CN-121809998-B - Sanitation job scheduling method based on Internet of things

CN121809998BCN 121809998 BCN121809998 BCN 121809998BCN-121809998-B

Abstract

The invention relates to the field of sanitation operation scheduling, in particular to a sanitation operation scheduling method based on the Internet of things, which comprises the steps of determining a load CEEMDAN model of an iteration stop condition through time local weighted variance based on load data so as to generate vehicle load characteristics; the method comprises the steps of calculating a fuel consumption CEEMDAN model of a fuel consumption channel residual signal based on statistical values of a load data residual and a fuel consumption data residual to generate vehicle fuel consumption characteristics, generating comprehensive vehicle carrying characteristics by extracting the model based on characteristics of a multi-layer perceptron, a gating mechanism and a cross-modal attention mechanism, and generating corrected vehicle garbage carrying capacity by based on a long-period memory network and a full-connection layer architecture. According to the invention, the efficient noise reduction and the accurate decomposition of the non-stable time sequence data of the oil consumption and the load are realized, the accurate processing of the internet of things data of the garbage transfer vehicle is realized, the operation efficiency of the garbage transfer vehicle is improved, and the operation cost is reduced.

Inventors

ZHAO YIWU
ZHANG MINGYU
TANG ZHENGYANG
CHENG ZHE
LIN LEILEI

Assignees

轩昂生态环境建设有限公司
北京轩昂智慧物业有限公司
轩昂环保科技股份有限公司

Dates

Publication Date: 20260508
Application Date: 20260310

Claims (8)

1. The sanitation job scheduling method based on the Internet of things is characterized by comprising the following steps of: Passing the load data of the internet of things of the vehicle stored by the sanitation work scheduling platform through a load CEEMDAN model to generate a vehicle load characteristic, wherein the load CEEMDAN model determines an iteration stop condition based on a time local weighted variance of the load data; the method comprises the steps that vehicle internet of things fuel consumption data stored by an environmental sanitation operation scheduling platform pass through a fuel consumption CEEMDAN model to generate vehicle fuel consumption characteristics, wherein the fuel consumption CEEMDAN model calculates fuel consumption channel residual signals based on load data residual and statistical values of the fuel consumption data residual; The vehicle oil consumption characteristics, the vehicle load characteristics and the internet of things vehicle garbage transfer operation data are subjected to a characteristic extraction model to generate comprehensive vehicle carrying characteristics, wherein the characteristic extraction model is constructed based on a multi-layer perceptron, a gating mechanism and a cross-modal attention mechanism; The comprehensive vehicle carrying characteristics are subjected to a mapping model based on a long-short-term memory network and a full-connection layer framework to generate corrected vehicle garbage carrying capacity, and garbage transfer truck operation scheduling is carried out according to the corrected vehicle garbage carrying capacity; the process of generating the vehicle fuel consumption characteristics through the fuel consumption CEEMDAN model comprises the following steps: the fuel consumption data of the vehicle Internet of things passes through a fuel consumption CEEMDAN initial operation layer to generate fuel consumption data residual errors; the load data of the vehicle internet of things passes through a load CEEMDAN operation layer to generate a load data residual error; Calculating a coupling coefficient based on the covariance of the oil consumption data residual and the load data residual and the variance of the oil consumption data residual; calculating oil consumption channel noise based on the coupling coefficient, the public noise signal, the oil consumption noise signal and the public noise amplitude; Passing the oil consumption channel noise and the oil consumption data residual error through an oil consumption CEEMDAN iteration operation layer to generate the vehicle oil consumption characteristic; the oil consumption CEEMDAN model comprises an oil consumption CEEMDAN initial operation layer and an oil consumption CEEMDAN iterative operation layer, and the load CEEMDAN model comprises a load CEEMDAN operation layer; the process of generating vehicle load characteristics from the load CEEMDAN model includes: calculating load channel noise based on the common noise signal and the common noise amplitude; The load data and the load channel noise of the vehicle Internet of things pass through a load CEEMDAN operation layer to iteratively calculate load iteration IMF components; Stopping iterative judgment of the fault through the local variance ratio by the load iterative IMF component and the load data of the vehicle internet of things so as to generate the load characteristics of the vehicle; The load CEEMDAN model comprises a load CEEMDAN operation layer and a local variance ratio stopping iteration judgment layer.
2. The internet of things-based sanitation job scheduling method of claim 1, wherein the process of generating the vehicle fuel consumption characteristics through the fuel consumption CEEMDAN iteration operation layer comprises the following steps: calculating a first L2 norm by using the high-frequency iteration IMF component generated by the oil consumption CEEMDAN iteration operation layer, calculating a second L2 norm by using the total sum of the iteration IMF components, and taking the ratio of the first L2 norm to the second L2 norm as the oil consumption characteristic of the garbage loading action; calculating a third L2 norm by using the intermediate frequency iteration IMF component generated by the oil consumption CEEMDAN iteration operation layer, and taking the ratio of the third L2 norm to the second L2 norm as the oil consumption characteristic of the congestion road section; Calculating a fourth L2 norm by using the low-frequency iteration IMF component generated by the oil consumption CEEMDAN iteration operation layer, and taking the ratio of the fourth L2 norm to the second L2 norm as the integral oil consumption characteristic of the garbage load; The vehicle oil consumption features comprise garbage loading action oil consumption features, congestion road section oil consumption features and garbage loading integral oil consumption features, and iteration times of Gao Pinci iteration IMF components, intermediate frequency iteration IMF components and low frequency iteration IMF components are sequentially increased.
3. The internet of things-based sanitation job scheduling method of claim 1, wherein the process of generating the vehicle load feature by the local variance ratio stop iteration judgment layer comprises: calculating a first local weighted variance sequence based on the load iterative IMF component, and calculating a second local weighted variance sequence based on the vehicle internet of things load data; calculating a local variance ratio sequence based on the ratio of the first local weighted variance sequence and the second local weighted variance sequence; Calculating local instantaneous oscillation frequency based on the time interval between two adjacent zero crossing points of the load iteration IMF component; And judging whether to stop iterative computation of the load iteration IMF component or not based on a comparison result of the first median of the local variance ratio sequence and a stop threshold value and a comparison result of the second median of the local instantaneous oscillation frequency and a loading action frequency range, and if so, computing the load characteristics of the vehicle based on the load iteration IMF component.
4. The internet of things-based sanitation job scheduling method of claim 1, wherein the process of generating the comprehensive vehicle carrying feature by the feature extraction model comprises: passing the internet of things vehicle garbage transfer operation data through a multi-layer perceptron to generate a transfer mapping vector; The transfer mapping vector is subjected to a gating mechanism to generate transfer gating weight, and the spliced vector of the vehicle oil consumption characteristic and the vehicle load characteristic is subjected to weighted calculation based on the gating weight to generate a weighted vehicle transfer quantity characteristic; the diversion map vector and the weighted vehicle diversion characteristics are passed through a cross-modal attentiveness mechanism to generate the integrated vehicle-carried characteristics.
5. The internet of things-based sanitation job scheduling method of claim 4, wherein the process of generating the comprehensive vehicle-carrying feature by a cross-modal attentiveness mechanism comprises: passing the weighted vehicle traffic through a standard convolution layer to generate vehicle traffic space-time features; Passing the vehicle transfer quantity space-time characteristics through a flattening mapping layer to generate an attention key vector and an attention value vector; Passing the diversion map vector through an attention map layer to generate an attention query vector; Passing the attention key vector, the attention value vector and the attention query vector through an attention calculation layer to generate attention weighted features, and taking a spliced vector of the attention weighted features and vehicle transportation space-time features as the comprehensive vehicle carrying features; Wherein the cross-modal attention mechanism includes a standard convolution layer, a flattening mapping layer, an attention mapping layer, and an attention calculation layer.
6. The internet of things-based sanitation job scheduling method of claim 4, wherein the process of generating the diversion map vector by the multi-layer perceptron comprises: Passing the internet of things vehicle garbage transfer operation data through a global average pooling layer to generate pooling features; Passing the pooled features through a first convolution layer to generate initial mapped features; passing the initial mapping feature through a second convolution layer to generate the transport mapping vector; The multi-layer perceptron comprises a global average pooling layer, a first convolution layer and a second convolution layer.
7. The internet of things-based sanitation job scheduling method of claim 1, wherein the process of generating the corrected vehicle garbage load capacity through the mapping model comprises: performing flattening operation on the comprehensive vehicle carrying features to generate comprehensive vehicle carrying feature vectors; Passing the comprehensive vehicle carrying feature vector through a long-short-term memory network to generate a total time sequence hiding state; and passing the total time sequence hidden state through a full connection layer to generate the corrected vehicle garbage carrying capacity.
8. The internet of things-based sanitation job scheduling method according to any one of claims 1 to 7, wherein the process of performing the job scheduling of the refuse transfer vehicle according to the corrected vehicle refuse carrying capacity comprises: and calculating the garbage load balance degree of the fleet based on the corrected vehicle garbage load capacity, and adjusting the operation scheduling of the garbage transfer vehicle according to the garbage load balance degree of the fleet.

Description

Sanitation job scheduling method based on Internet of things Technical Field The invention relates to the field of sanitation job scheduling, in particular to a sanitation job scheduling method based on the Internet of things. Background With the acceleration of the urban process, the urban household garbage production amount continuously rises, and garbage transfer is used as a core link of sanitation operation, so that the urban environment treatment efficiency and the operation cost are directly affected. The garbage transfer vehicle is used as key operation equipment, the scheduling rationality of the garbage transfer vehicle is not only related to the timeliness of garbage clearance and the avoidance of garbage accumulation and pollution, but also is closely related to the operation cost such as fuel consumption, equipment loss, manpower configuration and the like, so that the realization of intelligent and accurate scheduling of the sanitation operation vehicle has become the core requirement in the field of urban sanitation management. The traditional sanitation operation scheduling is based on manual experience to make a fixed route, lacks dynamic perception of real-time operation conditions of vehicles, cannot adjust operation tasks according to actual operation conditions of the garbage transfer vehicle, is easy to generate the phenomenon of unbalanced operation scheduling, causes resource waste when the operation is idle, and reduces the overall operation efficiency. Moreover, as the load sensor generates instantaneous fluctuation due to jolt and garbage impact during operation of the garbage transfer truck, misjudgment is easily caused only by single load data, and the oil consumption data is less influenced by mechanical impact and can reflect the real load state. Therefore, at present, deep mining and accurate analysis are needed to be performed on oil consumption, load and other data in the operation process through a deep learning algorithm based on the Internet of things, the optimal scheduling scheme is avoided by only relying on the statistical data of a load sensor, the error is obvious, the dynamic requirement of sanitation operation scheduling is difficult to adapt, and the optimal allocation of sanitation operation resources cannot be realized. And as the internet of things technology is gradually applied to the field of environmental sanitation operation management, internet of things equipment such as a fuel sensor, a load sensor and the like deployed on the garbage transfer truck can acquire internet of things information such as vehicle fuel consumption data, load data, operation track, running state and the like in real time through the internet of things technology and transmit the internet of things information to an environmental sanitation operation scheduling platform, so that data support is provided for optimization of a scheduling scheme. At the data processing level of the internet of things equipment, oil consumption data and load data of the garbage transfer vehicle belong to non-stable time sequence data, and are influenced by multiple factors such as road conditions of an operation road section, garbage types, start-stop frequency, weather conditions and the like, and the data contain a large amount of noise and fluctuation components. In the prior art, part of schemes adopt Empirical Mode Decomposition (EMD) models to process non-stationary data, but the traditional empirical mode decomposition has the problem of mode aliasing, and the relevance characteristics of the traditional empirical mode decomposition are not combined in the targeted processing of oil consumption and load data, so that the data decomposition precision is insufficient, and reliable characteristic data cannot be generated. Therefore, how to accurately process and fuse the internet of things data of the multi-dimensional garbage transfer vehicle, and to accurately correct the vehicle dispatching of the vehicle load, so as to improve the operation efficiency and optimize the cost of the garbage transfer vehicle is a technical problem to be solved at present. Disclosure of Invention Therefore, the invention provides an environmental sanitation operation scheduling method based on the Internet of things, which realizes high-efficiency noise reduction and accurate decomposition of non-stable time sequence data of oil consumption and load through a double CEEMDAN model designed in a targeted manner, realizes effective fusion of multi-dimensional data through a multi-layer perceptron, a gating mechanism and a feature extraction model constructed by a cross-modal attention mechanism, corrects the vehicle garbage load capacity through a mapping model of a long-short-term memory network and a full-connection layer architecture, realizes accurate fusion of the Internet of things data of the multi-dimensional garbage transfer vehicle, and accurately corrects the vehicle scheduling of the vehicle load, impr