Enterprise innovation power cloud evaluation and resource adaptation system and method

Abstract

The invention discloses an enterprise innovation force cloud evaluation and resource adaptation system and method, and relates to the technical field of enterprise innovation management, and the method comprises the following steps: periodically obtaining industry life cycle data, innovation sign data and external data source data; performing weighted average calculation on the industry life cycle data, the innovation sign data and the external data source data to obtain fused innovation sign data; and performing sliding window processing on the fused innovative sign data in the same period, and outputting a trend curve of the fused innovative sign data. According to the scheme, interference caused by industry periodic fluctuation is effectively eliminated through reference curve dynamic alignment and continuous window joint judgment, meanwhile, the abnormal starting time is positioned in combination with the trend curve, an accurate time anchor point is provided for subsequent causal analysis, the continuous abnormal state in the fused innovation sign data can be accurately recognized, and the accuracy and the reliability of the fused innovation sign data are improved. And misjudgment caused by single fluctuation is avoided.

Description

Technical Field

[0001] The present invention relates to the technical field of enterprise innovation management, and in particular to an enterprise innovation cloud assessment and resource adaptation system and method. Background Art

[0002] With the rapid development of global science and technology and industry, corporate innovation has become an important indicator for measuring the core competitiveness and sustainable development capabilities of technology-based enterprises. Especially under the impetus of digitalization and intelligentization, corporate innovation activities are becoming increasingly complex. The innovation process involves multi-dimensional factors such as technology research and development, market transformation, organizational collaboration, and capital operation. At the same time, the impact of the industry life cycle, external policy environment, technological trends, and market changes on innovation is becoming increasingly significant. Therefore, building a scientific, dynamic, and multi-source data-oriented innovation assessment and resource adaptation system has become an important research direction in the current field of corporate management and scientific and technological services.

[0003] However, in the process of implementing the technical solutions of the invention in the embodiments of this application, it was found that the above technology has at least the following technical problems:

[0004] Existing technologies are unable to effectively integrate industry life cycles, enterprise innovation indicators, and external data sources, resulting in insufficient comprehensiveness and dynamism in innovation assessments, and a lack of trend analysis of enterprise innovation indicator data, making it difficult to identify anomalies and outbreak nodes. This makes the subsequent generation of innovation resource intervention plans less targeted, and prone to lagging or excessive allocation of innovation resources, making it impossible to accurately support the improvement of enterprise innovation capabilities. Summary of the Invention

[0005] The purpose of the present invention is to provide an enterprise innovation cloud assessment and resource adaptation system and method to solve the problems raised in the above background technology.

[0006] In order to achieve the above object, the technical solution of the present invention is as follows:

[0007] In a first aspect, the present invention discloses a method for evaluating the innovation capability of an enterprise through cloud computing and resource adaptation, which is applied to the evaluation and resource adaptation of the innovation capability of technology-based enterprises at all stages, and includes the following steps:

[0008] Periodically obtain industry life cycle data, innovation sign data and external data source data;

[0009] Performing weighted average calculation on the industry life cycle data, innovation vital signs data and external data source data to obtain integrated innovation vital signs data;

[0010] Performing sliding window processing on the fusion innovation vital sign data within the same period, and outputting a trend curve of the fusion innovation vital sign data;

[0011] Calling a preset reference curve, comparing the trend curve with the preset reference curve, and obtaining a deviation value;

[0012] Determine whether the deviation value in N consecutive sliding windows is less than a preset threshold value of one, and if so, determine that the fusion innovation vital sign data corresponding to the deviation value is abnormal fusion innovation vital sign data, and then determine the abnormal time point in combination with the trend curve;

[0013] Determine whether the deviation value in a single sliding window is greater than a preset threshold value 2, and if so, determine that the fusion innovation vital sign data corresponding to the deviation value is the outbreak fusion innovation vital sign data, and then determine the outbreak time point in combination with the trend curve;

[0014] Connect abnormal time points and outbreak time points within the same cycle on the timeline to generate an innovation event chain;

[0015] According to the innovation event chain, the preset knowledge reasoning and resource matching model is called to output the innovation resource intervention plan.

[0016] In a second aspect, the present invention discloses an enterprise innovation cloud assessment and resource adaptation system, including:

[0017] Data acquisition module, used to periodically acquire industry life cycle data, innovation sign data, and external data source data;

[0018] A fusion innovation vital sign data calculation module is used to perform weighted average calculation on the industry life cycle data, innovation vital sign data and external data source data to obtain fusion innovation vital sign data;

[0019] a deviation value calculation module, configured to perform sliding window processing on the fused innovative vital sign data within the same period and output a trend curve of the fused innovative vital sign data;

[0020] Calling a preset reference curve, comparing the trend curve with the preset reference curve, and obtaining a deviation value;

[0021] A time point determination module is used to determine whether the deviation value in N consecutive sliding windows is less than a preset threshold value of one. If so, the fusion innovation vital sign data corresponding to the deviation value is determined to be abnormal fusion innovation vital sign data, and then the abnormal time point is determined in combination with the trend curve;

[0022] Determine whether the deviation value in a single sliding window is greater than a preset threshold value 2, and if so, determine that the fusion innovation vital sign data corresponding to the deviation value is the outbreak fusion innovation vital sign data, and then determine the outbreak time point in combination with the trend curve;

[0023] The innovation resource intervention plan output module is used to connect abnormal time points and outbreak time points in the same cycle on the timeline to generate an innovation event chain;

[0024] According to the innovation event chain, the preset knowledge reasoning and resource matching model is called to output the innovation resource intervention plan.

[0025] Compared with the prior art, the present invention has the following beneficial effects:

[0026] 1. This solution effectively eliminates interference caused by industry cycle fluctuations through dynamic alignment of benchmark curves and joint judgment of continuous windows. At the same time, it combines trend curves to locate the start time of anomalies, providing a precise time anchor point for subsequent causal analysis. It can accurately identify persistent abnormal states in fusion innovation vital sign data, avoid misjudgments caused by single fluctuations, and enable subsequent causal chain analysis to focus on external events in specific time periods, improving the efficiency of anomaly root cause analysis.

[0027] 2. By integrating correlation screening with Bayesian network inference, this solution not only identifies external events related to anomalies, but also quantifies their direction and intensity, transforming anomaly attribution analysis from qualitative judgment to quantitative calculation. It can accurately identify the external driving factors that lead to anomalies in innovation vital sign data, avoid misjudgments caused by ignoring the influence of the external environment, provide data support for the formulation of precise resource intervention plans, and effectively solve the problem of resource allocation deviation caused by the lack of external event correlation analysis in existing technologies.

[0028] 3. By integrating the knowledge reasoning engine and the rule engine, this solution can combine the temporal relationship in the event chain with external causal factors to achieve dynamic optimization of resource matching. Through abnormal-external event causal chain analysis, resource demand can be predicted in advance and the intervention timing can be adjusted. It can improve the pertinence and timeliness of innovative resource intervention plans and solve the problem of delayed resource allocation caused by the lack of causal analysis and dynamic matching in existing technologies. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] The disclosure of the present invention is described with reference to the accompanying drawings. It should be understood that the drawings are for illustrative purposes only and are not intended to limit the scope of protection of the present invention. In the drawings, the same reference numerals are used to refer to the same components. Among them:

[0030] Figure 1 This is a flowchart of the steps of the enterprise innovation cloud assessment and resource adaptation method of the present invention;

[0031] Figure 2 A schematic diagram of a process for generating a trend curve provided by the present invention;

[0032] Figure 3 A schematic diagram of the process of generating an innovative event chain provided by the present invention;

[0033] Figure 4 A schematic diagram of the process of outputting the innovative resource intervention plan provided by the present invention;

[0034] Figure 5 The module function schematic diagram of the enterprise innovation cloud evaluation and resource adaptation system provided by the application is shown in the figure. DETAILED DESCRIPTION

[0035] It is easy to understand that, according to the technical scheme of the application, those skilled in the art can propose a plurality of structure modes and implementation modes which can be replaced with each other without changing the essential spirit of the application. Therefore, the following specific embodiments and drawings are only exemplary descriptions of the technical scheme of the application, and should not be regarded as the whole or as a limitation or restriction on the technical scheme of the application.

[0036] SUMMARY

[0037] In the prior art, the innovation of science and technology enterprises is mainly evaluated by relying on static index system and single dimension data, which is difficult to dynamically reflect the influence of industry cycle change and external environment on enterprise innovation; the traditional method usually adopts fixed time interval data collection method, which lacks the fusion processing ability of multi-source heterogeneous data, resulting in that the evaluation result lags behind the actual innovation process, for example, a certain intelligent hardware enterprise in the product research and development stage, due to the influence of supply chain fluctuation and policy adjustment on research and development investment, leading to the disconnection between innovation resource allocation and market demand, resulting in the extension of research and development cycle and the waste of resources.

[0038] In order to solve the above problems, it is found that the existing technology has the core defects of insufficient data integration ability, lack of trend analysis and lagging resource adaptation; by analyzing the correlation between enterprise innovation and industry life cycle and external events, a dynamic data fusion mechanism is constructed, combining industry stage characteristics with real-time data; further research shows that the use of sliding window processing can effectively capture the continuous change trend of innovation signs, and the deviation detection based on the reference curve can identify abnormal fluctuations and explosive growth nodes; finally, through the innovation event chain connected with key time points, combined with knowledge reasoning and resource matching model, accurate intervention scheme is generated.

[0039] After introducing the basic idea of the application, the embodiments of the application will be specifically introduced with reference to the drawings.

[0040] Embodiment one:

[0041] Please refer to Figures 1-4 , the enterprise innovation cloud evaluation and resource adaptation method is applied to the innovation cloud evaluation and resource adaptation of science and technology enterprises at each stage, including the following steps:

[0042] Periodically acquire industry life cycle data, innovation sign data and external data source data;

[0043] Performing weighted average calculation on the industry life cycle data, innovation vital signs data and external data source data to obtain integrated innovation vital signs data;

[0044] Performing sliding window processing on the fused innovative vital sign data within the same period, and outputting a trend curve of the fused innovative vital sign data;

[0045] Calling a preset reference curve, comparing the trend curve with the preset reference curve, and obtaining a deviation value;

[0046] Determine whether the deviation value in N consecutive sliding windows is less than a preset threshold value of one, and if so, determine that the fusion innovation vital sign data corresponding to the deviation value is abnormal fusion innovation vital sign data, and then determine the abnormal time point in combination with the trend curve;

[0047] Determine whether the deviation value in a single sliding window is greater than a preset threshold value 2, and if so, determine that the fusion innovation vital sign data corresponding to the deviation value is the outbreak fusion innovation vital sign data, and then determine the outbreak time point in combination with the trend curve;

[0048] Connect abnormal time points and outbreak time points within the same cycle on the timeline to generate an innovation event chain;

[0049] According to the innovation event chain, the preset knowledge reasoning and resource matching model is called to output the innovation resource intervention plan.

[0050] Among them, industry lifecycle data refers to structured data obtained through the industry database interface, including the industry stage and its corresponding weight parameters. Specifically, it can be implemented by calling standardized data packages through the API interface to reflect the constraints imposed by the overall development trend of the industry on innovation activities;

[0051] Innovation vital signs data covers operational data in dimensions such as technology research and development and market transformation. It can be collected in real time through the enterprise ERP system and used to quantify the core indicators of the enterprise's innovation capabilities.

[0052] External data sources include heterogeneous data such as policy documents and market intelligence. Specifically, web crawlers can be used to capture public information sources to capture the impact of the external environment on corporate innovation.

[0053] Sliding window processing refers to calculating the mean of time series data by intercepting it at a fixed time length. Specifically, a 30-day window length can be used for sliding calculation to eliminate data noise and extract trend features.

[0054] The preset benchmark curve refers to the standard change curve of innovation signs generated based on historical data training. Specifically, it can be generated by modeling the data of industry benchmark enterprises using a machine learning model to provide a dynamic comparison benchmark.

[0055] During the specific implementation process, enterprises periodically collect three types of data: industry life cycle data, innovation sign data, and external data source data, and use weighted average calculation to obtain integrated innovation sign data;

[0056] Subsequently, a sliding window method (e.g., a window length of 3 months and a step length of 1 month) was used to perform trend analysis on the fusion innovation vital sign data, and the data was compared with a preset benchmark curve of the same data type as the fusion innovation vital sign data to calculate the deviation value;

[0057] If the deviation values ​​of N consecutive windows (such as 4) are all lower than the preset threshold of one (such as -8), it is identified as abnormal fusion innovation vital sign data, and the abnormal time point is determined by searching the position of the abnormal fusion innovation vital sign data in the trend curve; if the deviation value in a single sliding window is higher than the preset threshold of two (such as +10), it is identified as burst fusion innovation vital sign data, and the burst time point is determined by searching the position of the burst fusion innovation vital sign data in the trend curve;

[0058] The anomalies and outbreak time points are linked together to form an innovation event chain, and based on the knowledge reasoning and resource matching model, customized innovation resource intervention suggestions (such as special fund allocation, industry-university-research cooperation, introduction of external experts, etc.) are output.

[0059] Beneficial Effects: This application effectively solves the problem of multi-source data integration and enables dynamic correlation analysis between industry characteristics and enterprise innovation. Through a deviation detection mechanism, it accurately identifies abnormal stagnation and sudden growth points in the innovation process, providing a precise time window for resource allocation. The event chain-based knowledge reasoning and resource matching model can establish a mapping relationship between innovation fluctuations and resource demand, avoiding blindness and lag in resource allocation and improving the innovation efficiency of technology-based enterprises.

[0060] This application further proposes to periodically obtain innovation vital signs data through the enterprise's internal management system. Such innovation vital signs data include technology research and development data (such as the amount of R&D investment, the number of R&D projects, the number of patent applications and authorizations, etc.), market conversion data (such as patent conversion rate, product market share, number of new product launches, market feedback scores, etc.), organizational collaboration data (such as team collaboration frequency, number of inter-departmental collaborative projects, talent turnover rate for key positions, number of internal training sessions, etc.), and financial resilience data (such as cash flow, risk of capital chain rupture, credit line utilization rate, etc.);

[0061] By connecting to the application programming interface of the industry database, industry life cycle data is periodically obtained. The industry life cycle data includes the current life cycle stage of the industry (such as germination, growth, maturity, and decline) and the weight parameters of the innovation vital signs data corresponding to each stage (such as the weight parameter of technology R&D data in the growth stage is 0.3, and the weight parameter of financial resilience data in the decline stage is 0.5);

[0062] Data from external data sources are periodically acquired through web crawlers. The data from external data sources include structured data related to the enterprise and its industry (such as industry public financial reports, market statistical reports), semi-structured data (such as news information, policy announcements, public opinion monitoring results), and unstructured data (such as expert advice, online comments, forum discussion content, etc.).

[0063] The internal management system refers to the digital platform used by the enterprise for daily operations management. It can be implemented using an ERP system or a customized data collection system to integrate real-time data generated by internal innovation activities such as technology research and development and market transformation.

[0064] The application programming interface of the industry database refers to a programming interface that provides standardized data access. Specifically, it can be implemented using a RESTful API or a GraphQL interface to dynamically obtain the industry lifecycle stages and their corresponding weight parameters;

[0065] A web crawler refers to a program that automatically collects public data from the Internet. It can be implemented using the Python-based Scrapy framework or a distributed crawler system to capture external data sources such as policy documents, market reports, and social media, expanding data coverage.

[0066] During the specific implementation process, the company's internal management system extracts core innovation data such as technology research and development data and market transformation data at a fixed period to ensure the real-time and integrity of internal data;

[0067] The application programming interface of the industry database obtains information on the industry life cycle stages at fixed intervals, such as the start-up phase, growth phase, maturity phase, or decline phase, and synchronizes the corresponding weight parameters for each phase. For example, technology R&D data may be given a higher weight in the growth phase.

[0068] Web crawlers configure differentiated collection strategies for different external data sources. For example, they use direct database connection for structured data and natural language processing technology for semantic analysis of unstructured data, ultimately forming a unified format of external data source data sets. Through the collaboration of the above three data acquisition methods, it can cover multi-dimensional information on the company's internal innovation capabilities, industry dynamics and external environment, providing comprehensive input for subsequent integrated computing.

[0069] By the technical solution, the application solves the problems of single data source and weight update lag in the prior art, realizes dynamic integration of multi-dimensional data, for example, when an industry enters a mature period, the weight of technical research and development data is automatically reduced, and the weight of market conversion data is increased, so that the fusion innovation sign data is more in line with the innovation characteristics of enterprises at this stage; at the same time, the introduction of external data sources can capture the influence of policy changes or market emergencies on innovation, avoid the deviation of evaluation results from the actual situation due to external environment mutation, and thus provide a reliable data basis for subsequent trend analysis, anomaly detection and resource adaptation.

[0070] The application further proposes linear calculation of the industry life cycle data, the innovation sign data and the external data source data to obtain the fusion innovation sign data, which specifically includes:

[0071] According to the preset weight parameter of the innovation sign data, the innovation sign data is weighted and calculated, and weighted innovation sign data is output;

[0072] The external data source data is filtered and processed;

[0073] The weighted innovation sign data and the filtered external data source data are normalized;

[0074] The weighted innovation sign data and the filtered external data source data are normalized.

[0075] The weighted innovation sign data and the filtered external data source data are normalized.

[0076] The filtering process refers to quality filtering of the external data source, which can be realized by confidence calculation combined with a set value, for example, after feature extraction and normalization processing, data with a confidence lower than a set value is removed, so as to exclude the interference of noise or low confidence data on the fusion result.

[0077] The normalization process refers to converting data of different dimensions or orders of magnitude to a uniform scale, which can be realized by maximum-minimum value standardization or Z-score standardization method, for example, the weighted innovation sign data and the external data source data are mapped to the [0, 1] interval, so as to eliminate the influence of data distribution difference on weighted average calculation.

[0078] The weighted average calculation refers to calculating the mean value after weighting different sources of data, which can be realized by using fixed weight or dynamic weight allocation strategy, for example, the normalized weighted innovation sign data and the screened external data source data are weighted and summed according to a preset proportion to form a comprehensive evaluation index.

[0079] The specific implementation process is as follows:

[0080] First, the weight parameters of the preset innovation sign data are obtained, for example: technical research and development data 0.35, market conversion data 0.3, organizational synergy data 0.2, and financial resilience data 0.15; the innovation sign data is weighted and calculated by using the preset weight parameters of the innovation sign data to obtain the weighted innovation sign score, so as to ensure that the evaluation system can fully highlight the outstanding performance of enterprises in technology and market, and take into account the stability of organization and funds;

[0081] Subsequently, the external data source data is screened;

[0082] In the data normalization stage, the range normalization method is used to normalize the weighted innovation sign data and the screened external data source data to the interval [0, 1] to eliminate the dimensional and scale differences and ensure the fairness of the subsequent weighted average process;

[0083] After normalization, the fusion weight of the weighted innovation sign data and the screened external data source data (such as weighted innovation sign data 0.7 and screened external data source data 0.3) is set, and the weighted average is performed to finally form the fusion innovation sign data reflecting the innovation power of the enterprise. The fusion data can not only dynamically respond to industry policies and market trends, but also can make up for the limitations of single internal evaluation, and realize the improvement of objectivity and forward-looking of innovation power evaluation.

[0084] Through the above technical solution, the present application solves the problem of fusion result deviation caused by complex data sources and uneven quality in the prior art. Through phased weighting, screening and normalization processing, the accuracy and consistency of multi-source data fusion are improved, providing a reliable data foundation for subsequent trend analysis.

[0085] The present application further proposes that the screening processing of the external data source data specifically includes:

[0086] The external data source data is subjected to data formatting, deduplication, and content normalization processing;

[0087] The processed external data source data is subjected to feature extraction, the extracted features are subjected to normalization processing, and linear weighting calculation is performed thereon to obtain the confidence of the external data source data;

[0088] It is determined whether the confidence level is lower than a set value, and if so, the external data source data corresponding to the confidence level is eliminated.

[0089] Data formatting refers to converting structured, semi-structured, and unstructured data from different sources into a unified format. This can be achieved using JSON or XML standard format conversion tools to eliminate data heterogeneity.

[0090] Deduplication refers to identifying and deleting duplicate data entries through hash algorithms or similarity matching algorithms. For example, the SimHash algorithm is used to calculate text similarity to eliminate redundant data interference.

[0091] Content normalization refers to the segmentation, part-of-speech tagging, and entity recognition of unstructured text, for example, using an NLP toolkit to extract standardized semantic features.

[0092] Feature extraction refers to extracting key indicators related to innovation from normalized data. For example, the TF-IDF algorithm is used to extract keyword weights to quantify the impact factors of external data.

[0093] Confidence refers to the degree of credibility of characteristic indicators calculated through linear weighting, such as the use of hierarchical analysis to determine the weight coefficient of each characteristic, which is used to evaluate data reliability;

[0094] The set value refers to a predefined confidence threshold, such as a critical value obtained through historical data training, which is used to dynamically screen valid data.

[0095] The specific process is as follows:

[0096] First, the data from external data sources is formatted uniformly, and duplicate data is automatically identified and removed using hashing algorithms and intelligent text comparison technology. For data content, the natural language processing (NLP) module is used to normalize descriptive text, such as unifying terminology, removing irrelevant information, and standardizing time and unit expressions.

[0097] After initial cleaning, data features are extracted and normalized to eliminate the effects of different dimensions or scales. The normalized feature data is linearly weighted based on historical experience and expert knowledge to determine the confidence level of each external data source.

[0098] The confidence level of all data from external data sources is determined by a set value. If the confidence level of data from a certain external data source is lower than the set value (such as 0.6), the corresponding data will be automatically eliminated to ensure that the data basis for subsequent innovation assessment is highly accurate and reliable.

[0099] Beneficial effects: This application solves the technical problem of distortion in fusion calculations caused by uneven quality of external data sources; by constructing a complete processing chain including data cleaning, feature quantification and dynamic screening, it effectively eliminates invalid data and noise interference, such as avoiding the inclusion of contradictory industry report data in the analysis, ensuring the reliability and consistency of data input into the weighted average calculation link, and providing a high-quality data foundation for subsequent trend analysis and resource matching.

[0100] This application further proposes to perform sliding window processing on the fusion innovation vital sign data within the same period, and output a trend curve of the fusion innovation vital sign data, specifically including:

[0101] Sort all types of fusion innovation vital sign data within the same period by timestamp;

[0102] Starting from the starting point of the cycle, the fusion innovation vital sign data is intercepted with the set sliding window length as the fusion innovation vital sign data within the sliding window;

[0103] Calculate the mean of the fused innovation sign data within each sliding window;

[0104] With the center moment of the sliding window as the horizontal axis and the mean value of the fusion innovation vital sign data in each sliding window as the vertical axis, all sliding windows are spliced ​​in sequence to form trend curves of various types of fusion innovation vital sign data.

[0105] The sliding window length refers to the length of the time range used to intercept data. It can be implemented using a fixed time period such as 30 days or a dynamically adjusted window. Its role is to balance the sensitivity of data fluctuations with the stability of trends.

[0106] The center time of the sliding window refers to the midpoint of the window coverage period. It can be calculated by taking the arithmetic mean of the window start and end times. Its function is to eliminate the deviation of the window edge time points from the trend positioning.

[0107] Trend curve splicing refers to connecting the means corresponding to the center points of each window in chronological order. It can be implemented by linear interpolation or piecewise function. Its function is to form a continuous and analyzable data change trajectory.

[0108] Specifically, first, sort the various types of fusion innovation vital signs data within the same period by timestamp; set the sliding window length to 3 months and the window step to 1 month, and intercept the various types of fusion innovation vital signs data in sequence starting from January; in each sliding window, calculate the mean of each type of fusion innovation vital signs data, such as January-March, February-April, March-May, and April-June; in order to enhance the precision of the analysis, set the window center moment (such as February, March, April, and May) as the horizontal coordinate point of the trend curve, and the mean as the vertical coordinate, and draw the trend curve of each type of fusion innovation vital signs data.

[0109] Through the above technical solution, the present application effectively solves the problems of coarse granularity and phase misalignment in trend analysis in the prior art; through the combination of sliding window and central moment positioning, it can accurately capture the short-term fluctuation characteristics in the fused data, and provide a high-precision trend benchmark for subsequent anomaly detection and outbreak point identification. For example, when detecting anomalies in market conversion data, the trend curve generated by this method can clearly show the continuous downward trend of the conversion rate within a specific 3-day window, while traditional monthly statistics cannot capture such short-term abnormal signals in a timely manner.

[0110] The present application further proposes to determine whether the deviation value in N consecutive sliding windows is less than a preset threshold value of one, and if so, determine that the fusion innovation vital sign data corresponding to the deviation value is abnormal fusion innovation vital sign data, and then determine the abnormal time point in combination with the trend curve, specifically including:

[0111] Calling preset benchmark curves corresponding to various types of fusion innovative vital sign data;

[0112] Aligning the preset benchmark curve with the trend curve of the same type of fusion innovative vital sign data in terms of time period;

[0113] For each time point, take the trend curve value and the baseline curve value respectively;

[0114] The deviation value is calculated based on the trend curve value and the reference curve value. The specific calculation formula is as follows:

[0115]

[0116] Where, Indicates the deviation value, Indicates the trend curve value, Indicates the reference curve value;

[0117] For all sliding windows, detect the deviation values ​​in chronological order;

[0118] Determine whether the deviation value in N consecutive sliding windows is less than a preset threshold value one, and if so, determine that the fusion innovation vital sign data corresponding to the deviation value is abnormal fusion innovation vital sign data; the preset threshold value one is set according to the type of fusion innovation vital sign data;

[0119] Then, in combination with the trend curve, the starting and ending sliding windows corresponding to the abnormal fusion innovative vital sign data are located, and the starting window time point of N consecutive sliding windows is used as the abnormal time point.

[0120] The deviation value refers to the difference between the trend curve value and the reference curve value. Specifically, it can be calculated using Euclidean distance or standardized mean square error to objectively reflect whether the data fluctuation exceeds a reasonable range.

[0121] The specific steps are as follows:

[0122] Using data from the company's best historical innovation performance period, we fit benchmark curves for various types of innovation activity data for comparison and reference. We then align the trend curves of the same type of innovation activity data in the current cycle with the benchmark curves on the timeline to ensure comparability of innovation activities at each point in time.

[0123] Extract the trend curve value and the reference curve value respectively, and calculate the deviation value between the two;

[0124] Determine whether the deviation values ​​in N consecutive (such as 4) sliding windows are all less than the preset threshold value (such as -8). If so, the innovation sign data of the corresponding window will be determined as abnormal fusion innovation sign data. Further combined with the trend curve, the abnormal start and end windows are located, and the start window time point of the abnormal interval is used as the abnormal time point to facilitate subsequent resource intervention and strategy adjustment.

[0125] Through the above technical solution, the present application can accurately identify persistent abnormal states in the integrated innovation vital sign data, avoid misjudgment due to single fluctuations, and effectively distinguish between sporadic events and systematic anomalies through the sliding window continuous detection mechanism, providing a reliable time-series positioning basis for innovation resource intervention; the precise determination of the abnormal time point enables subsequent causal chain analysis to focus on external events in a specific period of time, thereby improving the efficiency of abnormal root cause analysis.

[0126] The present application further proposes that after determining the abnormal time point in combination with the trend curve, the method further includes calculating the correlation between the abnormal fusion innovative vital sign data and the external data source data at the abnormal time point, specifically including:

[0127] Calculate the correlation between each type of abnormal fusion innovative vital sign data and the external data source data at the abnormal time point. The specific calculation formula is as follows:

[0128]

[0129] In the formula, represents the correlation calculation result of the abnormal fusion innovative sign data and the external data source data of the abnormal time point, represents the abnormal time point, represents the center time point of the sliding window, represents the time range of the sliding window, represents the value of the abnormal fusion innovative sign data at time , represents the value of the external data source data at time , represents the mean value of the abnormal fusion innovative sign data within the sliding window, represents the mean value of the external data source data within the sliding window;

[0130] The total correlation score is calculated by weighted fusion of each correlation calculation result;

[0131] It is judged whether the total correlation score is greater than a preset correlation value. If yes, the external data source data of the abnormal time point is input into the pre-constructed Bayesian network, and the influence probability difference value of the abnormal fusion innovative sign data is output;

[0132] According to the influence probability difference value, the influence direction and influence strength of the external data source data of the abnormal time point on the abnormal fusion innovative sign data are determined, and an abnormal-external event causal chain is generated;

[0133] The influence direction is determined according to the positive and negative of the influence probability difference value, and the influence strength is determined according to the absolute value of the influence probability difference value;

[0134] The Bayesian network takes each type of abnormal fusion innovative sign data and external data source data of the abnormal time point as a node variable, and obtains the dependency relationship between nodes through deep learning inference.

[0135] The correlation calculation refers to quantifying the correlation degree between abnormal data and external events by statistical methods, which can be realized by Pearson correlation coefficient or grey correlation degree algorithm, and is used for screening external events that have potential causal relationship with abnormal data fluctuations;

[0136] The weighted fusion calculation refers to comprehensive evaluation of multiple correlation indicators, which can use entropy weight method or analytic hierarchy process to determine the weight parameter, and is used to eliminate the deviation of a single correlation indicator;

[0137] The Bayesian network refers to a causal relationship inference tool based on a probabilistic graphical model, which can be realized by training the conditional probability table between nodes through historical data, and is used for quantitative analysis of the dynamic influence of external events on abnormal data;

[0138] The impact probability difference refers to the difference in the probability of abnormal data occurring before and after the introduction of an external event in the Bayesian network. It can be calculated by the difference between the posterior probability and the prior probability, and is used to characterize the driving force of the external event on the abnormal data.

[0139] The specific implementation process is as follows: when an abnormal time point is detected, a multi-dimensional correlation analysis is first performed on the abnormal fusion innovation sign data and the external data source data at the abnormal time point. For example, if a company's technology R&D data continuously deviates from the baseline curve, it is necessary to extract external data source data such as policy changes and market fluctuations during that time period. By calculating the correlation scores of various types of abnormal fusion innovation sign data and the external data source data at the abnormal time point, the total correlation score is obtained through weighted fusion calculation of each correlation score;

[0140] Furthermore, when the total correlation score exceeds a preset correlation value, the external data source data at the corresponding abnormal time point is input into a pre-constructed Bayesian network. This Bayesian network constructs a probabilistic dependency relationship between various types of abnormal fusion innovation vital sign data and the external data source data based on historical data, and can output the difference in the probability of the external data source data causing the abnormal fusion innovation vital sign data to occur.

[0141] Based on the positive and negative signs of the impact probability difference, it can be determined whether the external data source data suppresses or aggravates the abnormal fusion innovation sign data, and the absolute value reflects the intensity of its effect. The final causal chain clearly marks the correlation pattern between the external data source data and the abnormal fusion innovation sign data, providing a traceable decision-making basis for subsequent resource adaptation.

[0142] Through the above technical solution, the present application can accurately identify the external driving factors that lead to abnormal innovation vital signs data, and avoid misjudgment caused by ignoring the impact of the external environment. For example, when abnormal capital resilience data is detected, it can be quickly linked to the bank credit policy adjustment event in the same period, and the specific value of the policy that increases the probability of capital abnormality can be calculated; this quantitative analysis of causal relationships provides data support for the formulation of accurate resource intervention plans, and effectively solves the problem of resource allocation deviation caused by the lack of external event correlation analysis in the existing technology.

[0143] This application further proposes to connect abnormal time points and outbreak time points within the same cycle on a timeline to generate an innovative event chain, specifically including:

[0144] Label the specific type of fusion innovation vital signs data for each abnormal fusion innovation vital signs data and burst fusion innovation vital signs data detected in the same cycle;

[0145] Record the occurrence sliding window and duration of each abnormal fusion innovation vital sign data and burst fusion innovation vital sign data;

[0146] According to the recording results, the relationship between the abnormal fusion innovation vital signs data and the burst fusion innovation vital signs data of the same type and adjacent is analyzed from the starting point of the cycle to obtain data chains of different data types, for example:

[0147] If the outbreak fusion innovation vital signs data occurs before the abnormal fusion innovation vital signs data, it is determined to be a "post-outbreak fallback" data chain;

[0148] If abnormal fusion innovation sign data appears first, followed by burst fusion innovation sign data, it is determined to be an "innovation repair" data chain;

[0149] If two burst fusion innovative vital signs data appear consecutively, it is determined to be a "continuous burst" data chain;

[0150] If two abnormal fusion innovation sign data appear consecutively, it will be determined as an "innovation exhaustion" data chain;

[0151] Starting from the starting point of this cycle, data links of different data types are numbered in the same way according to the timeline;

[0152] For the same data chain number, calculate the correlation of data chains of different types of fusion innovative vital sign data;

[0153] For example, when the data chain number is 5, the technology R&D data is a "continuous outbreak" data chain, while the market transformation data is a "innovation exhaustion" data chain. In this case, a "R&D-Transformation" cross-type event chain is constructed, and it is speculated that the failure of R&D results to be effectively transformed has led to a decline in market performance.

[0154] Determine whether the correlation calculation result is greater than the preset correlation value. If so, merge the data chain of the corresponding type of fusion innovation vital sign data into an innovation event chain.

[0155] Type labeling refers to labeling each abnormal or explosive data point with the innovation indicator category it belongs to, such as technological research and development or financial resilience. This can be achieved using a label matching algorithm to distinguish innovation events of different dimensions.

[0156] The occurrence sliding window and its duration refer to the start and end time range of recorded data anomalies or outbreaks. This can be achieved through timestamp tracking and window boundary calculation to quantify the duration characteristics of events.

[0157] The correlation of data chains refers to the degree of interaction between different types of data on the timeline. It can be achieved by calculating the Pearson correlation coefficient or the probability of event co-occurrence, and is used to identify potential connections between cross-dimensional innovation events.

[0158] The specific implementation process is as follows:

[0159] For each abnormal fusion innovation sign data and burst fusion innovation sign data, the data type (technology research and development data, market transformation data, organizational collaboration data, financial resilience data) is automatically labeled, and the sliding window time and duration of its occurrence are recorded. All abnormal and burst fusion innovation sign data within the same period are arranged in chronological order. By analyzing the abnormal fusion innovation sign data and burst fusion innovation sign data with the same data type and adjacent, a data chain is automatically constructed;

[0160] Each data chain is numbered (such as T1, T2, T3, etc.). For the same data chain, correlation analysis methods (such as the Pearson correlation coefficient) are further used to calculate the correlation between different types of data chains (such as the "innovation repair" data chain of technology research and development data and the "continuous outbreak" data chain of market transformation data). If the correlation calculation result of two data chains is greater than the preset correlation value (such as 0.8), they will be merged into an innovation event chain.

[0161] Through the above technical solution, this application solves the problem of fragmented innovation event identification in the existing technology and realizes the serial analysis of cross-dimensional anomalies and outbreak events. For example, when there is a strong correlation between the stagnation of technological research and development and the fluctuation of capital resilience, the system can accurately identify the impact of their synergy on overall innovation, thereby providing an integrated intervention basis for resource allocation and avoiding the program deviation caused by single-dimensional resource allocation.

[0162] This application further proposes to call a preset knowledge reasoning and resource matching model based on the innovation event chain to output an innovation resource intervention plan, specifically including:

[0163] Calling a preset knowledge reasoning and resource matching model; the knowledge reasoning and resource matching model includes a knowledge reasoning engine, a knowledge graph, a resource library, and a rule engine;

[0164] The innovation event chain and the anomaly-external event causal chain are used as inputs to the knowledge reasoning and resource matching model. The knowledge reasoning engine analyzes the type, temporal structure, and causal relationship of the innovation event chain and the anomaly-external event causal chain, and outputs the judgment results.

[0165] Through the knowledge graph, the judgment results are matched with the innovation elements available in the resource library, and combined with expert rules, the optimal resource combination and intervention path are inferred;

[0166] The rule engine generates multi-level intervention recommendations based on the judgment results, and then uses the abnormal time point and outbreak time point as the intervention timing for each intervention recommendation;

[0167] Comprehensively integrate the optimal resource combination, intervention path, intervention suggestions and intervention timing to output innovative resource intervention plans.

[0168] The knowledge reasoning engine refers to a computing module for analyzing the logical relationship of the event chain and the cause-effect chain, and can be specifically implemented by using an inference algorithm based on semantic analysis. The key nodes in the innovation event chain and their relevance are identified.

[0169] The knowledge graph refers to a structured database for storing innovation elements and their associated relationships, and can be specifically implemented by using a graph database technology. The innovation resources are dynamically matched with the requirements in the event chain.

[0170] The resource library refers to a data set containing innovation elements such as technology, funds, and talents, and can be specifically implemented by using a distributed storage system. The resource library provides a resource base that can be called.

[0171] The rule engine refers to a program module for generating intervention suggestions according to preset logic, and can be specifically implemented by using a rule system based on a decision tree. The rule engine converts complex cause-effect relationships into executable intervention strategies.

[0172] Specifically, after the innovation event chain and the abnormal-external event cause-effect chain are input into the knowledge reasoning and resource matching model, the knowledge reasoning engine first analyzes the time sequence association and the cause-effect relationship in the innovation event chain and the abnormal-external event cause-effect chain. Then, the knowledge graph filters the matched innovation elements from the resource library based on the analysis results. For example, for the problem of technology research and development lag, the patent database and the research and development team resources are matched. The rule engine generates hierarchical intervention suggestions according to the priority rules set by experts. For example, the research and development investment ratio is adjusted first, and the fund injection is triggered at the abnormal time point. Finally, the system integrates the resource matching results, the intervention path, and the triggering time into an executable scheme.

[0173] In some specific embodiments, the knowledge reasoning engine can use natural language processing technology to analyze the cause-effect relationship in unstructured data, such as extracting external factors affecting technology research and development from policy documents. The knowledge graph can dynamically update industry technology trend data, such as real-time access to patent disclosure data to supplement the resource library. The rule engine can configure multiple sets of expert rule libraries, such as setting different intervention priorities for different industry life cycle stages.

[0174] Through the above technical solutions, the application can improve the pertinence and timeliness of the innovation resource intervention scheme, and solve the resource allocation lag problem caused by the lack of cause-effect analysis and dynamic matching in the prior art. Through the cooperation of the knowledge graph and the rule engine, the innovation elements are accurately called, and the problem of excessive investment or mismatch of resources is avoided, thereby improving the overall efficiency of enterprise innovation capability evaluation and resource adaptation.

[0175] In order to better understand the above embodiments, an application scenario example is given as follows:

[0176] Taking a certain technology company (mainly engaged in intelligent sensor modules and application integration) as an example, it will enter the transition period of industry growth and maturity in 2024;

[0177] Three types of data are collected periodically:

[0178] The industry database interface obtains industry life cycle data (growth-maturity stage, technology weight 0.28, market weight 0.32, synergy weight 0.22, capital weight 0.18);

[0179] The enterprise ERP system automatically aggregates data on innovation indicators such as technology research and development, market transformation, organizational collaboration, and financial resilience (such as R&D investment, number of patents, number of products launched, market share, team collaboration, cash flow, etc.);

[0180] The web crawler captures data from external data sources such as policy documents, market reports, and public opinion trends (including structured and unstructured data, automatic formatting, deduplication, NLP feature extraction and confidence screening, with a set value of 0.65).

[0181] The data were weighted (internal innovation sign 0.7, external data source 0.3), normalized (range mapping [0,1]), and processed with a sliding window (window 3 months, step size 1 month) to generate a trend curve for the fusion innovation sign data. Taking the technology R&D data as an example, the mean values ​​of the sliding windows from 2024.01 to 2024.06 were [0.78, 0.75, 0.70, 0.65, 0.67, 0.73]. Compared with the baseline curve [0.80, 0.79, 0.77, 0.76, 0.75, 0.74], the deviation values ​​were calculated and it was found that the deviations of the four consecutive sliding windows from 2024.03 to 2024.06 were all less than -0.08 (the preset threshold value is -0.08). The fusion innovation sign data corresponding to the deviation values ​​were determined to be abnormal fusion innovation sign data, and the abnormal time point was 2024.03.

[0182] Furthermore, by correlating the data from external data sources at the abnormal time points, we found that there was policy news about "core component import restrictions" between February and March 2024. Correlation analysis (Pearson coefficient 0.82, Bayesian network influence probability difference -0.37) determined that policy was the main driving factor, generating a "technological innovation anomaly-policy external causal chain." At the same time, a single window in May 2024 discovered an explosion in market conversion scores (deviation value of +0.12 > preset threshold two +0.10), corresponding to the "product new function launch" news, with a correlation of 0.74, determining it as a "market explosion-external innovation event chain."

[0183] Ultimately, the innovation event chain connects to form a composite link of "technological innovation anomalies - policy shocks - collaborative optimization - market explosion." The knowledge reasoning and resource matching model is invoked, and the event chain and causal chain are input. The knowledge reasoning engine analyzes the phenomenon that "policy shocks lead to continuous anomalies in technological innovation, while the market explosion benefits from collaborative optimization." The knowledge graph matches the analysis results to the resource library (technical cooperation, university think tanks, funding guidance, market promotion, and other factors). Based on the analysis results, the rule engine outputs the following intervention suggestions:

[0184] Intervention timing: 2024.03 (technical anomaly), 2024.05 (market outbreak);

[0185] Resource combination: Introducing university think tanks for joint research + policy consulting services + special fund support + marketing promotion team;

[0186] Intervention path: Technical bottleneck → Initiation of patent research project; Supply chain risk → Policy advisor intervention; Market explosion → Increased promotion investment

[0187] Multi-level suggestions: in the short term (1 month), complete the cooperation with universities; in the medium term (2-3 months), promote domestic substitution of core components; in the long term (6 months), establish a normalized collaboration and public opinion monitoring mechanism between industry, academia and research.

[0188] Example 2:

[0189] See also Figure 5 , the enterprise innovation cloud assessment and resource adaptation system includes:

[0190] Data acquisition module, used to periodically acquire industry life cycle data, innovation sign data, and external data source data;

[0191] A fusion innovation vital sign data calculation module is used to perform weighted average calculation on the industry life cycle data, innovation vital sign data and external data source data to obtain fusion innovation vital sign data;

[0192] a deviation value calculation module, configured to perform sliding window processing on the fused innovative vital sign data within the same period and output a trend curve of the fused innovative vital sign data;

[0193] Calling a preset reference curve, comparing the trend curve with the preset reference curve, and obtaining a deviation value;

[0194] A time point determination module is used to determine whether the deviation value in N consecutive sliding windows is less than a preset threshold value of one. If so, the fusion innovation vital sign data corresponding to the deviation value is determined to be abnormal fusion innovation vital sign data, and then the abnormal time point is determined in combination with the trend curve;

[0195] Determine whether the deviation value in a single sliding window is greater than a preset threshold value 2, and if so, determine that the fusion innovation vital sign data corresponding to the deviation value is the outbreak fusion innovation vital sign data, and then determine the outbreak time point in combination with the trend curve;

[0196] The innovation resource intervention plan output module is used to connect abnormal time points and outbreak time points in the same cycle on the timeline to generate an innovation event chain;

[0197] According to the innovation event chain, the preset knowledge reasoning and resource matching model is called to output the innovation resource intervention plan.

[0198] This embodiment has the same technical effects as the first embodiment.

[0199] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus.

[0200] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. The enterprise innovation cloud assessment and resource adaptation method is applied to the innovation cloud assessment and resource adaptation of technology-based enterprises at all stages, and is characterized by: The following steps are involved: Periodically obtain industry life cycle data, innovation sign data and external data source data; Performing weighted average calculation on the industry life cycle data, innovation vital signs data and external data source data to obtain integrated innovation vital signs data; Performing sliding window processing on the fusion innovation vital sign data within the same period, and outputting a trend curve of the fusion innovation vital sign data; Calling a preset reference curve, comparing the trend curve with the preset reference curve, and obtaining a deviation value; Determine whether the deviation value in N consecutive sliding windows is less than a preset threshold value of one, and if so, determine that the fusion innovation vital sign data corresponding to the deviation value is abnormal fusion innovation vital sign data, and then determine the abnormal time point in combination with the trend curve; Determine whether the deviation value in a single sliding window is greater than a preset threshold value 2, and if so, determine that the fusion innovation vital sign data corresponding to the deviation value is the outbreak fusion innovation vital sign data, and then determine the outbreak time point in combination with the trend curve; Connect abnormal time points and outbreak time points within the same cycle on the timeline to generate an innovation event chain; According to the innovation event chain, the preset knowledge reasoning and resource matching model is called to output the innovation resource intervention plan.

2. The enterprise innovation cloud assessment and resource adaptation method according to claim 1 is characterized by: Periodically obtain industry life cycle data, innovation sign data, and external data source data, including: Periodically obtain innovation vital signs data through the enterprise's internal management system; the innovation vital signs data includes technology research and development data, market transformation data, organizational collaboration data, and financial resilience data; By connecting to the application program interface of the industry database, industry life cycle data is periodically obtained; the industry life cycle data includes the current life cycle stage of the industry and the weight parameters of the preset innovation vital signs data corresponding to each stage; The external data source data is periodically acquired through a web crawler; the external data source data includes structured data, semi-structured data and unstructured data related to the enterprise and the industry in which it is located.

3. The enterprise innovation cloud assessment and resource adaptation method according to claim 2 is characterized by: Linear calculation is performed on the industry life cycle data, innovation vital signs data and external data source data to obtain the integrated innovation vital signs data, which specifically includes: Performing weighted calculation on the innovative vital signs data according to preset weight parameters of the innovative vital signs data, and outputting weighted innovative vital signs data; Screening the data from the external data source; Normalizing the weighted innovative vital sign data and the screened external data source data; The normalized weighted innovation vital signs data and the screened external data source data are weighted averaged to obtain the fused innovation vital signs data.

4. The enterprise innovation cloud assessment and resource adaptation method according to claim 3 is characterized by: The filtering process of the external data source data specifically includes: Performing data formatting, deduplication, and content normalization on the external data source data; Perform feature extraction on the processed external data source data, normalize the extracted features, and perform linear weighted calculation on them to obtain the confidence level of the external data source data; It is determined whether the confidence level is lower than a set value, and if so, the external data source data corresponding to the confidence level is eliminated.

5. The enterprise innovation cloud assessment and resource adaptation method according to claim 1 is characterized by: Performing sliding window processing on the fusion innovation vital sign data within the same period and outputting a trend curve of the fusion innovation vital sign data specifically includes: Sort all types of fusion innovation vital sign data within the same period by timestamp; Starting from the starting point of the cycle, the fusion innovation vital sign data is intercepted with the set sliding window length as the fusion innovation vital sign data within the sliding window; Calculate the mean of the fused innovation sign data within each sliding window; With the center moment of the sliding window as the horizontal axis and the mean value of the fusion innovation vital sign data in each sliding window as the vertical axis, all sliding windows are spliced ​​in sequence to form trend curves of various types of fusion innovation vital sign data.

6. The enterprise innovation cloud assessment and resource adaptation method according to claim 1 is characterized by: Determining whether the deviation value in N consecutive sliding windows is less than a preset threshold value of one, and if so, determining that the fusion innovation vital sign data corresponding to the deviation value is abnormal fusion innovation vital sign data, and then determining the abnormal time point in combination with the trend curve specifically includes: Calling preset benchmark curves corresponding to various types of fusion innovative vital sign data; Aligning the preset benchmark curve with the trend curve of the same type of fusion innovative vital sign data in terms of time period; For each time point, take the trend curve value and the baseline curve value respectively; Calculating a deviation value based on the trend curve value and the reference curve value; For all sliding windows, detect the deviation values ​​in chronological order; Determine whether the deviation value in N consecutive sliding windows is less than a preset threshold value one, and if so, determine that the fusion innovation vital sign data corresponding to the deviation value is abnormal fusion innovation vital sign data; the preset threshold value one is set according to the type of fusion innovation vital sign data; Then, in combination with the trend curve, the starting and ending sliding windows corresponding to the abnormal fusion innovative vital sign data are located, and the starting window time point of N consecutive sliding windows is used as the abnormal time point.

7. The enterprise innovation cloud assessment and resource adaptation method according to claim 1 is characterized by: After determining the abnormal time point in combination with the trend curve, the method further includes calculating the correlation between the abnormal fusion innovative vital sign data and the external data source data at the abnormal time point, specifically including: Calculate the correlation between various types of abnormal fusion innovative vital sign data and external data source data at the abnormal time point; The total correlation score is obtained by weighted fusion calculation of each correlation calculation result; Determine whether the total correlation score is greater than the preset correlation value. If so, input the external data source data at the abnormal time point into the pre-built Bayesian network and output the probability difference of the impact on the abnormal fusion innovative vital sign data; According to the impact probability difference, the direction and intensity of the impact of the external data source data at the abnormal time point on the abnormal fusion innovation vital sign data are determined, thereby generating an abnormality-external event causal chain; The impact direction is determined by the positive or negative impact probability difference; the impact intensity is determined by the absolute value of the impact probability difference; The Bayesian network uses various types of abnormal fusion innovative vital sign data and external data source data at abnormal time points as node variables, and is obtained by inferring the dependency relationship between nodes through deep learning.

8. The enterprise innovation cloud assessment and resource adaptation method according to claim 7 is characterized by: Connecting abnormal time points and outbreak time points within the same cycle on the timeline to generate an innovation event chain specifically includes: Label the specific type of fusion innovation vital signs data for each abnormal fusion innovation vital signs data and burst fusion innovation vital signs data detected in the same cycle; Record the occurrence sliding window and duration of each abnormal fusion innovation vital sign data and burst fusion innovation vital sign data; According to the recording results, the relationship between the abnormal fusion innovation sign data and the burst fusion innovation sign data of the same type and adjacent is analyzed from the starting point of the cycle to obtain data chains of different data types; Starting from the starting point of this cycle, data links of different data types are numbered in the same way according to the timeline; For the same data chain number, calculate the correlation of data chains of different types of fusion innovative vital sign data; Determine whether the correlation calculation result is greater than the preset correlation value. If so, merge the data chain of the corresponding type of fusion innovation vital sign data into an innovation event chain.

9. The enterprise innovation cloud assessment and resource adaptation method according to claim 7, characterized in that: According to the innovation event chain, the preset knowledge reasoning and resource matching model is called to output the innovation resource intervention plan, which specifically includes: Calling a preset knowledge reasoning and resource matching model; the knowledge reasoning and resource matching model includes a knowledge reasoning engine, a knowledge graph, a resource library, and a rule engine; The innovation event chain and the anomaly-external event causal chain are used as inputs to the knowledge reasoning and resource matching model. The knowledge reasoning engine analyzes the type, temporal structure, and causal relationship of the innovation event chain and the anomaly-external event causal chain, and outputs the judgment results. Through the knowledge graph, the judgment results are matched with the innovation elements available in the resource library, and combined with expert rules, the optimal resource combination and intervention path are inferred; The rule engine generates multi-level intervention recommendations based on the judgment results, and then uses the abnormal time point and outbreak time point as the intervention timing for each intervention recommendation; Comprehensively integrate the optimal resource combination, intervention path, intervention suggestions and intervention timing to output innovative resource intervention plans.

10. Enterprise Innovation Cloud Assessment and Resource Adaptation System, characterized by: include: Data acquisition module, used to periodically acquire industry life cycle data, innovation sign data, and external data source data; A fusion innovation vital sign data calculation module is used to perform weighted average calculation on the industry life cycle data, innovation vital sign data and external data source data to obtain fusion innovation vital sign data; a deviation value calculation module, configured to perform sliding window processing on the fused innovative vital sign data within the same period and output a trend curve of the fused innovative vital sign data; Calling a preset reference curve, comparing the trend curve with the preset reference curve, and obtaining a deviation value; A time point determination module is used to determine whether the deviation value in N consecutive sliding windows is less than a preset threshold value of one. If so, the fusion innovation vital sign data corresponding to the deviation value is determined to be abnormal fusion innovation vital sign data, and then the abnormal time point is determined in combination with the trend curve; Determine whether the deviation value in a single sliding window is greater than a preset threshold value 2, and if so, determine that the fusion innovation vital sign data corresponding to the deviation value is the outbreak fusion innovation vital sign data, and then determine the outbreak time point in combination with the trend curve; The innovation resource intervention plan output module is used to connect abnormal time points and outbreak time points in the same cycle on the timeline to generate an innovation event chain; According to the innovation event chain, the preset knowledge reasoning and resource matching model is called to output the innovation resource intervention plan.

Images