Standard-based fuel cost impact analysis system

Abstract

The invention discloses a standard-based fuel cost impact analysis system, which comprises an information acquisition and storage module, a cost calculation module, a cost prediction module and a cost evaluation module, the information acquisition and storage module is used for acquiring and storing processing cost information and fuel cost information of fuel; the cost calculation module is used for calculating first processing cost of the processing cost information and first fuel cost of the fuel cost information to obtain actual comprehensive cost; the cost prediction module is used for analyzing the cost influence data of the first fuel cost and the first processing cost, and constructing a comprehensive cost prediction model to obtain predicted comprehensive cost; and the cost management and control module is used for analyzing the predicted cost difference value to obtain a comparison result, and sending a cost early warning signal command according to the comparison result, so that the change of the fuel cost can be predicted by analyzing the change of the fuel cost influence factor, and a reliable adjustment index is provided for subsequent fuel cost management and control.

Description

Technical Field

[0001] The present invention relates to the technical field of cost impact analysis, and in particular to a standard-based fuel cost impact analysis system. Background Art

[0002] In recent years, with the rapid development of transportation fields such as automobile manufacturing, aviation manufacturing and shipbuilding industries, industrial energy production fields, living construction fields and service fields, the demand for fuel has continued to increase. Fuel is one of the most important energy resources in industrial production. In the manufacturing industry, fuel is widely used in boilers, heat treatment, drive equipment, etc. to provide energy support. Fuel is widely used in the field of transportation. Fuel is used in power plants to generate a large amount of electricity, which in turn drives electric locomotives and railway systems in the transportation industry. Fuel is of great significance to the development and progress of society. However, when factories use fuel, the cost of losses is often high, which has caused huge adverse effects on the economy and the environment. Therefore, it is necessary to analyze the cost impact of fuel.

[0003] Traditional fuel plant operating cost calculation methods are based on static models and are usually calculated under fixed conditions. They cannot fully adapt to rapidly changing market demand, fuel prices, environmental regulations and other factors, and are difficult to cope with the real-time fluctuations and complexity of the fuel market. They only focus on a single goal, such as minimizing fuel costs or maximizing electricity production, and ignore the trade-offs and interrelationships of multiple influencing elements, which may lead to less comprehensive decisions. Summary of the invention

[0004] The technical problem solved by the present invention is that the operating costs of traditional fuel plants are usually calculated under fixed conditions, which cannot fully adapt to rapidly changing market demands, fuel prices, environmental regulations and other factors, and it is difficult to cope with the real-time fluctuations and complexity of the fuel market. It is also impossible to predict the subsequent trend of changes in fuel costs based on the common changes of multiple influencing elements.

[0005] To solve the above technical problems, the present invention provides the following technical solutions: a standard-based fuel cost impact analysis system, including an information collection and storage module, a cost calculation module, a cost prediction module and a cost evaluation module;

[0006] The information collection and storage module is used to collect and store fuel processing cost information and fuel cost information;

[0007] The cost calculation module is used to calculate the first processing cost of the processing cost information and the first fuel cost of the fuel cost information to obtain the actual comprehensive cost;

[0008] The cost prediction module is used to analyze the cost impact data of the first fuel cost and the first processing cost, construct a comprehensive cost prediction model, and obtain a predicted comprehensive cost;

[0009] The cost control module is used to analyze and predict the cost difference, obtain a comparison result, and issue a cost warning signal command according to the comparison result.

[0010] As a preferred solution of the standard-based fuel cost impact analysis system described in the present invention, the information collection and storage module is used to collect and cache the processing cost information and fuel cost information of the fuel, and record the cache capacity of the processing cost information and the fuel cost information. When the cache capacity reaches the set capacity value, the cached processing cost information and fuel cost information are compressed into a comprehensive package and sent to the cost calculation module.

[0011] As a preferred solution of the standard-based fuel cost impact analysis system of the present invention, wherein: the processing cost information includes processing personnel information, processing equipment information, and processing energy consumption information;

[0012] The processing personnel information includes the number of personnel and the corresponding personnel's salary;

[0013] The processing equipment information includes the number of processing equipment and the corresponding equipment purchase funds;

[0014] The processing energy consumption information includes unit energy consumption quantity and corresponding unit energy consumption funds;

[0015] The fuel cost information includes the remaining fuel amount of the fuel warehouse corresponding to the single generator set when the set capacity value starts, and the remaining fuel amount of the fuel warehouse corresponding to the single generator set when the set capacity value ends.

[0016] As a preferred solution of the standard-based fuel cost impact analysis system of the present invention, wherein: the first processing cost of the processing cost information specifically includes total expenditure on personnel salaries, total expenditure on equipment costs and total expenditure on energy consumption;

[0017] The total expenditure on personnel salaries = number of personnel * corresponding personnel salaries;

[0018] The total equipment cost expenditure = number of processing equipment * corresponding equipment purchase funds;

[0019] The total energy consumption expenditure = unit energy consumption quantity * corresponding unit energy consumption funds;

[0020] The first fuel cost of the fuel cost information is = unit fuel funds * number of generator sets * (the remaining fuel amount of the fuel warehouse corresponding to a single generator set when the set capacity value ends - the remaining fuel amount of the fuel warehouse corresponding to a single generator set when the set capacity value starts);

[0021] The actual comprehensive cost is = first processing cost + first fuel cost.

[0022] As a preferred solution of the standard-based fuel cost impact analysis system of the present invention, wherein: the cost impact data are merged according to time characteristics, and the cost impact data are preprocessed, and the preprocessing includes missing value filling, outlier processing and invalid data removal;

[0023] Extract trend characteristic parameters from the pre-processed cost impact data in unit time;

[0024] Dividing the cost impact data into a training set and a test set according to a certain ratio;

[0025] A comprehensive cost prediction model for predicting comprehensive cost is trained, wherein the comprehensive cost prediction model takes actual comprehensive cost as a training target, and the comprehensive cost prediction model is tested until the accuracy reaches a first threshold.

[0026] As a preferred solution of the standard-based fuel cost impact analysis system of the present invention, the method of extracting trend characteristic parameters from the pre-processed cost impact data according to unit time specifically includes:

[0027] The cost impact data of the first fuel cost and the first processing cost include supply and demand relationship data, external market data, environmental data and policy data;

[0028] Input the supply and demand relationship data, external market data, environmental data and policy data into a linear regression model respectively to obtain corresponding residual data and regression coefficients of the supply and demand relationship data, external market data, environmental data and policy data;

[0029] The mathematical expression of the linear regression model is as follows:

[0030] y=β0+β1x+∈

[0031] Among them, y represents the comprehensive cost per unit time, x represents the supply and demand relationship data, external market data, environmental data and policy data per unit time, β0 represents the residual data per unit time, β1 represents the regression coefficient per unit time, and ∈ represents the error data per unit time;

[0032] The corresponding residual data and regression coefficients of the supply and demand relationship data, external market data, environmental data and policy data are recorded as Y1, Y2, Y3 and Y4, and the corresponding trend characteristic parameters per unit time are obtained. The trend characteristic parameters are subjected to feature interaction to obtain interactive feature data;

[0033] The interactive feature data is used as input of the comprehensive cost prediction model, and the output is the predicted comprehensive cost.

[0034] As a preferred solution of the standard-based fuel cost impact analysis system described in the present invention, the time length of a unit time is the time point when the set capacity value ends minus the time point when the set capacity value starts.

[0035] As a preferred solution of the standard-based fuel cost impact analysis system of the present invention, wherein: the cost control module is used to analyze the predicted cost difference to obtain a comparison result, and the comparison result includes an increase result, a decrease result and a constant result;

[0036] When the predicted cost difference / actual comprehensive cost is greater than 0, the comparison result is an increasing result, and the comprehensive cost increases;

[0037] When the predicted cost difference / actual comprehensive cost is less than 0, the comparison result is a decreasing result, and the comprehensive cost decreases;

[0038] When the predicted cost difference / actual comprehensive cost is equal to 0, the comparison result is an unchanged result, and the comprehensive cost remains unchanged.

[0039] As a preferred solution of the standard-based fuel cost impact analysis system of the present invention, the comprehensive cost prediction model is constructed based on the LSTM neural network, and the process of constructing the comprehensive cost prediction model specifically includes:

[0040] The dimensions of the input layer, hidden layer and output layer of the LSTM neural network are set. The dimension of the hidden layer of the LSTM neural network is smaller than the dimension of the input layer. The supply and demand relationship data, external market data, environmental data and policy data are described with a smaller dimension. The LSTM neural network performs a dimensionality reduction operation from the input layer to the hidden layer.

[0041] Set the dimension format of the input data to batch size, number of time steps, number of features, and use the interactive feature data as the input of the LSTM layer;

[0042] Pass the input data to the hidden layer, set the number of hidden layers and the number of LSTM units in each layer;

[0043] Get the output of the last time step from the hidden state as the input of the fully connected layer;

[0044] The output of the fully connected layer is passed to the Softmax layer to obtain the final output probability distribution to obtain the predicted comprehensive cost. The LSTM neural network is trained by a back propagation algorithm to minimize the error between the predicted comprehensive cost and the actual comprehensive cost.

[0045] As a preferred solution of the standard-based fuel cost impact analysis system of the present invention, wherein: the predicted cost difference = predicted comprehensive cost - actual comprehensive cost;

[0046] The LSTM unit includes a memory cell, an input gate, a forget gate and an output gate.

[0047] The beneficial effects of the present invention are as follows: by collecting and storing fuel processing cost information and fuel cost information, cost information of fuel cost can be obtained, fuel cost is divided into two cost parts, namely, a first processing cost generated in the fuel processing process and a first fuel cost of the fuel itself, the first processing cost is added to the first fuel cost to obtain the actual comprehensive cost, cost impact data affecting the first processing cost and the first fuel cost are collected, the cost impact data include supply and demand relationship data, external market data, environmental data and policy data, changes in the cost impact data will lead to changes in the fuel cost, the cost impact data per unit time is obtained, a regression model is used to calculate the change characteristic parameters of each cost impact data per unit time, the change characteristic parameters include residual data and regression coefficients, a comprehensive cost prediction model is established according to the change characteristic parameters of the cost impact data and the corresponding change in the actual comprehensive cost, the change characteristic parameters of the supply and demand relationship data, the external market data, the environmental data and the policy data are subjected to feature interaction, the interaction of the cost impact data on the actual comprehensive cost after the mutual change is obtained, the subsequent actual comprehensive cost is predicted, a reliable basis is provided for subsequent cost control, the cost change trend under the influence of multiple cost impact data such as overall environmental supply and demand changes, external market changes, environmental changes and policy changes can be predicted, and a reference value of great significance is provided for the control of fuel cost. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 A basic flow chart of a standard-based fuel cost impact analysis system provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0049] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings. It is obvious that the described embodiments are only part of the embodiments of the present invention, but not all of the embodiments.

[0050] Example, see Figure 1, as one embodiment of the present invention, provides a standard-based fuel cost impact analysis system, including an information collection and storage module, a cost calculation module, a cost prediction module and a cost evaluation module;

[0051] The information collection and storage module is used to collect and store fuel processing cost information and fuel cost information;

[0052] The cost calculation module is used to calculate the first processing cost of the processing cost information and the first fuel cost of the fuel cost information to obtain the actual comprehensive cost;

[0053] The cost prediction module is used to analyze the cost impact data of the first fuel cost and the first processing cost, construct a comprehensive cost prediction model, and obtain a predicted comprehensive cost;

[0054] The cost control module is used to analyze and predict the cost difference, obtain a comparison result, and issue a cost warning signal command according to the comparison result.

[0055] The information collection and storage module is used to collect and cache fuel processing cost information and fuel cost information, record the cache capacity of the processing cost information and fuel cost information, and when the cache capacity reaches a set capacity value, compress the cached processing cost information and fuel cost information into a comprehensive package and send it to the cost calculation module.

[0056] In this embodiment, the financial information database of the factory is called up to read the processing cost information and fuel cost information. The financial information database can automatically record and track all relevant cost information, including the processing cost information and fuel cost information of the fuel, and can grasp the specific expenditure of each cost item. The processing cost information and fuel cost information are cached. When the processing cost information and fuel cost information reach a preset amount of information, the cached processing cost information and fuel cost information are compressed into a comprehensive package and sent to the cost calculation module.

[0057] The processing cost information includes processing personnel information, processing equipment information, and processing energy consumption information;

[0058] The processing personnel information includes the number of personnel and the corresponding personnel's salary;

[0059] The processing equipment information includes the number of processing equipment and the corresponding equipment purchase funds;

[0060] The processing energy consumption information includes unit energy consumption quantity and corresponding unit energy consumption funds;

[0061] The fuel cost information includes the remaining fuel amount of the fuel warehouse corresponding to the single generator set when the set capacity value starts, and the remaining fuel amount of the fuel warehouse corresponding to the single generator set when the set capacity value ends.

[0062] The first processing cost of the processing cost information specifically includes total expenditure on personnel salaries, total expenditure on equipment costs and total expenditure on energy consumption;

[0063] The total expenditure on personnel salaries = number of personnel * corresponding personnel salaries;

[0064] The total equipment cost expenditure = number of processing equipment * corresponding equipment purchase funds;

[0065] The total energy consumption expenditure = unit energy consumption quantity * corresponding unit energy consumption funds;

[0066] The first fuel cost of the fuel cost information is = unit fuel funds * number of generator sets * (the remaining fuel amount of the fuel warehouse corresponding to a single generator set when the set capacity value ends - the remaining fuel amount of the fuel warehouse corresponding to a single generator set when the set capacity value starts);

[0067] The actual comprehensive cost is = first processing cost + first fuel cost.

[0068] Merging the cost impact data according to time characteristics, and preprocessing the cost impact data, wherein the preprocessing includes filling missing values, processing outliers, and removing invalid data;

[0069] Extract trend characteristic parameters from the pre-processed cost impact data in unit time;

[0070] Dividing the cost impact data into a training set and a test set according to a certain ratio;

[0071] A comprehensive cost prediction model for predicting comprehensive cost is trained, wherein the comprehensive cost prediction model takes actual comprehensive cost as a training target, and the comprehensive cost prediction model is tested until the accuracy reaches a first threshold.

[0072] In this embodiment, the cost impact data within the same time period with a fixed duration is collected, and missing values ​​are filled, outliers are processed, and invalid data is removed for the cost impact data, which can ensure the integrity of the data, reduce the deviation of model prediction, and improve the quality of the data, so that the subsequent analysis and modeling process will not be interrupted due to missing data. The trend characteristic parameters of the cost impact data are extracted, and the characteristics of the changing trend of the cost impact data within a unit time can be obtained. The cost impact data is divided into a training set and a test set in a certain ratio of 7:3, which can enhance the robustness of the model. The trend characteristic parameters are input into the comprehensive cost prediction model, and the predicted comprehensive cost under the trend characteristic parameters is output. The comprehensive cost prediction model is trained using the training set until the accuracy reaches 0.95 and stops.

[0073] The trend characteristic parameters extracted from the pre-processed cost impact data in unit time include:

[0074] The cost impact data of the first fuel cost and the first processing cost include supply and demand relationship data, external market data, environmental data and policy data;

[0075] Input the supply and demand relationship data, external market data, environmental data and policy data into a linear regression model respectively to obtain corresponding residual data and regression coefficients of the supply and demand relationship data, external market data, environmental data and policy data;

[0076] The mathematical expression of the linear regression model is as follows:

[0077] y=β0+β1x+∈

[0078] Among them, y represents the comprehensive cost per unit time, x represents the supply and demand relationship data, external market data, environmental data and policy data per unit time, β0 represents the residual data per unit time, β1 represents the regression coefficient per unit time, and ∈ represents the error data per unit time;

[0079] The corresponding residual data and regression coefficients of the supply and demand relationship data, external market data, environmental data and policy data are recorded as Y1, Y2, Y3 and Y4, and the corresponding trend characteristic parameters per unit time are obtained. The trend characteristic parameters are subjected to feature interaction to obtain interactive feature data;

[0080] The interactive feature data is used as input of the comprehensive cost prediction model, and the output is the predicted comprehensive cost.

[0081] In this embodiment, changes in supply and demand relationship data, external market data, environmental data and policy data can all cause changes in fuel costs. The corresponding residual data and regression coefficients of the supply and demand relationship data, external market data, environmental data and policy data are calculated using a linear regression model and recorded as corresponding trend characteristic parameters. The corresponding residual data and regression coefficients can characterize the relationship between changes in the supply and demand relationship data, external market data, environmental data and policy data within a unit time and changes in the comprehensive cost within a unit time. The characterization relationship at this time is one-to-one, that is, changes in the supply and demand relationship data within a unit time cause changes in the comprehensive cost within a unit time. The trend characteristic parameters of the supply and demand relationship data, external market data, environmental data and policy data are subjected to feature interaction to obtain interactive feature data. The characterization relationship at this time is many-to-one, and the cost change trend under the influence of multiple cost influencing data such as overall environmental supply and demand changes, external market changes, environmental changes and policy changes is obtained.

[0082] The time length of a unit time is the time point when the set capacity value ends minus the time point when the set capacity value starts.

[0083] In this embodiment, when the set capacity value ends, the cache capacity of the collected processing cost information and material cost information reaches the preset information capacity. When the set capacity value starts, the last information collection stage has just passed one cycle. The end time point and the start time point of one cycle of the information collection stage are subtracted to obtain the unit time length of this information collection stage.

[0084] The cost control module is used to analyze the predicted cost difference and obtain a comparison result, which includes an increase result, a decrease result and a constant result;

[0085] When the predicted cost difference / actual comprehensive cost is greater than 0, the comparison result is an increasing result, and the comprehensive cost increases;

[0086] When the predicted cost difference / actual comprehensive cost is less than 0, the comparison result is a decreasing result, and the comprehensive cost decreases;

[0087] When the predicted cost difference / actual comprehensive cost is equal to 0, the comparison result is an unchanged result, and the comprehensive cost remains unchanged.

[0088] In this embodiment, when the difference between the predicted cost and the actual comprehensive cost is greater than 0, the warning signal indicator light is red, indicating that the comprehensive cost control is dangerous;

[0089] When the difference between the predicted cost and the actual comprehensive cost is less than 0, the warning signal indicator turns green, indicating that the comprehensive cost control is excellent;

[0090] When the predicted cost difference / actual comprehensive cost is equal to 0, the early warning signal indicator is yellow, indicating that the comprehensive cost control is general. When the early warning signal indicator is yellow and green, there is no need to adjust the cost control. When the early warning signal indicator is red, it indicates that the subsequent fuel cost will increase, and the cost control needs to be adjusted according to the actual situation to keep the fuel cost within the normal range.

[0091] The comprehensive cost prediction model is constructed based on the LSTM neural network. The process of constructing the comprehensive cost prediction model specifically includes:

[0092] The dimensions of the input layer, hidden layer and output layer of the LSTM neural network are set. The dimension of the hidden layer of the LSTM neural network is smaller than the dimension of the input layer. The supply and demand relationship data, external market data, environmental data and policy data are described with a smaller dimension. The LSTM neural network performs a dimensionality reduction operation from the input layer to the hidden layer.

[0093] Set the dimension format of the input data to batch size, number of time steps, number of features, and use the interactive feature data as the input of the LSTM layer;

[0094] Pass the input data to the hidden layer, set the number of hidden layers and the number of LSTM units in each layer;

[0095] Get the output of the last time step from the hidden state as the input of the fully connected layer;

[0096] The output of the fully connected layer is passed to the Softmax layer to obtain the final output probability distribution to obtain the predicted comprehensive cost. The LSTM neural network is trained by a back propagation algorithm to minimize the error between the predicted comprehensive cost and the actual comprehensive cost.

[0097] In this embodiment, the dimension of the hidden layer of the LSTM neural network encoder is smaller than the dimension of the input layer. The encoder can compress the data volume of the input image, convert high-dimensional image data into low-dimensional feature representation, retain key feature information, reduce the amount of calculation in subsequent processing, and improve processing efficiency.

[0098] The predicted cost difference = predicted comprehensive cost - actual comprehensive cost;

[0099] The LSTM unit contains a memory cell, an input gate, a forget gate and an output gate.

[0100] In this embodiment, the LSTM neural network can avoid gradient disappearance and gradient explosion, capture the relationship between cost impact data in time series data and fuel cost data with long-term trends and periodic fluctuations, and process fuel cost information and processing cost information at different time steps. The unit time of the information collection stage depends on the time difference between the cache capacity of the collected processing cost information and material cost information and the preset information capacity. The LSTM neural network has higher flexibility and higher prediction accuracy when processing time series data.

[0101] By collecting and storing fuel processing cost information and fuel cost information, the cost information of fuel cost can be obtained, and the fuel cost is divided into two cost parts: the first processing cost generated by the fuel processing process and the first fuel cost of the fuel itself. The first processing cost is added to the first fuel cost to obtain the actual comprehensive cost, and the cost impact data affecting the first processing cost and the first fuel cost are collected. The cost impact data include supply and demand relationship data, external market data, environmental data and policy data. Changes in cost impact data will lead to changes in fuel costs. The cost impact data per unit time is obtained, and a regression model is used to calculate the change characteristic parameters of each cost impact data per unit time. The change characteristic parameters include residual data and regression coefficients. According to the change characteristic parameters of the cost impact data and the corresponding changes in the actual comprehensive cost, a comprehensive cost prediction model is established, and the change characteristic parameters of the supply and demand relationship data, external market data, environmental data and policy data are interactively characterized to obtain the interactive effect of the cost impact data on the actual comprehensive cost after mutual changes, and the subsequent actual comprehensive cost is predicted to provide a reliable basis for subsequent cost control. It can predict the cost change trend under the influence of multiple cost impact data such as overall environmental supply and demand changes, external market changes, environmental changes and policy changes, and provide a reference value of great significance for the control of fuel costs.

[0102] It should be understood by those skilled in the art that the embodiments of the present invention can be provided as methods, systems or computer program products. Therefore, the present invention can take the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media containing computer-usable program codes. Among them, the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (Static Random Access Memory, referred to as SRAM), electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, referred to as EEPROM), erasable programmable read-only memory (Erasable Programmable Read Only Memory, referred to as EPROM), programmable read-only memory (Programmable Red-Only Memory, referred to as PROM), read-only memory (Read-Only Memory, referred to as ROM), magnetic memory, flash memory, disk or optical disk. These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0103] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A standards-based fuel cost impact analysis system, characterized by: include: Information collection and storage module, cost calculation module, cost forecasting module and cost control module; The information collection and storage module is used to collect and store fuel processing cost information and fuel cost information; The cost calculation module is used to calculate the first processing cost of the processing cost information and the first fuel cost of the fuel cost information to obtain the actual comprehensive cost; The cost prediction module is used to analyze the cost impact data of the first fuel cost and the first processing cost, construct a comprehensive cost prediction model, and obtain a predicted comprehensive cost; The cost control module is used to analyze and predict the cost difference, obtain a comparison result, and issue a cost warning signal command according to the comparison result.

2. The standards-based fuel cost impact analysis system of claim 1, wherein: The information collection and storage module is used to collect and cache fuel processing cost information and fuel cost information, record the cache capacity of the processing cost information and fuel cost information, and when the cache capacity reaches a set capacity value, compress the cached processing cost information and fuel cost information into a comprehensive package and send it to the cost calculation module.

3. The standards-based fuel cost impact analysis system of claim 2, wherein: The processing cost information includes processing personnel information, processing equipment information, and processing energy consumption information; The processing personnel information includes the number of personnel and the corresponding personnel's salary; The processing equipment information includes the number of processing equipment and the corresponding equipment purchase funds; The processing energy consumption information includes unit energy consumption quantity and corresponding unit energy consumption funds; The fuel cost information includes the remaining fuel amount of the fuel warehouse corresponding to the single generator set when the set capacity value starts, and the remaining fuel amount of the fuel warehouse corresponding to the single generator set when the set capacity value ends.

4. The standards-based fuel cost impact analysis system of claim 3, wherein: The first processing cost of the processing cost information specifically includes total expenditure on personnel salaries, total expenditure on equipment costs and total expenditure on energy consumption; The total expenditure on personnel salaries = number of personnel * corresponding personnel salaries; The total equipment cost expenditure = number of processing equipment * corresponding equipment purchase funds; The total energy consumption expenditure = unit energy consumption quantity * corresponding unit energy consumption funds; The first fuel cost of the fuel cost information is = unit fuel funds * number of generator sets * (the remaining fuel amount of the fuel warehouse corresponding to a single generator set when the set capacity value ends - the remaining fuel amount of the fuel warehouse corresponding to a single generator set when the set capacity value starts); The actual comprehensive cost is = first processing cost + first fuel cost.

5. The standards-based fuel cost impact analysis system of claim 4, wherein: Merging the cost impact data according to time characteristics, and preprocessing the cost impact data, wherein the preprocessing includes filling missing values, processing outliers, and removing invalid data; Extract trend characteristic parameters from the pre-processed cost impact data in unit time; Dividing the cost impact data into a training set and a test set according to a certain ratio; A comprehensive cost prediction model for predicting comprehensive cost is trained, wherein the comprehensive cost prediction model takes actual comprehensive cost as a training target, and the comprehensive cost prediction model is tested until the accuracy reaches a first threshold.

6. The standards-based fuel cost impact analysis system of claim 5, wherein: The trend characteristic parameters extracted from the pre-processed cost impact data in unit time include: The cost impact data of the first fuel cost and the first processing cost include supply and demand relationship data, external market data, environmental data and policy data; Input the supply and demand relationship data, external market data, environmental data and policy data into a linear regression model respectively to obtain corresponding residual data and regression coefficients of the supply and demand relationship data, external market data, environmental data and policy data; The mathematical expression of the linear regression model is as follows: y=β0+β1x+∈ Among them, y represents the comprehensive cost per unit time, x represents the supply and demand relationship data, external market data, environmental data and policy data per unit time, β0 represents the residual data per unit time, β1 represents the regression coefficient per unit time, and ∈ represents the error data per unit time; The corresponding residual data and regression coefficients of the supply and demand relationship data, external market data, environmental data and policy data are recorded as Y1, Y2, Y3 and Y4, and the corresponding trend characteristic parameters per unit time are obtained. The trend characteristic parameters are subjected to feature interaction to obtain interactive feature data; The interactive feature data is used as input of the comprehensive cost prediction model, and the output is the predicted comprehensive cost.

7. The standards-based fuel cost impact analysis system of claim 6, wherein: The time length of a unit time is the time point when the set capacity value ends minus the time point when the set capacity value starts.

8. The standards-based fuel cost impact analysis system of claim 7, wherein: The cost control module is used to analyze the predicted cost difference and obtain a comparison result, which includes an increase result, a decrease result and a constant result; When the predicted cost difference / actual comprehensive cost is greater than 0, the comparison result is an increasing result, and the comprehensive cost increases; When the predicted cost difference / actual comprehensive cost is less than 0, the comparison result is a decreasing result, and the comprehensive cost decreases; When the predicted cost difference / actual comprehensive cost is equal to 0, the comparison result is an unchanged result, and the comprehensive cost remains unchanged.

9. The standards-based fuel cost impact analysis system of claim 8, wherein: The comprehensive cost prediction model is constructed based on the LSTM neural network. The process of constructing the comprehensive cost prediction model specifically includes: The dimensions of the input layer, hidden layer and output layer of the LSTM neural network are set. The dimension of the hidden layer of the LSTM neural network is smaller than the dimension of the input layer. The supply and demand relationship data, external market data, environmental data and policy data are described with a smaller dimension. The LSTM neural network performs a dimensionality reduction operation from the input layer to the hidden layer. Set the dimension format of the input data to batch size, number of time steps, number of features, and use the interactive feature data as the input of the LSTM layer; Pass the input data to the hidden layer, set the number of hidden layers and the number of LSTM units in each layer; Get the output of the last time step from the hidden state as the input of the fully connected layer; The output of the fully connected layer is passed to the Softmax layer to obtain the final output probability distribution to obtain the predicted comprehensive cost. The LSTM neural network is trained by a back propagation algorithm to minimize the error between the predicted comprehensive cost and the actual comprehensive cost.

10. The standards-based fuel cost impact analysis system of claim 9, wherein: The predicted cost difference = predicted comprehensive cost - actual comprehensive cost; The LSTM unit includes a memory cell, an input gate, a forget gate and an output gate.

Images