As a policy tool to limit greenhouse gas emissions using market-based instruments, carbon emissions trading is being adopted by an increasing number of countries and regions. By 2021, a total of 33 carbon trading systems have been put into operation worldwide, covering a wide range of industries such as power, industry, aviation, and construction. And by analyzing the data generated by the exchange, carbon quotas can be allocated more rationally, so that companies can use their quotas more fully, unfortunately, the analysis of the data may result in the leakage of company data. For example, since there is more than one carbon exchange, the analysis will be done after aggregating the transaction data of all exchanges, which may cause data leakage and create the problem of malicious competition between companies. Therefore, we searched for an efficient way to perform carbon allowance allocation analysis so that the data of trading companies are protected and do not lose their usability. We propose the locally differential privacy-carbon allocation model, which perturbs the data locally in the exchange to a certain extent, and the data center removes the impact of the perturbation as much as possible for data estimation, finally, uses it for carbon quota allocation share prediction. This allows for better protection of the trading data while still allowing for a reasonable allocation of carbon quotas. The final experimental results show that the allocation results are effective and reasonable under the premise of satisfying a certain degree of privacy protection.