for identifying the operating status of equipment in a pollutant-discharging enterprise according to claim 5, characterized in that: Also includes: Determining the degree of fluctuation of the electricity consumption data based on an analysis of the real-time electricity consumption data sequence; Dynamically adjust the length of the sliding window according to the degree of fluctuation of the electricity consumption data: calculate the degree of fluctuation of the electricity consumption data in the sliding window; when the degree of fluctuation of the electricity consumption data exceeds a set threshold, reduce the sliding window according to a preset step size; when the degree of fluctuation of the electricity consumption data is lower than the set threshold, increase the sliding window according to a preset step size.

8. A device for identifying the operating status of equipment in a sewage-discharging enterprise, characterized in that: Includes the following modules: The data acquisition module is configured to collect electrical parameters of electrical equipment of pollutant-discharging enterprises in real time and generate a real-time electricity consumption data sequence; The statistical calculation module is configured to construct a sliding window, perform statistical analysis on the real-time power consumption data sequence within the sliding window, and calculate the power consumption change rate, power factor, and start-stop frequency within the corresponding window; A model building module is configured to build a device operation status recognition model using a dynamic Bayesian network, wherein the nodes of the dynamic Bayesian network include the device start and stop status at the current moment, the current power consumption, the power consumption change rate, the power factor, the start and stop frequency, and the device start and stop status at the previous moment; The parameter learning module is configured to train the conditional probability distribution of a dynamic Bayesian network using the following parameter learning mechanism: Historical data of equipment in pollutant-discharging enterprises is used to identify stages and distinguish them into multiple process stages; a sub-model is independently constructed for each process stage to identify the behavioral characteristics of the equipment in each process stage and perform parameter learning on each stage; an output module configured to input the multi-dimensional continuous features consisting of the real-time power consumption data sequence, the power consumption change rate, the power factor, and the start-stop frequency into a dynamic Bayesian network, perform state inference based on the dynamic Bayesian network, and output the start-stop state of the device at the current moment; The device further includes: a threshold adjustment module configured to dynamically adjust the threshold used to determine the start / stop state of the device according to changes in the device operation characteristics and the recognition results.

9. A device for identifying the operating status of equipment in a sewage-discharging enterprise, characterized in that: The method comprises a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the method for identifying the operating status of equipment in a pollutant-discharging enterprise according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium, characterized in that A computer program is stored thereon, and when the computer program is executed by a processor, the method for identifying the operating status of equipment in a pollutant-discharging enterprise according to any one of claims 1 to 7 is implemented.

Images