Asset full life cycle management system

Abstract

The invention belongs to the technical field, and discloses an asset full life cycle management system comprising a data acquisition module used for acquiring asset information data, asset economic data and external environment data; the data processing module is used for preprocessing and fusing the acquired asset information data, asset economic data and external environment data to obtain a comprehensive feature data set; the asset full-life-cycle prediction module is used for training according to the comprehensive feature data set to obtain an asset full-life-cycle prediction model, and predicting the asset state change rate through the asset full-life-cycle prediction model; the resource scheduling module is used for judging whether asset risk early warning information is generated or not according to the asset state change rate; if the asset risk early warning information is generated, a state change rate trend is extracted through statistical analysis, and resource allocation is dynamically adjusted based on the state change rate trend; the resources can flow to assets needing to be supported more accurately, and resource waste is avoided.

Description

Technical Field

[0001] The present invention relates to the technical field, and more specifically, to an asset full life cycle management system. Background Art

[0002] The patent publication number CN113888293A discloses an asset life cycle management system, where the acquisition terminal is used to acquire asset information; the server is used to establish a unique identity for the asset, asset valuation, asset change assessment, asset change expectation analysis, financial management and data query during the asset life cycle; each blockchain node records the status information and change information of each asset during its asset life cycle; the management terminal is used to interact with the server and also to obtain the status information and change information of the asset information from each blockchain node. Asset management is achieved during the asset life cycle by adopting a remote interaction method; the server data management method reduces manpower input; the blockchain node is used for data storage to achieve asset data tamper-proof; and the server can actively perform automated asset change assessment and asset change expectation analysis, improve the degree of automation and the computing efficiency of large data volumes, and avoid the possibility of asset waste.

[0003] The existing asset life cycle management system has the following main problems:

[0004] Failure to consider the correlation between features and target variables may result in the selected features being unable to accurately reflect changes in the target variables, thereby reducing the generalization ability of the model; failure to consider the redundancy between features may result in the model containing a large number of redundant features, which not only do not provide additional information, but may also increase the complexity of the model, thereby increasing the risk of overfitting; failure to set a redundancy tolerance threshold or failure to make adaptive adjustments based on the redundancy correction function may result in the feature selection process being too rigid and unable to adapt to the feature distribution of different data sets; failure to dynamically adjust the redundancy tolerance threshold based on the relationship between the number of currently selected features and the total number of features may result in the feature selection process being unable to flexibly adjust as the data set changes;

[0005] If resource allocation is not adjusted dynamically, resources may be wasted on assets with stable status or low demand, while assets that degrade quickly or have high demand will not receive sufficient resource support, and limited resources cannot be optimally allocated, reducing overall utilization efficiency. If assets that degrade quickly are not identified through the status change rate trend, resource input will not be increased in a timely manner, resulting in serious degradation and even irreversible failure. The failure of key assets may trigger a chain reaction and affect the operation of the overall system. The lack of real-time monitoring and the inability to respond quickly to changes in asset status may lead to the accumulation of problems, which will ultimately require higher costs for remediation. The inability to adapt to changes in demand at different stages of the asset life cycle will result in asset scrapping ahead of schedule and increase asset replacement costs.

[0006] In view of this, the present invention proposes an asset life cycle management system to solve the above problems. Summary of the invention

[0007] In order to overcome the above-mentioned defects of the prior art and to achieve the above-mentioned purpose, the present invention provides the following technical solution: an asset life cycle management system, comprising:

[0008] Data collection module, used to collect asset information data, asset economic data and external environment data;

[0009] The data processing module is used to pre-process the collected asset information data, asset economic data and external environment data to obtain an asset information feature data set, an asset economic feature data set and an external environment feature data set; and to fuse the asset information feature data set, the asset economic feature data set and the external environment feature data set to obtain a comprehensive feature data set;

[0010] The asset life cycle prediction module is used to obtain the asset life cycle prediction model based on the comprehensive feature data set training, and to predict the asset status change rate through the asset life cycle prediction model;

[0011] The resource scheduling module is used to determine whether to generate asset risk warning information based on the asset status change rate; if asset risk warning information is generated, the status change rate trend is extracted through statistical analysis, and resource allocation is dynamically adjusted based on the status change rate trend; each module is connected by wired and / or wireless means.

[0012] Furthermore, the asset information data includes the type, specifications, purchase date and usage status of the asset; the asset economic data includes the purchase cost, depreciation and maintenance costs of the asset; and the external environment data includes market interest rates, industry development trends and changes in policies and regulations.

[0013] Furthermore, the method of preprocessing the collected asset information data, asset economic data and external environment data to obtain the asset information feature data set, the asset economic feature data set and the external environment feature data set includes:

[0014] The missing values ​​in the asset information data, asset economic data and external environment data are filled by the mean filling method, and the outliers in the asset information data, asset economic data and external environment data are detected and eliminated by the box plot method; and the standard deviation is normalized and converted into a standard normal distribution;

[0015] The improved mRMR algorithm is used to analyze the correlation between asset information data, asset economic data, external environment data and asset status change rate, while minimizing the redundancy between the features in asset information data, asset economic data and external environment data, thereby obtaining asset information feature data set, asset economic feature data set and external environment feature data set; the specific steps are as follows:

[0016] Define the target variable Y, where Y is the asset status change rate; calculate the mutual information between each feature f in the asset information data, asset economic data and external environment data and the target variable Y by maximizing the correlation formula;

[0017] The formula for maximizing correlation is: Among them, p(f,Y) is the joint probability distribution of feature f and target variable Y, which indicates the probability of feature f and target variable Y occurring simultaneously; p(f) is the marginal probability distribution of feature f, which indicates the probability of feature f occurring independently; p(Y) is the marginal probability distribution of target variable Y, which indicates the probability of target variable Y occurring independently; I(f,Y) is the mutual information between feature f and target variable Y, which is used to measure the correlation between feature f and target variable Y;

[0018] The redundancy between features is calculated by minimizing the redundancy formula, that is, the mutual information between feature f and feature f′; the minimizing redundancy formula is: Among them, p(f,f′) is the joint probability distribution between feature f and feature f′, which indicates the probability of feature f and feature f′ occurring simultaneously; p(f) and p(f′) are the marginal probability distributions of feature f and feature f′, respectively, which indicate the probability of feature f and feature f′ occurring independently; I(f,f′) is the mutual information between feature f and feature f′, which is used to quantify the redundancy between feature f and feature f′;

[0019] By improving the optimization objective function, the most relevant and least redundant features are selected to form the feature subset S; the improved optimization objective function is: Among them, J(S) is the improved optimization objective function; |S| is the number of features in the feature subset S; g(f,f′) is the redundancy correction function, which is dynamically adjusted according to the redundancy tolerance between features;

[0020] Set a redundancy tolerance threshold, dynamically adjust the redundancy tolerance between features according to the redundancy correction function, and adaptively adjust the set redundancy tolerance threshold;

[0021] The redundancy correction function is: Among them, τ is the set redundancy tolerance threshold.

[0022] Furthermore, the method for adaptively adjusting the set redundancy tolerance threshold includes:

[0023] The set redundancy tolerance threshold is adaptively adjusted through the redundancy tolerance threshold adjustment formula. The redundancy tolerance threshold adjustment formula is: Among them, τ′ is the adjusted redundancy tolerance threshold; n cr is the number of features currently selected; n tl is the total number of features; τ max is the maximum redundancy tolerance threshold.

[0024] Furthermore, the method of fusing the asset information feature data set, the asset economic feature data set and the external environment feature data set to obtain a comprehensive feature data set includes:

[0025] The normalized asset information feature data set, asset economic feature data set and external environment feature data set are fused through a weighted model to obtain a comprehensive feature data set; the asset information feature data set is recorded as K1, the asset economic feature data set is recorded as K2, and the external environment feature data set is recorded as K3;

[0026] The weighted model is: ZC=K1·ω1+K2·ω2+K3·ω3; wherein ω1 is the weight coefficient of the asset information feature data set; ω2 is the weight coefficient of the asset economic feature data set; and ω3 is the weight coefficient of the external environment feature data set.

[0027] Furthermore, the training method of the asset life cycle prediction model includes:

[0028] The data set is divided into a training set, a validation set, and a test set; the data set includes a historical comprehensive feature data set and a corresponding asset status change rate; a sliding window method is used to generate a sample set, which is a subset of the data set, and each sample set includes a historical comprehensive feature data set and a corresponding asset status change rate within a period of time;

[0029] The asset life cycle prediction model is constructed using the LSTM model, which specifically includes an input layer, an LSTM layer, a fully connected layer, and an output layer. The input layer is used to input a historical comprehensive feature data set. The number of neurons in the input layer matches the number of features in the historical comprehensive feature data set. The LSTM layer is used to receive data from the input layer and capture the temporal characteristics of the asset state changes during the life cycle. The learning ability of the model is adjusted by changing the number of LSTM units and layers. The fully connected layer is used to map the output of the LSTM layer to the target prediction value. The output layer uses linear as the activation function to output the asset state change rate.

[0030] Use mean square error as the loss function to measure the error of model prediction; train the asset lifecycle prediction model on the training set and minimize the loss function through the Adam optimizer; use the validation set to evaluate the generalization ability of the model, and tune the model's hyperparameters based on the evaluation results until the preset stopping condition is reached, thus obtaining the final trained asset lifecycle prediction model;

[0031] The current comprehensive feature data set is input into the trained asset life cycle prediction model to predict the corresponding asset status change rate.

[0032] Furthermore, the method for determining whether to generate asset risk warning information according to the asset status change rate includes:

[0033] If the predicted asset status change rate is greater than or equal to the preset asset status change rate threshold, the asset is determined to be in an abnormal state and asset risk warning information is generated;

[0034] If the predicted asset status change rate is less than the preset asset status change rate threshold, the asset is determined to be in a healthy state and no asset risk warning information is generated.

[0035] Furthermore, if the asset risk warning information is generated, the method of extracting the state change rate trend through statistical analysis includes performing fluctuation analysis and second-order derivative calculation on the asset state change rate, and then obtaining the change rate fluctuation amplitude, change rate fluctuation cycle and asset degradation acceleration; based on the change rate fluctuation amplitude, change rate fluctuation cycle and asset degradation acceleration, the state change rate trend is reflected.

[0036] Furthermore, the method for dynamically adjusting resource allocation based on the state change rate trend includes constructing a resource allocation adjustment formula according to the state change rate trend to dynamically adjust resource allocation.

[0037] Furthermore, the method for constructing a resource allocation adjustment formula according to a state change rate trend includes:

[0038] The resource allocation adjustment formula is: Among them, α is the resource allocation adjustment coefficient; ΔR is the fluctuation amplitude of the change rate; T is the fluctuation period of the change rate; A is the acceleration of asset degradation; k1 and k2 are adjustment parameters; e is the base of the exponential function;

[0039] A resource allocation adjustment coefficient threshold β is preset. When β≤α≤1, the resource allocation ratio of the current asset needs to be increased; when 0≤α<β, the resource allocation ratio of the current asset needs to be reduced; the obtained resource allocation adjustment coefficient is normalized and the resource allocation ratio is recalculated to perform resource allocation.

[0040] Technical effects and advantages of an asset life cycle management system of the present invention:

[0041] The present invention not only considers the correlation between features and target variables, but also considers the redundancy between features by improving the mRMR algorithm, so that the most relevant feature subset with the lowest redundancy can be selected; it is helpful to simplify the model, improve the generalization ability of the model, and reduce the computational complexity; the selected feature subset is more representative, which helps to better understand the relationship between the asset state change rate and asset information, asset economy and external environment; a redundancy tolerance threshold is set, and adaptive adjustment is performed according to the redundancy correction function, so that the feature selection process is more flexible and adaptive; the redundancy tolerance threshold is dynamically adjusted according to the relationship between the number of currently selected features and the total number of features; this means that as the feature selection process proceeds, the system can more flexibly deal with the redundancy in the data set;

[0042] By analyzing the trend of the status change rate, resource allocation is adjusted dynamically so that resources can flow more accurately to the assets that need support; avoid wasting resources and prioritize the allocation of limited resources to assets with faster status degradation; improve resource utilization efficiency and reduce redundant allocation of resources to low-demand assets; use dynamic adjustment formulas and real-time monitoring mechanisms to quickly respond to changing trends in asset status and adjust resource allocation ratios; adjust resources according to real-time changes in asset status to prevent greater losses due to failure to respond in time to status deterioration; the threshold mechanism ensures that resource allocation adjustments are reasonable and will not change frequently due to short-term fluctuations. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 This is a schematic diagram of the structure of an asset full life cycle management system of the present invention;

[0044] Figure 2 The figure is a flow chart of an asset life cycle management method of the present invention. DETAILED DESCRIPTION

[0045] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0046] Embodiment 1

[0047] See also Figure 1 As shown, the asset life cycle management system described in this embodiment includes:

[0048] Data collection module, used to collect asset information data, asset economic data and external environment data;

[0049] The data processing module is used to pre-process the collected asset information data, asset economic data and external environment data to obtain an asset information feature data set, an asset economic feature data set and an external environment feature data set; and to fuse the asset information feature data set, the asset economic feature data set and the external environment feature data set to obtain a comprehensive feature data set;

[0050] The asset life cycle prediction module is used to obtain the asset life cycle prediction model based on the comprehensive feature data set training, and to predict the asset status change rate through the asset life cycle prediction model;

[0051] The resource scheduling module is used to determine whether to generate asset risk warning information based on the asset status change rate; if asset risk warning information is generated, the status change rate trend is extracted through statistical analysis, and resource allocation is dynamically adjusted based on the status change rate trend; each module is connected by wired and / or wireless means.

[0052] The asset information data includes the type, specifications, purchase date and usage status of the asset; the asset economic data includes the purchase cost, depreciation and maintenance costs of the asset; the external environment data includes market interest rates, industry development trends and changes in policies and regulations.

[0053] The method of preprocessing the collected asset information data, asset economic data and external environment data to obtain the asset information feature data set, the asset economic feature data set and the external environment feature data set includes:

[0054] The missing values ​​in the asset information data, asset economic data and external environment data are filled by the mean filling method, and the outliers in the asset information data, asset economic data and external environment data are detected and eliminated by the box plot method; and the standard deviation is normalized and converted into a standard normal distribution;

[0055] The improved mRMR algorithm is used to analyze the correlation between asset information data, asset economic data, external environment data and asset status change rate, while minimizing the redundancy between the features in asset information data, asset economic data and external environment data, thereby obtaining asset information feature data set, asset economic feature data set and external environment feature data set; the specific steps are as follows:

[0056] Define the target variable Y, where Y is the asset status change rate; calculate the mutual information between each feature f in the asset information data, asset economic data and external environment data and the target variable Y by maximizing the correlation formula;

[0057] The formula for maximizing correlation is: Among them, p(f,Y) is the joint probability distribution of feature f and target variable Y, which indicates the probability of feature f and target variable Y occurring simultaneously; p(f) is the marginal probability distribution of feature f, which indicates the probability of feature f occurring independently; p(Y) is the marginal probability distribution of target variable Y, which indicates the probability of target variable Y occurring independently; I(f,Y) is the mutual information between feature f and target variable Y, which is used to measure the correlation between feature f and target variable Y;

[0058] The redundancy between features is calculated by minimizing the redundancy formula, that is, the mutual information between feature f and feature f′; the minimizing redundancy formula is: Among them, p(f,f′) is the joint probability distribution between feature f and feature f′, which indicates the probability of feature f and feature f′ occurring simultaneously; p(f) and p(f′) are the marginal probability distributions of feature f and feature f′, respectively, which indicate the probability of feature f and feature f′ occurring independently; I(f,f′) is the mutual information between feature f and feature f′, which is used to quantify the redundancy between feature f and feature f′. A higher mutual information means that the two features provide similar information, so the redundancy is higher.

[0059] By improving the optimization objective function, the most relevant and least redundant features are selected to form the feature subset S; the improved optimization objective function is: Among them, J(S) is the improved optimization objective function; |S| is the number of features in the feature subset S; g(f,f′) is the redundancy correction function, which is dynamically adjusted according to the redundancy tolerance between features;

[0060] Set a redundancy tolerance threshold, dynamically adjust the redundancy tolerance between features according to the redundancy correction function, and adaptively adjust the set redundancy tolerance threshold;

[0061] The redundancy correction function is: Among them, τ is the set redundancy tolerance threshold;

[0062] For example, the redundancy tolerance threshold is set to 0.6, and the mutual information I(f,f′) between feature f and feature f′ is 0.8; because 0.8>0.6, according to the redundancy correction function g(f,f′)=1, this means that the redundancy of these two features is high, and one of the features should be reduced;

[0063] If the mutual information I(f,f′) between feature f and feature f′ is 0.4, according to the redundancy correction function This indicates that the redundancy of these two features is low and their coexistence can be tolerated.

[0064] The method for adaptively adjusting the set redundancy tolerance threshold comprises:

[0065] The set redundancy tolerance threshold is adaptively adjusted through the redundancy tolerance threshold adjustment formula. The redundancy tolerance threshold adjustment formula is: Among them, τ′ is the adjusted redundancy tolerance threshold; n cr is the number of features currently selected; n tl is the total number of features; τ max is the maximum redundancy tolerance threshold;

[0066] For example, if the total number of features is 50, the number of currently selected features is 30, and the maximum redundancy tolerance threshold is 0.8, then the adjusted redundancy tolerance threshold is

[0067] The total number of features is 50, the number of currently selected features is 45, and the maximum redundancy tolerance threshold is 0.8. Then the adjusted redundancy tolerance threshold is At this time, the redundancy tolerance threshold is 0.72, which allows more redundant features to exist and adapts to a wider range of feature selection.

[0068] The method of fusing the asset information feature data set, the asset economic feature data set and the external environment feature data set to obtain a comprehensive feature data set includes:

[0069] The normalized asset information feature data set, asset economic feature data set and external environment feature data set are fused through a weighted model to obtain a comprehensive feature data set; the asset information feature data set is recorded as K1, the asset economic feature data set is recorded as K2, and the external environment feature data set is recorded as K3;

[0070] The weighted model is: ZC=K1·ω1+K2·ω2+K3·ω3; wherein ω1 is the weight coefficient of the asset information feature data set; ω2 is the weight coefficient of the asset economic feature data set; and ω3 is the weight coefficient of the external environment feature data set.

[0071] The training method of the asset life cycle prediction model includes:

[0072] The data set is divided into a training set, a validation set, and a test set; the data set includes a historical comprehensive feature data set and a corresponding asset status change rate; a sliding window method is used to generate a sample set, which is a subset of the data set, and each sample set includes a historical comprehensive feature data set and a corresponding asset status change rate within a period of time;

[0073] The asset life cycle prediction model is constructed using the LSTM model, which specifically includes an input layer, an LSTM layer, a fully connected layer, and an output layer. The input layer is used to input a historical comprehensive feature data set. The number of neurons in the input layer matches the number of features in the historical comprehensive feature data set. The LSTM layer is used to receive data from the input layer and capture the temporal characteristics of the asset state changes during the life cycle. The learning ability of the model is adjusted by changing the number of LSTM units and layers. The fully connected layer is used to map the output of the LSTM layer to the target prediction value. The output layer uses linear as the activation function to output the asset state change rate.

[0074] Use mean square error as the loss function to measure the error of model prediction; train the asset lifecycle prediction model on the training set and minimize the loss function through the Adam optimizer; use the validation set to evaluate the generalization ability of the model, and tune the model's hyperparameters based on the evaluation results until the preset stopping condition is reached, thus obtaining the final trained asset lifecycle prediction model;

[0075] The current comprehensive feature data set is input into the trained asset life cycle prediction model to predict the corresponding asset status change rate.

[0076] The method for determining whether to generate asset risk warning information according to the asset status change rate includes:

[0077] If the predicted asset status change rate is greater than or equal to the preset asset status change rate threshold, the asset is determined to be in an abnormal state and asset risk warning information is generated;

[0078] If the predicted asset status change rate is less than the preset asset status change rate threshold, the asset is determined to be in a healthy state and no asset risk warning information is generated.

[0079] If the asset risk warning information is generated, the method of extracting the state change rate trend through statistical analysis includes performing fluctuation analysis and second-order derivative calculation on the asset state change rate, thereby obtaining the change rate fluctuation amplitude, change rate fluctuation cycle and asset degradation acceleration; based on the change rate fluctuation amplitude, change rate fluctuation cycle and asset degradation acceleration, reflecting the state change rate trend;

[0080] By counting the discrete degree of the asset status change rate, the fluctuation amplitude of the change rate is defined as the standard deviation of the asset status change rate, which reflects the volatility of the asset status; the change rate fluctuation period is calculated by Fourier transform to measure the regularity of asset status changes; the asset degradation acceleration is obtained by calculating the second-order derivative of the asset status change rate to reflect the speed of asset status deterioration.

[0081] The method for dynamically adjusting resource allocation based on the state change rate trend includes constructing a resource allocation adjustment formula according to the state change rate trend and dynamically adjusting resource allocation; assets with high volatility need to prioritize resource allocation to cope with uncertainty; short-cycle assets need to be frequently monitored to avoid sudden changes; assets with high degradation acceleration require more resources to slow down the degradation trend; and resource allocation is readjusted by comprehensively considering the current asset's rate of change fluctuation range, rate of change fluctuation cycle, and asset degradation acceleration to ensure accurate allocation of resources and improve system efficiency and reliability.

[0082] The method for constructing a resource allocation adjustment formula according to a state change rate trend includes:

[0083] The resource allocation adjustment formula is: Among them, α is the resource allocation adjustment coefficient; ΔR is the fluctuation amplitude of the change rate; T is the fluctuation period of the change rate; A is the acceleration of asset degradation; k1 and k2 are adjustment parameters; e is the base of the exponential function;

[0084] A resource allocation adjustment coefficient threshold β is preset. When β≤α≤1, it indicates that the asset is in a period of aggravated degradation or unstable state, and the resource allocation ratio of the current asset needs to be increased; when 0≤α<β, it indicates that the asset is in a stable period, and the resource allocation ratio of the current asset needs to be reduced; the obtained resource allocation adjustment coefficient is normalized, the resource allocation ratio is recalculated, and resource allocation is performed.

[0085] For example, suppose there are three existing assets, D1, D2, and D3, the total number of resources is 100, and the resource allocation adjustment coefficient threshold β = 0.6. The current state change rate trend is as follows:

[0086] assets Rate of change fluctuation Change rate fluctuation period Acceleration of asset degradation D1 0.8 2.0 0.4 D2 0.6 1.5 0.3 D3 0.4 3.0 0.2

[0087] Adjust parameters k1 = 0.1, k2 = 0.2; According to the resource allocation adjustment formula, we can get:

[0088]

[0089] Therefore, asset D1 will receive a larger resource allocation ratio; assets D2 and D3 will have their resource allocation ratios reduced accordingly; taking asset D1 as an example, the result of reallocation of resources is: That is, the number of resources obtained by asset D1 is 44.

[0090] In this embodiment, by improving the mRMR algorithm, not only the correlation between the features and the target variable is considered, but also the redundancy between the features is considered, so that the most relevant feature subset with the lowest redundancy can be selected; it is helpful to simplify the model, improve the generalization ability of the model, and reduce the computational complexity; the selected feature subset is more representative, which helps to better understand the relationship between the asset status change rate and asset information, asset economy and external environment; a redundancy tolerance threshold is set, and adaptively adjusted according to the redundancy correction function, so that the feature selection process is more flexible and adaptive; the redundancy tolerance threshold is dynamically adjusted according to the relationship between the number of currently selected features and the total number of features; this means that as the feature selection process proceeds, the system can more flexibly deal with redundancy in the data set;

[0091] By analyzing the trend of the status change rate, resource allocation is adjusted dynamically so that resources can flow more accurately to the assets that need support; avoid wasting resources and prioritize the allocation of limited resources to assets with faster status degradation; improve resource utilization efficiency and reduce redundant allocation of resources to low-demand assets; use dynamic adjustment formulas and real-time monitoring mechanisms to quickly respond to changing trends in asset status and adjust resource allocation ratios; adjust resources according to real-time changes in asset status to prevent greater losses due to failure to respond in time to status deterioration; the threshold mechanism ensures that resource allocation adjustments are reasonable and will not change frequently due to short-term fluctuations.

[0092] Embodiment 2

[0093] See also Figure 2 As shown, the part not described in detail in this embodiment is described in Example 1, which provides an asset life cycle management method, including:

[0094] S1. Collect asset information data, asset economic data and external environment data;

[0095] S2. Preprocess the collected asset information data, asset economic data and external environment data to obtain an asset information feature data set, an asset economic feature data set and an external environment feature data set; merge the asset information feature data set, the asset economic feature data set and the external environment feature data set to obtain a comprehensive feature data set;

[0096] S3. Obtain an asset life cycle prediction model based on the comprehensive feature data set training, and obtain the asset status change rate through the asset life cycle prediction model prediction;

[0097] S4. Determine whether to generate asset risk warning information based on the asset status change rate; if asset risk warning information is generated, extract the status change rate trend through statistical analysis, and dynamically adjust resource allocation based on the status change rate trend.

[0098] Since the electronic device introduced in this embodiment is an electronic device used to implement an asset life cycle management system in the embodiment of this application, based on the asset life cycle management system introduced in the embodiment of this application, the technical personnel of this field can understand the specific implementation of the electronic device of this embodiment and its various variations, so how the electronic device implements the method in the embodiment of this application is not introduced in detail here. As long as the technical personnel of this field implement the electronic device used in the asset life cycle management system in the embodiment of this application, it belongs to the scope of protection of this application.

[0099] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters and thresholds in the formula are set by technicians in this field according to actual conditions.

[0100] The above is only a preferred embodiment of the present invention, and the protection scope of the present invention is not limited to the above embodiments. All technical solutions under the concept of the present invention belong to the protection scope of the present invention. It should be pointed out that for ordinary technical users in this technical field, some improvements and modifications without departing from the principle of the present invention should also be regarded as the protection scope of the present invention.

Claims

1. An asset life cycle management system, characterized in that: include: Data collection module, used to collect asset information data, asset economic data and external environment data; The data processing module is used to pre-process the collected asset information data, asset economic data and external environment data to obtain an asset information feature data set, an asset economic feature data set and an external environment feature data set; The asset information feature data set, the asset economic feature data set and the external environment feature data set are integrated to obtain a comprehensive feature data set; The asset life cycle prediction module is used to obtain the asset life cycle prediction model based on the comprehensive feature data set training, and to predict the asset status change rate through the asset life cycle prediction model; The resource scheduling module is used to determine whether to generate asset risk warning information based on the asset status change rate; If asset risk warning information is generated, the state change rate trend is extracted through statistical analysis, and resource allocation is dynamically adjusted based on the state change rate trend; each module is connected by wired and / or wireless means.

2. An asset life cycle management system according to claim 1, characterized in that: The asset information data includes the type, specifications, purchase date and usage status of the asset; the asset economic data includes the purchase cost, depreciation and maintenance costs of the asset; the external environment data includes market interest rates, industry development trends and changes in policies and regulations.

3. An asset life cycle management system according to claim 2, characterized in that: The method of preprocessing the collected asset information data, asset economic data and external environment data to obtain the asset information feature data set, the asset economic feature data set and the external environment feature data set includes: The missing values ​​in the asset information data, asset economic data and external environment data are filled by the mean filling method, and the outliers in the asset information data, asset economic data and external environment data are detected and eliminated by the box plot method; and the standard deviation is normalized and converted into a standard normal distribution; The improved mRMR algorithm is used to analyze the correlation between asset information data, asset economic data, external environment data and asset status change rate, while minimizing the redundancy between the features in asset information data, asset economic data and external environment data, thereby obtaining asset information feature data set, asset economic feature data set and external environment feature data set; the specific steps are as follows: Define the target variable Y, where Y is the asset status change rate; calculate the mutual information between each feature f in the asset information data, asset economic data and external environment data and the target variable Y by maximizing the correlation formula; The formula for maximizing correlation is: Among them, p(f,Y) is the joint probability distribution of feature f and target variable Y, which indicates the probability of feature f and target variable Y occurring simultaneously; p(f) is the marginal probability distribution of feature f, which indicates the probability of feature f occurring independently; p(Y) is the marginal probability distribution of target variable Y, which indicates the probability of target variable Y occurring independently; I(f,Y) is the mutual information between feature f and target variable Y, which is used to measure the correlation between feature f and target variable Y; The redundancy between features is calculated by minimizing the redundancy formula, that is, the mutual information between feature f and feature f′; the minimizing redundancy formula is: Among them, p(f,f′) is the joint probability distribution between feature f and feature f′, which indicates the probability of feature f and feature f′ occurring simultaneously; p(f) and p(f′) are the marginal probability distributions of feature f and feature f′, respectively, which indicate the probability of feature f and feature f′ occurring independently; I(f,f′) is the mutual information between feature f and feature f′, which is used to quantify the redundancy between feature f and feature f′; By improving the optimization objective function, the most relevant and least redundant features are selected to form the feature subset S; the improved optimization objective function is: Among them, J(S) is the improved optimization objective function; |S| is the number of features in the feature subset S; g(f,f′) is the redundancy correction function, which is dynamically adjusted according to the redundancy tolerance between features; Set a redundancy tolerance threshold, dynamically adjust the redundancy tolerance between features according to the redundancy correction function, and adaptively adjust the set redundancy tolerance threshold; The redundancy correction function is: Among them, τ is the set redundancy tolerance threshold.

4. An asset life cycle management system according to claim 3, characterized in that: The method for adaptively adjusting the set redundancy tolerance threshold comprises: The set redundancy tolerance threshold is adaptively adjusted through the redundancy tolerance threshold adjustment formula. The redundancy tolerance threshold adjustment formula is: Among them, τ′ is the adjusted redundancy tolerance threshold; n cr is the number of features currently selected; n tl is the total number of features; τ max is the maximum redundancy tolerance threshold.

5. An asset life cycle management system according to claim 4, characterized in that: The method of fusing the asset information feature data set, the asset economic feature data set and the external environment feature data set to obtain a comprehensive feature data set includes: The normalized asset information feature data set, asset economic feature data set and external environment feature data set are fused through a weighted model to obtain a comprehensive feature data set; the asset information feature data set is recorded as K1, the asset economic feature data set is recorded as K2, and the external environment feature data set is recorded as K3; The weighted model is: ZC=K1·ω1+K2·ω2+K3·ω3; wherein ω1 is the weight coefficient of the asset information feature data set; ω2 is the weight coefficient of the asset economic feature data set; and ω3 is the weight coefficient of the external environment feature data set.

6. An asset life cycle management system according to claim 5, characterized in that: The training method of the asset life cycle prediction model includes: The data set is divided into a training set, a validation set, and a test set; the data set includes a historical comprehensive feature data set and a corresponding asset status change rate; a sliding window method is used to generate a sample set, which is a subset of the data set, and each sample set includes a historical comprehensive feature data set and a corresponding asset status change rate within a period of time; The asset life cycle prediction model is constructed using the LSTM model, which specifically includes an input layer, an LSTM layer, a fully connected layer, and an output layer. The input layer is used to input a historical comprehensive feature data set. The number of neurons in the input layer matches the number of features in the historical comprehensive feature data set. The LSTM layer is used to receive data from the input layer and capture the temporal characteristics of the asset state changes during the life cycle. The learning ability of the model is adjusted by changing the number of LSTM units and layers. The fully connected layer is used to map the output of the LSTM layer to the target prediction value. The output layer uses linear as the activation function to output the asset state change rate. Use mean square error as the loss function to measure the error of model prediction; train the asset lifecycle prediction model on the training set and minimize the loss function through the Adam optimizer; use the validation set to evaluate the generalization ability of the model, and tune the model's hyperparameters based on the evaluation results until the preset stopping condition is reached, thus obtaining the final trained asset lifecycle prediction model; The current comprehensive feature data set is input into the trained asset life cycle prediction model to predict the corresponding asset status change rate.

7. An asset life cycle management system according to claim 6, characterized in that: The method for determining whether to generate asset risk warning information according to the asset status change rate includes: If the predicted asset status change rate is greater than or equal to the preset asset status change rate threshold, the asset is determined to be in an abnormal state and asset risk warning information is generated; If the predicted asset status change rate is less than the preset asset status change rate threshold, the asset is determined to be in a healthy state and no asset risk warning information is generated.

8. An asset life cycle management system according to claim 7, characterized in that: If the asset risk warning information is generated, the method of extracting the state change rate trend through statistical analysis includes performing fluctuation analysis and second-order derivative calculation on the asset state change rate, thereby obtaining the change rate fluctuation amplitude, change rate fluctuation period and asset degradation acceleration; Reflects the state change rate trend based on the change rate fluctuation amplitude, change rate fluctuation cycle and asset degradation acceleration.

9. An asset life cycle management system according to claim 8, characterized in that: The method for dynamically adjusting resource allocation based on the state change rate trend includes constructing a resource allocation adjustment formula according to the state change rate trend and dynamically adjusting resource allocation.

10. An asset life cycle management system according to claim 9, characterized in that: The method for constructing a resource allocation adjustment formula according to a state change rate trend includes: The resource allocation adjustment formula is: Among them, α is the resource allocation adjustment coefficient; ΔR is the fluctuation amplitude of the change rate; T is the fluctuation period of the change rate; A is the acceleration of asset degradation; k1 and k2 are adjustment parameters; e is the base of the exponential function; A resource allocation adjustment coefficient threshold β is preset. When β≤α≤1, the resource allocation ratio of the current asset needs to be increased; when 0≤α<β, the resource allocation ratio of the current asset needs to be reduced; the obtained resource allocation adjustment coefficient is normalized and the resource allocation ratio is recalculated to perform resource allocation.

Images